DragonFly commits List (threaded) for 2008-12
[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index][Thread Index]

DragonFly-2.1.1 annotated tag v2.1.1 created

From:	Simon Schubert <corecode@xxxxxxxxxxxxxxxxxxxxxxx>
Date:	Tue, 2 Dec 2008 19:59:11 -0800 (PST)

at 3d7add7439dab6cdebf9adfa8faa2bf24f1faa87 (tag)
tagging 02a112e67533ef38cdb99bd93b46b3232f0f0a44 (commit)
tagged by Simon Schubert
on Wed Dec 3 04:47:05 2008 +0100

- Log -----------------------------------------------------------------
DragonFly 2.1.1
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (DragonFly)

iEYEABECAAYFAkk2AToACgkQr5S+dk6z85plAQCgtI1g8rwywzLIOq1xpSQgTRms
WocAoPOzXOY8Dua3Y9rnDeYV5I8w7W05
=EZ8C
-----END PGP SIGNATURE-----

Aggelos Economopoulos (12):
Let's try and start a tradition here. This is book is now a classic.
The i386/amd64 abi specifies that the direction flag must be clear
do early copyin / delayed copyout for socket options
update unix domain socket option retrieval
Provide MP_LOCK_HELD() for UP builds as well
Update bluetooth ctloutput functions not to use copy{in,out}
Don't do copy{in,out} in ctloutput any more
More copy{in,out} removal for ctloutput routines
Do copy{in,out} early in the {g,s}etsockopt system call paths
Fix sockopt syscalls for the Nth time.
Set fp to NULL on EINVAL
Use cred from componentname, don't access curthread

Charlie (1):
test

Chris Pressey (299):
Pre-test spot removal solutions in an inconspicuous area before use.
Practice commit: learn how to use 'cvs remove', so that, should I need
Fix minor formatting errors in cpdup(1)'s man page.
Clean up the code in bin/ to reduce warnings from GCC3.
Clean up the code in bin/ to reduce warnings from GCC3.
Clean up the code in bin/ to reduce warnings from GCC3.
Clean up the code in bin/ to reduce warnings from GCC3.
Clean up the code in bin/ to reduce warnings from GCC3.
Style(9) cleanup.
Style(9) cleanup.
Style(9) cleanup.
Style(9) cleanup.
Style(9) cleanup.
Style(9) cleanup.
Style(9) cleanup.
Style(9) cleanup.
Style(9) cleanup.
Style(9) cleanup.
Correct obvious typo in comment.
Four new features and a bugfix.
Style(9) cleanup.
Style(9) cleanup: remove ``register'' keywords.
Correct a typo that was introduced in revision 1.2.
Correct misspelling of "orphan" and fix up comment structure.
Style(9) cleanup.
Update to style(9) guidelines.
Style(9) cleanup.
Style(9) cleanup.
Style(9) cleanup.
Style(9) cleanup to src/sys/vfs, stage 1/21: coda.
Remove `-ly' and `${LIBY}' from our Makefiles. Linking to liby is not
Clarify the purpose of liby:
Correct type slippage in previous commit: a u_int was accidentally
Style(9) cleanup to src/sys/vfs, stage 2/21: deadfs.
Style(9) cleanup.
Style(9) cleanup to src/sys/vfs, stage 3/21: fdesc.
Style(9) cleanup to src/sys/vfs, stage 4/21: fifofs.
Style(9) cleanup.
Style(9) cleanup to src/sys/vfs, stage 5/21: ext2fs.
Style(9) cleanup to src/sys/vfs, stage 6/21: hpfs.
Style(9) cleanup.
Style(9) cleanup to src/sys/vfs, stage 7/21: isofs.
Style(9) cleanup to src/sys/vfs, stage 8/21: mfs.
Make a comment less misleading. rpcbind_enable (portmap) isn't
Style(9) cleanup to src/sys/vfs, stage 9/21: msdosfs.
Style(9) cleanup to src/sys/vfs, stage 10/21: nfs.
Fix code typo in previous commit to this file, thus allowing make(1)
Style(9) cleanup to src/sys/vfs, stage 11/21: ntfs.
Clean up typos and punctuation in comment.
Remove two unused GEOM function prototypes. They were hangovers from
Style(9) cleanup to src/sys/vfs, stage 12/21: nullfs.
Style(9) cleanup.
Split the suggested invocation of fdisk (which is failing for some
Style(9) cleanup to src/sys/vfs, stage 13/21: nwfs.
Style(9) cleanup.
Style(9) cleanup to src/sys/vfs, stage 14/21: portal.
Clean up style(9) issues that were missed in previous commit to this
Style(9) cleanup.
Merge with FreeBSD (RELENG_4) src/sys/dev/syscons/syscons.c,
Add missing function prototype.
Style(9) cleanup to src/sys/vfs, stage 15/21: procfs.
Style(9) cleanup to src/sys/vfs, stage 16/21: smbfs.
Style(9) cleanup to src/sys/vfs, stage 17/21: specfs.
Merge with FreeBSD (RELENG_4) src/bin/mv/mv.c, revision 1.24.2.6:
Merge with FreeBSD (RELENG_4) src/sys/msdosfs/msdosfs_vfsops.c,
Style(9) cleanup to src/sys/vfs, stage 17/21: udf.
Make a small grammar correction: "...most forms of the disklabel
Fix a bug in bsd.port.mk that was causing 'make clean' to fail in some
Merge with FreeBSD (RELENG_4) src/usr.sbin/timed/timedc/timedc.c,
Three minor fixes:
Fix catman(1) so that it recognizes valid manpage names as valid.
Style(9) cleanup to src/sys/vfs, stage 19/21: ufs.
Style(9) cleanup to src/sys/vfs, stage 20/21: umapfs.
Style(9) cleanup to src/sys/vfs, stage 21/21: unionfs.
Style(9) cleanup to src/sys/netinet6:
Correct sort order of SUBDIRs.
Style(9) cleanup to src/usr.sbin: remove remaining `register' keywords.
Clarify how our `test -w' is a superset of POSIX' requirements.
The default shell for the user `man' is nologin(8). Make the examples
Style(9) cleanup:
Fix typo.
Fix typo in error message - there is no system call named "lchflags."
Add -u option to cpdup(1), which causes its -v[vv] logging output to
Merge with FreeBSD (RELENG_4) src/usr.bin/script/script.c, revision
Fix a bogus type declaration to remove a compiler warning - `group'
Merge with FreeBSD (HEAD) src/usr.sbin/kbdmap/*:
Merge with FreeBSD (HEAD) src/usr.sbin/adduser/*:
Make virecover more robust. Instead of aborting (and causing
Perl is no longer user during kernel build, so remove the Perl scripts
Correct the FreeBSD attribution on this file.
Correct the FreeBSD attribution on this file.
Synch to modifications in the installer scripts:
Don't enable CAPS unnecessarily.
Remove temporary files immediately after they're no longer needed.
Change the REQUIRE from `inetd' to the more sensical `mountoptional'.
Update the URL from which the installer packages can be obtained.
Clarify behaviour of the -s option.
In dumpon(8) usage, `off' is a literal keyword, not a parameter name.
Style(9) cleanup:
Import from NetBSD: `pgrep' and `pkill' utilities for finding and
DragonFly-ize pgrep/pkill:
Correct an inaccuracy in disklabel(8)'s manual page: the default
Update list of installer packages to match installer version 1.1.
Hook pkill up to the build.
Update installer packages to 1.1.1, fixing a couple of minor bugs:
Correct an inaccurate statement. According to my testing, cpdup never
Update installer to 1.1.2.
Add various improvements to the "pre-flight installer":
Style(9):
Document the slightly surprising behaviour of 'ifconfig -alias' when
Bump WARNS to 6.
Bump WARNS to 6:
Bump WARNS to 6.
Oopsie; use WARNS?=, not WARNS=.
Bump WARNS to 6.
Bump WARNS to 6:
Style(9):
Bump WARNS to 6:
Bump WARNS to 6:
In install_* targets, honour any EXTRA_PACKAGES and EXTRA_ROOTSKELS
Improve seperation between kernel code and userland code by requiring
Bump WARNS to 6 and apply some style(9):
Bump WARNS to 6.
Merge with FreeBSD revision 1.16 (sheldonh):
Merge with FreeBSD revision 1.18 (jmallett):
Raise WARNS to 7 and apply some style(9):
Apply some style(9):
Merge with FreeBSD revision 1.20 (markm):
Make an example follow the recommendation given in the immediately
Merge with FreeBSD revision 1.21 (kientzle):
Clean up the usage message, both in usage() and in the main program
Use real getopt() handling instead of the hand-rolled and IOCCC-worthy
Raise WARNS to 6:
Style(9): the return type of strcmp() is not a boolean, so don't
Raise WARNS to 6.
Raise WARNS to 6:
Raise WARNS to 6.
Raise WARNS to 6.
There is no such thing as WARNS=7 (yet,) so use WARNS=6.
Apply style(9):
Raise WARNS to 6:
Raise WARNS to 6:
Raise WARNS to 6:
Raise WARNS to 6.
Raise WARNS to 6.
Raise WARNS to 6:
Raise WARNS to 6:
Raise WARNS to 6.
Raise WARNS to 6.
Raise WARNS to 6:
Raise WARNS to 6.
Raise WARNS to 6.
Raise WARNS to 6:
Raise WARNS to 6.
Raise WARNS to 6.
Clarify a run-on sentence by splitting it into two. snprintf() and
Style(9): remove `register' keywords.
Style(9) cleanup:
Raise WARNS to 6:
Raise WARNS to 6:
Raise WARNS to 6:
Raise WARNS to 6.
Raise WARNS to 6.
Raise WARNS to 6:
Raise WARNS to 6.
Raise WARNS to 6.
Raise WARNS to 6:
Raise WARNS of newfs to 6:
Style(9):
Clarify/clean up code, make fewer assumptions about types:
Raise WARNS to 6:
Raise WARNS to 6:
Further cleanup:
Raise WARNS to 6:
Raise WARNS to 6 and generally clean up:
Raise WARNS to 6:
Raise WARNS to 3:
Reduce warnings when compiled under WARNS=6:
Raise WARNS to 6.
Raise WARNS to 6:
Raise WARNS to 6:
mkdep(1) seems to want the current directory on the include path,
Fix typo in error message.
Raise WARNS to 6.
Raise WARNS to 6:
Raise WARNS to 6:
Raise WARNS to 6.
Raise WARNS to 6:
Raise WARNS to 6.
Raise WARNS to 6:
Raise WARNS to 6.
Raise WARNS to 6.
Unlike printf(3), fwrite(3) doesn't stop at the first NUL character
Use memchr(3) instead of a hand-rolled loop.
Raise WARNS to 6:
Raise WARNS to 6:
Raise WARNS to 6:
Raise WARNS to 6:
Raise WARNS to 6:
Raise WARNS to 6:
Raise WARNS to 6.
Raise WARNS to 6:
Raise WARNS to 6:
Clean up:
Raise WARNS to 6:
Raise WARNS to 6:
Partial merge with recent revisions[1] from FreeBSD.
Clean up:
Partial merge with recent revisions from FreeBSD:
Apply less dodgy range checks.
Fix up a bug that crept in during the strcpy->strlcpy conversion;
Change type of find_compare() so that it doesn't need to be casted
Partial merge with FreeBSD: revisions 1.23 through 1.34.
Raise WARNS to 6:
Merge with FreeBSD, revision 1.27 (fenner):
Merge with FreeBSD, revision 1.29 (imp):
Merge with FreeBSD, revision 1.30 (markm):
Further cleanup:
Import (slightly modified) ru.koi8-r.win.kbd:1.1 from FreeBSD (fjoe):
Hook ru.koi8-r.win.kbd up to Makefile and INDEX.keymaps.
Raise WARNS to 6.
Add "exit" command as a synonym for "quit".
Make pgrep(1) print a newline at EOL, as expected, instead of at the
Style(9):
Style(9):
In the parent process, close the slave file descriptor at the earliest
Raise WARNS to 6:
Style(9):
Don't have binsrch() alter a global variable if it fails. Instead,
Constify filename variable.
Raise WARNS to 6:
Merge with FreeBSD, revision 1.16 (stefanf):
Raise WARNS to 6.
Sync with FreeBSD, random.c:1.17 and random.6:1.7-1.8 (ru):
Improve data validation:
Raise WARNS to 6:
Add WARNS?=6 to Makefile.
Update installer to version 1.1.4. Highlights include:
Remove 'pristine' file tree. It is no longer needed now that there is
Update cdrtools package requirement to version 2.01.
Add the pfi_backend variable, which can be used to specify which
Reduce warnings as part of WARNS=6 cleaning, stage 1/6:
Reduce warnings as part of WARNS=6 cleaning, stage 2/6 (or so):
Reduce warnings as part of WARNS=6 cleaning, stage 3/6 (or so):
Reduce warnings as part of WARNS=6 cleaning, stage 4/6 (or so):
Reduce warnings as part of WARNS=6 cleaning, stage 5/6:
WARNS=6 cleaning, stage 6/6:
Style(9):
Style(9):
Style(9):
Style(9):
Style(9):
Small step towards WARNS=6:
Raise WARNS to 6:
Unbreak the pfi_autologin option by fixing a broken invokation of sed
- Tack pfi.conf onto the end of rc.conf later (only after pfi.conf
If a get_authorized_hosts file exists on the pfi media, copy it into
Raise WARNS to 6:
When rendering the routename() or netname() of a struct sockaddr
When dumping the contents of a struct sockaddr with an unknown address
Raise WARNS to 6.
De-spaghettify:
Followup to previous commit: use uint8_t instead of u_char, for a real
We have _DIAGASSERT now; might as well enable it here.
Style(9):
Style(9):
Style(9):
Style(9):
Raise WARNS to 6:
Raise WARNS to 6.
Raise WARNS to 6:
Add the pfi_curses_escdelay variable. When using the curses frontend,
Update installer to version 1.1.5. Highlights include:
Update the information regarding the installer and the pfi
Punctuation, formatting, grammar, and spelling nits.
Back out revision 1.31. The 'release' target alone should be
The error message issued when a requested package cannot be found has
When installing packages into an ISO-image-to-be file tree:
Reduce warnings that appear under higher WARNS levels:
Style cleanup: instead of seperate #defines for NULL cast to different
Fix grammar (associate -> associated) and start sentence on new line.
Make comments about bpf_validate()'s behaviour reflect reality. The
Style(9) cleanup: use ANSI format for function definitions.
Merge with FreeBSD vary.c:1.16 and date.1:1.68 (yar):
Merge with FreeBSD netdate.c:1.17 (charnier): udp/timed -> timed/udp
Merge with FreeBSD 1.44 (dds): report and exit on write error.
Raise WARNS to 6:
Sync with all revisions up to FreeBSD date.1:1.72 (ru), notably:
Style(9):
Reduce diffs with BSD Installer repo:
Introduce a make variable PACKAGE_SITES to improve package fetching in
Document non-obvious behaviour of missing 'a' line in fdisk script.
Clarify examples by differentiating between literals and arguments.
Make 'make upgrade' work from the LiveCD, stage 1/5 or so:
Make 'make upgrade' work from the LiveCD, stage 2/5 or so:
Make 'make upgrade' work from the LiveCD, stage 3/6 or so:
Make 'make upgrade' work from the LiveCD, stage 4/5 or so:
Make 'make upgrade' work from the LiveCD, stage 5/5:
Two small alterations to 'make upgrade':

Chris Turner (2):
testing, 123
Add '-l' support to vnconfig(8) and supporting VNGET ioctl to vn(4).

Dave Hayes (5):
Testing a commit per man page
Installer import
Installer import into contrib (real import this time)
Merge from vendor branch BSDINSTALLER:
* Allow nrelease/Makefile to build the local installer in usr.sbin

David P. Reese, Jr. (34):
Change the split syscall naming convention from syscall1() to kern_syscall()
Make the linux emulator use the newly split listen(), getsockname(),
Create an emulation/43bsd directory and move the recently modified
Separate all of the send{to,msg} and recv{from,msg} syscalls and create
Split getsockopt() and setsockopt().
Modify kern_{send,recv}msg() to take struct uio's, not struct msghdr's.
Modify linux_{send,recv}msg() and linux_{set,get}sockopt() to use the
Introduce the function iovec_copyin() and it's friend iovec_free().
Fix a bug in the last commit where sendfile() would forget to drop a file
Implement socket() and shutdown() using the in-kernel syscalls. This
Rename do_dup() to kern_dup() and pull in some changes from FreeBSD-CURRENT.
I wasn't properly checking for rollover in iovec_copyin(). Remove code
Create kern_readv() and kern_writev() and use them to split read(), pread(),
Create the kern_fstat() and kern_ftruncate() in-kernel syscalls.
makesyscalls.sh wants comments to be on their own line. Move the Checkpoint
Remove the FreeBSD 3.x signal code. This includes osendsig(),
Split wait4(), setrlimit(), getrlimit(), statfs(), fstatfs(), chdir(),
Unbreak svr4 damage from last changes to kern/kern_resource.c.
The recent lseek() fix didn't touch olseek(). I don't see why, as olseek()
The big syscall split commit broke utimes(), lutimes() and futimes() when
Split execve(). This required some interesting changes to the shell
Remind myself and others that kern_readlink() isn't properly split yet.
Split mkfifo().
Kill two more stackgap allocations. One is in linux_sysctl(). The other
Move ogethostname(), osethostname(), ogethostid(), osethostid(), and
Move ogetkerninfo() to the compat tree without any changes.
Split mmap().
Remove an unused variable from last commit.
Sync with FreeBSD-5.x MODULE_DEPEND() and DECLARE_MODULE() broke viapm.
Implement linux_truncate64() and linux_ftruncate64().
Implement linux_truncate64() and linux_ftruncate64().
Implement linux_mmap2().
Fix a bug in signal translation introduced in last revision. This effects
Fix linux_getrlimit() and linux_old_getrlimit() which weren't copyout()'ing

David Rhodus (277):
Null terminate the new argv[] array when converting 'tar zcf' to
Fix for trap 12 on KVA space exhaustion.
Fix CPU stats percentages formatting
Try to distinguish PCS encoding error events
Fix a problem that occurs when truncating files on NFSv3 mounts: we need
Don't map LINUX_POSIX_VDISABLE to _POSIX_VDISABLE and vice versa for
Allow the caller to get an error directly if we sent the packet immediately.
Fix compile error.
LINT breakage fix.
The iBCS2 system call translator for statfs(2) did not check the
Fix for insufficient range checking of signal numbers.
Fix a bug which caused signals on YUV images to fail.
Add or correct range checking of signal numbers in system calls and
* Fix the type of the NULL arg to execl()
Fix a bug that could cause dc(4) to m_freem() an already freed
Add function so Control-D erases the screen.
regen
Allow the optional setting of a user, primary group, or grouplist
Update man page to reflect the optional setting of user, primary
Cleaned up dead declarations.
Sync with FreeBSD
M_PREPEND() can fail, meaning the returned mbuf
The game rogue(6) has a potential buffer overflow allowing
Add USE_SLAB_ALLOCATOR into the config.
#include reorganization
Add a few more build options.
Inital cleanup work to make NETNS compile again
Add NETNS to the LINT config file.
Do not record expanded size before attempting to reallocate
Clean up.
Correct address parsing bug that is believed to be remotely exploitable.
* de- __P()
* de- __P()
* de-__P()
* Intel ACPI 20030228 distribution with local DragonFly changes.
* Finish up last commit.
Forgot another one.
Looks like we can't have comments on the same
* Add #include <sys/buf2.h>
* Add the Promise Supertrack to the build
Correct a case in readv(2) where the file descriptor reference count
Introduce a uiomove_frombuf helper routine that handles computing and
* Turn on debugging options by default
* Try to make gencat(1) recognize.
* Correct several integer underflows/overflows in linprocfs
* Fix a typo that was introduced from the last change.
* Add this nice filesystem testing tool that I've recently
* Move variable 'p' into a more proper place.
* Add missing function declaration that was some how
* Pull is some security checks from sendmail 8.12.10.
* Un-break buildworld
* Nuke ~771 __P()'s from our tree.
* Fix a that showd up on 64 bit systems. It was actually
* Change the buffer lenght test in NEEDSP() so that it does not
* Add a comment to the file pertaining to the last commit
* Handle realloc() failure correctly.
* Call crypt() directly instread of taking a detour through makekey.
* Fix BBS buffer overflow in makeargv().
* Fix two buffer overflows caused by off-by-one errors: avoid writing
* Do not set the No_CRC bit in the Mode Control Register.
* Fix a potential race condition in crfree.
* Temporarily back out last change because of a problem reported
* Add ACPI to the LINT build.
* s/FreeBSD/DragonFly/ So I know what system I'm booting up now.
* Fix some __P() mess up.
* Fix a locking issue inside the inode hashing code which
* How did these get here.
* Add quirk for LEXAR 256MB JUMPDRIVE
* Prevent leakage of wired pages by setting start_entry
* s/FreeBSD/DragonFly at the boot2 prompt.
* buildworld doesn't need to look at nrelease.
* Attempt to save the last dregs of emacs users' sanity by saving the
* don't fclose() a bogus FILE *.
* Document the fact that send(2) can return EPIPE (like when a
* Allow a return of 0 from __sys_write() to exit the loop in libc_r's
* Add support for gb18030 encoding.
* Add GBK encoding support.
* Fix problem where initgroups would silently truncate groups with
* Sync comment with code's reality.
* Might help if we built gbk.c and gb18030.c
* Remove a few extra $FreeBSD$ tags
* Add a part of the AGP update that was missed.
* Print out a warning if /dev/null doesn't exist.
* Add GB18030 directories.
* GBK locale directories
* Try and match README guideline better.
* Update Sendmail to version 8.12.10
* Sync Changes over from FreeBSD
* Add the POSIX.1-2001 header file <cpio.h>
* Cleanup compiler warnings and other style cleaning while here.
* Add the pst(4) driver to the GENERIC kernel build.
* Merge From FreeBSD
* Merge From FreeBSD
* Implement POSIX (XSI)'s header file re_comp.h
* Change __guard back into a multi-character int
* Add ending comment to line as to avoid warning messages.
* Merge From FreeBSD
* Merge From FreeBSD
* Remove references to deprecated options.
* Welcome the DragonFly project into to new year by bumping
* Add in some basic parameter values that we will use for
* Merge in changes from FreeBSD with some modification
* Merge in the copyright that DragonFly is being distributed under.
* Add usb_mem.c back into the build process so the usb module
* Don't install the freefall gnats files into /etc as
* Remove the HTT option from the kernel as we can now
* Remove HTT option from kernel config files.
* Update so as to reflect structure name changes to upcall.h
* Perform some man page maintenance bringing the miibus related
* Use id(1) instead of grep(1) to detect the presence of the smmsp
* Add in support for the IBM ServeRAID controller.
* Change SUBDIR sort order.
* Add in ips.4 to the Makefile so it gets installed
* Add in 'device ips' to the LINT kernel config.
* Fix libcaps build by syncing up with kernel work.
* Drop the directory outline for building gcc3.
* Starting using GCCVER when needed and other things
* Fixup gcc3 build process to use correct install path.
* Fix man page names.
* Remove extra compile flag to clean up redefine warnings.
* DragonFly<-FreeBSD name change in boot loader code.
* Change FreeBSD ifdefs to DragonFly
* Change FreeBSD ifdef over to DragonFly
* DragonFly<-FreeBSD Name clean ups
* Add kern_systimer.c to the kernel build process.
* Sync pages with FreeBSD 4.x
nectar 2004/02/05 10:00:35 PST
* Update function defines to match up with the work from
* Remove ufs_disksubr.c from kernel build files.
* Return if array is out of bounds in icu_setup function.
* Fix placement of return value.
* Remove instances of PRIBIO because DragonFly no longer
* Correct end value calculation. This should fix the numerous problems
* Add in support for the Silicon Image SATA controller.
Start removing the old build infrastructure for the a.out
Hook in UDF to the build process.
Should have been on last commit.
Change which(1) over from a perl script to a C program.
Change whereis(1) over from a perl script to a C program.
Next time I'll run cvs update to make sure I've added
Attach mount_udf to the buildworld process now.
Update rc.d scripts to use the correct path for the named pidfile.
Change this vnode check inside of the VFS_BIO_DEBUG
* Fix an off-by-one problem.
* Change the offset alignment in vn_rdwe_inchunks()
Make sure the ELF header size is not too large. This fixes a potential over
Move vm_fault_quick() out from the machine specific location
The existing hash algorithm in bufhash() does not distribute entries
Update tag creation paths to reflect the cvs
Add back in basic ACPI support into the LINT kernel build.
Place 'device acpica5' into the TINDERBOX kernel config so
Add in the twa(4) driver. This adds in support for the 3Ware
Add in kernel config file options that were
Define that Dragonfly HAS isblank() as to fix the
Fix an symlink ordering issue that was causing the install
Fix symlink ordering issue with the less man page.
Count statistics for exec calls.
Add in the new acpica5 to the device build path.
Remove the VREF() macro and uses of it.
General update:
Cosmetic changes.
Leave out acpica5 from the device build path for a little longer, well
Cosmetic changes.
Point the remote site to use gobsd.com so the remote option now works.
Cleanup pass. Removed code that is not needed anymore.
Remove unneeded typecast.
oops.
Add the SysKonnect ethernet driver to the GENERIC kernel config.
Default the MMX/XMM kernel optimizations to the on position. Also make a
Fix installworld by making it aware of the rconfig examples.
Create twed0 entries in /dev
Drop to the debugger if unable to sync out the buffers during shutdown.
Unhook gcc-3.3 from the buildworld process. This also removes the gcc3
Use correct copy-write notation for The DragonFly project copy-writes.
Sync in Copyright notice change from FreeBSD.
Update some copyright notices to become more legal compliant.
Fix compile warning.
Remove unused variable.
Merge in FreeBSD-SA-04:13.linux
Fix warning when building with c99.
Fix memory leak.
Spelling.
Fix a typo in last commit as to fix the EHIC compile.
Bump the rev to 1.1-CURRENT.
Implement advisory locking support for the cd9660 filesystem.
Perform some basic cleanups. Change some types over to C99 standard
Remove __FreeBSD__ defines which were not needed and caused compile
Cleanout __FreeBSD__ defines which we don't need.
Cleanout __FreeBSD__ defines which are not needed.
DragonFly MP support, not fbsd anymore.
Emulate __FreeBSD__ till 3rd party applications later add in DragonFly
POSIX.1-2003: Changing the group ID is permitted to a process with an
Correct line wrap.
Correct getpriority() and setpriority() function definitions to
We will need more resource-limiting compliance work done before these
Correct path location search when pulling packages from GoBSD.
Document that there seems to be a problem with the syncer not running
Remove unused variable.
Plug in missing brelse calls as to fix a bug in the FFS reload code.
Check error return value when creating the IPC service.
Minor cleanups. Make a type usage correction. Remove unneeded cast when
Add a few notes.
Spelling.
Clearly I need to wakeup all the way before starting to make changes.
Fix a problem introduced by the last commit.
Change the array of char to an array of struct pollfd to avoid an
Avoid leaving an obsolete pointer while the interface is detaching
yar 2004-08-28 12:49:58 UTC
Add in support for the iRiver iFP MP3 player.
add in support Frontier Labs NEX IA+ Digital Audio Player with USB CF
Bring in fix from gzip 1.3.3 to avoid crashes when processing certain
do not send icmp response if the original packet is encrypted.
Fix typo.
Fix printf example.
I guess I should bumb document date too.
Print link level address on vlan interfaces using ether_ntoa(), to make
Fix a problem with the NSIG check which could lead to an exploitable
Fix another incorrect usage of NSIG.
Add support saving both user/group permission on symlinks(-p)
Remove register keyword usage.
Check that the authpf user and group are installed before starting the
Since there isn't an authpf user we'll have to grep for the authpf
Declare some of the local variables static.
Update the 3ware card comment to better reflect which cards are supported.
Sync up items from the GENERIC config and a few more items to make
Spelling.
Initialize 'version' variable to silence compiler warning.
Quiet tinderbox:
Spelling: Filesystem is one word.
Handle all types of interrupts when operating the uhci(4) controller in
Remove the old system perl binary when performing 'make upgrade'.
ANSI and whitespace cleanup. No operational changes.
Fix typo.
re-fix typo. Maybe its time to start drinking coffee in the morning. :-)
Add information about return code for immutable files.
Add information about return code for immutable files.
Remove call to readdir() that was missed durning the libc namespace
Connect our 1:1 threading library to the build.
Compensate for off by one bugs in disk firmware for 48BIT addressing cutover.
no need to have break; after return;
Fix some signed/unsigned comparisons.
Clean up main() arguments.
Unbreak the pkg_add -r which the last commit broke.
Testing a new commits monitoring script.
test
Correct a pair of buffer overflows in the telnet(1) command:
Merge from vendor branch HEIMDAL:
Correct a pair of buffer overflows in the telnet(1) command:
Remove some uses of the SCARG macro.
Remove the telnet program from the heimdal source as it contains
Remove the telnet program from the heimdal source as it contains
Merge from vendor branch HEIMDAL:
Add in a 'livecd' taget for those who don't want to include the
Clean up some register keyword usage.
Remove some uses of register keywords.
Fix kernel build issue. Missing function argument.
Replace spl with critical sections.
Remove reference to devd.
Fix typo.
Remove scheduler define which was never used.
Make print statements more verbose.
ANSI-fy some functions, remove some duplicate CVS tags and add in some
Remove some register keywords and add in some missing $DragonFly$ tags.
The packages have been moved to the fireflybsd server now as it is a
scm test.
Change the MAX_BUTTONS count to 31. This allows us to recognize more than
Add -P to the cvs example commands.
Merge in security fix from FreeBSD.
Fix typeo.
Add missing code needed for the detection of IPSec packet replays.
NetBSD merge.
NetBSD merge.
Correctly identify the user running opiepasswd(1) when the login name
Properly copy in userland scheduler name via copyinstr.
Mark a few more system calls MPSAFE.
Merge from FreeBSD.
Hold MP lock for getppid(). As noted by Dillon getppid() is not MP safe.

David Xu (96):
My first commit.
Import initial version of 1:1 pthread library.
Implement sigtimedwait and sigwaitinfo syscalls.
Implement sigwait.
generate __sys_sigtimedwait and __sys_sigwaitinfo to allow
Implement cancellation points for sigwait, sigtimedwait and sigwaitinfo.
Regen.
Fix timeout verification bug. Matt Dillon asked to check tv_sec,
Remove debug code.
Remove unused macro.
Set initial thread's stack to 2M on 32 bits platform, 1M for default stack.
Use new kernel tls interface.
With new tls interface, sysarch.h is no longer needed.
Code cleanup, remove unneeded includes.
Remove alpha support.
Style fixes.
Add barrier prototypes.
Enable barrier support.
In kern_sigtimedwait, do not mask SIGKILL because a stopped process
Merge from FreeBSD:
Add more relocation types.
Import tls support code for static binary from FreeBSD.
Add tls functions prototype.
Implement _set_tp which is used to set TLS pointer into
Oops, fix license format(copy and paste problem).
Compile tls support code.
Compile tls.c
1. use __weak_reference to define weak symbol.
Initialize tls support code for static binary.
Add clockid to _thr_umtx_wait, so that clockid attribute of condition
Fix comments.
unconstify a modifiable parameter.
Add prototypes for pthread_condattr_getclock and pthread_condattr_setclock.
Add following prototypes:
Fix incorrect comment.
Add types prgregset_t and psaddr_t which will be used by
Add support for TLS.
Use rtld's TLS interface to allocate tcb.
Don't restart a timeout wait, one can periodically send signal to
Pass exact number of threads to thr_umtx_wake.
tcb is now managed by rtld, caching it is not correct,
Remove unused strong reference.
Fix brokeness by using _pthread_xxx functions instead.
Backout revision 1.5, the pthread->error was used as internal
Fix comment of error member, now errno is TLS based.
Define type lwpid_t, it will be used for thread debugger code.
Add prototypes of proc service API, every debugger must provide the
Add libthread_db, a thread debugging support library.
1. Fix symbols needed by libthread_db.
Remove unused function.
Sort names.
Add locking for FILE.
Eliminate PS_FILE_WAIT state, instead use FILE locking code in libc.
o Fix obsolete comment.
Override _flockfile_debug.
I am lost. if thread is blocked on mutex, _thread_kern_sched_state()
Fix off-by-one error.
Replace THR_FLAGS_SUSPENDED with THR_FLAGS_NEED_SUSPEND, this
Introduce pthread_timedjoin_np().
Introduce pthread_timedjoin_np.
put pthread_timedjoin_np in right order.
Add pthread_mutexattr_setpshared and pthread_mutexattr_getpshared.
Export following functions:
Make usleep as a weak symbol, so thread library can override it.
Make usleep as a cancellation point.
Remove a redundant tls_get_curthread() call.
Remove unneeded tls_get_curthread() call.
Clear return code to zero if joiner sucessfully waited joinee.
MFBSD:
If we got a SIGKILL signal in kern_sigtimedwait, call sigexit to
Close a race for thread detach.
copy flag DETACHED.
Rewrite mutex_init, get rid of compile warnings.
Oops, disable debuging code.
Make thread suspension really work.
Revamp the algorithm to acquire a contested lock, the intention
Update UMTX_LOCKED and add UMTX_CONTESTED macro.
Convert weak reference to strong reference so that static library
Pull in all symbols needed for static binary.
Unbreak buildworld.
Move some global variables into its module, remove priority mutex code
Add function prototypes: pthread_atfork, pthread_attr_getguardsize,
Add all pthread functions into libc namespace.
WARNS level 4 cleanup.
s/long/int
Unlock recursive mutex in pthread_cond_wait, though this is arguable.
Tweak source code a bit to make gcc to generate better code.
Add compiler branch prediction hint macros, obtained from FreeBSD.
Use the branch prediction macros in sys/cdefs.h.
1) Use int for m_flags of pthread_mutexattr.
namespace cleanup.
Seperate _mutex_cv_unlock from _mutex_unlock_common.
Allow userland to bind a process to specific CPUs. The initial
Oops, the usched_set syscall prototype should be updated.
Regenerate.
Use type lwpid_t for lwp_tid.

Eirik Nygaard (204):
Add some missing $DragonFly$ keywords.
Remove __P macros from src/usr.bin and src/usr.sbin.
Remove some __P macros in sbin that my script missed.
Fix typo in cloud -> in a cloud.
Remove send-pr from the build.
Make truss check if procfs is mounted, and where it is mounted.
Reattact send-pr to the build and change the mail address to bugs@crater.dragonflybsd.org
Nuke some more __P macros.
* Add a missing $DragonFly$ keyword
Remove the rest of the __P macros in src/usr.sbin
* Parse arguments with getopt
* then -> they
Add missing $DragonFly$ keywords
* Remove some __P macros from gnu/usr.bin/
* __P removal from games/
* Get rid of of the _BSD_XXX type usage.
* Add missing $DragonFly$ keywords.
* Removed the __P macros from lib/
* K&R function cleanup
* K&R function cleanup
* K&R function cleanup
* K&R function cleanup
* K&R function cleanup
* Added tabs instead of spaces for readability and consistency with NetBSD
* Add DragonFly as an possibility to a case statement.
Fix a bug which causes wrong filename being written into the syslog
* Make pkg_add fetch the packages from /packages-4-stable instead of
* K&R function cleanup
* Remove unused #includes.
* Merge fixes from libc to libcr.
* Indentation cleanup
* Let mergemaster look for make.conf in the correct path.
* Fix typo: becuase -> because
* K&R function removal
* Remove the use of the kern.devname sysctl since it does not exist.
* Remove the historical r in front of the old raw character devices in /dev
* Default to a calendar file that actually exists (calendar.all)
* Add devname_r(3) function, which takes a buffer and buffer length as
* Nuke calendar.freebsd
We don't have the ACPI module, so don't try to load it at boot.
Last commit was completely wrong. Reverting to old revision of file.
Catch up with recent libc updates.
His name is Fred.
Let the man page know about varsym links.
Remove genassym and gensetdefs from the cross build as well.
Readd ending '\' so it compiles the cross build tools
Remove stale text in front of the DragonFly and FreeBSD keywords
Add missing */ on the end of a comment.
Replace K&R style functions with ANSI C style functions.
Fix comment to show the munlockall function, and not the mlockall one.
Remove unused settings
Update FILES section
Use MALLOC_DEFINE.
style(9) cleanup:
style(9) cleanup:
style(9) cleanup:
Make the comment a bit clearer.
this -> the fix in comment.
Move the ASSERT_VOP_LOCKED and ASSERT_VOP_UNLOCKED macros into its own
Remove perl from the i386 kernel build.
Update libcr with recent libc updates.
* Remove ``register'' keywords.
De-perlify.
Add lock.9 man page, it is also symlinked to: lockcount.9, lockinit.9,
We are DragonFly not FreeBSD, so rename the name in GENERIC, and remove the
Tell awk where the arguments for the program ends and where the arguments for
Remove gawk from the build and let one-true-awk be our default awk from now
Fix a core dump caused by a .DEFAULT target with no commands.
Merge from vendor branch DIFFUTILS:
Import of diffutils 2.8.1
Remove vim swap file, which should not have been imported.
Update diff, diff3 and sdiff to use diffutils 2.8.1
Merge from vendor branch AWK:
Import of awk 20040207
Update to awk 20040207.
Merge from vendor branch LESS:
Import of less 381
Update less to version 381.
Merge from vendor branch LIBPCAP:
Import of libpcap 0.8.3
Merge from vendor branch TCPDUMP:
Import of tcpdump 3.8.3
Update libpcap to version 0.8.3
Update tcpdump to version 3.8.3
Remove two mis-added .orig files.
Remove unused error variable.
Change mbug allocation flags from M_ to MB_ to avoid confusion with malloc
Add ref counting to the checkpoint handler and unload function so an unload
Split the __getcwd syscall into a kernel and an userland part, so it can be
Print the entire path to where the checkpoint file is being created to avoid
Use MAXPATHLEN as the size of the checkpoint filename sysctl variable instead
Swap order of first and second argument in bcopy, forgot this when changing
Update tcpslice to version 3.8.
This is less, correct name in upgrade file.
Make sure the xe driver found a supported card type, if it didn't then bail
Some laptops return other values for working toucpads. Allow test_aux_port to
-{h,k} are mutually exclisive. So only pay attention to the last of the two
Speed up hardlink detection by using a self-sizing hash table rather than the
Change machien to machine.
Rearrange the machine/cpufunc.h header and add it where needed to make libcaps
Play catchup with libc.
Fix formatting error.
The syscall number lives in ms_cmd.cm_op now.
* Remove one of the two err.h include statements.
Forced commit to note that this fixes was submitted by: "Liam J. Foy"
Fix build with gcc34.
Remove a not needed main() definition.
style(9) cleanup:
style(9) cleanup.
Add missing va_end(ap);
Don't depend on pollution in <limits.h> for the definition of <stdint.h> macros.
Add WARNS?=6 and remove CFLAGS=.
Remove local main() definition.
Fix gcc 3.4 build.
Fix gcc 3.4 build.
Fix gcc 3.4 build.
Since we are DragonFly we want to use the DragonFly version instead of the
Since we are DragonFly we want to use the DragonFly version instead of the
Fix gcc 3.4 build.
Add message passed syscall's.
Use strchr instead of index, and strrchr instead of rindex because the str*chr
Use strchr instead of index, and strrchr instead of rindex because the str*chr
Redo the sysmsg test programs to match the changes done in the kernel to
sendsys.h is no longer needed.
Move the simple pager from db_ps.c into the more obvious file db_output.c and
Add noops for vga_pxlmouse_direct and vga_pxlmouse_planar in the case where
Add KTR, a facility that logs kernel events to help debugging. You can access
Add two more header in the !_KERNEL case so libcaps knows about
* Fix spelling.
* static functions
* pid should be pid_t not int
style(9) cleanup.
Update the sysperf tools to match the current sysmsg functions.
#endif does not take an argument, comment it out.
Consify some variables.
style(9)
Remove not needed void casts.
Constify VAR.
Remove xpt_release_simq_timeout(), this function has not been in use since the
Back out constification. This caused a SIGBUS becaus a functon wanted to write
WARNS= 6 cleanup.
Fix traceroute.
Move away from GNU traceroute and use the BSD licensed one.
Make the code WARNS= 6 clean.
It is sys_checkpoint.2 not checkpoint.2
Fix type. It is pkgs and not pksgs.
Add directory /usr/include/kadm5.
Merge from vendor branch HEIMDAL:
Add heimdal-0.6.3
Update the kerberos5 build framework to work with heimdal 0.6.3.
Kerberos no longer comes with it's own telnet, so always build the one in
Kerberos no longer comes with it's own telnet. Always build the one in
Whitespace cleanup.
Move libutil up above KERBEROS5 libs and remove the duplicate entry when
No longer used.
Fix kerberos5 build by adding some headers files. These should be generated on
style(9)
Allow numbers to be used after a letter in the modifier section after a
Add per cpu buffer for storing KTR information.
I have added a per cpu buffer for ktr, so this note can be scraped.
Merge from vendor branch NCURSES:
Add ncurses 5.4 source code.
Be consistent in the use of BOOLEAN.
Remove Ada95 files and man files.
Merge from vendor branch NCURSES:
Remove Ada95 files and man files.
Update ncurses to version 5.4.
There is no need to set *entry on each entry traversed in the red-black tree
Fix nested comments. Vim added another */ while I copy/pasted some of the lines
A few more instances of leaf that should be leap.
getopt is WARNS 6 clean.
Remove *spl() from dev/disk/{advansys,aha,ahb,aic7xxx,amd} replacing them with
Remove *spl() from netinet6 replacing them with critical sections.
* Include string.h to get strlen() prototype.
More spl_* removal from dev/disk/, replacing them with critical sections.
Replace the hand rolled linked list with a SLIST.
We are in callout_stop, not callout_reset. Tell that to the world.
Commit untouched SCTP files from KAME originally written by Randall Stewart.
Add DragonFly to the #ifdef mess.
Our mbuf allocation flags are prefixed with MB_ not M_.
We have to declare the parent node for a sysctl.
Add va_arg handling for DragonFly.
Remove the second argument to ip_stripoptions(), it was never used.
Properly handle mbuf copying on DragonFly when we try to reduce the size of a
Initialize a few more timers.
Add a forth argument to soreserve(). Pass just a NULL for now.
Add libsctp.
Convert spl* to critical sections.
Remove forgotten debug printf.
Call suser() with the correct number of arguments.
Don't return right after a goto.
Tie SCTP into the kernel, this includes adding a new syscall (sctp_peeloff).
Hook libsctp into the build.
Update manpages to reflect the changes joerg did that added const to some
WARNS?= 6 and style(9) cleanup.
Remove libraries libtermcap and libcompat from the linking.
Make wakeup, tsleep and friends MP-safe.
Back out last change since Matt has issues will it. Will reimplement it using
Use errx(), no errno is set in this case.
SUSv3 states that the type builtin should return a value > 0 if the completion
Make SCTP compile when IPSEC is enabled.
Add SCTP to LINT.
Remove syscall-args. It is not needed now that libcr has been removed.
Add locale(1).

Gregory Neil Shapiro (28):
Test CVS commit.
Merge from vendor branch SENDMAIL:
Import sendmail 8.13.4 into a new contrib directory as the first step
Bring DragonFly's sendmail infrastructure up to date. This commit includes:
Remove installation of the dragonfly{,.submit}.cf files since users may
Make links for hoststat(8) and purgestat(8) man pages.
Merge from vendor branch SENDMAIL:
Import sendmail 8.13.6
Slight cleanup on the DragonFly README
Adjust build infrastructure for sendmail 8.13.6
Merge from vendor branch SENDMAIL:
Import sendmail 8.13.7
Add README.DRAGONFLY for sendmail 8.13.7
Hook sendmail 8.13.7 into the build
Make the patch apply cleanly with sendmail 8.13.7's source
Merge from vendor branch SENDMAIL:
Import sendmail 8.13.8
Add DragonFly instructions file to new version directory
Upgrade to sendmail 8.13.8
Merge from vendor branch SENDMAIL:
Import sendmail 8.14.1
Add DragonFly instructions file to new version directory.
Change build infrastructure over to sendmail 8.14.1.
Merge from vendor branch SENDMAIL:
Bring in sendmail.org code from the future 8.14.2 release which restores
Import sendmail 8.14.2
Merge from vendor branch SENDMAIL:
sendmail 8.14.2 has been imported

Hasso Tepper (287):
Testcommit.
Although defined in sys/time.h we don't have CLOCK_VIRTUAL and CLOCK_PROF
Bring kernel threads back into top.
Revert intial IPv6 routing header type 0 processing fix.
More agressive fix for IPv6 routing header type 0 issue.
My round of spelling corrections in /share/man.
ICMP extensions for MPLS support for traceroute(8).
Although our linker supports pie, our elf loader doesn't.
Update ping(8) code and manpage to the newest ones from FreeBSD.
Add implementations of the inet6_opt* and inet6_rth* functions (RFC3542).
Add ip6 and icmp6 displays to systat(1).
Fix rewrite error which appeared in rev 1.3.
Bring in manpages from RELENG_6. Manpages related to sound imported in
Bring in the latest sound changes from RELENG_6.
Add mpls-in-ip. Bring in some fixes from IANA and FreeBSD in progress.
Nuke "is is" stammering.
Clean up sys/bus/usb/usb_port.h. Remove not used/dead/old code.
- Fix headphone jack sensing support for Olivetti Olibook 610-430 XPSE.
malloc -> kmalloc
One callout_stop() is enough.
Nuke USBDEV().
Nuke the code specific to NetBSD/OpenBSD/FreeBSD at first. I doubt anyone
Use kernel functions. I don't understand how I could miss these ...
Nuke device_ptr_t, USBBASEDEVICE, USBDEVNAME(), USBDEVUNIT(), USBGETSOFTC(),
Remove duplicate.
Nuke SIMPLEQ_* and logprintf.
Nuke usb_ callout macros.
Fix KASSERT messages.
Nuke PROC_(UN)LOCK, usb_callout_t, usb_kthread_create* and uio_procp.
Nuke USB_MATCH*, USB_ATTACH* and USB_DETACH* macros.
Nuke USB_GET_SC and USB_GET_SC_OPEN macros.
There is no need to have ETHER_ALIGN define here.
Nuke USB_DECLARE_DRIVER and USB_DECLARE_DRIVER_INIT macros.
Nuke the code specific to other BSDs.
Nuke USB_DO_ATTACH and remove device_t dv, since it is no longer needed.
Fix typo.
Remove last usb_port.h defines usages from the tree - selwakeuppri(),
Fix stupid mistake. Sorry.
Reduce diff with FreeBSD where it makes sense - add a lot of vendors and
Regenerate usbdevs.h and usbdevs_data.h and fix affected drivers to use new
Better chips distinguishing code for uplcom(4).
Add more and fix some IDs, all related to uplcom(4).
Add support for many new devices into uplcom(4). IDs are obtained from
Minimal (relatively) patch to make my Nokia 9300 smartphone which uses
Some trivial fixes obtained from NetBSD:
Add references to the uftdi(4), umct(4), and umodem(4).
ICMP Extensions for MPLS is porposed standard now - RFC4950.
Fix driver_t.
Fix warning.
Reomve unnecessary sys/vnode.h include.
ttyclose() increments t_gen. Remove redundant increments from drivers.
There is no need to explicitly call ttwakeup() and ttwwakeup() after
Magic Control Technology (MCT) USB to serial converters are not handled by
Add devices based on Silicon Laboratories USB-UART bridge.
Add uslcom(4) driver which provides support for USB devices based on
Add uslcom(4) into LINT.
uslcom(4) works with devices based on CP2103 chip. Tested by me with CP2103
Add some new uslcom(4) devices found in Linux driver.
Add some new uslcom(4) device ids found in Linux driver.
- Correct SYNOPSIS section in USB serial manpages.
Add uark(4) driver which supports Arkmicro Technologies ARK3116 chip found
Hardware flow control support for uslcom(4).
Add the ID of USB serial interface used in HandyTech Braille displays.
Add support for HandyTech's Braille displays into ubsa(4) (ID found in
Remove reference to the nonexistant uhub(4).
Add some devices based on Qualcomm HSDPA chips.
Add umsm(4) driver for EVDO and UMTS modems with Qualcomm MSM chipsets.
Make functions static.
Fix setting 115200 baudrate.
Add Smart Technologies USB to serial adapter.
Add Smart Technologies USB to serial adapter.
Use device_printf() where it makes sense.
There is no reason to be so verbose.
Bring in latest uftdi(4) driver from FreeBSD.
Add usbdi(9) manpage.
Nuke ARCnet support.
Nuke token ring support. This also means one blob less in DragonFly.
Nuke FDDI support.
Nuke ARCnet, Token Ring and FDDI related headers and manpages during upgrade.
Remove remainings of the oltr(4).
Nuke fla(4). It's known to be buggy, supports very limited set of obsolete
Remove fla(4) manpage during upgrade.
Nuke nv(4), we have nfe(4) which replaces it.
Missed this file in previous commit.
Simplify the way how chip type is determined. Instead of managing insane
Handle baudrate requests algorithmically with newer chips (not old SIO),
Better setpgid(2) documentation.
Add bus_alloc_resources() and bus_release_resources() functions to allow to
Update the agp(4) code to the latest one from FreeBSD HEAD. This brings in
Update the agp(4) manpage. General description and example based on NetBSD
Oops, remove comma.
uftdi(4) related usbdevs work:
Add many devices to the uftdi(4). Sources of the info are mainly Linux and
Add info about Vitesse Semiconductor Corporation VSC8601 PHY and Realtek
Add support for Vitesse VSC8601 and Realtek 8211B PHYs. Patches are obtained
Remove duplicates.
Unconstify members of the lconv structure to make it conform to the C89 and
Fix typo.
Remove terminating semicolons from SYSCTL_ADD_* macros. This will allow to
Add support for newer ICH SMBus controllers. Also corrected ICH4 entry in
Fix warning.
Hardware sensors framework originally developed in OpenBSD and ported to
Coretemp(4) driver for Intel Core on-die digital thermal sensor with patch
lm(4) and it(4) drivers for hardware sensors used in many motherboards. Ported
Fix synopsis (reminded by Constantine A. Murenin) and history.
Dragonfly always passes a flag for every IO operation depending whether
Simplify the code a lot - don't try to be too clever and handle chips with
According to RFC2711 routers shouldn't treat all packets with a Router
If answer to the repeated probe (same TTL as in the previous probe) packet
Update named.root to the version from 1 November 2007 from ftp.internic.net.
Sync with FreeBSD - add OpenBSD 4.2.
Kill usage of USB_VENDOR_FOO and USB_PRODUCT_BAR defines mostly using two
Nuke usbdevs and references to it.
Kill devinfo handling in drivers, set device description in one place -
- Add support for 230400 baud rate.
Add uticom(4) driver for Texas Instruments TUSB3410 USB to serial chips
Add moscom(4) - the driver for MosChip Semiconductor MCS7703 USB to
Add uchcom(4) - the driver for WinChipHead CH341/CH340 chips.
Add missing USB to serial drivers.
Update the uftdi(4) manpage.
Fix LINT build.
Remove 386 CPU support from the runtime linker.
-x was removed long time ago.
Fix typo.
Add SATA ATAPI support for AHCI controllers.
Fix the fix.
Fix typos.
Allow for any baud rate within a range rather than having a fixed list of
Add support for Intel 7221's and 845M GMCH controllers.
Fix no-sound issues with ASUS A9T notebook.
- Merge input/microphone support for ASUS A8N-VMCSM series.
- Add codec id for Realtek ALC268.
- malloc M_NOWAIT -> M_WAITOK.
- Gigabyte G33-S2H fixup, due to the present of multiple competing
Enable headphone jack-sense for HP nx6100 with AD1981B AC'97 codec,
Remap and virtualize mixer controls for HP nx6110 with AD1981B AC97 codec,
Add support for trimmed down version of ATI SB600 AC97 audio
Limit total playback channels to just 1, for ALi M5451. The reliability
* Fix support for followings:
- Add missing MCP65 id which was accidentally removed in previous commit.
Some trivial changes from FreeBSD that allow to use kgdb on /dev/fwmem0.0.
Hifn 7955/7956 support to the hifn(4) driver.
- Add a '-k' option which does not remove input file, like bzip2(1) do.
Add ID for ICH8M in compatibility mode. This makes Thinkpad X61s report
Remove references to drivers which don't exist in DragonFly.
Don't show in netstat(1) output without the -a switch TCP socket in LISTEN
Kernel part of bluetooth stack ported by Dmitry Komissaroff. Very much work
Pass all ATAPI commands through. Fixes detecting capabilities of DVD
When attached to a high-speed device, report a more appropriate
Make NO_GETMAXLUN quirk really do something useful.
Add missing ';'.
Pay attention to the timeout value passed down by the upper layer. This
Add bluetooth userspace libraries - bluetooth(3) and sdp(3).
Add /etc/bluetooth/ with common files.
Add bluetooth(3) and sdp(3) libraries. Adjust indenting in progress.
Add btconfig(8) - the utility used to configure Bluetooth devices.
Add sdpd(8) (Bluetooth Service Discovery Protocol daemon) and sdpquery(1)
Fix id of the 945GME chip.
Add few more usb devices. 0-5 are used in any modern machine and user might
Don't supress attach messages from devices other than first one while
- Fix compiling umsm(4) with UMSM_DEBUG
Add support for EVDO/UMTS card found in X61s Thinkpads.
Make sure we really do only the software part if we're dying. Fixes panic
Huawei UMTS/HSDPA adapetrs are already handled by umsm(4).
Increase size of the umsm(4) buffers and tty(4) input buffer to allow high
Defaults for btconfig and sdpd rc.d scripts.
Speed up uhub attachment considerably. Rather than powering up each port
Import libevent-1.3e.
Merge from vendor branch LIBEVENT:
Add READMEs.
Build libevent.
Add bthcid(8) - Bluetooth Link Key/PIN Code Manager and btpin(1) Bluetooth
Add rfcomm_sppd(1) - RFCOMM Serial Port Profile daemon.
Document bthcid(8) related variables in rc.conf(5).
Nuke wicontrol(8).
More wicontrol(8) removal.
Nuke the ntpd(8).
Short manpage for ubt(4). Most of it is commented out for now because sco
Implement net.bluetooth sysctls.
Add bluetooth(4) manpage.
Implement SCO related sysctls.
SCO sysctls are implemented now in ubt(4).
Mobe btpin(1), rfcomm_sppd(1) and sdpquery(1) where they really should be -
Remove btpin(1), rfcomm_sppd(1) and sdpquery(1) from /usr/sbin.
umsm(4) -> ugensa(4) as it makes much more sense - there is nothing Qualcomm
Update to the version 2008020400 which adds IPv6 addresses for six root
- Install bthcid.conf.
Add some Sierra Wireless devices found in Linux sierra driver to ugensa(4).
Add more device id's to the ugensa(4) taken mostly from option USB serial
Add _SC_NPROCESSORS_CONF and _SC_NPROCESSORS_ONLN variables to the
Fix pf and ipfilter module loading checks.
Fix typo.
Remove #ifndef __cplusplus around wchar related stuff in include/wchar.h
Fix buffer overflow in ppp command prompt parsing (OpenBSD errata 2008-009).
Make sure lo0 is brought up before any other interfaces to avoid problems
10Base-TX -> 10Base-T and 1000Base-TX -> 1000Base-T. Although 1000Base-TX
Regenerate miidevs.h.
Protect macros with "do { } while(0)" where needed.
Sync Bluetooth stack with NetBSD.
Decrease the number of reported stray interrupts from 100 to 10. Problems
DRM update to git snapshot from 2008-01-04.
Add double_t and float_t typedefs for both i386 and amd64 as required by C99.
Fix ifdefs to make it possible to use time.h in standards compilant code.
Sync Citrus iconv support with NetBSD.
Add libc support for gcc stack protector. Compatibility with gcc34 propolice
Add OMNIKEY CardMan 4040 smartcard reader.
Regenerate.
Add a driver for Omnikey CardMan 4040 smartcard reader - cmx(4).
Add support for cmx(4) devices.
Add useconds_t and suseconds_t used for time in microseconds.
Assorted fixes to ugen(4) from FreeBSD.
Remove udbp(4) form tree. It was never connected to the build and supports
Remove fortran from base.
Make use of interrupt endpoint to increase responsiveness.
Update the traceroute(8) to the newest code from FreeBSD HEAD. Besides many
Merge error fix.
Add support for for the AI_NUMERICSERV getaddrinfo(3) flag. While pulling
Add pthread_atfork() implementation to libc_r. libthread_xu has it already,
Unbreak buildworld.
According to SUSv3 including just regex.h must be enough. Fixes build of
Remove superfluous recursive lock. This little change makes possible (safe)
Remove the code which disables port status change interrupts for 1 second
Make sure host controller interrupts are not enabled until all
Fix probable cut-n-paste error.
Link libarchive against libbz2 and libz to be compatible with upstream.
Move timeval struct into its own header and include it from headers where
Change suseconds_t to long as it is in most of systems. Fixes a lot of
Add objc to the gcc-4.1.2.
Merge from vendor branch GCC:
Build objc support.
We are using 4.1.2 really.
Fix [gs]etsockopt(IP_HDRINCL) which allows mere mortals like me to obtain
Detach correctly so there is no need to panic during reattach.
Cleanup err/error mess in the uticom_download_fw().
Use IPv6 documentation prefix (RFC3849) instead of 6bone prefixes in
Fix interrupt pipe processing to treat a ugensa(4) interrupt message
Remove some useless variables and assignments from USB code.
Fix some NULL pointer dereferences, most of the in debug code though.
Make Huawei E220 change the mode from device with single umass interface to
Some agp(4) fixes:
Add some methods to ACPI to handle embedded controllers and device matching.
Move acpi_toshiba.c, it's not pc32 specific.
Add ACPI support module for IBM/Lenovo Thinkpad laptops. Work in progress,
Add STAILQ_FOREACH_MUTABLE.
Add acpi_video(4) - a driver for ACPI video extensions.
Document STAILQ_FOREACH_MUTABLE macro.
Add acpi_video(4) manpage and move acpi_toshiba(4) manpage out from man4.i386.
acpi_thinkad(4) manpage.
Unbreak build.
Remove dhcp-3.0 from base and import dhclient from OpenBSD. Porting work
Upgrade pieces for new dhclient and dhcpd/dhcrelay removal.
Remove dhcpd and dhcrelay remainings.
Make BOOTP server in installer work with dhcp server from pkgsrc.
Handle (unit == -1) case in the device_find_child() function as already used
Fix coretemp(4) to provide temperatures from all cores (instead of reading
Make pkgsrc/wip cvs checkout use -P.
The result of the "RFC3542 support" SoC project by Dashu Huang.
Bring in newer ping6(8) from KAME via FreeBSD lying in my disk quite some
Forgot this file in the "RFC3542 support" SoC project commit.
acpi_cpu(4) update. It's now possible to use higher (lower power usage) C
Quite minimal patchset to help to save some more power - put unused PCI
Sync pci_[gs]et_powerstate_method with FreeBSD which makes things a little
Make acpi support modules depend on acpi module.
Turn power off for detached (module unloaded) PCI devices. No power down is
Put IPV6_RTHDR_TYPE_0 define back until the world fixes itself.
Add _SC_PAGE_SIZE as alias to _SC_PAGESIZE to make modern software pieces
Attempt to fix the crash if accessing usb device(s) after unloading usb.ko.
Update acpi_battery(4) related code to the latest one from FreeBSD HEAD.
The devinfo(3) library provides userspace access to the internal device
Welcome devctl(4) and devd(8).
devctl(4)/devd(8) support in acpi_thinkpad(4).
If a neighbor solictation or neighbor advertisement isn't from the
Correctly handle Intel g33 chips and add support for g45 chips.
Don't allocate space for empty banners. Makes me able to connect various
How buggy this little piece of code could be? Repair strnvis() buffersize
Add code to parse the utrace(2) entries generated by malloc(3) in a more
Update sensorsd(8) to the latest code.
Fix CVE-2008-3831. Affects the Intel G33 series and newer only.
Bring in some fixes from FreeBSD. Amongst other fixes, like panics in debug
We don't have /dev/audio support any more, but make it symlink to /dev/dsp
Hopefully more bulletproof workaround to fix problems with SATA ATAPI
Make apps using '#define _POSIX_C_SOURCE' compile.
Install acpiio.h.
Add hardware type value define for IP over firewire. Not used yet.
Unbreak installworld.
Sync libusbhid with other BSDs (breaks API compatibility). Sync usbhidctl
Remove /usr/include/libusbhid.h.

Hidetoshi Shimokawa (11):
Add dcons(4), a pseudo console driver for FireWire and KVM interface.
Use opt_dcons.h.
Add support for eui64(5) to libc.
dconschat - user interface to dcons(4)
Add dcons(4) related manpages.
Hooks to build dcons(4)/dcons_crom(4).
Add dcons(4).
Update FireWire device nodes.
Add eui64.5.
Preserve dcons(4) buffer passed by loader(8).
Sync with FreeBSD-current:

Hiroki Sato (6):
Test commit.
Replace IPv6 related manual pages that may have violated
- Nuke #ifdef SCOPEDROUTING. It was never enabled and is useless now[1].
Move to ND_IFINFO().
Query A records before AAAA records in getaddrinfo() when AF_UNSPEC
Fix a bug which can allow a remote attacker to cause denial

Hiten Pandya (452):
Add a handy macro, called FOREACH_PROC_IN_SYSTEM(p) for better
Consolidate usage of MIN/MAX().
mdoc(7) assorted fixes:
mdoc(7) assorted fixes:
Oops, fix build before anyone realises. :-)
Merge from FreeBSD:
Fix building of vm_zone.c in the case of INVARIANTS.
DDB updates:
Bring the malloc/mbuf flags up-to-date with FreeBSD.
Bring us in sync with 4.8-STABLE.
Assorted mdoc(7) fixes:
Kernel Police:
Kernel Police:
Comment out the unused proc/thread declarations.
No need to comment out unused thread/proc declarations (prev. revision),
Remove INVARIANT_SUPPORT conditional pre-processing on zerror(). Just
style(9) police:
Consolidate MIN() usage across kern/ tree.
- Make `vmstat -i' working after Matt's interrupt related
LINT breakage fix, part one:
Generalise, and remove SI_SUB_VINUM; use SI_SUB_RAID instead.
Get LINT to build.
Use FOREACH_PROC_IN_SYSTEM() throughout.
LINT cleanup: add <sys/systm.h> for printf()
LINT cleanup: fix MAX difinition.
Define HAVE_PPSRATECHECK macro, now that we have merged the
LINT cleanup: remove redundant ``struct proc *p'' declarations.
Add backtrace() prototype.
DELAY() does not belong in a MD include file, instead, move
Move the backtrace() function from kern_subr.c to kern_debug.c.
Um, ok. I should have slept a little more.
Backout my ``proc cleanup'' brain damage.
Remove the documentation/ folder from under 'src' tree.
Add the TINDERBOX kernel configuration file.
Remove `YOUR'.
Add two useful macros that I have been meaning to add for quite
Fix style issue.
Use addalias() to track the vnode if it not of a regular type.
Optimize split(1) by using dynamic allocation for buffers.
Add a `-q' option to killall(1). This is helpful when you don't
Fix the case when `-p' option of truss can be passed the pid of
Fix vmstat(1) diagnostic output.
Fix sorta critical bugs in fsdb(8); the @mtime and @atime reporting
Add device indentification for Intel 82801DC (ICH4) SMBus Controller.
Add device identification for Intel 82801EB (ICH5) SMBus Controller.
1) Add new tunable, kern.syncdelay:
Respect ps_showallprocs when using the Proc file system.
Use vm_page_hold() instead of vm_page_wire().
Return a more sane error code, EPIPE. The EBADF error code is
Introduce a new poll operation bit, `POLLINGIGNEOF'. It is used for
Fix style nit.
Pass only one argument to vm_page_hold() as a sane person would do.
Check when M_PREPEND returns an empty mbuf.
Change my e-mail.
Fix logic, flow and comments for previous (rev 1.9) change.
OK, fix build, while I am there, I will get some screw for my head
Add MLINK mount_ufs(8) to mount(8).
Mdoc cleanup of section 4:
Add the linux(4) manual page.
Fix IPFW2 build.
K&R style function removal. Update functions to ANSI style.
Sort Copyright order.
Fix L2 internal cache reporting when it is an AMD Duron rev. A0 chip.
K&R style function removal. Update functions to ANSI style.
Rename:
K&R style function removal. Update functions to ANSI style.
Fix a spelling mistake.
Security Fix:
Correct unsafe use of realloc().
Fix compile when GUPROF is defined.
Fix kldload(2) error return when a module is rejected becaue it is
AMI MegaRAID Crash-Dumps support.
Fix an ordering issue, call vm_map_entry_reserve() prior to locking
Nuke the zalloci() and zfree() stuff sky-high. We no longer have
Remove zalloci/zfreei from the Makefile too.
Major contigmalloc() API cleanup:
Second contigmalloc() cleanup:
Fix build by not attempting to compile libc's malloc directly.
Add SysV IPC regression suite.
Use vnconfig(8) instead of mdconfig(8) to enable ${swapfile},
Update sysinstall's NFS module:
Add the pim(4) and multicast(4) manual pages, following
Fix a ``missing free''.
Un-initialise (i.e free) the HPFS inode hash; previously, unloading
Fix two bugs in split in split revealed after my optimization in
Remove unneeded XXX, to rename lwkt_create() as it was named
Add a prototype for if_indextoname().
Fix long interface name handling.
Support for conflict (pkg_add -C) checking and pkg_tools (pkg_add -P).
Add pkgwrap.c, which was missed in my last commit.
Improve device identification strings.
Fix typo in a comment.
Fix build of the PPS driver.
Remove the pca driver from the build for now, due to the revamp
Merge from FreeBSD:
Merge from FreeBSD:
Merge from FreeBSD:
Merge from FreeBSD:
Merge from FreeBSD:
Merge from FreeBSD:
Merge from FreeBSD:
Merge from FreeBSD:
Merge from FreeBSD:
Update to devlist2h.awk and friends:
Include Makefile.miidevs, so we can just do:
Merge from FreeBSD:
Merge from FreeBSD:
NEWCARD: change `device card' to `device pccard'.
Update maintainer contact information.
Score a duh-point for myself. Change the remaining lines for the
Merge from FreeBSD:
Merge: FreeBSD (RELENG_4) uipc_socket.c rev. 1.68.2.24
Merge: FreeBSD (RELENG_4) isp_ioctl.h 1.1.2.5
Merge: FreeBSD (RELENG_4) i386/isa/psm.c rev. 1.23.2.7
Merge: FreeBSD (RELENG_4) netstat/inet.c rev. 1.37.2.11
Update the Broadcom Gigabit Ethernet driver and the Broadcom
Remove my middle initial.
Correct a filename typo, there is no such thing as machine/pcpu.h,
Bring the cue(4) and miibus(4) manual page in line with
Include thread.h if _KERNEL_STRUCTURES is defined.
Do not print a warning about PIM sysctl node (net.inet.pim.stats)
I just scored a few duh-points for myself. I committed an older version
Bring the BFE(4) manual page up-to-date with FreeBSD RELENG_4.
Merge from FreeBSD:
Merge from FreeBSD:
Merge from FreeBSD:
Document that kldload(2) can also return EEXIST.
Fix violating usage of M_DONTWAIT in calls to malloc() by replacing
Adjust IPFW to use M_WAITOK instead of M_NOWAIT. The M_NOWAIT flag on
Fix spelling mistake, s/itnerrupts/interrupts/.
Convert the code to ANSI style, and remove 'register' keywords.
Linux emulation system call update.
Mega mdoc(7) update:
Replace a manual check for a VMIO candidate with vn_canvmio() under
Use info->td instead of curthread in ffs_reload_scan1(); although
Collapse the if(...) block in pim_stats added by my previous commit
Correct typo in comment.
Turn TDF_SYSTHREAD into TDF_RESERVED0100 since the flag is never used
Integrate remaining part of the network interface aliasing
Integrate the remaining parts of the network interface aliasing
Fix loading of the SMBFS kernel module. The KMODDEPS line in the SMBFS
Give the VFS initialisation functions an update:
Add manual page for the busdma(9) API. It has detailed information on
Merge: FreeBSD (HEAD) sys/kern/sysv_sem.c rev. 1.69
Merge: FreeBSD (RELENG_4) ip_fw2.c rev. 1.6.2.19
Correct a bug in vm_page_cache(). We should make sure that a held
Per-CPU VFS Namecache Effectiveness Statistics:
The globaldata houses a pointer and not an embedded struct for nchstats;
Add Makefile for the netif/ie ISA NIC driver.
Adapt the netisr message handlers to accomodate the available error
Garbage-collect unused variable.
Add `device atapicam' to unbreak TINDERBOX config.
Merge: FreeBSD (RELENG_4) aac_pci.c rev. 1.3.2.19
Merge: FreeBSD (RELENG_4) kern_descrip.c rev. 1.81.2.19
Merge: FreeBSD (RELENG_4) msdosfs_vfsops.c rev. 1.60.2.9
Merge: FreeBSD (RELENG_4) kern_event.c rev. 1.2.2.10
Merge: FreeBSD (RELENG_4) vfs_syscalls.c rev. 1.151.2.19
Add a KKASSERT to mount(2) to make sure we have a proc pointer.
Rename the sysctl handler for nchstats to reflect reality; I named it
The "Hashed Timers and Hierarchical Wheels: Data Structures for the
Fix compilation of profiling.
1) Move the tcp_stats structure back to netinet/tcp_var.h.
Make IP statistics counters per-CPU so they can be updated safely.
Clean warnings under DIAGNOSTIC.
Use the correct cast, ns_ifra_addr -> ns_ifaddr.
Move around some #ifdefs to silence warnings.
Add a forward declaration of 'struct uidinfo'.
Bring the I4B layer up-to-speed with 64-bit physical addressing.
Handle UIO_USERISPACE (just fallthrough to UIO_NOCOPY), to silence
Just pass NULL to sync(), no need to create a `dummyarg'.
Catch up with if_ioctl prototype changes (see rev. 1.10 of net/if_var.h).
Bring le(4) up-to-speed with 64-bit physical addressing.
netif/cx/cx.c:
Correct pre-processor conditional surrounding mmxopt's declaration by
Add bus_alloc_resource_any(9).
KKASSERT that we require inp->inp_pcbinfo, in in_pcbinswildcardhash().
Add a readonly sysctl for the `kern.mmxopt' loader tunable (same name).
Remove redundant newline in a call to panic(9).
Remove newline from panic(9) message, it is redundant.
Remove newline from panic(9) message, it is redundant.
Add MLINK busdma(9) which points to bus_dma(9).
Update the tsleep(9) manual page about our reality.
Update the DELAY(9) manual page about the header file where
Update the KASSERT(9) manual page to reality.
Update the suser(9) manual page about reality.
Correct mdoc(7).
Remove unneeded empty line to silence mdoc(7) warnings.
Mdoc(7) police:
Mdoc(7) police:
Remove erroneous use of the `Fl' mdoc macro and replace it with
Correct the use of the .Dx/.Fx macro.
Add entry for the CAPS IPC library. It is now possible to refer to the
Correct the usage of the .Dx macro to avoid mdoc errors.
Remove extraneous `.El' macro.
Correct the usage of the .Dx macro.
Correct usage of the `.Em' macro.
Do not specify a macro as first argument to the literal macros
In sodealloc(), use do_setopt_accept_filter() to free an accept filter
Add a read-only sysctl for observing the maximum number of
Remove an extra comma at the end of .Nm list.
Add a manual page which documents the generic hash routines, i.e.
Add an MLINK for KKASSERT(9).
Document the ``resource management'' (rman) abstraction in rman(9).
Quickly fix an MLINK while no one is looking...
Update the mlock(2) manual page:
It is DragonFly BSD, not FreeBSD. Correct config(8) files.
Document the pmap_kenter_quick(9) function. While I am here, fix
Fix SYSCTL description style.
Wrap the VM MAP locking routines with _KERNEL, user-land has no
Consolidate SYSCTL_DECL(_kern_ipc), move it to sys/sysctl.h as
First pass at updating top(1):
Adjust include path so that the machine.h in ${.OBJDIR} is used
Force commit to clarify that the previous revision should have been
Remove a stale comment: PG_DIRTY and PG_FILLED were removed in
Remove an unimplemented advisory function, pmap_pageable(); there is
Do not use the display function if the -o (opaque) or -x (hexdump)
Remove '-*- nroff -*-'.
Correct spelling.
Merge from FreeBSD, RELENG_4 branch, revision 1.250.2.26.
Plug a memory leak when the kernel initialiazes config_devtab resources
Remove some long gone #defines, PFCLUSTER_BEHIND and PFCLUSTER_AHEAD;
Quotactl(2) should set the uid correctly, based on the QUOTA type supplied.
Cleanup the textvp_fullpath() function; summary of changes:
VM Resident Executables update:
Cleanup the manual page:
Remove the compat macro textvp_fullpath(), and use vn_fullpath()
Surround a multi-line conditional block with braces for readability.
Lock manipulation of the 'exec_res_list', i.e., the list of resident
Add a manual page which describes the vn_fullpath(9) function.
Flush cached access mode after modifying a files attributes for NFSv3.
Deprecate use of m_act, which is an alias of m_nextpkt; just use
Allow top(1) to toggle display of processes+threads or *only* threads
Re-arrange the 'Executable' column for the '-l' option so that long
Add a manual page describing the LSI Fusion family of devices with a
Discard the first 1024 bytes of output as suggested by
Now that we have clients that use m_getcl(9), set the default mcl_pool_max
Do not use the installed include files, instead, set the include path
Implement POSIX.1-2001 (XSI)'s ulimit(3) library call.
Add a reference to the ulimit(3) manual page.
Register keyword removal. No operational changes.
ANSI-fication. No operational changes.
Dissolve use of the 'register' keyword.
Use correct sentence structure.
Remove useless __STDC__ junk.
Avoid a memory leak if vfprintf(3) by always calling va_end(3); this
Use ANSI C prototypes and remove the !__STDC__ varargs compatibility
FUNLOCKFILE(fp) *after* modifying the FILE pointer's fields.
Set the return value properly for fgetpos(3).
Document security issues with gets(3) in a proper manner.
mdoc(7) corrections; use .Dv instead of .Em etc, fix grammar.
C99 update: freopen(3) with NULL 'path' argument so that it opens the
Re-order include, 'sys' includes first, 'vm' includes after.
Conditionally include the essential header files, sys/queue.h and
Remove an accessory function called msf_buf_xio(); it is unnecessary for
Replace the use of specially reserved pbufs in NFS's nfs_getpages() and
IEEE Std. 1003.1-2001 dictates that fileno(3) behave as it locked the
IEEE Std. 1003.1-2001 wants feof(3) to behave as if it locked the FILE
If handed a bad file pointer that we can't write to, set the errno value
Revert previous commit about FUNLOCKFILE(fp), it causes ill (and weird)
Add a 'big fat comment' about the FUNLOCKFILE(fp) implementation and why
Add some whitespace for clarity. No operational changes.
Revert the locking of feof(3) for now; there is possibility of ill
Do not produce a warning if the sysctl does not exist. This happens when
If there was a cache hit, and the msf->m_flags was set to SFBA_ONFREEQ,
Make the procfs_validfile() function static.
Fix spelling in comment.
Spell 'written' properly.
Add a helper utility, called asf(8): Add Symbol File.
Fix prototype of usage().
Add the csplit(1) utility, which splits files based on context, as
IEEE Std. 1003.1-2001 (SUSv3):
Hook up the recently added utilities [1] to the build.
Add the POSIXv2 asa(1) utility; it interprets FORTRAN carriage-control
Hook up the asa(1) utility to the build system.
Zero the interval timers on fork(2) rather than copying them to the
Zero-out the whole pstats structure and then copy the relevant fields,
Move the 'p_start' field from struct pstats (Process Statistics) into the
Use the kern.boottime sysctl for retrieving the system boot time as a
Check kp (struct kinfo_proc *kp) against NULL and not 0, because it is a
Remove references to FreeBSD in the comments.
Fix indentation.
Merge rev. 1.7 of FreeBSD's src/sys/dev/twa/twa_freebsd.c.
Turn on the DDB_TRACE config option, so that we will get a trace as soon
Merge from FreeBSD-4, revision 1.115.2.20:
Merge from FreeBSD-4, revision 1.1.2.9:
Add a debug directory under src/test, where we will house all of our debug
Document the 'running_threads' GDB macro.
Update our xlint(1) to work with recent preprocessor changes.
Randomize ephermal source ports.
Generalize a comment, remove 'FreeBSD' from the comment because we are
Remove the advertising clause from this file.
Use the official devised license for the first time, starting with
Fix a small but important mistake.
Bunch of mdoc(7) and various file path fixes.
Add a bunch of .Fx (FreeBSD) version numbers.
Correct mdoc(7) errors:
Use the .Dv directive for marking up PROT_* constants.
A variety of mdoc(7) and grammar changes:
Correctly use the .Bd directive, i.e., present it with the -literal
Just use the .Ev directive for printing an environment variable,
Use the official 3-clause license for the MSFBUF header file.
Update list of FreeBSD version numbers, for use with manual pages.
BUF/BIO work, for removing the requirement of KVA mappings for I/O
Correct reference to buf->b_xio.xio_pages in a comment.
Append necessary information to the package name for portupgrade(8) to
Remove an erroneous '+' symbol at start of 'rand_irqs'.
Update a stale comment about lwkt_replymsg().
POSIX update and cleanups for getopt(3):
Correct mdoc(7) processing errors; the .Bl directive should be provided
Correct mdoc(7) for basename(1) and passwd(1) manual pages.
Merge changes from FreeBSD:
Stop depending upon an implicit 'int' as the return type of main().
Annotate the b_xio field member of the BUF structure.
Change all files that I own to use the official DragonFly Project
Merge revision 1.25 of src/usr.bin/time/time.c from FreeBSD.
Merge revision 1.26 of src/usr.bin/time/time.c from FreeBSD, and an
Eliminate hard sentence breaks.
Eliminate hard sentence breaks.
Double semi-colon police!
BUF/BIO stage 2:
Remove an erronous 'static' in front of pci_alloc_resource(9).
Add a prototype for isab_attach(), which is used by the ACPI-5 ISA
Bring definition of va_list and friends, so that ACPI module actually
Eliminate hard sentence breaks.
Merge revision 1.16 of src/usr.bin/rusers/rusers.c from FreeBSD.
Minor KNF/style cleanups. No operational changes.
Display proper information when the verbose flag (-v) is passed to
Aesthatical change, move 'All rights reserved' to the same line as
Include <sys/types.h> and <netinet/in.h>, so that parse.y gets the correct
Fix compiler warnings; include <sys/types.h> and <netinet/in.h> to satisfy
Mechanically kill hard sentence breaks.
Respect locale settings from the environment.
Add new options to select different registries, basically synch'ing us
Assorted spelling fixes from Christian Brueffer <brueffer@freebsd.org>.
It is 'estcpu', not 'setcpu'.
Add the C99 utility, now that we have a decent C99 compiler in base, i.e.
The return-path is optional in a headline, therefore don't skip a message
Merge mdoc(7) corrections from FreeBSD -CURRENT.
Merge revision 1.8 and 1.9 from FreeBSD -CURRENT, i.e., add a section on
Aesthetic changes:
Add MOD_SHUTDOWN to be processed by the module event handling function.
Fix generation of opt_inet.h and opt_ipx.h by providing their targets
Remove UIO_USERISPACE, we do not support any split instruction/data
Remove VAX conditionalized code.
Add the ieee80211(9) API manual pages.
KNF/style changes. Match scanfiles() function prototype with the rest
Minor cleanups to bring us on-par with FreeBSD's cat(1):
Use the correct header file, which is located in netproto/802_11.
KNF/style and warnings clean up. ANSI style prototype for printb(), and
Major cleanup of the base IPFilter:
Readd the $DragonFly$ Id tag which I removed by mistake in previous
Correct spelling.
Test Commit
Add a SECURITY section which describes the kern.ckptgroup sysctl among
Reorder included headers to an acceptable standard. System headers
Tell the reader about sys/buf.h as well.
Document lockcountnb(9) which is the non-blocking counterpart of
Slap 2005 into the COPYRIGHT. Happy New Year!
Slap 2005 into the copyright. Happy New Year!
Improve the IPFilter rc.d file, mainly bring it in line with changes
Move the MALLOC_DECLARE into sys/msfbuf.h header file.
Remove a comment that does not make sense; it was just a remnant of
Fix whitespace for function prototypes.
Header include protection from userland.
Rename the flags for sf_buf_alloc(9) to be in line with FreeBSD:
Add a typedef msf_t for struct msf_buf *.
Cleanup build warnings, cast and type modifications.
Remove 'rttrash', it has been long removed from the routing code; former
Remove stale inclusion of struct disklabel and other bits from
Remove contents of include files from their associated manual pages thus
Fix GCC3 related pre-processor issues.
Fix mistakes from previous commit, cleanup _NCH_ENT macro.
Conditionalise include of "opt_ktr.h" under _KERNEL.
Fix display of code in the EXAMPLE section; do not enter vertical space
Bring in the ktrdump(8) utility from FreeBSD.
Update manual page with the version in FreeBSD, r1.8.
Update manual page with FreeBSD's r1.7.
Fix $FreeBSD$ ident string.
Add a new function to the nlookup API, called nlookup_set_cred(9); this
Provide a better annotations for KTR_VOP and KTR_NFS trace classes.
Add manual page for closefrom(2) system call and hook it to the build.
Add entry for DragonFly 1.2.
Add standard entries for libcaps, libkinfo and libkcore; they can now be
Add 'mycpuid', a #define for accessing mycpu->gd_cpuid.
Mechanical cleanup of TCP per-cpu statistics code, better naming etc.
Remove the '_GD' macro hack:
Forced commit to note previous change, of introducing 'mycpuid' was
Mechanical cleanup of IP per-cpu statistics code, better naming etc.
Ick! Seems I was under a placebo, correct a last minute typo.
Use the correct macro for printing TCP statistical counters when we only
Add bpf_mtap_family(9) which allows the client to specify an address
Fix whitespace in copyright dates.
Be consistent for preventing redundant header inclusion.
Change CPU time statistics (cputime) to be accounted on a per-CPU basis.
Adapt the KINFO library to aggregate per-cpu cputime statistics.
KINFO library cleanups:
Fix breakage.
Adapt kcore_get_sched_cputime(3) to retrieve and present an aggregated
Minor adjustment of types for consistency.
Use reallocf(3) and cleanup some NULL checks.
Add some useful comments to interface functions; required a little bit of
Networking routing statistics on a per-CPU basis:
Correct typo in comment for vshiftl().
Covert netproto/ipsec into using critical sections instead of SPL ops.
The BLIST API is just as usable in userland as it is in the kernel; and
Minor word-smithing.
Use _KERNEL macro for wrapping kernel-only code.
Add code comments to improve object type documentation.
Add counters for recording Token/MPlock contention, this would help in
Correct a typo, v_rbdirty_tree is for dirty buffers.
Remove conditional bits about other operating systems, they are not
Update the unifdef(1) utility to the latest code from FreeBSD, and the
Remove outdated information with regard to old tinderbox.
Clean the VFS operations vector and related code:
Add some useful GDB macros that have been sitting in my local tree,
Add minimal manual page explaining use of bread(9) and bwrite(9) kernel
Remove items from TODO list that have been implemented in last major
*** empty log message ***
Forced commit to ensure revision 1.18 fixed a build break introduced
Update the physio(9) manual page to reflect reality. Clean up the
Update description of msfbufs sysctls.
Update filename in comments.
BUF/BIO cleanup 2/99:
Add flag for compiling the tests without GCC -O2 optimisations, quite
Add test for comparing following types of conditionals:
Add a workaround to make 3COM cardbus cards to propagate the device
BUF/BIO cleanup 3/99:
Bring name of an unused flag field in line with the rest.
Put unused flag space definitions back to their original position in
Initialize buf->b_iodone to NULL during bufinit(9) stage.
BUF/BIO cleanup 4/99:
BUF/BIO cleanup 5/99:
Update copyright notice. Use standard DF copyright.
Add 'debug.sizeof.buf' sysctl for determining size of struct buf on a
BUF/BIO cleanup 6/99:
A better description for 'debug.sizeof' sysctl.
BUF/BIO cleanup 7/99:
Add minimal utility that is able to make sense of the per-cpu load
Move the bswlist symbol into vm/vm_pager.c because PBUFs are the only
Remove stale comment about vm_mem_init's arguments.
Make a few PRINTF lines readable, break them up if necessary.
Whitespace cleanup.
Re-word some sysctl descriptions, make them compact.
Wrap 'pqtype' variable with INVARIANTS so annoying GCC warnings are
Move bio_lblkno (logical blockno in a file) field back to its rightful
Remove PV_* flags from PMAP MD header files; these were made useless
Update the TWE 3ware/AMCC driver code, bringing in all the fixes made
Remove NO_B_MALLOC preprocessor macro, it was never turned on, and
Add more documentation comments to disk_create() and dscheck().
Style: break line into two, so it fits nicely in 80-column mode.
Add a comment on top of ad_start, mentioning that it is called with
Document the dscheck(9) function and explain how it affects the slice-

Jeffrey Hsu (278):
Add support for RFC 3390, which allows for a variable-sized
Use relative directory, rather than /sys, as base directory for "make tags".
Implement the Eifel Dectection Algorithm for TCP (RFC 3522).
New statistics to keep track of supurious retransmits.
Fix spurious spelling within comments.
Make the logic clear on when to use Eifel detection or fall back
Add statistics to disambiguate how a spurious retransmit was detected.
Decouple slow-starting an idle connection from Nagle's algorithm.
Non-semantic-changing cosmetic transformation. Gets rid of unnecessary
Properly handle an error return from udev2dev().
Fix typos in comments.
Add support for Protocol Independent Multicast.
Add support for Protocol Independent Multicast.
Account for when Limited Transmit is not congestion window limited.
Differentiate between send and receive window variables.
Introduce the DDB_TRACE kernel config option to automatically print a stack
Non-semantic changing cleanups:
Centralize if queue handling.
Add missing interface queue.
Leftover netisr consolidation cleanups.
Leftover netisr consolidation cleanups.
Leftover netisr consolidation cleanups.
Leftover netisr consolidation cleanups.
Reset the retransmit counter when setting the timer on a failed
Optimize out an unneeded bzero().
For performance reasons, kernel sends should not be subject to sockbuf
Unroll obfuscated loop.
Unravel a nested conditional.
Remove illegal identifier after #endif.
This patch improves the performance of sendfile(2). It adds a hash
Pull the sf_buf routines and structures out into its own files in
Merge from FreeBSD:
Merge from FreeBSD rev 1.43 (original commit made by tjr@FreeBSD.org).
Merge from FreeBSD:
Remove dead code.
Remove unused local variable.
Cosmetic code cleanup.
Relax a KASSERT condition to allow for a valid corner case where
Cosmetic changes.
Split out wildcarded sockets from the connection hash table.
A UDP socket is still bound after it is disconnected, so we need to
Introduce access methods for making protocol requests.
Once we distribute socket protocol processing requests to different
Propagate curproc removal changes to files compiled by LINT.
Remember the next lowest power of 2 of "npus" in "ncpus2".
Use power of 2 masking to make packet hash function fast.
Verify code assumption on number of processors with a kernel assertion.
Dispatch upper-half protocol request handling.
Dispatch upper-half protocol request handling.
Correct double increment of the inp generation count.
Use 0 for integer value rather than NULL.
Change the "struct inpcbhead *listhead" field in "struct inpcbinfo"
Eliminate the use of curproc in route_output() by passing down the process id
Remove unused second argument to ip_stripoptions().
Send UDP packets out without a temporary connect.
Cosmetic changes.
Implement Early Retransmit.
Print out Early Retransmit statistics.
Include <sys/types.h> for autoconf/automake detection.
To comply with the spec, do not copy the TOS from the outer IP
Partition the TCP connection table.
Clarify strange ipfw byte ordering convention.
Make tcp_drain() per-cpu.
Make tcp_drain() per-cpu.
Ifdef out unused variable.
Cosmetic cleanup.
Consolidate length checks in ip_demux().
Eliminate use of curthread in if_ioctl functions by passing down the
Do all the length checks before returning even if "ip_mthread_enable"
Eliminate use of curproc and curthread by propagating thread pointer down
Need header file to deference proc structure.
Directly call pru_control until copyin problem is resolved.
Give UDP its own sosend() function.
Pull out m_uiomove() functionality from sosend().
Change sendfile() to send the header out coaleseced with the data.
Only enter wildcard sockets into the wildcard hash table.
Only enter into wildcard hash table if bind succeeds.
Only enter into wildcard hash table if bind succeeds.
Remove the ip_mthread_enable sysctl option. Enough code has been converted
Consolidate length checks in ip_demux().
Fix byte-order.
Dispatch reassembled fragment.
Consistently use "foreign" and "local", which are invariant on the
Cosmetic changes.
Workaround for not having a proc context. Use the thread0 context when
Push the lwkt_replymsg() up one level from netisr_service_loop() to
Fix typo with last minute change in last commit.
Add header file to pull in the setting of the TCP_DISTRIBUTED_TCBINFO option.
Send connects to the right processor.
Add predicate message facility.
Make the declaration of notifymsglist visible outside #ifdef _KERNEL
Always send the sendfile header out even if the file has no data.
Fix compilation errors with missing header files and misnamed formal parameter.
Silence warning about missing prototype.
Silence compiler warning by adding include files.
Create another entry point into ip_input() so MT_TAGs will work.
Don't need opt_tcp_input.h for TCP_DISTRIBUTED_TCBINFO anymore.
Cosmetic changes.
Allow an inp control block to be inserted on multiple wildcard hash tables.
Pass more information down to the protocol-specific socket dispatch function
Use a message structure off the stack for a synchronous call.
Drop packet if the length checks fail in ip_demux().
Replicate the TCP listen table to give each cpu its own copy.
The default protocol threads also need the check for
Cosmetic changes.
Remember if an inpcb was entered into the wildcard table to save
Add restrict keyword to string functions.
Move accounting of sendfile header bytes sent down one level to handle
Trade off more writes for a simpler check for when to pull up snd_recover.
Put snd_recover in the same cache line as snd_una. Make room in the
Panic in udp_output() if a socket is found in an inconsistent state.
Allow an inp control block to be inserted on multiple wildcard hash tables.
Detect and foil optimistic ACK attack with forced slow-start
Close race condition in accept(2).
Try the ELF image activator first.
Update some of my copyright notices before we officially publish
Add the standard DragonFly copyright notice to go along with mine.
Add the standard DragonFly copyright notice to go along with mine.
Increase the size of the nfsheur hash table as pointed out by
Readability changes, mostly removing the option to not do NewReno,
Fix bug with tracking the previous element in a list.
Get cosmetic changes out of the way before committing SACK.
Move a comment to the right place.
Allow the syncache to run lock-free in parallel on multiple processors
Update includes now that the Fast IPSec code has moved to netproto/ipsec.
From KAME freebsd4/sys/netinet/ip_input.c rev 1.42:
Remove duplicate comment.
Separate out the length checks from IP dispatch and also do them along
Correct use of the flags argument to the recvmsg system call.
Implement SACK.
Fix bug with wrong length being used when coalescing out-of-order segments.
Properly propagate the FIN flag from the following to-be-coalesced segment.
We have to replicate listening IPv6 sockets in the wildcard table
Handle window updates inside header prediction to increase the hit rate.
Accept resets sent while the receive window is zero.
Cache a pointer the last mbuf in the sockbuf for faster insertion.
Merge from FreeBSD:
Clean up routing code before I parallelize it.
Clean up routing code before I parallelize it.
Patch up user/kernel space difference with boolean types.
Fix whitespace.
Clean up the routing and networking code before I parallelize routing.
Fix problem with last commit that was breaking tcpdump.
Clean up the networking code before I parallelize the routing code.
Fix off-by-one error with the range check for PRC_NCMDS.
FreeBSD PR: kern/54874
Fix buffer overflow bug involving inet_ntoa().
Forced commit to say the previous commit wasn't really a buffer overflow
Fix compile error.
Add the new ND6_IFF_ACCEPT_RTADV flag to control whether to accept
Add the "accept_rtadv" interface option to specifiy whether to accept
Document the "accept_rtadv" interface flag.
Back out port randomization. FreeBSD users report problems with it under load.
Fix double-free problem when sysctl net.inet.ip.rtexpire=0.
Fix double-free problem when sysctl net.inet.ip.rtexpire=0.
Correct a byte-order bug with fragment header scanning.
Set ip6_v6only to true by default. The administrators who want to use
Cosmetic cleanups.
Move a global variable into local scope for MP safety.
Fix compile error.
Now that I understand the poorly written BSD routing code and what
Catch up to recent rtlookup() changes.
Fix copyright notice.
Remove the sysctl options for altering the initial TCP congestion window size.
Remove the sysctl options for altering the initial TCP congestion window size.
Increase the default TCP maximum segment size from 512 to 1460.
Instead of explicitly initializing "fp" to NULL in kern_sendfile(),
Now that we generate the ethernet header in place in the mbuf instead
Code cleanup. Refactor some functions. Push some globals into local scope.
If dhclient fails, an interface could be left with an IP address of 0.0.0.0.
Temporarily disable non-working Path MTU discovery pending real fix.
Now that 'so_pcb' is properly declared as a 'void *', remove a layer of
Eliminate conditional check for initialized 'fp' on error in kern_sendfile().
Strip away convoluted route reference counting logic.
Clear up confusion about negative route reference counts.
Use malloc(M_ZERO) in preference to separate bzero() after allocation.
Convert the struct domain next pointer to an SLIST.
None of the callers of rtredirect() want to know the route that was modified,
Remove (void) cast before a function call with an unused return value.
Readability changes.
Change a 'char *' to a 'void *' because that field is not accessed
Cosmetic changes only.
Prefer rtlookup() to rtalloc() when not saving the result of the look up.
The route in a syncache entry is cleared if the connection was successfully
Minimal patch that allows Path MTU discovery to be turned back on, but
Give some guidelines on when to turn on Path MTU Discovery.
Take into account the number of SACKed bytes skipped when slow-starting
Use fixed-width type to ensure correct wraparound for
Fix confusion with wrong route reference count being decremented.
Better byte packing for struct tcpopt.
Ensure that Limited Transmit always sends new data, even after a
When doing Limited Transmit, don't retract snd_nxt if it was previously
We can only do upper-layer protocol length checks on the first fragment.
Eliminate a redundant variable assignment.
Keep a hint for the last packet in the singly-linked list of packets
A kludge to always give the driver a second chance to attach the cbb device.
Use a larger initial window size when restarting after a long idle period.
Minor cosmetic cleanups. Use the canonical name "ro" for a variable
Defer assigning to the forwarding route variable until the forwarding
Prefer TAILQ_EMPTY() to null-check on the result of TAILQ_FIRST().
sbappend() is called by stream-oriented as well as record-oriented
Strip away a layer of indirection. Now that we properly declare
Cosmetic changes, mostly changing zeros to NULLs.
Clean up some of the sockbuf append code.
Implement TCP Appropriate Byte Counting.
Remove redundant assignment.
Prefer m_getcl() to separate calls to MGETHDR() and MCLGET() in order to
Fix typo with last commit.
Deprecate MCLGET() in favor of m_getcl() or m_getl() in order to
Use m_getl() to get the right sized mbuf.
Rename local variable for clarity.
Simplify the interface to m_uiomove().
Deprecate MCLGET() in favor of m_getcl() or m_getl() in order to
Prefer the clearer m_getc() API over m_getm().
Generic cache of pre-initialized objects. It uses per-cpu caches
Re-implement the mbuf allocator using the object cache.
Convert to use m_getl() in order to take advantage of cluster caching and
Allocate the right type of mbuf to begin with rather than switching types
Should have allocated a mbuf packet header to begin with.
Only allow packet headers to be copied into a packet header mbuf.
Get an mbuf packet header to begin with instead of getting an mbuf and then
Only duplicate packet headers into mbuf packet headers.
A packet header without any packet tags is still a packet header.
The header type of a mbuf doesn't change when appended onto a chain.
Preserve the target M_EXT_CLUSTER flag when duplicating a packet header.
Also preserve all the non-copied flags in the target mbuf when duplicating a
Deprecate MCLGET() in favor of m_getl() in order to take advantage
Fix typo that turns out to be harmless by accident, as MT_HEADER and
Deprecate MCLGET() in favor of m_getcl() or m_getl() in order to
Use m_gethdr() instead of m_get() to get a mbuf header.
Deprecate MCLGET() in favor of m_getcl() or m_getl() in order to
Replace the linear search in file descriptor allocation with an O(log N)
Document the standard and historical behavior that open(2) returns
The proper way to check for a normal mbuf cluster is with the
A machine-independent spinlock implementation. It has the advantages of
Add a space to the output for legibility.
Check the IP length first to avoid a memory leak later.
Fix indentation with previous commit.
Now that the C language has a "void *", use it instead of caddr_t.
Give each CPU its own taskqueue thread so per-cpu data can be
Make zalloc() and zfree() non-blocking for ZONE_INTERRUPT zones.
Prefer the general purpose m_getl() routine for mbuf allocation.
Zero out stack memory before copying out to requesting process.
Correct test for fragmented packet.
Explicitly mark places in the IPv6 code that require a contiguous buffer.
We're guaranteed m_pkthdr.fw_flags is already zero on allocation.
Combine two allocations into one.
Clear up mbuf usage statistics.
Remove unneeded assignments.
Fix missing malloc -> kmalloc conversions.
Never dereference a NULL pointer.
OpenBSD rev 1.66:
Cosmetic cleanups.
Refactor internal ip6_splithdr() API to make it more of a pure function
Eliminate a macro layer of indirection to clear up the control flow.
Cosmetic changes.
Cosmetic changes.
Remove nested block.
Fix compilation error with IPSEC.
Localize some variables.
Clean up code.
Fix userland compilation error.
Apply FreeBSD rev 1.9:
Apply FreeBSD rev 1.15:
Apply FreeBSD rev 1.16:
Apply FreeBSD rev 1.6:
Apply FreeBSD rev 1.17:
Clean up code.
Add support for SSDT tables.
Same as FreeBSD rev 1.27:
Same as FreeBSD rev 1.19:
Add some comments. Same as FreeBSD revs 1.28 through 1.30.
Just bump version numbers due to null offsetting changes from FreeBSD:
Merge FreeBSD revs 1.20 through 1.21: update list of supported tables.
Sync up to FreeBSD rev 1.23:
Add support for parsing MCFG tables.
Add split on whitespace functionality.

Jeremy C. Reed (9):
testing cvs commit
Create a manual page link of passwd.5 for new master.passwd.5.
Add master.passwd as a name of this manual page.
Only define atop for _KERNEL or _KERNEL_STRUCTURES.
Document that the caret works like the excalamation mark
Fix typo or mispelling.
Fix typo.
Comment out line about UDF specific mount options -- none were listed
Add patch from 9.3.5 to 9.3.5-P1. This is for adding randomization

Jeroen Ruigrok/asmodai (539):
Clean up two lint warnings, by adding /* NOTREACHED */ in the appropriate
Remove kerberosIV from the build.
Correct spelling of Matt's surname.
Install math.h.
Remove math.h from Makefile, it gets installed through lib/msun/Makefile.
Change __signed to signed.
Remove kerberosIV a.k.a. eBones. It has served well, but definately
Change __volatile and __const into volatile and const.
Add, hopefully, support for the ICH5 sound hardware.
First stab at our own UPDATING file.
Remove reference to FreeBSD's documentation about their source trees.
Add README for the documentation directory.
Get rid of __const.
Update the manual page to reflect the normal definitions.
Remove more __const occurences.
Add the RIPEMD-160 header file.
Fix dependency that long is always 32 bits. This does not work on (I)LP64
Remove 'register' keyword while I am here.
Get rid of __P, the days of K&R support are long past.
Make sure a DiskOnKey never sends a sync_cache. This fixes the error
Linux emulation has been working well for a while now, remove notice.
Allow NVIDIA's nForce 2 chipset to use proper ATA DMA modes.
Add OHCI support for the NVIDIA nForce 2 chipset.
Add NVIDIA's nForce2.
Reorder to alphabetical order.
Add the AMD-766, Apple KeyLargo, and SiS 5571.
Add a bunch of Intel USB controllers.
Add NVIDIA nForce3 OHCI USB support.
Add the NVIDIA nForce3 chipset.
Change 'nvidianf2' to 'nforce2' to be more explanatory.
Make our manual pages report that they are part of DragonFly.
Get rid of DDB, INVARIANTS, and INVARIANT_SUPPORT in the boot floppies.
Change # comments to /* */ equivalents.
Properly detect and print the nForce2 Host to PCI bridge and PCI-PCI
Add nForce AGP support, taken from FreeBSD with some minor changes to get
Add identifier for the nForce2 PCI to ISA bridge.
Remove the matching for the host to PCI bridge, since this is actually
Add another vendor who delivers PL 2303 derived cables/convertors.
Add -ifstat functionality to systat. It shows the network interfaces and
Add protocols 134 and 135.
Add Subversion's port number.
Add Veritas NetBackup.
Make sure to skip the PC98 architecture on the OPTi detection, since it
Properly document getenv()'s return values using a slightly altered patch.
Add __DragonFly_version, set to 100000, which basically means:
Change wrapping definition from ending in DECLARED_ to DECLARED.
Convert files from DOS to Unix.
Add FreeBSD's makeobjops.awk.
Rework the logic in the kernel config files.
Add forgotten semi-colon.
Factor out the object system from new-bus so that it can be used by
Fix misplacement of code. Due to additional DF code everything got
Add details on how to get a "checked out" source tree.
Spell initialise correctly.
Use unsigned integers for the counters, since they cannot be negative
Remove the archaic wd(4) driver and its dependencies.
Remove the archaic wd(4) driver.
Properly spell compatible and compatibility.
Use malloc() + M_ZERO instead of malloc() + bzero().
Remove haveseen_iobase(), it is not in use in the kernel.
Add proper $FreeBSD$ identifier.
Add two more awk kernel build scripts from FreeBSD.
Remove KTR damage.
Also revert the VI_UNLOCKED damage.
Change $FreeBSD$ to $DragonFly$ on the output files.
Get rid off the POSIX real-time extensions as well as the System V IPC and
Temporary hack out release.9, which creates the floppies.
Add support for the ICH 4 mobile chipset.
Get rid off FreeBSD mirrors and add our own three (in Germany, Ireland, and
Add forgotten newline in debug output.
Add PFIL_HOOKS functionality. This allows us to plug in many firewalling
Add the packet filtering files.
Comment PFIL_HOOKS since it should not be needed in GENERIC.
Add entry for the EHCI controller, pretty common these days (especially
Replace the boring beastie with a dragonfly.
Synch up with FreeBSD 5 with clean up changes.
Add support for SoundBlaster Audigy and Audigy 2.
Add PCI identifier to match against a Dell OEM version of the SB Live!
Make sure we identify our toolchain as DragonFly, not FreeBSD. This is
Make sure the vendor identifier matches the right project. The binutils
Get rid of PZERO. This was removed about the first of August.
Spell 'separate' and its siblings the way it is supposed to.
Spell 'weird' the way English expects it.
Synchronise partially with NetBSD's umodem.c v1.46:
Synchronise partially with NetBSD's v1.85:
Make sure cvs reports the proper OS.
Update for the new version of CVS.
K&R -> ANSI C conversion.
Update config.h.
Update with new source file targets.
Fix typo: CAPF_ANYCLIENT->CAPS_ANYCLIENT.
Synchronise with FreeBSD:
Add support for the following AC'97 codecs:
Get rid of question mark, the vendor ID is correct.
Fix name of 'Silicon Laboratories'.
Recognise the Asahi Kasei AK4544A and AK4545 AC'97 codecs.
Recognise the Realtek ALC850.
Add support for the Wolfson WM9711L and WM9712L.
Add support for the Wolfson WM9709.
Fix typo to make the TLC320 work.
Add Texas Instruments' TLV320AIC27 AC'97 codec.
Add support for the Conexant SmartDAA 20463.
Fix the entry for the CX20468. The base default is 28h, not 29h.
Fix misunderstood commit. The CX20463 is not the AC'97 codec. It _is_
Add missing comma.
Add recognition code for the SiS 645DX.
Add detection code for the SiS 746 ATA133 controller.
Update to latest version.
Cut our umbilical cord from mother FreeBSD.
Add missing . before An macro at the end.
Add -S flag for C99 support.
Synchronise with NetBSD: get rid of __STDC__ selective compilation.
Synchronise with NetBSD: ANSIfy.
Synchronise with NetBSD:
ANSIfy.
Add support for CS4294.
Add ATI Radeon RV280 9200.
Forced commit to, belatedly, note that this (rev 1.1) was:
Add proper entropy pool scripts and rc.conf lines, as it was.
Add a clarification comment stating that Avance Logic's IC products are
Add detection support for the Avance Logic (Realtek) ALC203 and ALC250.
Sync with FreeBSD v1.16:
Add support for the Texas Instruments IEEE 1394 controllers designated by:
Change identifier display text to include all models supported.
Add detection code for the Intel 82372FB IEEE 1394 OHCI controller.
Add support for the Adaptec AIC5800 based IEEE 1394 cards.
Detect the National Semiconductor Geode CS4210 OHCI IEEE 1394 controller.
Add the SiS "7007" OHCI IEEE 1394 controller.
Fix the vendor id string of NatSemi to its correct one.
Add detection support for the Intel ICH6 chipset.
Add support for the 82801FB UHCI/ICH6 controller.
Add Intel 82801FB EHCI/ICH6 controller.
Update per Intel 82801FB errata:
Add Intel 82801FB/FBW/FR/FRW PCI-LPC detection code.
Update string to show this device is the Hub-PCI bridge for ICH2, 3, 4, 5,
Document addition of ICH6 UHCI.
Clarify VIA Fire II identifier string.
The device in question is the Sony CXD3222, not the CX3022.
Add the Sony CXD1947, which seems to be used in some Sony Vaios.
Clarify the Sony strings.
Correct dumb copy/paste of existing line mistake to correct name.
Update comment to point to the right file (src/sys/bus/pci/pcireg.h).
Add rndcontrol.
Update the PCIS_ definitions per the PCI 2.3 specification.
Add PCIS_ definitions per PCI specification 3.0.
Correct some PCIS to PCIC and PCIP prefixes.
Update pciconf to print the recently added categories.
Add identifiers for Serial Attached SCSI per dicussion with the PCI SIG
Inspired by rev 1.33 of FreeBSD, but which was not documented:
Document PCIY_xxx.
Make unit_adjust() static as per its prototype.
Garbage collect RETAINBITS, not used anywhere.
Use static on file scoped function prototypes and variables.
Make f_stream static.
Make scanfiles() static since it is limited to the file scope.
Also mark the function itself static, not just the prototype.
Make setthetime() static per the prototype.
Garbage collect two unused variables (SaveFs, CountCopiedBytes).
Mark the variables defined at file scope static.
Mark filescope functions static.
Actually mark the function itself static as well.
Rework the WARNS levels per FreeBSD's CURRENT source.
Make usage() static.
Raise WARNS to level 6.
Correct usage of the cvsup command.
Use err() instead of a perror()/exit() combination.
Move a perror() to warn().
Use proper ANSI function definitions.
Fix two sign comparison mistakes.
Make the second argument of interval() const, since it is not changed
Initialise to NULL to silence gcc.
Add WARNS and set to level 3.
Hook c99 up to the build for usr.bin.
Bump version number for the 1.1-CURRENT tree.
Remove stray fr_checkp() declaration.
Use stronger wording against using 'register' and '__P()'.
Get rid of varargs.h.
Get rid of varargs.h.
Get rid of the CFLAGS with traditional-cpp, it compiles fine without.
Get rid of varargs.h.
Add Pentium 4 Thermal Control Circuit support.
Get rid of varargs.h.
Actually add the main file for Pentium 4 Thermal Control Circuit support.
Get rid of varargs.h.
Update mk files list with the current supplied one.
Add the latest source versions of OpenBSD's traceroute program (which came
Add proper prototypes for dump_packet() and pr_type().
Use proper ANSI prototypes and make sure all arguments are supplied in the
Use proper EXIT_FAILURE/EXIT_SUCCESS for exit() calls.
Make all the local functions static.
Split off code in print_all_info() into print_batt_life(), print_batt_stat()
Remove question mark (?) case statement.
Fix function definition {} placement according to our own style.
Add the long overdue ehci(4) manual page.
Update to reflect DragonFly reality.
Add a temporary hack to avoid the local files to be picked up.
Add a commented out STRIP variable to show people how to make sure installed
Clarify the 's' and 'msg' arguments and note how send() and sendto()'s msg
Remove a redundant call to gettimeofday().
Rework the wording in a different way. For send() and sendto() change
Change rcvar from "mixer" to `set_rcvar`.
Add updated Lithuanian locale (lt_LT-ISO8859-13).
Update zoneinfo database with the latest information.
Correct typo: mv -> nv.
Get rid of COPY, its functionality has been superseded by install -C.
Add first stab at Faroese locale.
Get rid of the alpha entries.
Forced commit to note that site_perl/5.005/i386-freebsd is now:
Merge from vendor branch OPENSSL:
Add OpenSSL 0.9.7d.
Use proper filenames for the Faroese locales.
Add missing entries for fo_FO, lt_LT, Argentina.
-1, not -15.
Add 136: UDPLite [UDP for error prone networks]
Merge from vendor branch OPENSSL:
Add OpenSSL 0.9.7d.
MLINK sata(4) to ata(4).
First stab at getting the Silicon Image (SiI) SATA controllers 3112 and
Synchronize with our current code.
Fix it properly.
Switch from OpenSSL 0.9.7a to OpenSSL 0.9.7d.
Commit manual pages after running 'man-update' and add new manual pages.
Update per latest manual pages.
Update per latest manual pages after running 'man-update'.
Update per latest manual pages after 'man-update'.
Change hackers@freebsd.org to kernel@crater.dragonflybsd.org.
Add comment markers to avoid the same stupid mistake as I made.
Get rid off the host.conf to nsswitch.conf conversion.
Update per recent newsletters of the ISO3166 committee.
Belatedly remember that in Nordic languages the aa, ae, and similar letters
From NetBSD 1.11:
Merge from vendor branch BIND:
Add BIND 9.2.4rc7.
Remove a '+', remnant of patchset.
Switch to 9.2.4rc7.
Fix typo of->pf.
Unbreak addump().
Synchronise with FreeBSD-CURRENT as of 2004-09-26.
Fix spammage introduced by dillon's commit in r1.5.
Add WARNS, set to 3.
Make WARNS ?= instead of = per all the other Makefiles.
Bump WARNS to 6.
Add WARNS, set to 6.
Add WARNS and set to 6.
Add WARNS and set to 3.
Add, if not already present, WARNS and set to 6.
Set WARNS to 6.
Add WARNS and set to 5.
Set NCURSES_CONST to const.
Make ls compile under WARNS 6.
Bump hostname to WARNS 6 by initialising silen.
Add, if not present, WARNS and set to 6.
Bump WARNS to 6.
Add WARNS, set to 0.
Bump WARNS to 6.
Bump WARNS to 6.
Regenerate.
Fix author name with .An macro. Add missing period.
Bump to WARNS 6.
Bump WARNS to 6.
Forced commit:
Add Wacom Graphire 3.
Regenerate.
Bump to WARNS 6.
Be consistent in the white space usage.
Add some more vendors and some HP devices.
Regenerate.
Change SYNOPSYS to SYNOPSYS2 and regenerate.
Change to use USB_PRODUCT_LINKSYS2_USB200M.
Add the Unicode 3.2 ctypes table.
Prepare for the locale additions.
Add en_GB.UTF-8.
Add UTF-8 time definitions.
Be a coward and add ja_JP.EUC back in for now.
Add la_LN.UTF-8.
Add collation definition for UTF-8.
Fix consistent mistake: CP1252 -> CP1251.
Add zh_CN.GBK.
Add Hye-Shik's UTF monetary defitions.
Add Hye-Shik's UTF-8 message definitions.
Incorporate Hye-Shik's work for numeric UTF-8 definitions.
Correct BASE_LOCALEDIR to ..
Correct the Polish locales per the submission of Bodek <bodek@blurp.org>
Add sr_YU.ISO8859-2/5.
Correct non-matching comment.
Add definitions for _CTYPE_SW[0-4MS].
Add more locales:
Use rune.h instead of runetype.h.
Add omitted space after 'while'.
Add NetBSD 2.0 and DragonFly 1.1.
Add iso-C alias and p1003.1-2004 definition.
Document ENOMEM error case, note how strdup() is 1003:2004 sanctioned.
Remove leftover 'is'.
Merge FreeBSD SA 04:16 security fix.
Add IDs for the following:
Regenerate.
Add more detail to the 845, 865, and 915 family.
Regenerate.
Major cleanup and expansion of the NVIDIA id section.
Regenerate.
Merge from vendor branch OPENSSL:
Add OpenSSL 0.9.7e.
Add some nForce2 identifiers.
Regenerate.
Remove obsolete/unused file.
Merge from vendor branch FILE:
Add file 4.12.
First stab at file's libmagic.
Use the more commonly used SRCDIR instead of SOURCEDIR, does not seem to
Add libmagic and reorder/restructure the list of the libraries to be build.
Add LIBMAGIC.
Use a space instead of a tab.
Use spaces after the CFLAGS assignment as well.
Simplify file to be a binary linking to libmagic (which contains the real
Move the Magdir handling from usr.bin/file to here and simplify.
Add libmagic to _prebuild_libs
Revert from previous commit, seems to have to be solved elsewhere.
Merge from vendor branch BINUTILS:
Add binutils 2.15.
Switch from OpenSSL 0.9.7d to 0.9.7e.
Regenerate the manual pages after the OpenSSL update to 0.9.7e.
Move from K&R function declaration to ANSI.
Let the allocation of registers be done by compilers nowadays. The average
Merge from vendor branch CVS:
Add CVS 1.12.11.
Add missing backslash.
Remove file from build_tools, since it serves no build tool purpose.
Add binutils 2.15 directories.
First stab at bmake glue for binutils 2.15.
Match GCC's configured target.
Regenerate manual pages.
Add ICH5 10/100 Mbit interface id.
Get rid of unused hardware targets.
Retire Alpha bits.
Get rid of hardware architectures we do not use.
Add abi.h which defines particulars for x86 and AMD64.
Make sure to include header files from -I${ARCH} directories, in this case
Replace the files with the NetBSD ones, which are the rewritten ones by
libm hasn't been used in ages. msun has taken over its place for a long
Add more files to the architecture specific section.
Use ANSI C and get rid of the __STDC__ and other wrapping.
Prefix -I's path addition with ${.CURDIR}.
Get rid of an Alpha specific part.
Prefix function names with __generic_ as per the rest, to remove the
Add NetBSD's ieee754.h header file.
Include sys/types.h for now.
Add ieee.h files for amd64 and x86.
Add lrint() and associated functions.
Add matching line for ATI Radeon RV280 9200SE.
Get rid of the register keyword.
Add icmp6 alias.
Update comments for the Belarussian locale.
GCC supports two pseudo variables to get the function name, __FUNCTION__
Get rid of the #define mess for other platforms. We're just Unix.
Get rid off conditionals for hpux, AIX, THINKC, TURBOC, MS_DOS, VMS.
Get rid of the Alpha specific manual pages.
Do not create Perl 5.00503 directories anymore since it is removed from base.
Get rid of [cat|man]1aout.
Remove [cat|man]1aout directories.
Fix HISTORY by using a proper .Dx macro.
Fix extraneous spacing of .Dx by changing two unneeded tabs to spaces.
Add the additional space to make mount_udf.8's .Dx macro work for real.
Get rid off the PC98 support.
Use .Dx macro.
Get rid off PC98 conditional code.
Get rid off PC98 conditional code.
Get rid off Alpha mentions.
Add va_copy() implementation (thanks to Chris Torek's comp.lang.c post).
Replace Turtle references.
Merge from vendor branch CVS:
Add CVS 1.12.12.
Change EXIT STATUS to DIAGNOSTICS. We have standardised on the latter.
Add gperf 3.0.1.
Merge from vendor branch GPERF:
Update to 3.0.1.
Synch our GNU_PREREQ() macro with FreeBSD:
__pure__ is supported from 2.96 onward, not 3.0.
Get rid off the wrappers around __va_copy(), they serve no real purpose.
Remove dllockinit.3 from the Makefile.
Back out getloadavg()'s change from int to size_t. This breaks gmake, for
Detail thread-safety conformance.
ANSIfy.
Add fsblkcnt_t and fsfilcnt_t.
Add blkcnt_t and blksize_t per IEEE Std 1003.1, 2004 Edition.
Add id_t, a general identifier type, per IEEE Std 1003.1, 2004 Edition.
Add first stab at a statvfs.h.
Sync to FreeBSD 1.14/1.15:
Fix last two return() calls to comply to style.
Bump FreeBSD identifier to 1.16 to signal to which version we synchronised.
Document thread-safety.
Add include file protection wrapper.
Add NO_PKGTOOLS to disable building of the pkg_* tools during world.
Remove NO_PKGTOOLS wrapper, it existed in top-level Makefile already.
Remove extraneous closing brace.
.Fx -> .Dx.
Fix two installworld mtree warnings.
Rename the variable PROG to LDR to remove a warning message about
Add SHF_TLS and STT_TLS to complete the ELF ABI for TLS.
Expand e_type with OS and processor-specific ranges.
Seriously expand e_machine.
Expand e_machine per ELF ABI of Dec 2003.
EM_ALPHA has been assigned number 41 nowadays, reflect this fact.
Demarcate e_machine reserved ranges.
Retire EM_486.
Add ELFOSABI for OpenVMS, HP Non-Stop Kernel, and Amiga Research OS.
Add SHN_LOOS, SHN_HIOS and SHN_XINDEX.
Add SHT_INIT_ARRAY, SHT_FINI_ARRAY, SHT_PREINIT_ARRAY, SHT_GROUP and
Add SHF_MERGE, SHF_STRINGS, SHF_INFO_LINK, SHF_LINK_ORDER,
Add section group flags (GRP_*).
Add STB_LOOS, STB_HIOS, STT_COMMON, STT_LOOS, STT_HIOS, STV_DEFAULT,
Add PF_MASKOS and PF_MASKPROC and realign the comment section.
Add NetBSD's nls(7).
Reorder alphabetically.
Remove part about wscons. This will need information about our console
Get rid off extraneous spaces in function prototypes.
Add sshd example, might be useful for jails.
Add manual page.
Add DIAGNOSTICS.
Add EXAMPLES.
Actually add a copyright name.
Add SUSv3 information.
Document the fact we do not conform to 1003.1-2004/SUSv3.
Document that we do not conform to SUSv3.
Document that dirname does conform to SUSv3.
Remove part about 89 and 1989, we live in the 21st century now.
Add DIAGNOSTICS.
Use consistent wording, blank has been changed to empty (which is POSIX
Document SUSv3 conformity.
Use POSIX wording of the -R option, the previous one was really unclear
Add -f to the non-standard list.
Pull -h to its own synopsis line to avoid possible confusion of it being
Take the -h use case separate to make it clearer.
Remove trailing /.
Synchronise with NetBSD v1.18:
Add vendor ids for ATi and Philips.
Update with the NetBSD code (which can include FreeBSD/OpenBSD changes):
Update FreeBSD tag to what the source code has.
Sync with NetBSD:
Place && at the right place.
Synchronise with NetBSD (some come from OpenBSD):
Synchronise with NetBSD:
Clean up manual page.
Add DragonFly release numbers.
Fix PROCDURE -> PROCEDURE.
Merge from vendor branch BIND:
Merge from vendor branch BINUTILS:
Merge from vendor branch CVS:
Fix PROCDURE -> PROCEDURE.
Merge from vendor branch GCC:
Merge from vendor branch NCURSES:
Fix PROCDURE -> PROCEDURE.
Fix PROCDURE -> PROCEDURE.
Fix PROCDURE -> PROCEDURE.
Fix PROCDURE -> PROCEDURE.
Fix PROCDURE -> PROCEDURE.
Use FreeBSD's HEAD tag.
Merge from vendor branch TEXINFO:
Add texinfo 4.8, appropriately stripped down.
Switch to texinfo 4.8, which is needed for a lot of new texi files.
Commit a missed change.
fdl.texi is needed.
Merge from vendor branch TEXINFO:
Merge from vendor branch GROFF:
Add groff 1.19.1, stripped down appropriately.
Merge from vendor branch GROFF:
Add groff 1.19.1, stripped down appropriately.
Merge from vendor branch GROFF:
Update and reorder.
Remove mdoc.local, we need to make this truly local.
Merge from vendor branch GROFF:
Remove mdoc.local, we need to make this truly local.
Taken from FreeBSD-HEAD:
Retire old perl 5.00503 optional manpath and replace with a manpath to
Update to groff 1.19.1.
Get rid of the old texinfo.
Get rid of the old groff.
Add patch to add Dx macro definition to doc-syms.
Use groff 1.19.1 supplied file.
Culminate all our local changes into one file:
Remove these files, we augment with mdoc.local what groff delivers us
Fix for the error
Get rid of xditview, since by default we do not have X installed, it makes
Update to groff 1.19.2.
Merge from vendor branch GROFF:
Update to groff 1.19.2.
Update to groff 1.19.2.
Add pdfmark to the build.
Unhook cvsbug from the build.
Get rid of the Makefile.
No need to prototype main().
Synchronise with FreeBSD:
Move the termwidth declaration over to extern.h, where is make more sense
Get rid of the third clause from the UCB license.
Make len size_t, since strlen() returns a size_t, case i to size_t later
Use warnx() instead of home-rolled fprintf() constructions.
ls.1: Add y to SYNOPSIS line.
Update FreeBSD Id.
Synchronise with FreeBSD:
Make sure err() does not use a NULL format string, this way we at least
Synchronise with FreeBSD v1.65:
Synchronise with FreeBSD:
Fix long standing logic bug in basename() introduced in 1.4.
Update FreeBSD Id and synchronise with FreeBSD:
Move the Ids down to where they should be.
Fix one last nit: compare the result of strlcpy() only to see if it is
Rework dirname() with the same logic Joerg introduced for basename():
Restore the MAXPATHLEN comparison in basename.c, don't forget that strlcpy()
Welcome to 2005:
Partial synch with FreeBSD v1.74:
Add inttypes.h for intmax_t.
Partial synch with FreeBSD 1.74:
Bump FreeBSD Id.
Synchronise with FreeBSD:
Synchronise with FreeBSD:
Synchronise with FreeBSD up to and including v1.86.
Synchronise with NetBSD:
Reflect type change and remove third clause.
Get rid of the third clause.
Fix broken comment.
Get rid of the third clause where we can.
Bump OpenBSD Id.
Synchronise with v1.73:
Synchronise with FreeBSD:
Fix strmode()'s parameters to reflect the reality it had been in for a long
Pull in sys/types.h if it has not been parsed yet.
Actually add a manual page for xe(4).
Actually hook up ipw.
Document and order the wlan devices.
Fix the function declaration.
Enable wide character support in ncurses, since we have it, better make
Fix accidental reversal of assignment for suffix/suffixlen.
Revert last commit for two reasons:
Add the wide character files for ncurses.
Add usr.bin/stat to bootstrap-tools:
Use echo instead of ls to test for files, change test logic slightly for
Replace use of ls with echo and tr.
Get rid of ls and use basic sh/echo constructs.
Remove bin/ls from the bootstrap-tools.
getchar() is void in its prototype.

Joe Talbott (11):
* testing
Make vkernel compile with 'options SMP'. Most functions are stubs that
Fix files that included the posix scheduling headers that were merged earlier.
Add break after parsing the -n option.
Add usched_set() manpage.
Add support to vkernel for locking virtual CPUs to real CPUs.
Let the user know if they attempt to start a vkernel with vm.vkernel_enable
Make mbuf allocator statistics SMP safe.
Fix conditional so that the linux module is loaded.
Add support for Cisco-Linksys WUSB54GC which has a different vendor ID
Test commit

Joerg Sonnenberger (2002):
Move the FreeBSD 2.2 and 3.x PCI compatibility code into pci_compat.c and let it
Add/uncondionalize the sixt argument of set_memory_offset to ease the migration
Add card_if.h dependency
Split off the PCCARD specific driver parts of fd and sio and remove last
Add hw.firewire.sbp.tags to control tagging for SBP devices. The default
Add PCIBUS to nexus
Add rman_get_device and rman_get_size, use macros in nexus.c
Make hw.firewire.sbp.tags tunable
Add black hole device for system PnP IDs
Fix indentation to tabs, no functional changes
Drop chip driver and merge the functionality into pci_probe_nomatch.
Sync pci_cfgreg.c with FreeBSD 5, rev. 1.101. This makes the PCI interrupt
Fix linker issues with /usr/libexec/elf/ld not using rtld's search path
Add pcib_if.m
Add pcib interface methods.
Remove HOSE support which should be implemented in the bus function when
Use nexus_pcib_read_config instead of pci_cfgread. Do some cleanup.
Add comment for nexus_pcib_write_config.
Replace pci_cfgread and pci_cfgwrite with PCIB_READ_CONFIG and
Fix compile errors introduced with last commit
Fix PCI deadlock on boot
Add DragonFly specific headers. Add __DragonFly__ and correct the specs.
This is Fred, not Beastie
Add a dummy dependency for each target make(1) is called with when
Add lib/gcc3/csu
Fix tconfig.h dependency
Add lib/gcc3/libgcc and lib/gcc3/libgcc_r
Import libstdc++ from GCC 3.3.3-pre 20031106.
Merge from vendor branch LIBSTDC++:
Add directories for lib/gcc3/libstdc++
Conditionalize use of vfscanf
Implement __cxa_atexit and __cxa_finalize as requested by the cross-platform
Add C++ runtime libraries for gcc3
Fix missing .. from last commit
Rename GCCVER to CCVER and prepend gcc to the former values
Clean multiline string literal for gcc3
GCC 3.3 doesn't support #pragma weak A=B, when A is defined. This patch
Fix multiline string literals
Fix more multiline string literals
Fix compilation with GCC 3.3 and a few warnings
Fix GCC 3.3 compilation and style(9)
Include ${.CURDIR} before bin/ls to get the right extern.h
Use C99 syntax for variadic macros
Add -I{.CURDIR} to CFLAGS
Add missing \ to string
Add -I${.CURDIR} to Makefile
Remove bogus sys_nerr defines
Fix prehistoric C
Fix prehistoric C
Remove some prehistoric C crap
Add support for ${BINUTILSVER} to objformat. This defaults to gcc2
Adjust program names by removing the additional 3
Remove generated file from repo
Force make to export CCVER for buildworld
Both cc and cc3 were built using CCVER=$(CCVER) for certain compiler path.
GCC3 doesn't inspect #line to allow inclusion relative to the source
Add -I${BIND_DIR}/bin/nslookup
Fix prehistoric C
The __cxa_atext/__cxa_finalize patch is based on FreeBSD OR bin/59552
Remove bogus -fstrict-prototypes which isn't support for gcc3.
Use new style variadic functions
Add GCC3 to buildworld.
Add -ffreestanding to CFLAGS for gcc3
Split multiline string literal
Don't use non-existing DragonFly override ports for ports using
Cleanup the fix for dependency handling. This explicitly checks wether
Replace K&R style function declarations with ANSI style one.
Rename bsd.cpu.mk into bsd.cpu.gcc2.mk and add the proper Makefile magic
Always re-export CCVER
Fix make complaining about CCVER being not defined
Always include bsd.init.mk to fix CPUTYPE evaluation
Remove C++ from libc_r. Use __attribute__((constructor)) instead.
Really remove C++ from libc_r
To let the crazy Perl Makefiles work, add the support for ../Makefile.sub
Rename Makefile.inc to Makefile.sub
Rename Makefile.in to Makefile.sub
Remove Makefile.inc
Rename Makefile.inc to Makefile.sub
Add support for the AMD 8111 chipset
Change functin definitions to conform to style(9)
Fix various warnings and other glitches.
Move binutils from /usr/libexec/gcc2/{aout,elf} to
Add getopt_long from NetBSD
Build lwkt_process_ipiq_frame only for the kernel, since
Fix various warnings.
Update style(9) to reflect current code practise.
Sync DragonFly and FreeBSD-current's FireWire driver.
Add missing sbp.h to complete the firewire sync
Second part of the firewire sync. Add defined(__DragonFly__) or
Fix bug when doing backquote expansion.
Allows option to be specified on the command line when mount with -a.
De-K&R-ify source, remove register keywords.
Adjust infrastructure for NEWCARD
Add a tunable hw.pci_disable_bios_route to work around broken PCI-BIOSes.
Add lost -D__FreeBSD__
Initial backport of NEWCARD from FreeBSD 5.
Allow choosing different GCC versions for buildworld and buildkernel as
Fix gcc3 compilation
Add defined(__FreeBSD__) and defined(__DragonFly__) where appropiriate.
Add defined(__FreeBSD__) and defined(__DragonFly__) where appropriate
Fix gcc3 compilation
Sync with FreeBSD's pccarddevs
__FreeBSD__ to __DragonFly__
Probe via CIS lookup
Add accessor for CIS4 and change some functions to take const char* arguments
Add __DragonFly__
Add __DragonFly__
Always include net/bpf.h
Add __DragonFly__
Add __DragonFly__
Add __DragonFly__
Add __DragonFly__
Add __DragonFly__
Add __DragonFly__
Add __DragonFly__
Add __DragonFly__
Add __DragonFly__
Fix wrong conditional from last commit
Add __DragonFly__
Fix broken string literals
Fix # style comment in file using the C prepocessor
Add __DragonFly__
Add __DragonFly__
Sync if_ed with FreeBSD current
Fix warnings about casting const pointers.
Add device_is_attached to allow a driver to check wether a given device
Add PCIR_BAR and PCIR_BARS for FreeBSD 5 compatibility
Add pci_get_ether and pci_set_ether for FreeBSD 5 compatibility
Add BUS_DMA_ZERO flag to bus_dmamem_alloc.
Install getopt_long.3
pmap_zero_page expects an address, not a page number
Fix typo.
Add proper match routines for PCCARD nics.
Remove duplicate line for if_ray
Add PCCARD match routines sio and the disk drivers.
Add PCCARD match function to ata. Code taken from FreeBSD.
Remove some unneeded #include
PCCARD has a central device database in bus/pccard/pccarddevs, add one
Add generated pcidevs files. Fix a small typo in devlist2h.awk.
Add supfile to fetch only dfports
INTR_TYPE_AV is used by FreeBSD 5 code and was defined to INTR_TYPE_TTY
Fix compilation with -fno-common
Certain port Makefiles expect variables like ARCH or HAVE_GNOME to be set
Add missing make_dev
DFports cleanup part(1)
Split off the PCI-PCI bridge and the PCI-ISA bridge code from
Add support for the kernel printf conversion specifiers %b, %D and %ry.
Use -fformat-extensions when building kernel with GCC2.
Hide ISA compatibility layer under COMPAT_OLDISA
Use ovbcopy instead of bcopy to match prototyp
Fix some warnings
Fix a typo and include <sys/random.h>
Remove unused static variable
Add prototype for bootpc_init
Remove unused static declarations
Fix format string
Fix argument order for snprintf, the size is the second argument
Fix format string
Fix spurious warning about ANSI trigraphs
Conditionalize filll_io and filll, they are only used with VGA_NO_MODE_CHANGE
#ifdef0 pst_shutdown, it is not used
edquota(8) should honour MAXLOGNAME.
Remove mixerctl script from NetBSD and add a replacing mixer script.
Style(9) cleanup. Remove K&R style prototyps and use __BEGIN_DECLS/i
Fix use after free / double free bugs. Return an error in error conditions.
Adjust mixer script to depend on mixer_enable="YES" and default to NO.
Cleanup emujoy_pci_probe
Don't use parameter names for kernel prototyps
Remove parameter names.
Remove the entry for pccard and allow src/include/Makefile to properly
Remove parameter names, adjust white spaces in prototyps and remove
Remove unused and undocumented strhash files.
Add missing return from last commit
Use ifp->xname instead of if_name(ifp)
There are historically two families of fixed size integers, u_intX_t and
Add bfe(4) support from FreeBSD.
De-K&R-ify function prototyps and remove register keyword.
Merge FreeBSD rev. 1.8:
Adjust indentation, use uint32_t and line up comments.
Make subr_bus.c more consistent with regard to style(9) and itself.
Use M_WAITOK instead of M_WAIT to get memory.
We have /etc/rc.subr, don't add the dependency
The slab allocator has been in for while now. Change the printf for invalid
Add patch infrastructure for contrib/ and similiar directories.
Propolice for GCC 3.3 based on:
From FreeBSD:
Fix panic in acd_report_key when ai->format==DVD_INVALIDATE_AGID.
Handle failure in atapi_queue_cmd correctly
While converting ATA to use MPIPE allocations, ata_dmaalloc was changed
Remove a debug printf added with the last commit.
Use local cpu tcbinfo
Initialize the interface name for if_de
Initialize all fields in MALLOC_DEFINE and VFS_SET to fix warnings.
Add a short cut for DVD_INVALIDATE_AGID to simplify the rest
Think before commit and remove some more cruft
The free(9) implementation based on the slab allocator doesn't handle
Merge the kernel part of UDF support from FreeBSD 5.
Add userland UDF support based on mount_cd9660
Merge from FreeBSD 5:
Add convient functions for the bus interface: child_present,
Add BPF_TAP and BPF_MTAB macros from FreeBSD
Fix warning about missing prototyp for psignal
Remove macro definitions for BPF_MTAP
Remove unused BSDI I4B interface
BPF has been in the kernel for ages and is supported by all NICs but I4B.
Remove BPF_MTAP definition
Some drivers depend on the link layer address in ac_enaddr.
Add a new function ether_ifattach_bpf which can be used by NICs not using
Merge FreeBSD's rev. 1.81:
Add support for building dependent modules automatically by
Cleanup sis(4):
Add default case of error = EINVAL to ether_ioctl
Add device IDs of BCM5788, BCM5901 and BCM5901A2. Fix some spelling mistakes.
Change to PCI_VENDOR_DELL
Replace the Perl scripts makewhatis(1), makewhatis.local(8) and catman(1)
Add functionality to binutils 2.14's ld to scan /var/run/ld-elf.so.hints
kern_sysctl.c
Remove the old locking based on memory flags by lockmgr based code.
Revert last commit. This should not have happened.
Add SI_SUB_LOCK as sysinit priority for the initialisation of tokens and
M_NOWAIT => M_INTWAIT conversion. This subsystems are way too crucial to
Remove unused obsolete drivers.
Fix warning about unused variable
Add the "struct ucred *" argument to the remaining nic ioctls in LINT.
KObj extension stage I/III
KObj extension stage II/III
Convert sis(4) from vtophys to busdma.
Allocate the DMA segment array in bus_dma_tag_create instead of using a
Correct C++ header handling for gcc2 and lex.
Readd _G_config.h and the missing std headers. This brings C++ back to where
Since GCC 2.95.4 is known to produce bad code for higher optimization
Adjust the C++ preprocessor to include /usr/include/c++ by default for
Add support for AC'97 codec of the AMD-8111 chipset.
This is _SYS_XIO_H, not _SYS_UIO_H.
nawk => awk
Explicitly build the boot2.c without propolice. We can't really handle
Remove unit from sis_softc and use device_printf and if_printf instead of
KObj extension stage IIIa/III
Don't print the recording sources to stderr, the manpage doesn't indicate
Do some style(9) cleanups and make add static.
Fix output format of "mixer -s", it is supposed to be =rec rdev1 rdev2 ...
Use "mixer -s" for saving the mixer settings and adjust messages since we
KObj extension stage IIIb/III
Move IFF_PROMISC and IFF_POLLING from ifnet.ipending to ifnet.if_flags,
Nuke unused fields in struct ifnet, if_done and if_poll_* hasn't been used
for() ==> TAILQ_FOREACH
Merge changes from FreeBSD 5:
In contrast to FreeBSD 4 and 5, our slab allocator does hand out cross-page
m_tag_alloc illegally passed the mbuf flags to malloc, hitting the
Partial sync with kernel to get libcaps compilable again.
Fix bsd.port.subdir.mk by adding the normal environment hacks
Small style fix
Sync libcr with libc.
Conditionalize accept_filter variable on defined(INET).
Move the Plug'n'Play BIOS support into a separate file. This is included
POSIX lock resource limit part 3/4
Fix races in lf_getlock and lf_clearlock when waiting for memory.
Serialize access to lockf via pool tokens.
Switch to the callout interface and rename the associated entry to sis_timer,
Add ifmedia infrastructure for the generic IEEE 802.11 support.
Two more defines from FreeBSD.
Fix panic due to the way change_ruid handles the userinfo.
Merge rev 1.10 from FreeBSD:
Fix two bugs in the lockf code. The first one is a missing reinitialization
Sync em(4) with FreeBSD current. Most important is the initial bus DMA support.
Add the lockf regression test from NetBSD, slightly modified to test
- remove em_adapter_list, it was not used for anything beside adding and
Readd em_read_reg_io and em_write_reg_io for workarounds in various
Add PCI IDs for i865 agpgart support.
Temporary switch cc3 to the old stabs debugging format to unbreak gdb5.
Update bktr(4) to FreeBSD current's version. This most importantly
Add dev/video/bktr and dev/video/meteor to the header file list.
Use struct thread for kernel threads, not struct proc.
Explicitly cast away the volatile for conversions and argument passings.
Explicitly cast-away volatile since it should be save here.
Initialize the magic cookie using real numbers instead of a multi-character
Fix some const warnings
The const call is linted, use proper cast to silence GCC
Use __DECONST to silence GCC. Make some private routines static.
Use volatile and __DEVOLATILE to silence gcc warnings.
set_lapic_isrloc depends on APIC_IO for the prototype to exist, it isn't
Improve the way error message from ALART are printed.
Include ns.h to get prototyp for ns_cksum
Hide unused function under #ifdef SMP
Hide unused functions to silence GCC
Don't cast away the const before dereferencing.
Move extern declaration to file scope to fix warning
Change pr_output's signature to take two fixed arguments and possible
Make pr_domain and pr_usrreqs pointers to const. The general stack is not
Make pr_input use variadic arguments for anything but the first mbuf.
Continue cleaning em(4).
Fix a small bug in the last commit. ether_ifdetach has to be called
em(4) assumes that bus_dmamap_destroy of bus_dmamap_load_mbuf maps
Begin implementing a -liberty replacement for binutils and GCC under
Fix the warranty, this is not UCB code
Add implemenation of lrealpath. This works like realpath, but returns
Add Makefile support for host programs (.nx) just like we support normal
Import BSD-licensed crtbegin/crtend support.
Switch from GCC-version specific crtbegin/crtend code to the version
Add the old CSU files to list of file to be deleted and keep the new ones.
Build infrastructure for GCC 3.4
Add directory entries for GCC 3.4.
Add CCVER=gcc34 support to bsd.cpu.mk. Also add the magic for AMD64 support
Nuke lib/csu/i386 (a.out support) and copy lib/csu/i386-elf there.
Really use the host compiler in bsd.hostprog.mk
Fix warning
Don't whine about malloc/realloc "poising" the code for YACC files.
The insn-conditions.c generated by stock GCC does some bad premature
Include ProPolice suport for GCC 3.4.
Include ProPolice suport for GCC 3.4.
First stage in cleaning the built-in pathes of gcc. Adapted patch from
Remove the evil inline ==> rpcgen_inline CPP hack, rename the variable
Fix GCC 3.4 build
Fix compilation "Label at end of compound statement" and some missing
Fix GCC 3.4 build.
Fix GCC 3.4 build
Redo the ProPolice patches, there were partly broken.
Manually recurse into ../cc_prepend for depend to workaround
Use our specs, not FreeBSD's.
Always set the _CPUCFLAGS in bsd.cpu.gcc2.mk to some known, safe
Define __DragonFly_cc_version for CPP
Don't include _CPUCFLAGS since the host compiler (aka NXCC) might not
Export HOST_CCVER via environment to fixate it to either the specified
Explicitly recurse into gnu/usr.bin/cc34/cc_prep for depend to ensure
For cc_tools, recurse into cc_prep when building dependencies to get
Add the STABS default output hack for GCC 3.4 too.
make => ${MAKE}
make ==> ${MAKE} ${.TARGET}
Merge from vendor branch GCC:
This is GCC 3.4, not 3.3
Add missing WI_UNLOCK
Fix a race in the initialisation of struct lock by moving the TAILQ_INITs
Print the correct list in _lf_print_lock
procfs_validfile does have a public prototyp, but doesn't seemed to be used.
Remove cast as lvalue
Remove cast as lvalue
Use const char * for string argument of _assert_sbuf_integrity and _assert_sbuf_state
Remove invalid tokens after #endif
Add a default initializer for data_sds. The warning from GCC is not correct,
Remove invalid tokens after #endif
Add the support for BSD format specifiers. This was adopted from the
Announce MAC address in ether_ifattach, not in each NIC indepently.
Add common functions for computing the Ethernet CRC on arbitrary length
Add ETHER_ALIGN for portability.
Include the header files from the current source tree and not from /usr/src.
Add IFCAP_POLLING for per-interface polling support.
Add re(4) as kernel module. After some feedback, this will be added to the
Add support for setting polling on a per-interface base.
Add re(4) to the list of manpages. Update the list of NICs supporting
Welcome BPF in the 21st century and remove all the pre-ANSI C, BSD < 1991
Add per-device polling support.
Add PDEBUG call for device_shutdown.
In lf_wakeup, once we got a range embedded in the unlocked range,
Stop using if_dname, use if_printf or ifp->if_xname instead.
Don't init sc->re_timer twice.
Fix grammatik error.
Don't return 0 from rl_probe, because rl(4) is not a best match for
Add if_broadcastaddr to struct ifnet to hold the link layer broadcast address.
Add struct ucred * argument to ng_fec_ioctl
Add llc_snap shortcut.
Add handling of patches to the module framework.
Remove unused variable _PATCHES.
Unify the input handling of the low-level network stack by introducing
Sync with FreeBSD 5-CURRENT.
Regenerate.
Add RC4 to the crypto module / device. This will be used by the generic
Instead of casting the function, cast the argument to the (correct) type.
Comment out extra token at end of #endif.
Remove usage of NTOHS / NTOHL / HTONS / HTONL.
Last commit changed a NTOHL to ntohs, correct this.
Extend the patch framework to handle non-compilable files. E.g. for
Add macro to test for broadcast / multicast Ethernet addresses.
Don't cast lvalues.
Don't use cast as lvalues.
Import generic 802.11 layer.
Add 802.11 include directory
Add 802.11 include directory
- turn a strcpy into a strlcpy to avoid overflow
Make raycontrol(8) WARNS=6 safe by adding const and fixing a signed/unsigned
Use netproto/802_11 includes, instead of net/if_ieee80211.h
Use netproto/802_11 includes instead of net/if_ieee80211.h.
Use netproto/802_11 includes instead of net/if_ieee80211.h
Use netproto/802_11 includes instead of net/if_ieee80211.h and
Use netproto/802_11 includes instead of net/if_ieee80211.h.
Refer to netproto/802_11/ieee80211{.h,_ioctl.h} instead of net/if_ieee80211.h.
Remove now obsolete header.
Sync with FreeBSD CURRENT.
Add two more 802.11 media types.
Add IF_QLEN and the ALTQ macros. This are only the lock-free versions with
NTOHL(x) ==> x = ntohl(x)
NTOHL / HTONL removal.
Sync with FreeBSD CURRENT (most white space cleanup and ordering).
Change (almost) all references to tqh_first and tqe_next and tqe_prev
Forced commit to annotate the (unrelated) changes from the last commit.
Release to correct ressource in re_detach, this is PCI_LOIO now.
Add re(4) to GENERIC.
Add re(4) to LINT as well.
Yet another hack for x11/xorg-clients.
The EISA attachment of vx does have a softc, export it's size correctly.
Add LIST_FOREACH_MUTABLE which works like TAILQ_FOREACH_MUTABLE.
Add a description for LIST_FOREACH_MUTABLE and TAILQ_FOREACH_MUTABLE.
Add MODULE_VERSION(pci, 1), e.g. agp(4) can't be loaded as module otherwise.
IOCTL mapping layer Part I/II
- remove '?' from getopt switch case
Add DFOSVERSION for ports to check for DragonFly and allow them to handle
- remove prototype for main
IOCTL mapping layer Part II/II
Make the addr parameter to kernacc and useracc const.
- WARNS ?= 6 clean
Add strotonum(3) into the !ANSI_SOURCE && !_POSIX_SOURCE && !__STRICT_ANSI
Minor style changes.
- use const for file names, static for local functions
Fix the fprintf statement for overlong domainnames.
Make libftpio WARNS=6 clean.
More constify.
Fix various buffer overflows.
In cmd(), after the vsnprintf is a strcat done to append a newline.
Make this WARNS?=6 clean by explicitly using __DECONST for the write
Add implemenation of splay tree and red-black tree.
Use sys/param.h instead of sys/types.h for endian macros.
- use WARNS?= 6 instead of a hard-wired list. -pedantic doesn't really work
sys/types.h ==> sys/param.h for ntohl
Fix spurious warning
Hide prototyp for loginit if ACULOG is false.
Fix prototype for signal handlers
sys/types.h ==> sys/param.h for endian macros
sys/types.h ==> sys/param.h for endian macros
- include sys/param.h for ntohl
- include sys/param.h for endian macros
- include sys/param.h for endian macros
- include sys/param.h for endian macros
- include sys/param.h for endian macros
sys/types.h ==> sys/param.h for endian macros
- include sys/param.h for endian macros
sys/types.h ==> sys/param.h for endian macros
rev 1.35:
Remove ASR_MEASURE_PERFORMANCE, it doesn't work anyway.
Improve the endian support for DragonFly by providing functions to convert
Remove pre-FreeBSD4 compability code.
Use sys/types.h here, since we don't want to restrict ourselves to the safe
Fix compilation of !PCI config.
Fix linker set creation for GCC 3.4 with -funit-at-a-time.
Add support for Nforce onboard ethernet. This is the content of
Fix the way warnings are printed. Use warnx(3) if the message is provided,
Remove the unit included in the softc, it is not used beside the call to
Create the dma maps before allocating the memory. Use WAITOK for the
Save current version of wi(4) as owi before switching to generic 802.11
Add const for argument of argmatch to fix warnings.
Import the new wi(4) driver based on the generic 802.11 layer.
Minor style cleanups.
Sync driver:
Add wlan and crypto for the new wi driver.
Sync with FreeBSD. Most importantly, add handling for -e, -H and -F
Make mark/unmark static. Use getprogname() instead of __progname.
Make edstop and writeback static.
Remove !TIOCSTI code and TIOCEXT conditionals. This have been in the
Remove void casts of function return values. Remove empty lines for missing
Transform the PCI probe ifs into the normal loop. Merge the controller
Don't sleep and commit. Fix the last commit.
Fix some warnings in the code about non-ISO C prototypes and void *
Add GID_MAX and UID_MAX.
From FreeBSD 5:
From FreeBSD 5:
Minor style cleanup. Use err(3) since we are interested in the error message.
Remove a duplicated PCIR_BAR definition.
Remove the unused PACKET_TAG definitions and document the structures used
Don't append '\n\0' to the return value of clnt_sperror if the string was
Make clnt_create take both host and proto as const char * arguments.
- statify functions
IP6A_SWAP is never set in our code base and this fragment doesn't even
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
Convert axe(4) to use splimp/splx instead of mutex calls, use the
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
Add axe(4).
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
Remove dead code which depends on timeout interface.
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
Convert timeout ==> callout_*.
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
SUS3 specified that netinet/in.h provides ntohl and may provide all the
The prefered location of the byteorder functions is arpa/inet.h. This
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
Remove unused consumer of timeout
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
Don't use dev/acpia/* includes for the ACPI5 infrastructure.
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
Replace the timeout based busy-loop with a DELAY(1000) based busy-loop.
timeout/untimeout ==> callout_*
if_clone_event should take a struct if_clone *, not a struct ifnet *.
timeout/untimeout ==> callout_*
- ISO C cleanup
Move the timer declaration a bit up. I don't know why GCC 3.4 works and
timeout/untimeout ==> callout_*
Split DN_NEXT into a version with and without cast. For the left side usages,
timeout/untimeout ==> callout_*, even though this is currently not used.
Don't include arpa/inet.h, which needs certain struct to be defined.
Remove unused defines
timeout/untimeout ==> callout_*
Don't include the PCI parts until the our system has involved to support
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
Fix some warnings / typos
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
Change the FreeBSD 5 jail sysctls to the correct DragonFly locations.
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
Fix typo.
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_* for p->p_ithandle
It is unlikely that NetBSD wants to take this code back or that DF will
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
Move the callout init below the softc allocation. *sigh*
timeout/untimeout ==> callout_*
Use ioctl_map_range instead of ioctl_map_cmd, as required by the mapping API.
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
Remove cast of lvalues.
Remove extra tokens at end of #undef.
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
Change getgrouplist(3) to take gid_t arguments for the groups.
The CIS vendor and product string routines return the string directly, not
Kernel part of PF
WARNS=6 cleanes
Also define global uid_t and gid_t variables and remove some warnings.
Install PF header files.
Add various ICMP defines used by PF.
Add a new option "TIMER_USE_1", which allows switching the primary heart
Use libcaps_free_stack instead of kmem_free for userland
Sync defines with sys/globaldata.h.
Include machine/cpu.h for userland to get clear_lwkt_resched.
PF uses -1 as rule number for the default entry. To make the value more
Uesrland part of PF
Use const char* for tcpstates
Add pidfile(3).
s/pidfile/my_pidfile/
This should read const char *, not char char *.
Use the protected names for BYTE_ORDER and co, the others are not defined
Typo.
Change the conditionals to use the protected _BYTE_ORDER defines.
BYTE_ORDER ==> _BYTE_ORDER
Fix typo
Add GPL-free patch. This is the extended version of Larry Wall's original
From FreeBSD:
Allow ip_output to be called with rt=NULL by making the FAST_IPSEC code
- make process() static and take const arguments
tsptype is an array of character constants, declare it as such.
EVENTHANDLER_REGISTER uses the name of the event as string and therefore
We now have pidfile in libutil, update kerberos5 accordingly.
Make the BSD patch WARNS=6 clean
Don't use patch -b, the flag has different meanings for BSD and GNU patch.
Switch patch(1) from GNU to BSD version.
Fix the setup for initial processing. Always use argv[0] and correctly
Add CTASSERT, a facility for simple compile time assertions. This is useful
Fix the code for the nxt != 0 case and use the assembler backend.
Fix locations of PF helpers.
Add some more functions for -liberty and make the library WARNS=6 clean.
GCC 3.4 doesn't check for NULL format-strings for __attribute__((printf))
Import zlib-1.2.2 using new-style contrib handling.
Merge from vendor branch ZLIB:
Import zlib-1.2.2 using new-style contrib handling.
Remove the GCC 3.3 build system.
Use SCRIPTS instead of a beforeinstall target, it's cleaner.
I have to specify SCRIPTSNAME too, otherwise bsd.prog.mk strips the extension.
Welcome GNU bc to the Attic.
Fix the references to libz sources scattered over the tree, make them
Import GNU readline 5.0.
Import GNU readline 5.0.
Merge from vendor branch READLINE:
Build framework for GDB 6.2.1.
Forced commit to add missing annotations.
Hide prototype for tilde_expand, it's incorrect now.
Add the info pages for GDB 6. The inc-hist.texinfo and rluser.texinfo are
Complete doc infrastructure.
Merge from vendor branch GDB:
Import GDB 6.2.1 as obtained from ftp.gnu.org without the files in
Good bye, GNU awk. RIP
- fix order problem in Makefile, Makefile.${ARCH} has to be included before
RIP GNU patch.
Remove the GCC 3.3 library build framework.
Remove GNU-CSU Makefile
Add gzip based on libz. This is faster for decompression and yields
Add support to specify the default compiler and binutil version in
gcc3 ==> gcc34
Change the way binutils versions are handled. Depending on CCVER,
Change the variable from CCVER_DEFAULT to OBJFORMAT_BUILTIN_CCVER_DEFAULT
Define PRId64 as "lld" for older systems laking support for it.
Switch to libz-based gzip.
Remove the old one-true-awk.
We have to support uncompress by default, because zcat is used e.g.
Move libstdc++3 into the attic. It can be removed from the repo later.
RIP gzip, we found a nicer playmate.
Remove a.out support from base. Bumb WARNS to 6 for ldconfig.
Fix a race condition in detach path of Ethernet devices. Most current
We don't have lkm, use aftermountlkm instead.
Small speedups:
Remove references to sysinstall, it vanished quite some time ago.
Fix the column width for IPv6 to correctly align the fields.
I don't know why I added use_bpf.h, it isn't needed of course.
Define IOV_MAX for userland too and make it available via sysctl and
Use the speedup patches, too
Import OpenNTPD 3.6 (the OpenBSD version, not the portable).
Merge from vendor branch NTPD:
Import OpenNTPD 3.6 (the OpenBSD version, not the portable).
Switch to gdb 6.2.1 and wait for new problem reports.
After another look at the current changes, switch directly to the in-tree
Merge from vendor branch NTPD:
After another look at the current changes, switch directly to the in-tree
Add SA_LEN patch for client.c, too
Close .if.
Install GNU tar as gtar now in preparation for adding bsdtar.
Import libarchive and bsdtar. The default tar (/usr/bin/tar) can be choosen
Merge from vendor branch BSDTAR:
Merge from vendor branch LIBARCHIVE:
Import libarchive and bsdtar. The default tar (/usr/bin/tar) can be choosen
Import libarchive and bsdtar. The default tar (/usr/bin/tar) can be choosen
Add libarchive to prebuilt libraries.
- style(9) cleanup
Add _ntp to the user database, correct the path to nologin for proxy
Initialize pseudointr_ch in fdc.
Change zconf.h and zlib.h to use underscored protection macros.
Fix systat -netstat and remove the dependency on KVM for this part.
Switch to OpenNTPD by default. For the moment, the documentation is
Remove ntpdate from FILES list as well.
Show the CPU used by the multi-threaded network stack to handle a socket.
Fix the condition under which yacc -o is used. This should fix the parallel
Remove left-over '{'.
Add solib-legacy.c, it is needed for proper attach support.
Merge from vendor branch GDB:
Add solib-legacy.c, it is needed for proper attach support.
If the first slot is empty, don't continue scanning. This fixes hangs
Fix an endian bug in pflog. The DragonFly version of PF uses the normal
-W is included in CFLAGS already.
-Wall -Wstrict-prototypes already in CFLAGS.
Remove CFLAGS?= 2
Remove the dependency on either asprintf or va_copy. This will be included
Fix a lhs-cast by casting first to unsigned char and then to the
Sync with FreeBSD. Most importantly, this removes the need for perl.
Remove ptx, it isn't maintained and hasn't been used by base for years.
The code and sources here haven't been used since FreeBSD 2.x, nuke
Add preprocessor handling for newer processor, they should get the better
Sync with FreeBSD. This removes the need for perl.
Remove the old release infrastruture. Certain parts e.g. sysinstall or
Cleanup send-pr and ptx on next update.
Add some more hacks for dfport handling, to allow make index to mostly
- use STAILQ, instead of hand grown implementation
Fix a bug in parsefmt where a string still referenced would be freed.
Override USE_GCC=3.4, we want to use our system compiler in that case.
Don't install the undocumented scriptdump Perl script.
Replace spkrtest script with a shell version.
Always use the width field from the varent, not the original width.
Use the YACC source for c-exp.y and f-exp.y, not the pre-built
Normally we want to warn if the local IP address is used by a different
Change the name entry in struct nlist to const char *.
Make pstat WARNS=6 clean.
Revert a part of the last commit. The changes to use sysctl interface
Add a new system library, libkinfo. It is intended to replace libkvm for
Include Makefile
Add kinfo_get_vfs_bufspace(). Remove some junk from Makefile.
Use kinfo_get_vfs_bufspace instead of kvm.
Replace lvalue cast.
Change the include order of archive_string.h and archive_prviate.h.
Comment out extra tokens after #else and #endif.
Add the basic of libkcore. Switch pstat to use kcore/kinfo backing,
Fix parameters for comparision function.
Fix warnings about yyerror, yylex and yy_unput
Don't define the .BEGIN with the override message twice for modules
Remove some !FreeBSD compat code, which produces warnings on DragonFly.
Fix a stupid fault in the last condition. The code should be skipped, if
Reorganise the nlist handling a bit, use an enum instead of preprocessor
Missing return-value check for malloc.
Cover the verbosity code in {}, otherwise the newline is printed everytime.
Allow "make index" to actually by fixing lots of smaller glitches all over
Convert some broken cases where Error is called, but we try to continue,
Set DFPORTSDIR. It will be used by "make index" soon.
main.c: 1.81->1.82
author: ru
dir.c: 1.31->1.32
Fix a bug that leads to a crash when binat rules of the form
job.h: 1.20->1.21
parse.c: 1.53->1.54
make.1: 1.67->1.68
Makefile:1.30->1.31
Don't include stddef.h from the kernel.
Remove struct ipprotosw. It's identical to protosw, so use the generic
Switch a strncpy into a strlcpy. I'm not sure why this triggers propolice,
Makefile: 1.31->1.32
Makefile: 1.32->1.33
make.1: 1.77->1.78
job.c:1.52->1.53
*** empty log message ***
Merge from vendor branch NTPD:
*** empty log message ***
Force commited to annotate missing log message.
Merge from vendor branch NTPD:
Force commited to annotate missing log message.
Change the default for ntpd back to -s, the bug which triggered this
Move ntp.org's ntp into the attic.
Add rdate(8). This supports both SNTP(RFC 2030) and RFC 868.
var.c: 1.42->1.43
var.c: 1.44->1.45
cond.c:1.27->1.28
compat.c:1.38->1.39
job.c: 1.51->1.52
Don't read userland pointers directly, copy them first into kernel land
Remove unused tcpdump sources.
Make rp(4) compilable again. Don't even think about unloading this stuff.
Replace div_pcblist / rip_pcblist / udp_pcblist with in_pcblist_global.
Return retval if the second sysctl failed, not NULL.
Don't include the "Hello, world" example in libz.
Mark the following stuff as depricated:
Implement generation counters as (at least) 64 bit counters. The increment
Add dependency for libcrypto to dc(1). Don't build bc and dc if
Merge from vendor branch NTPD:
Sync with OpenBSD.
Add an example ntpd.conf. This is not installed by default, because
Remove generic generating counting. I'm going to use a different approach
Add weak fake pthread functions, which always fail. This is needed to
Remove the userland visible part of the socket generation counting.
Consistently use /:/boot:/modules as search path in the loader, the kernel
USER_LDT has been removed ages ago.
Remove unecessary range check. start and num are unsigned anyway and
Define __arysize for FreeBSD 4 buildworlds.
Remove cruft for GCC 3.3
- Update GCC to version 3.4.3.
Merge from vendor branch GCC:
- Update GCC to version 3.4.3.
Don't even include the GDB build framework.
Install the GCC 2.95 main info pages as gcc2.info and cpp2.info.
Fix a regression with GCC 3.4.3 by using __packed instead of mode(byte).
Change type of len to size_t.
Use size_t for len.
Fix NO_OBJC knob.
Depend on _KERNEL or _KERNEL_STRUCTURES.
Remove support for ancient (FreeBSD) kernel, which don't set profrate.
Trust your intuition. If something feels wrong, it often is wrong.
Add digi driver to simplify testing. This should replace dgb.
Add new structures for exporting the cputime statistics via 64 bit counters
Remove my local patch again, it was still not meant to be commited.
Check that the ifnet_addrs entry is actually used before dereferencing it.
Ignore ENOENT when fetching the interface MIB. This can happen for
Don't build the gdb-related documents in binutils-2.15.
Mark old file as dead.
Replace sockstat(1) Perl script with the C version from FreeBSD 5.
Switch from binutils 2.14 to binutils 2.15.
Update isc-dhcp to 3.0.2rc3 using patch infrastructure.
Update isc-dhcp to 3.0.2rc3 using patch infrastructure.
Merge from vendor branch DHCP:
RIP
Don't install old GDB documentation.
Add ISO C99's _Exit, which is identical to _exit.
Merge FreeBSD rev 1.70:
WARNS=6 safeness:
Mark binutils-2.14 as dead.
Move sa_X macros under _KERNEL protection for now.
Unhook Perl from build.
Add _DIAGASSERT macro for library internal usage. This is not active by
style(9) cleanup.
Don't clobber match_str in buildmatch by strdup'ing the string internally.
ANSIfy. Fix various warnings. Statify.
It's WARNS?=6, not WARNS=6.
Don't create binutils 2.14 and perl directories. Remove perl and
RIP Perl.
Move short option parsing into its own function. Constify the local char *
Move the handling of '--' as argument into getopt_internal. Add a parameter
Add support for getopt_long_only. It allows using long options with a single
Fix two small bugs in getopt_long_only handling:
Better diagnostic for getopt_long_only.
Remove wx(4). It's been superseded by em(4).
Remove wx(4) man page as well.
Remove documentation of NOPERL, it's the default now. Don't create an
Remove perl's man path.
Add local_syms script. Nice for cleaning the kernel namespace.
Add splitpatch.
Mark OpenSSL 0.9.7d as dead.
Make newkey WARNS=6 clean.
WARNS=6 cleaness.
Switch chkey/newkey to use libcrypto instead of libmp for the internal
Add forgotten cast back. This code depends on BASE being short and
Remove some debugging printfs. Fix an error message.
Remove another Perl left over.
Convert to keyserv, telnetd and telnet to libcrypto's BIGNUM
Deorbit libgmp.
Remove LIBGMP and LIBMP.
Add gmp handling to upgrade_etc.
typo
Add default if_re_load line.
Don't write the name of the sysctl to a temporary variable and truncate
Correctly reset place in getopt_long_only the next time it is used.
Now that we have dhcpd and dhcrelay in base, fix the inherited rcNG scripts
RIP acpica-unix-20031203.
Explicitly initialize fp to NULL. If sendfile is called on a non-socket,
Don't include the kvm backend, it's not really working anyway.
Add "proc" command.
Initial jail support for varsyms.
Initial part of DEVICE_POLLING support for wi(4). Still some rough edges,
Replace temporary allocation from alloca with malloc/free.
strl* conversion.
Instead of messing around with sprintf, use asprintf to do the allocation
Back out part of rev 1.24. The intention the quoting backfires and bad
Don't print the override warning for package-name.
Support PORTDIRNAME and PKGORIGIN.
WARNS=6. The __DECONST is save, because execv doesn't mess with the strings,
Avoid possible copyright problems and add the copyright of the
Use PCIB_ROUTE_INTERRUPT instead of PCI_ROUTE_INTERRUPT, the latter is
Ignore zero length mbuf in bus_dmamap_load_mbuf.
Remove left-over.
From FreeBSD:
Add a simple tool to generate a sequence of numbers, without all the
Add missing 'by' in the license text of the man page.
Adjust calling convention for idp_input, idp_output and spp_input to match
Add forward declaration of struct vnode to fix compiler warning.
Use __DEQUALIFY, not only __DECONST to get rid of the volatile too.
Consolidate ifqueue macros for the upcoming ALTQ work.
Remove the support for lib/csu/${MACHINE_ARCH}-elf.
ether_input already handles BPF and dropping of
Use BPF_TAP and BPF_MTAP instead of the expanded versions where possible.
Mostly switch to the patch framework. For cvs.1.patch, it's easier to do by
Fix a type in rev. 1.16
Remove !_KERNEL parts.
Remove the _KERNEL parts.
- Add support for attaching alternative DLTs to an interface.
Add some spaces for typographical correctness.
Fix the quoting of MAKEFLAGS as noted in rev. 1.47.
Don't build a new mbuf and bpf_mtap the old one.
Use and define size_t as specified by SUSv3.
Add a default for Perl 5.8.5 to override the default for
Use M_INTWAIT, not M_NOWAIT. We don't really support fast interrupt
ANSIfication and minor style cleanups.
remove bad semicolon
We don't currently build ld-elf as binary for i386 and even if
Don't build sysctl as bootstrap tool. This helps us avoiding FreeBSD 4
Remove unused compat code like FXP_LOCK, a fake struct mtx.
In preparation for the jail commit, include sys/param.h instead of sys/types.h.
sys/types.h ==> sys/param.h
Split vn_fullpath into cache_fullpath and vn_fullpath. The former just
Uncomment the entry for kern_chrot in kern_syscall.h and change the
Add jail_attach syscall.
Regen.
Don't copy blindly MAXPATHLEN byte from fullpath. vn_fullpath
Cleanup last commit. Remove local ncp, it shadows the parameter, add
Regen.
sys/types.h ==> sys/param.h
sys/types.h ==> sys/param.h
Fix warning by adding explicit () around addition.
Use PRIx32 to fix compiler warning correctly.
Buglet in last commit, the first argument of bpf_ptap is the actual bpf_if.
Remove unused local variable.
Define in_inithead in net/route.h, it's purpose is the initialisation
Add missing "return(error)".
Include jail.h after proc.h to get struct ucred.
Add jail_attach support.
Remove GPLed fpemulation, old rp, old awe and pcic.
Remove unused variable.
Remove compat code for anything, but DragonFly.
Add prototype for ciss_print0 in place.
Make const correct.
Hide pure kernel threads from jailed processes.
Remove SCARG junk.
Always honor fdp->fd_nrdir as root. Once the loop reached NCF_ROOT,
*_load_pallete's second argument should be const.
Don't double assign -W -Wall and some more warning flags.
GCC 3.4 doesn't include a #pragma weak reference, if the symbol
Change the ALTQ macro stubs to actually work. For the current ALTQ less
Change default perl version to 5.8.6
Fix CPP buglet.
Don't cast int ==> unsigned char ==> char, int ==> char is enough.
Don't use the statfs field f_mntonname in filesystems. For the userland
Free temporary buffer in the buffer overflow case too.
GCC 1.x is dead.
Read eaddr in two parts (32 bit read and 16 bit read). This fixes a
No namespace pollution in sys/cdefs.h. Rename VM_CACHELINE_SIZE
Fully separate the kcore initialisation and the kinfo wrapper.
Don't define _Bool for GCC 3.0 and later. It was added was builtin
Remove debug flag which slipped into last commit.
Use M_ZERO instead of manual bzero.
exit() needs stdlib.h
strcmp() needs string.h
Forced commit to note that I'm removing the default VINUMDEBUG for now.
Remove the default VINUMDEBUG option for now. I'll make most of the
Provide a mechanism for dumping relocation information.
No need to zero fill memory, mmapped anonymously. Kernel will
Remove -DFREEBSD_ELF.
give out a little more information in case of a missing dependency
If we change obj_rtld.path after initialising __progname, make sure we
Do not depend on existence of _end symbol in obj_from_addr, use
Stop caring about GCC versions between 2.5 and 2.7, they are
Readd the copystr for f_mntfromname of root. It wasn't meant to be
Do initialise fp to NULL explicitly, the last comment wasn't enough.
- Add support for DT_FLAGS.
Add a macro SYSCTL_SET_CHILDREN. Use it to avoid lhs cast.
During init time, we can savely allocate the mbuf cluster with
Remove the conditionalized FreeBSD 5 code. Keep the capability assignment,
Add missing */.
Don't cast lvavalues.
- convert to bus_space macros
Include pci_private.h to get pci_class_to_string.
Set so->so_pcb directly to NULL to avoid lvalue cast.
Add -DIN_TARGET_LIB to prevent libobjc from depending on in-tree GCC code.
Directly use ifp->if_snd, it's type will change soon.
Don't assign &ifp->if_snd to a temporary variable, it's type will change
Use ifp->if_snd directly. Use IFQ_SET_MAXLEN.
Remove stale local variable ifq.
Add the 'All rights reserved.'. It's not entirely clear if this is still
Remove unused variable.
Import ALTQ support from KAME. This is based on the FreeBSD 4 snapshot.
Add ALTQ support to pfctl(8). Fix a small bug introduced earlier
altq include dir
install altq headers
Remove unnecessary packed attribute.
Remove unneeded packed attributes.
Remove extra token after #endif introduced in last commit.
Separate error handling path from normal return to avoid
GC strsize.
GC local variable size.
Clean-up.
Mark dead. Not supported and if we ever want to support it again, this needs
Move mac = NULL initialisation up and simplify second conditional.
Inline some users of SC_STAT to avoid lvalue cast.
link in altq support.
ALTQ support.
ALTQ support.
ALTQ support.
Mark it as ALTQ ready too.
Increase size of limit column by one character to better fit
Add support for ICH6 and some nForce AC97 chips.
Rename IFM_homePNA to IFM_HPNA_1 and IFM_1000_TX to IFM_1000_T.
ALTQ support.
Set ALTQ ready.
ALTQ support.
ALTQ support. Remove bogus check of ifq max len.
ALTQ support.
Introduce vnodepv_entry_t as type for the vnodeopv_entry functions.
Use ifq_is_empty to allow USB drivers to support ALTQ.
Be more careful when doing el_parse() - only do it when el is
ALTQ support.
Don't cast lvalues.
Avoid casts as lvalues.
Replace lvalue cast with explicit cast from u_short via int to void *.
Replace lvalue cast.
Make the second argument to ncp_conn_asssert a const char *.
ALTQ support.
Change second argument of dc_crc_le to c_caddr_t to fix warning.
Fix lvalue cast.
Replace list of checks with loop. Fix lvalue casts.
Fix lvalue cast.
ALTQ support.
ALTQ support.
Forgotten in last commit.
ALTQ support.
ALTQ support.
ALTQ support.
Add __pure as attribute. A __pure function can only depend on the
The old USB ethernet code utilized a netisr to hand packets over
Rename PACKAGES to REL_PACKAGES. PACKAGES is used by ports already.
Use -pthread instead of adding -lc_r for linking the thread library.
GCC supports two pseudo variables to get the function name, __FUNCTION__
ALTQ support.
ALTQ support.
ALTQ support.
ALTQ support.
ALTQ support.
Use ifq_set_maxlen and don't change the field directly.
ALTQ support.
ALTQ support.
ALTQ support.
Don't inline lnc_rint and lnc_tint, it's useless and GCC rightfully complains
ALTQ support.
Call the "cluster_save buffer" type just "cluster_save", it doesn't fit
ALTQ support.
ALTQ support.
ALTQ support.
Use ifq_set_maxlen instead of messing with the fields directly.
Depend on ntpd. ntpdate is gone and the behaviour is the default now.
ALTQ support.
ALTQ support.
ALTQ support.
ALTQ support.
Add m_defrag_nofree, which works like m_defrag, but doesn't free the
Fix a bug introduced earlier. We can't put packets back into
Prepare for ALTQ.
ALTQ support.
ALTQ support. Fixes a small bug which could result in packets showing
ALTQ support.
ALTQ support.
ALTQ support.
ALTQ support.
Move the m_freem from vr_encap to vr_start, making the passed mbuf effectively
ALTQ support.
ALTQ support.
ALTQ support
Replace dc_coal with m_defrag.
ALTQ support.
ALTQ support.
ALTQ support. The error handling in nv_ifstart should be reviewed,
Remove tulip_ifstart_one, always process the whole queue.
ALTQ support.
Remove altqd's init script, we don't have it anyway.
Don't depend on altqd.
We don't need syslogd support for chrooted ntpd.
Require NETWORKING, not DAEMON for ntpd. The former makes much more sense
Rename local variables and arguments "index" to "idx".
Some cleanup.
Use ether_crc32_le.
Expand some macros.
Use normal inb / outb / inw / outw / inl / outl functons.
Remove LE_NOLEMAC and LE_NOLANCE conditionals.
Remove TULIP_USE_SOFTINTR support. If someone really needs it, we can
TLUIP_HDR_DATA was always defined, unconditionalize.
GC __alpha__ support.
GC TULIP_PERFSTAT support.
GC TULIP_BUS_DMA.
GC !__DragonFly__ section.
GC TULIP_VERBOSE.
GC TULIP_DEBUG.
GC TULIP_NEED_FASTTIMEOUT
GC TULIP_CRC32_POLY, TULIP_ADDREQUAL and TULIP_ADDRBRDCST.
Add sigtimedwait.o and sigwaitinfo.o.
Remove some duplicate FreeBSD CVS IDs, move some IDs to better places.
More cleanup.
Define default value for PRId64 to keep FreeBSD 4 happy.
Conditionalize the source on INET and INET6 respectively.
Temporary switch the cardbus interrupt from INTR_TYPE_AV to
Split search for already loaded object into a helper function.
Sync changes from OpenBSD. Most importantly, this adds reverse proxy support.
Move the open, stat and recheck for already loaded objects into
Also bumb the reference counters when a object is already loaded.
Instead of messing with the internal name-to-oid conversion,
Make init(8) WARNS=6 clean.
Merge from vendor branch NTPD:
Sync OpenNTPD with OpenBSD.
Sync OpenNTPD with OpenBSD.
Don't define unit field on DragonFly.
We already have the unit number in fl_ifname.
lvalue cast.
Remove NO_WERROR assignment.
Remove -Werror from CFLAGS and fix the missing part for WARNS=6.
Use __DECONST for the linted interface violations.
Remove pre-ANSI malloc prototype.
Remove NO_WERROR.
lvalue casts.
Add some consts.
lvalue casts.
Mark symname const
Update Atheros entries.
Regen.
Remove extra tokens after #else and #endif.
Use u_char define from sys/types.h.
Rearm receiver, it was lost in the conversion.
Define __DECONST here for the sake of FreeBSD.
Simplify patches by defining SA_LEN in one place.
Rest of last commit. Define SA_LEN in one place.
Reduce foot-shooting potential by adding -dP for cvs update and -P for
WARNS=6 cleaness, some constification, some cleanups.
RIP Alpha libc.
Merge from vendor branch GCC:
Add C++ ctype based on NetBSD-style ctype.h.
Use NXCC to compile a runnable executable. The libc we are building might
Use the standard isxdigit instead of the non-standard ishexnumber.
Replace digittoint(c) with (c - '0'), we already know that c is a digit,
Use isdigit instead of isnumber.
Don't define isatty, it's already defined in unistd.h.
WARNS=6 cleanup.
Fix warning about pre-ANSI function types.
Fix prototypes. This functions actually need a parameter.
Use ULLONG_MAX, off_t is unsigned.
Correct a mixup between NLSGRP and NLSOWN. Add LOCALE{DIR,GRP,OWN,MODE}.
Remove bad line breaks.
Don't use stdint.h on FreeBSD.
Really use ALTQ.
Don't try to prepend to an ALTQ queue.
ALTQ support.
Initialize all fields in kobj_class.
Change Makefile to follow common white space rules. Hide -DALTQ and
Sync with FreeBSD.
Use getprogname() instead of depending on __progname where possible.
Add bus_dmamap_syncs before bus_dmamap_unloads.
typo
Add generic build framework for message catalogs.
Use bsd.nls.mk.
Use NLS framework.
Add data files for Character Set mappings and Encoding Scheme alias
Add locale and character set descriptions. This differs from the
forgotten file
iconv frontend.
Helper programs.
Add i18n directories.
I18N module build framework.
Add UTF7 build wrapper.
Add citrus backend code and iconv front end. This is intentionally
Add wchar and multibyte related man pages.
Remove MT_FTABLE, it is not used anymore. Remove the #if 0'd entries
Add ALTQ-style enqueue / dequeue / request functions for traditional
Always build altq_etherclassify.
Change PACKET_TAG_* ids to be consecutive.
Fix build. Move common CFLAGS additions up.
Don't build / install the various historic troff papers to
Don't create some directories which will now be empty.
Don't build groff as buildtool, it's not needed anymore.
Move ctype man page from locale to gen, they are locale-sensitive, but
Use getopt_long's getopt implementation.
Use libc's getopt_long, I've been building and using ports for a while with
Another man page which was moved to lib/libc/gen.
SUS says that getopt() should return -1 immediately if the argument is "-".
Correctly return -1 for "-" as argument as required by SUS.
Add '+' to argument list of find to prevent getopt from reordering
Back out switch to getopt_long implementation of getopt, it breaks
Merge r1.2 from NetBSD:
Move SHLIB_MAJOR / SHLIB_MINOR assignment up into Makefile.inc.
Use NXCC to build make_hash and make_keys.
Instead of using the non-standard conforming %+ format string,
GC
Stop building old GDB.
Pass const ** as second argument to the command functions,
Raise WARNS to 6:
WARNS=6.
Fix warnings.
idx is always used, shut GCC up by initialising it to 0.
Fix some warnings.
s/index/idx. Add an explicit default case to teach GCC that idx is
Make dktypnames const.
More stuff for Citrus.
More Citrus code.
Add Citrus files not conflicting with the current rune implementation.
Add some more path definitions as used by the Citrus framework.
Remove redundant declarations. Convert to ANSI C function declarations.
Use a common source for the string to integer conversion. Add
Fix warnings. Raise WARNS to 6.
WARNS=6.
WARNS=6.
Generic firmware support. Currently implemented is loading from
Add /etc/firmware, remove /etc/gnats.
Minor style changes.
Add kqueue overwrite for libc_r. We have to trace the opened descriptor.
Override _kevent, not kevent. This should fix the DNS issue, since
Change prototype of sys_set_tls_area and sys_get_tls_area to take
regen
Adjust userland prototypes as well.
int size --> size_t size
Fix reboot -k, it didn't truncate /boot/nextboot.conf.
No more Alpha support for RTLD.
Defer work from the signal handlers into the main loop.
Check for the signal handlers another time, they might not disrupt
Include sys/types.h to make it self-contained.
Fix handling of deallocation of dynamic TLS, the previous code could
Don't activate -fstrict-aliasing by default, not even with -O2+.
Use the correct type in va_arg call, char is promoted to int before calling
License typo.
Remove redundant panic.
Remove the pre-NEWPCM sound drivers and the speaker-based emulations.
Separate M_NULLOK from M_RNOWAIT.
Cleanup the TLS implementation:
New strcspn implementation, which is O(strln(str) + strlen(chars))
GNU getopt resets itself partially when the application sets optind to 0.
Fix warnings, use ISO prototype.
Remove tcb_size and flag argument for _rtld_allocate_tls,
Remove extern for functions, line up function names, remove option names.
Avoid discarding constness.
const correctness
Declare prototypes for all functions.
Fix non-BPF declaration of bpf_ptap, it was out-of-sync with the header.
Make it O(strlen(s) + strlen(charset)) like strcspn.
Back out part of last commit, optind has to be initialised to 1.
Move the processing of malloc flags out of the loop. The exception
Move the processing of flags out of the loop. The exception is M_WAITOK,
Remove reminders of VoxWare.
Remove VoxWare related entries, don't list drivers support by NEWPCM.
Split pcm into the generic framework (pcm) and the sound cards (snd).
WARNS=6.
WARNS=6.
WARNS=6.
const changes.
Remove pre-FreeBSD 3 compat conditionals.
WARNS=6
Add a macro to print the list of current processes independ of wait state.
When ALTQ was detached from an interface queue, the function pointers
Replace TIMER_USE_1 kernel option with hw.i8254.walltimer tunable.
Sync nv(4) with nvnet-src-20050312.
Since optind=0 has a special meaning for GNU getopt compatibility,
Revamp getopt(3) usage:
Add nv(4) man page.
Add closefrom(2) syscall. It closes all file descriptors equal or greater
Regen.
Add userland prototype for closefrom.
Use kernel's closefrom.
Tell the world that we have closefrom.
Don't just undefine USE_RC_SUBR, because it does define the list
Simply the loop based on the knowledge that fd <= fdp->fd_lastfile
First bunch of dump(8) cleanups.
Remove unused functions.
Compute buffer length only once per rmtcall.
Split atomic into two functions, atomic_read and atomic_write.
Warning fixes.
Move initialisation of wrote into the loop to prevent longjmp/setjmp
Remove argument names from prototypes. Adjust prototypes to deal
Don't write to tape, strdup it first.
Remove a cast, correct another.
Decommon a lot of variables, makes some static, fix a small bug in rmtcall
Remove another argument name in a prototype.
Simplified NTP kernel interface:
wether ==> whether
Use corrected system time internally, adjust old offsets after
Use corrected system time internally, adjust old offsets after
Merge from vendor branch NTPD:
Sync with OpenBSD. This fixes the bug of -s not working and allows
leap second, not leaf second. fix totally stupid, but consistent spelling
Ignore replies with a negative delay. Correct time of next query and
Fix a small stack disclosure in ifconf().
Extend a local buffer to workarond a buffer overflow for now.
Fix the parameter order of __stack_smash_handler. Also print the
GC OLDCARD userland tools.
Remove pccard RCNG script, the daemon config, the depency on pccard
Use the headers from src/include for ctype.h and runetype.h instead
Add snprintf and vsnprintf.
Make kern.ntp.permanent specify the frequency correction per second,
Switch IP divert from mbuf based tagging to mbuf tags.
Use the local runetype.h here too.
Sync with OpenBSD. The ntpd -s is included now.
Merge from vendor branch NTPD:
Sync with OpenBSD. The ntpd -s is included now.
WARNS=6 + minor style issues while here.
Sync in my paranoia check.
Merge from vendor branch NTPD:
Print "reply from ..." before "adjust local clock ..." in debug mode.
Merge from vendor branch NTPD:
Provide correct PLIST_SUB value for batch scripts.
Correct a bug in the positive 32bit overflow handling of ntp_tick_acc.
Add support for ServerWorks chipsets.
Use get_approximate_time_t().
basetime should be static, all access done via PKI.
Remove some commented out warnings.
Remove old beforeinstall hack. Should be not used for ages.
Un-disable stack protector. I take responsibility for all problems this
WARNS=6. The (int) case should be reevaluated when this code is converted
RIP compat libraries. Use misc/compatXX from ports instead.
Make GCC 3.4 the default compiler.
Add _DATE_FMT to get the locale-specific date(1) string.
Use new _DATE_FMT langinfo instead of '%+'.
Unhook PCVT from kernel build.
Stop building libkeycap, PCVT is gone.
Unhook GCC 2.95 and Binutils 2.12 from build.
date as bootstrap tool doesn't make sense, remove it.
Make osreldate.h building happing by explicitly using /bin/date.
Override gcc2 for buildworld. Necessary to handle the former default
As long as a peer has some trust left, use the aggressive retry interval
Consider only drifts smaller than 32 ms as negligible.
Bump major numbers in preparation for libc work.
Don't create directories for gcc2 and binutils212 anymore.
Build aicasm as host program, not via world's compiler.
Complete Citrus import. Import message catalog implement from
Use system version of getopt_long and basename for the bootstrapping tools.
Unhook gperf, it was only used by gcc2.
RIP PCVT userland.
Remove RC scripts for stuff we have don't have in our tree.
Add frequency correction support. The drift factor is currently
Don't install wscons.
Remove obsolete rcNG scripts as part of make upgrade.
Use normal variable names for rtsold, provide default values.
Move to CPU #0 in settime() to prevent races.
Back out last commit, we are not there yet and I have to study
Don't call cpu_mb1 after lwkt_setcpu_self, but call it internally
Remove MIPS bits.
GC more mips parts.
Make kern.ntp.delta preemption save.
Add forgotten SYSCTL_ADD_QUAD, move the SYSCTL_ADD_INT up where it belongs.
Add support for unsigned quads. Use strtoq and %qd / %qu.
Always use SYSCTL_OUT, sysctl doesn't print the value otherwise.
Fix stupid order bug. The code should ignore the first sample(s),
Properly create and destroy the DMA maps.
ANSIfy and fix function casts.
ANSIfy.
CISS quirk.
Ansify getcwd and declare the syscall prototype.
First stab at WARNS=6 cleaning. More will be done once I figure out
Fix warnings, ANSIfy, constify.
const correctness
Readd ypresp_allfn, now correctly typed.
Fix warnings.
Unconditionalize HAS_UTRACE, we never have NetBSD syscalls.
Remove compat junk, __getcwd always exists on DragonFly.
Remove more __NETBSD_SYSCALLS.
Fix warnings, remove unused headers. Remove additional { }.
Fix warnings. Sprinkle const. Use size_t in some places.
ANSIfy. Fix warnings.
Fix warnings, ANSIfy.
Fix warning.
vfork can clobber the local stack frame, use fork(). We might also
Fix warnings.
use uid_t / gid_t for prototype in stdlib.h, ANSIfy.
Use namespace mangler.
ANSIfy. Fix warnings.
Ensure that the directory fits into memory. Fix warnings.
Include sys/types.h to get uid_t and gid_t.
Include guard.
Fix warnings. Use sysctlbyname.
Really include patches, don't let them catch dust.
Fix warnings.
Fix warnings.
Change arc4random_addrandom to pass the more natural uint8_t * and
Add prototype for __creat.
Fix warnings. Use sysctlbyname.
Correct and improve __diagassert.
Work around restrict.
Explicitly initialize e. The code flow looks safe, but having
Fix warning.
Always use strlcpy, in the last case also check the return value to
Remove useless void * before free. Remove local prototype of
It's dead, Jim.
ANSIfy, no (void) before functions, include stdlib.h for prototype.
Correct types for devname[_r].
Remove dead code. Fix warnings.
Remove unionfs hack. DTF_NODUP is now a NOP.
const / sign correctnes
const correctness
Mark name as const char *, not char *.
ANSIfy, fix warnings.
Always use strlcpy, in the last case of possible truncation also check
Remove dllockinit, it's been deprecated and is a NOP anyway.
Remove dllockinit.
Fix warnings.
Use size_t for length, not int.
ANSIfy, use strlcpy from strncpy, fix most warnings.
Fix warnings.
Use size_t for the number of elements, use sysctlbyname instead of
Fix warnings, use strlcpy instead of strcpy + manual check.
When bitwise iterating over in_addr_t, use it for the loop variable too.
Include string.h, use strlcpy.
Use const for internal casts to not conflict with const pointers.
Remove unused local functions. Const correctness. Sign correctness.
DragonFly has decided to depend on char being signed, use it.
Declare __system first. ANSIfy.
Use __DECONST for interface const violation.
Fix the sign issue by reordering the operations. Use memcpy instead
Rename symlink to my_symlink to avoid global shadowing.
Fix warnings.
Declare environ on file scope. Don't declare __findenv as inline,
Always setup the initial TCB correctly. This saves us from having to
Add support for TLS.
For the initial thread, rtld has already created the TCB and TLS storage.
Slight correction for the last commit, thread TCB == NULL as error
Forced commit to correct imporant spelling error:
Always allocate static TLS space.
Readd explicit u_char casts for tolower().
Readd lost line.
PANIC for now, if the linker can't allocate TLS space.
Including errno.h and still declaring errno is BROKEN.
extern int errno considered harmful.
de-errno
de-errno
de-errno
de-errno
de-errno
de-errno
de-errno
Only install man page for libc.
Separate INTERNALLIB and INSTALL_PIC_ARCHIVE. We want to have the latter
Make a special libc version for RTLD which doesn't use TLS as it will
Prepare for thread-local errno by implementing full TLS support for
sys_set_tls_area is called from _init_tls path, it should not touch
Add NO_PKGTOOLS to not install the FreeBSD derived version of the tools.
Include __error in libc_rtld too, otherwise RTLD is left with an undefined
Link with --no-undefined to enfore the normal missing symbol check,
Remove sys_set_tls_area hack, doesn't matter when we fault in case
isprint() should be true only for characters in space, not blank.
Override closefrom() in libc_r to prevent it from closing the
Explicitly close low descriptors to keep the internal state
Forgotten major bump.
We have to allocate the TLS area for _thread_kern_thread too.
Remove PCVT related entries.
Make errno a thread-local variable and remove the __error function.
Spring-cleaning.
More spring-cleaning.
More spring-cleaning.
Refine USE_RC_SUBR / USE_RCORDER handling. The install-rc-script target
gnu/usr.sbin and gnu/libexec have been empty for ages, so save us
Move old locale sources into the attic.
Sync with FreeBSD. This adds read-only support for zip and ISO9660.
Sync with FreeBSD. This adds read-only support for zip and ISO9660.
Merge from vendor branch BSDTAR:
Merge from vendor branch LIBARCHIVE:
Sync with FreeBSD. This adds read-only support for zip and ISO9660.
Don't bother dealing with hidden syscalls, just do it for all.
Remove obsolete patches.
Allocate some additional space for dlopen'd libraries. Currently 256 byte,
Split libc and libc_r. -pthread now links aginst both libc and libc_r.
Typo, linking against -lc_p belongs into the -pg case.
Left-over from the old mbuf chain tagging.
ALTQ
Add handling of R_386_TLS_TPOFF32.
Bite the bullet and add real masks isprint and isgraph. Use this as
Nuke ctypeio.c and associated ctype to rune table conversion,
Catch up with ctype.h.
Replace ifq_handoff like code with a call to ifq_handoff.
atm_output did almost the same as ifq_handoff, it just skipped
Use ifq_handoff instead of handrolling it.
Use ifq_handoff instead of the hand-rolled version with inlined
0x7f is only a control character.
Typo in '(' description.
Replace __offsetof with similiar pointer expression and use "m"
Remove sc_unit and use if_printf / device_printf instead.
Remove obsolete comment.
Merge from vendor branch GCC:
Remove two unnecessary entries.
typo
Simplify code by using IF_DRAIN.
Replace local tulip_mbuf_compress with m_defrag call.
Remove unused second argument to tulip_intr_handler.
Use device_printf / if_printf and stop abusing the ifnet fields for
Don't activate -funit-at-a-time with -O2 and higher by default, when
WARNS=6 clean already.
Fix pointer arithemetic.
Add an explicit abort in printaddr to avoid unused variable warnings.
Move if_initname call before xl_reset to ensure correct initialisation.
Reorder initialisation by calling if_initname before vr_reset.
Remove an incorrect free. The code path should normally not be hit, but
Rework TX EOF handling. We have to always check for TX underruns,
if_printf/device_printf cleanup. Remove minor junk.
Use bus_alloc_resource_any. Use pci helper functions, don't roll them
Exploit bus_alloc_resource_any.
style(9)
Use ether_crc32_le instead of local hack.
Convert bge(4) to the new M_EXT API. This allows merging the dynamic
Spurious semicolon broke gcc2 build.
style(9) and nic style changes.
Cleanup bfe_probe.
device_printf / if_printf cleanup.
duplicate is
Fake DES was removed a while ago, remove the prototypes as well.
Remove spurious semicolon.
Merge from vendor branch GCC:
Update in-tree GCC to 3.4.4.
Update in-tree GCC to 3.4.4.
Merge from vendor branch GCC:
Update for GCC 3.4.4. Note that the prefered way to fetch just
Revert to old local unit allocation until I find a better solution
Use ether_crc32_be.
if_printf / device_printf.
No FreeBSD 5/6 support here.
No machine/clock.h needed.
Use ether_crc32_le.
Add a device ID for MPI350.
if_printf / device_printf.
Use bus_alloc_resource_any when possible.
Convert to new m_ext API.
style
Use ether_crc32_be.
if_printf / device_printf and some further cleanup.
Nuke further compatibility junk.
if_printf / device_printf and some more cleanup.
Use ether_crc32_le.
Remove compat junk.
Force jumbo buffers to be a multiple of 64bit.
Convert to new m_ext API.
Remove some compat junk, deindent a switch.
Add missing splx(s) in sk_detach.
style. remove some unused variables.
Use ether_ioctl for the default case in dc_ioctl. Merge SIOCSIFADDR,
Fall through to ether_ioctl and merge those cases which just called it.
Use ether_ioctl for the default case. Merge cases which just called it.
Don't call ether_ioctl first, check for errors and call it again
Don't cast command to int.
Use ether_ioctl for the default case and merge the cases which just
Minor reorder of the code to make it easier to deal with capabilities.
Remove nge_jpool_etrny, not used anymore.
Convert to new m_ext interface. This also fixes a memory leak, since
Use ether_crc32_le and ether_crc32_be.
style and minor cleanup
Use ether_crc32_be.
Forgot to assign ext_buf, resulting in a panic.
libc_r has to provide strong versions of the public symbols to override
if_printf / device_printf.
Use ether_crc32_be / ether_crc32_le.
Use PCI accessor functions instead of messing directly with the config
if_printf / device_printf cleanup.
if_printf / device_printf.
Remove bogus check if interface was already attached. The function is
Use PCI accessor functions instead of messing directly with the
Remove __inline hints, let the compiler figure out the details.
Let fxp_release take a device_t directly and change some device_printf
No need to bzero softc.
cleanup
Convert to new m_ext API.
If we want to abuse an API by providing a callback doing nothing, we
Remove M_EXT_OLD, rename m_ext.ext_nref.new to m_ext.ext_ref and
Don't match executables with ELFOSABI_NONE against the brand list.
Update to file 4.13. Put the contrib files into contrib/file-4 instead
Merge from vendor branch FILE:
Update to file 4.13. Put the contrib files into contrib/file-4 instead
Increment subversion to allow matching for ELF ABI tagging enabled kernels.
Teach file about DragonFly's ELF ABI tagging.
Add back support for SYSV binaries. Those are nasty since
Stop branding DragonFly binaries with the FreeBSD ABI.
Removing rest of debugging code which slipped into the commit.
if_printf / device_printf. remove stored device_t reference as it is
Back out last commit, this wasn't supposed to crep in.
Don't name arguments in prototypes. Nuke __STDC__ conditional.
Don't commit half of a change. Prefix parameter names and local arguments
Use IF_DRAIN.
reorder declarations
Add missing parameter.
Include sys/thread2.h to unbreak build.
Rename label to not collidate with local variable. Conditionalize error,
Expand itjc_bus_setup, it declares variables.
Add a new macro IF_LLSOCKADDR which maps a ifnet pointer to the
Another instance of IF_LLSOCKADDR.
Instead of checking for ifnet_addrs[ifp->index - 1] == NULL to detect
Instead of using ifnet_addrs and following ifa_ifp, use ifindex2ifnet
Add f_owner (user who mounted the filesystem), f_type (filesystem type ID),
Merge the pointer to the link-layer address into ifnet and remove
Make call to arc_ioctl the default case.
Remove splimp in xl_attach, the interrupt is created last and
Replace splimp with critical sections for now. Fix a bug in xl_init,
Always hook the interrupt up last, we don't have to worry about
Use WAITOK allocation, fix some arguments and remove a now unused
Let the system deal with device shutdown, don't do it yourself.
Convert splimp to criticial sections for now. Cleanup the critical
Fix detach order: We have to unhook the interrupt first and leave the
Reorder initialisation to make protection unnecessary.
Fix some bugs in the last commit. We have to call ether_ifdetach if we
device_printf / if_printf and some minor nits.
Use PCI helper functions instead of hand-rolling them. Remove now
Use ether_crc32_be.
Don't bzero the softc, it is already zero.
Use M_WAITOK for contigmalloc now that the attach path is interrupt-save.
Explicitly note that updating from pre-1.2 to PREVIEW or HEAD is not
Convert to critical sections, move timer reset into the protection.
Convert to critical sections. No need to protect the interupt from racing
Add /usr/pkg/etc/rc.d to the rcNG search list.
Reorder critical sections to be as short as possible by moving invariants out.
if_printf / device_printf
Use the PCI helper functions and nuke now unused macros.
Move the callout_reset into the critical section.
Use epic_detach for error cleanup during attach. Make attach interrupt
Remove a useless assignment. Move callout_reset into critical section.
We know that tl_probe is run first, so turn the check for a bad device
Don't bzero softc. Setup interrupt last to get tl_attach ISR race free.
Really use M_WAITOK.
Typo.
Convert from spl* to critical sections.
Remove compat junk.
Convert splimp to critical sections.
Stop abusing splbio simply because others do the same. Use critical
Switch to critical sections, fix some possible minor nits with ISR
- convert to critical sections
- setup interrupt last
Only delete the miibus if it was attached first.
Convert to critical sections.
Convert to critical sections.
Convert to critical sections.
Convert from splhigh to critical sections.
Convert to critical sections. Fix an early return without restoring
Convert to critical sections.
- convert to critical sections
Convert to critical sections. Remove compat junk. Hook up interrupt
Missed one splnet.
Convert to critical sections.
- convert to critical sections
No interrupts, no M_NOWAIT. Use M_WAITOK instead.
Remove compatiblity code.
Convert to critical sections.
Convert to critical sections.
Convert to critical sections.
Convert to critical sections. Add a missing ether_ifdetach when
Convert to critical sections. Use bge_detach as common error path
Convert to critical sections.
Fix build.
Convert to critical sections.
Convert to critical sections.
- convert to critical sections
Use critical sections.
Convert to critical sections.
Convert to critical sections.
- convert to critical sections
Convert to critical sections.
Remove old attempt at locking, it was incomplete and partially incorrect.
Convert to critical sections.
Convert to critical sections.
Nuke code dealing with empty address list or trying to set the link-layer
Enable port and memory-mapped IO in the PCI layer when the associated
device_printf / if_printf.
Use PCI accessor functions.
Use ether_crc32_be.
Nuke compat code.
device_printf / if_printf
Use PCI accessor functions.
Use local storage instead of ac_enaddr in tl_attach.
Use local storage instead of ac_enaddr.
static != extern
device_printf / if_printf
Use PCI accessor functions.
Fix an uninitialised variable I introduced earlier.
Hook up interrupt last. Use ti_detach as common error path.
No interrupts, no M_NOWAIT.
Use local storage instead of ac_enaddr in ti_attach.
Resource allocate now turns on port / memory bit in the PCI command reg,
Convert to critical sections.
Convert to critical sections.
- convert to critical sections
Use PCI accessor functions and nuke the port / memory enabling.
Convert to critical sections.
Convert to critical sections.
Convert to critical sections.
Convert to critical sections.
Add missing breaks.
- convert to critical sections
- convert to critical sections
Convert to critical sections.
Convert to critical sections. Be a bit more conservative for now.
simplify.
Use if_printf for TUNDEBUG.
Convert to critical sections.
Use if_printf.
Convert to critical sections.
Forgotten from the ALTQ conversion.
Convert to critical sections.
Convert to critical sections. Stop updating interrupt masks.
Convert to critical sections.
typo
Nuke compatiblity parts.
Missing ether_ifdetach in error path.
Fix another bunch of missing ether_ifdetach calls.
Convert to critical sections. Drop a splz.
Convert to critical sections. Stop messing with the interrupt masks.
Convert to critical sections.
Convert to critical sections.
Convert to critical sections. Rename a local variable from s to i,
Convert to critical sections.
Convert to critical sections.
Remove now unused variables.
Convert to critical sections.
Convert to critical sections.
Convert to critical sections.
Convert to critical sections.
Let the compiler decide what code should be inlined and what not.
Convert to critical sections.
Convert to critical sections.
Convert to critical sections.
Convert to critical section.
Convert to critical sections.
Merge from vendor branch GCC:
Include the fortran library sources from GCC 3.4.4.
Unhook old g2c from build and start building the GCC 3.4.4 version.
Remove old libf2c.
Fix some glitches.
Also include files in the .OBJDIR.
- initialise interface name early
Use PCI accessor functions. Stop changing port / memory bits manually.
Use ether_crc32_be.
Use vr_detach as common error path.
- pass dev directly to txp_release_resources, call it only once
Use PCI accessor functions, don't change memory / port bit manually.
Use queue(3).
Use ether_crc32_be.
Use local storage for MAC instead of arpcom.ac_enaddr.
Explicitly set error before jumping to fail.
Common PCI probe style.
Setup interrupt last in txp_attach. Protect against concurrent interrupts
Use queue(3) macros for if_multiaddrs.
queue(3) for if_multiaddrs.
Restore copy of revid in softc.
Add BCM5751.
Regen.
Add BCM5751.
SuSE includes a special .note.SuSE section descriping the version of
Defancy the infinite loop. Fix a bug where the loop wasn't left when
Instead of emulating a userland system call via stackgap, use
Remove redundant verbosity. The description of the parent just costs space
Hack in support for ar(4) based devices.
Use if_printf, especially for DPRINTF. Merge two DPRINTF statements
Remove now unnecessary messing with PCI command register.
sort
don't include regression tests, we don't use them anyway. more than enough
Tear down interrupt in wi_free when necessary.
Add PCI ID for BCM4401-B0.
Regen.
Add support for Broadcom BCM4401-B0.
Push device_t's down to vxattach, not the softc.
Remove some unused macros. Constify. Minor style changes.
Remove a bunch of breaks after return, merge a return into a default case.
Fix some stupid style bugs. Call bus_setup_intr after vxattach,
Slightly chance the order of interrupt handling. First hook the
Deorbit Alpha.
Use local storage for MAC address.
Use M_ASSERTPKTHDR. (Obtained-from: FreeBSD)
Allow inclusion of Citrus modules in statically linked binaries.
Declare module for mapper_parallel explicitly to allow static linking.
Sync with NetBSD:
Sync with NetBSD:
Update to file-4.14. Remove merged patches.
Merge from vendor branch FILE:
Update to file-4.14. Remove merged patches.
Add forgotten wcstoull.
Remove explicit int casts for the array index. While it doesn't
Sync with master copy. This is necessary to allow installworld from
Use pcidevs. Use common probe style and PCI helper functions.
It's dead, Jim.
Convert to NEWBUS, remove ISA compat.
Allow disabling of unused parameter warnings. This is handy for third
Allow radio and statistic dump to use arbitrary interface names.
Merge from vendor branch OPENPAM:
Import OpenPAM Figwort.
Add /usr/lib/security, to be used for PAM modules.
Build the PAM modules under lib/pam_module, since they are not
Catch up with Perl version.
Revert part of the ALTQ conversion. It unintenionally removed code which
Import current pam_tacplus from FreeBSD HEAD.
Also install the pam_tacplus(8) man page.
Import current pam_opie(8) from FreeBSD HEAD.
DPADD is currently broken since LIBDIR != /usr/lib.
Import current pam_krb5 from FreeBSD HEAD.
Import current pam_opieaccess from FreeBSD HEAD.
Import current pam_radius from FreeBSD HEAD.
Import current pam_ssh from FreeBSD HEAD.
Hook up remaining PAM modules. Don't use DPADD for now, it's conflicting
Switch to OpenPAM. The PAM modules are now installed in /usr/lib/security
Define struct in_addr in both, arpa/inet.h and netinet/in.h,
A lot of software depends on netinet/in.h being self-contained,
Prepare for moving from /etc/pam.conf to /etc/pam.d.
Remove unused junk.
Change to common PCI probe style.
Use ether_crc32_be.
Use if_printf most of the time and remove the device_t stored in softc.
Move PCCARD attachment into separate file. Use the NEWCARD helper
Call bus_setup_intr in xe_attach, not xe_activate. This prevents
PAM is dead, long live PAM!
fix typo, SRCS should contain the .c file.
Prepare for using the "official" PAM support.
Merge from vendor branch OPENSSH:
Prepare for using the "official" PAM support.
MODULE_DIR must include the final '/'.
pam_skey is not supported anymore, remove it from the default config.
Instead of duplicating the Kerberos tools, use a single version.
Switch to auth-pam.c from OpenSSH to unbreak Kerberos 5 build.
Fix missing initialisation of big numbers. BN_hex2bn behaves
de-nop part introduced in the last commit.
Remove the minigzip example from libz. Since the real gzip is based on
Switch to zlib 1.2.3, imported under contrib/zlib-1.2 to reduce impact
Switch to zlib 1.2.3, imported under contrib/zlib-1.2 to reduce impact
Merge from vendor branch ZLIB:
Mark zlib 1.2.2 as dead.
Some more junk to remove after manual installation.
Add emulation of statvfs and fstatvfs based on statfs / fstatfs.
Split monolithic /etc/pam.conf into separate files for each service
pthread_self is used by stdio.
Add shlock(1), a program for the safe creation of lock files from shell
First step to cleaning up stdio. This breaks the libc ABI, all programs
Merge __sFILEX into __FILE. Let __fpending handle the ungetc buffer
Remove partial NetBSD support. It's pointless to have an emulation of
regen.
Fix C++.
Just threat all ELF dynamic objects as shared libraries, don't verify
Add description of FPU for floating point conversion functions.
Add most parts of the wide char stdio support. This is not hooked up
GC unused header from BSD libm.
Unhook rc.d/network, it should be unused.
Add support for HP hn210e usb ethernet.
Install pam.d's README and convert.sh as part of upgrade_etc.
Don't panic. Also races in the attach path.
Drop GCC < 1 support, simplify inline assembly and use proper #error for
Add a new feature-test macro __DF_VISIBLE for those functions we want to
Simplify conditional by making use of __DF_VISIBLE and __ISO_C_VISIBLE.
FreeBSD and NetBSD both use derivates of Sun's math library. On FreeBSD,
Fix typo.
Merge revision 1.38 from FreeBSD:
cvtstat doesn't exist.
Factor out an_detach, since the implementation for all busses is
Remove useless .PATH entries inherited from FreeBSD.
Use pcidevs and common PCI probe style.
Remove unnecessary initialisations. Return ENXIO instead of 1 in
Sync with FreeBSD(if_anreg.h 1.1.2.9, if_an.c 1.2.2.14 and
Sync with FreeBSD(if_an.c 1.2.2.15, if_aironet_ieee.h 1.1.2.9)
While setting up a transmit packet disable interupts on the card then
Remove break after return.
Eliminate USEGNUDIR and instead allow the target directory to be
Instead of overriding LIBDIR, override the new TARGET_LIBDIR variables.
Use TARGET_LIBDIR variables instead of overriding LIBDIR.
Fresh installations don't have libtelnet installed and older FreeBSD
Restore NCURSES_CONST constness.
Stop installing profiling libraries as /usr/lib/lib${LIB}_p.a, because
Partially backout last commit. /usr/lib is added by GCC anyway and the
Since we have a table for each device anyway, we can also use it to
GC unused variable.
Move ostat definition from sys/stat.h into emulation43bsd/stat.h.
Regen.
Fix macro name in comment.
Add passwd manipulation code based on parts of vipw and chpass.
Import pam_deny, pam_permit and pam_unix from FreeBSD, use them instead
Import pam_deny, pam_permit and pam_unix from FreeBSD, use them instead
Merge from vendor branch OPENPAM:
Add new option COMPAT_DF12, used for ABI compatibility with DragonFly
libutil.h hsa to go after pwd.h.
Instead of always defining FLOATING_POINT to get floating point, define
Constify xdigs argument to __u[lj]toa. Cast u_long and u_quad_t to
Add missing bit of the wchar stdio support and hook the whole thing in.
Make nlink_t 32bit and ino_t 64bit. Implement the old syscall numbers
Bump version to 1.3.4 for stat changes.
Require 1.3.4 before installworld.
Enforce COMPAT_DF12 for now, this can be overriden via NO_COMPAT_DF12
Regen.
Readd fix for FreeBSD PR/30631.
Catch up with reality, this is GCC 3.4.4.
Clean up search directories to what we really use.
Really support rpath only linking. Add a special option -nolibc which
Remove unused include of sys/dirent.h.
Add SYSCTL_NODE_CHILDREN.
Add rman_set_device.
ANSIfy.
Don't depend on struct dirent == struct direct, but fully separate the
Remove redundant assignment.
Add ethernet port of JVC MP-PRX1.
Move up CVS IDs, first must be DragonFly.
Reimport devlist2h.awk from FreeBSD, this version actually works. It
_PC_NAME_MAX is NAME_MAX, so use that for the storage allocation as
Instead of trying to compute the local storage based on maximum entry
Instead of MAXNAMELEN, use NAME_MAX for now. This should be revisited
Don't check for zero-length direntries, expect the system to handle
Don't match entries by hand, just use strcmp. It is efficient enough for
MAXNAMELEN ==> PATH_MAX
Both file and dp->d_name are NUL-terminated, so it pointless to first
Use NAME_MAX instead of MAXNAMLEN and strlcpy, since dp->d_name is
Match "." and ".." with strcmp.
Make it actually compile without warnings.
Use NAME_MAX instead of MAXNAMELEN, replace a strncpy with strlcpy.
Use NAME_MAX instead of MAXNAMLEN. Cast ino_t to uint64_t + proper
Add vn_get_namelen to simplify correct emulation of statfs with maximum
Use vn_get_namelen to reduce bogusnes.
Kill stackgap in (f)statvfs(64).
Don't hide errors from kern_statfs by overwriting error, check if it was
Use vn_get_namelen to provide correct f_namemax field.
fsfind should use direct here, not dirent.
When allocating memory for the index file, query the filesystem for the
Just expect either the kernel or libc to drop empty dirents.
Pass the direction to kern_getdirentries, it will be used by the
- Propagate error code from various bus_dma functions in bfe_dma_alloc.
Revive multicast support, it got lost in the initial import.
Set both, CRC32 generation and LED modes. Clear powerdown control bit.
If possible, use builtin constants for newer GCC versions, but fallback
Add wcsftime(3).
Sync with recent libc and libm changes.
Add vop_write_dirent helper functions, which isolates the caller from
Set baudrate to 100Mbps and advertise VLAN.
Break long commits. Don't ask for transfers when there's nothing to
Merge from vendor branch GCC:
Update GCC 3.4 to current 3.4.5 pre-release.
Update GCC 3.4 to current 3.4.5 pre-release.
Use new vop_write_dirent function.
Use vop_write_dirent.
Add _DIRENT_NEXT, which is for now only used in the kernel to skip to
Replace the 4.3BSD getdirentries compat function with something which is
Use vop_write_dirent. Allocate a temporary buffer for the name for now,
Fix merge bug. d_namlen is used by GENERIC_DIRSIZ, when it isn't
We want to separate dirent and the maximum directory entry size.
Rip off PROCFS_ZOMBIE, it wasn't even a knob to play with.
Split the two parts of linprocfs_readdir into subroutines.
Split the two parts of procfs_readdir into subroutines.
Honor process visibility for jailed processes and ps_showallprocs for
Sprinkle some const.
Convert to vop_write_dirent.
Utilize vop_write_dirent. Slightly change the order by writing to
Use vop_write_dirent. Correctly handle the case of spare fd tables, e.g.
Improve C++ support.
Convert to vop_write_dirent.
Add some new Broadcom IDs.
Regen.
5705K, 5714C, 5721, 5750, 5750M, 5751M, 5789 support.
Atomically load and clear the status block. This makes the bge
Enable the memory arbiter before turning off the PXE restart. This
Prevent spurious link state changes.
HPFS != UFS, so use the right constant for directory entries. Check
Create a kernel option BGE_FAKE_AUTONEG for IBM/Intel blade servers,
GC unused macro.
Also document BGE_FAKE_AUTONEG in LINT.
Don't assume that ttys are always located directly in /dev. This
Also allocate PATH_MAX for the threaded case.
sendmail tried to limit directory names to MAXPATHLEN - MAXNAMLEN in an
Back out accidental commit.
We have to copy the pam.d entries after running mtree, otherwise
Don't define infinite macros when we want to define infinity.
No need to forget wcswidth.
Make struct dirent contain a full 64bit inode. Allow more than 255 byte
Sync GCC 3.4's propolice with Etoh's official version. This fixes the
Remove space before '('. Correctly align some function names. No need to
Smoke something else and revert the use of ssize_t, I've put it there
Add real function versions of the _unlocked family. Use putc for putchar
Retire old sendmail.
Don't add files patches via *.no_obj.patch back to SRCS, make them a
Update __DragonFly_version as well. Don't let this slip, please.
GC openssh-3.9
Sync with FreeBSD HEAD.
Add strnvis, which is orthogonal to strvisx by bounding dst, not src.
Welcome OpenSSH 4.2.
Merge from vendor branch OPENSSH:
Welcome OpenSSH 4.2.
Add some words about PAM.
Add /usr/pkg/info to info search path for pkgsrc users.
Move the modification of hspace out of the realloc call to make this a
When testing whether a negative delta is smaller than the default
Call resettodr on shutdown to ensure that RTC and wall clock gets
Too many windhills in the third party software world to fight them all.
Avoid text-relocations, binutils don't like them on AMD64.
Add rcng script to start ftpd stand-alone.
Add an option -H to override gethostname.
Also install the new ftpd script.
Add missing atoll.
Fix a very, very old bug in the man page, lbl-csam.arpa doesn't
Make the syslog message string static const, saves the copy.
Don't depend on the DragonFly keyword anymore, all unsupported scripts
Don't depend on DragonFly keyword for shutdown scripts as well.
Honor NOFSCHG.
Set POSIX feature test macros to the correct value as mendated by SUS.
toascii(3) got lost during ctype conversion.
Revert last change and add toascii.c to the right variable.
Rework nrelease framework for pkgsrc. The way the bootstrap kit was
Add g77 link.
Add g77 here as well.
Add a simple library for the default g77 linkage, providing main.
Provide proper offsetof macro for C++. Prefer __size_t here, since
Enable wchar usage.
Remove useless define.
nextpid is not public, so don't use it. It's not really useful anyway,
Sleep before commit, remove trailing ,
Honour NOSFCHG for the kernel installation as well, allowing to install
Allow static compilation of linprocfs using a LINPROCFS option.
Further cleanup of GCC's builtin directory list in an attempt to
When using visibility macros, sys/cdefs.h must be included first.
Merge OpenBSD r1.104:
Add a fix for CAN-2005-3001, via pkgsrc from Ubunto.
Sync vfscanf with FreeBSD, which makes it almost symmetrical to
pam_unix.so needs -lutil. Fixes qpopper issues as seen by Adrian Nida.
Fix initialisation of wide char support in FILE. Problem reported
Apply post-install correction of +CONTEXT files. nrelease now also
Be more jail friendly:
Add missing callout_init.
Improve portability of patch(1):
Merge bug fix from binutils main line:
Introduce a new variable to hold the size of buf.
Use localhost instead of 127.0.0.1, since IPv6 should work as well.
Teach kdump a handy new trick: -p $pid selects the records of
Include unistd.h to get isatty(). Has been lurking in my release tree
Don't depend on POSIX namespace pollution with u_char from sys/types.h.
Add wide char support for printf and friends. Fix a possible
Fix format string to process all arguments. Noticed by Trevor Kendall
Merge rev 1.96 of NetBSD's net/if_spppsubr.c:
Move atomic_intr_t to machine/stdint.h and predent __ to reduce
nullfs_subr doesn't exist anymore.
Ensure that exit_group actually returns a sane value, not some random
Add some more PCI IDs for the SATA300 controllers of newer Intel boards.
Fix buffer overflow in config parser.
Don't try to guess if the hunk was already applied if it doesn't have
Check, complain about and accumulate errors when writing reject files
Fix one inversed condition.
Remove bootstrap code -- you should really have stdint.h by now.
Just remove the README file, noone bothers with it anyway.
_PATH_MKDIR is unused, nuke it.
Make comment more readable by putting the copyright on a separate line.
Merge revision 1.21 and 1.22 from OpenBSD. Add a missing "the" to the
Fix typos in copyright.

Justin C. Sherrill (14):
Sync with FreeBSD 4-STABLE manpages.
New logo of the DragonFly logo instead of the BSD Beastie. The old
Added short descriptions for kern.acct_suspend, kern.acct_resume, and
Added mention of using MAKEDEV for when the target slice for installation
whereis will find a given application in various locations, including
Remove my changes. PATH_PORTS is not checked for multiple entries as
A cvsup file that pulls the "checked out" version of source. I'm referencing
Removed freebsd.mc line, added ports/arabic line as found in the FreeBSD
Testing auth forwarding. Me am idiot!
Testing commits.
Test commit of new file.
Create DragonFly fortune files, based on either the FreeBSD fortune files
This commit sets new users to see the DragonFly-tips fortunes instead
Making dntpd server list match the recommendations at:

Liam J. Foy (190):
Test commit! Thanks eirikn!
-Remove main proto
- Remove main() proto
- Remove registers
- Fix some sentences
-Setmode will deal with the octals, so we dont need to.
- Add $DragonFly$
- Bump WARNS to 6
- Remove unnecessary (void) casts
Add special case for the german whois nameserver. Without the
-Set WARNS to 6
- Remove all registers in bin/. This saves us doing it each time we come
Fix cpdup man page. The option -v[vvv] should be -v[vv].
Whoops! setmode(3) returns a void *, not mode_t *.
The gr_gid member of the struct group should be gid_t
Needed to add sys/types.h for previous commit. Spotted by Mr Dillon.
- Allow ipcs to display information about IPC mechanisms owned by
- Remove unnecessary selection statement to see if buf == NULL.
Sync with FreeBSD rm(1) with my modifications. Some things have been
- Remove (void) casts that are not necessary.
- Remove space
Small patch to fix the output of apm. It called print_batt_life twice
- Remove main proto
- Remove sys/time.h
- Set the clnt error functions protos from char * -> const char *
- Set the clnt error functions 's' argument to const char *
- Use the correct error functions
- Add $DragonFly$ tag
- Add WARNS?= 6
- Import newgrp(1) from FreeBSD
- Add newgrp(1) into the build
- Add new -vv option. If the -v flag is specified more than once,
- Remove unnecessary casts
More cleans
- Fix setmode. Setmode can fail due to malloc. We should print the correct
- Sync with FreeBSD
- Constify most of the function
- Remove unnecessary casts
- Remove registers from rcorder
- strings.h -> string.h
- Remove signal.h
- We should always check the setenv() call. Why? because it uses
- Check the return value of setenv(). We should check this value since
- Whoops! I missed some from the previous commit. My bad, sorry.
- We should warn with argv[i] not argv[1]
- Remove both errno.h and string.h. Both are unnecessary.
- Restructure the code
- Complete re-write/re-structure of rev(1). The previous code was ugly!
- Add $DragonFly$ tag
- Change manpage to show the change from M_ -> MB_
- Deregister strlcpy
- Remove unneccessary cast
- Add WARNS?= 6
- Make *user, *group and *groupset local, constify them and initialise
- Detect and exit on write errors (from FreeBSD, but our code)
Sync daemon(8) with FreeBSD:
- string.h is unnecessary, remove it
-Setmode can also be caused to fail because of malloc().
- Remove *ttyname()
- Static functions/globals
- Clean sbin/dump of all registers.
- Both <unistd.h> and <sys/types.h> can go - unnecessary
- Make modestr const and initialize it (stop GCC moaning)
Equivalent to: FreeBSD rev 1.82
- Bump WARNS to 6
- Remove duplicate headers
- Static functions/variables
- UNIX conformance: If -r -f on non-existent directory, don't emit error.
- Remove both sys/wait.h & sys/types.h - unnecessary
- Add -k option for whois.krnic.net (hold details of IP address
- Add WARNS 6
- Add WARNS 6 and make WARNS 6 clean
- Remove unnecessary headers
- Clean up wc (remove (void) etc)
- Complete re-write of sasc.
- Remove a duplicate unnecessary check on errno. strtol(3) will only ever set
General clean up of lptcontrol. Similar to FreeBSD 1.15
- Add -v option (verbose). Taken from FreeBSD with my modifications.
Minor Patch
- Initial import of the battd utility.
- Tie battd into the build
- Do not depend on stat_warn_cont when executing commands.
- Large cleanup/changes to man page.
- Improve how we handle the APM device. Much cleaner code.
- Reduce the size of msg (1024 -> 80)
- Remove unnecessary header (time.h)
- Remove mode argument from open().
- Remove two unnecessary headers
More cleaning:
- Improve the debugging functionality in battd.
- Fix some spelling mistakes.
- Fix a few grammar 'nits'.
- Remove unnecessary headers (string.h, sys/uio.h, utmp.h)
- Use strcasecmp() for upper and lower cases.
- isupper(x) can be true for x in [128..255], but since tolower(x) is
- Better handling of a given(-p) invalid port number.
- Add battd rcng script
- Add battd to rc.conf(5)
- Remove unnecessary header
- Remove unnecessary casts
- SUSV3 states on error we should return > 1
- Fix some xdrproc_t warnings (FreeBSD)
- Remove unnecessary headers
- The maximum value for a time_t is assumed to be <= INT_MAX, not >= INT_MAX
- Remove an unnecessary call to stat(). We can just use lstat before
- Static
- Back out rev 1.6 noticed by Matt Dillon. However, still keep the use
- Use variable 'len' instead of making an unnecessary call to strlen()
- Make sure we free() 'modep' after setmode()
- Back out part of previous commit. I thought free() was the prefered
- Remove unnecessary header
- Write status information to stdout instead of stderr.
- perror() -> warn()
- Update mesg(1) to confirm to SUSv3 changes previously committed.
- Remove unnecessary header
Minor patch:
- Validate the -i option (Idea from FreeBSD)
- Correct use of the err() family (err() -> errx())
- Validate -i option
- Good english
- Sync the example with the correct sentense.
Clean up:
- Remove sys/stat.h, sys/signal.h, varargs.h and ctype.h
- Correct usage of gethostname(3)
- Make sure we call exit() after clnt_pcreateerror()
- Style(9)
- sysctl(3), sethostname(3) and malloc(3) set errno, use it!
- Remove unused variable
- WARNS 6
- Correct usage of gethostname(3)
- warn() -> warnx(). The global variable errno will not be set.
- Cosmetic
- Clearly state that errno is set
- WARNS -> WARNS?
- Let receiver know if the sender accepts replies
- Use errno to report why freopen(3) failed
- Check return value of setenv(3)
- Add further functionality to check for invalid characters
- ANSI
- ANSI
- Improve option handling
- sprintf -> snprintf
- Use socklen_t
- Add missing include <stdlib.h> for exit()
- WARNS 6
- When calling syslog(3), use %m for errno and lose strerror(3)
- Use socklen_t
- Produce more informative output to syslog. Generally improve/clean error handling
- The AC Line state can be returned as 2 meaning 'backup power'. The previous
- Actually make the debug code work
- WITHOUT_ERRNO should be 0, silly mistake.
- More fixes for the debug code
- util.h -> libutil.h
- Use pidfile(3). The pid file can be used as a quick reference if the process
- Fix memory leak if realloc(3) failed. Use reallocf(3).
- Minor restructure. Don't bother calling makemsg() unless we can create
- Kill keyword 'register'
- int -> pid_t for pid
- We should use inet_ntoa which returns an ASCII string of representing the
- Small message
- Make rwhod now use poll(2) instead of signals
- Document mined(1) key bindings
- Introduce new -g option. This allows for the broadcast time to be reduced
- Whoops, some test code slipped in. Remove it
- Don't role our own - use err(3)
- Fix return values to conform with SUSv3
- Use MAXPATHLEN
- Don't flag NIS entries at invalid
- If at least one call to onehost() fails, return 1
It's actually 11 minutes when the machine is assumed to be down and removed
- State default
- In the ICMP debug code, use %d over %x. This makes it much easier when
Remove the hack that varied the first character of the output file name
- Do not allocate memory to entries in /var/rwho which have been down for
Fix a bug introduced which causes chkgrp to coredump on certain group file
- Use pidfilei(3) to write a pid file in /var/run
- Don't write our own pid file, just use pidfile()
Remove undocumented historic support for treating "-" as an option
Include the option-arguments in the description and remove a non-existant
Minor typo
This program requires at least one option. Clean up the checking and report
Remove the sleeper crap and use MAXPATHLEN
Kill unused variables
perror() -> err()
Zap unused variable!
K&R style function removal. Update functions to ANSI style. Also a few
Don't declare a struct just for sizeof(). Just use sizeof().
- Check for execvp returning ENOTDIR
- Dont pass a complete interface struct to icmp_error, just pass the mtu.
- Update to new icmp_error signature

Matthew Dillon (3856):
Initial import from FreeBSD RELENG_4:
Add the DragonFly cvs id and perform general cleanups on cvs/rcs/sccs ids. Most
Add the DragonFly cvs id and perform general cleanups on cvs/rcs/sccs ids. Most
Remove UUCP support. Note: /usr/src/gnu/libexec/uucp and /usr/src/libexec/uucpd
Retarget makesyscalls.sh from FreeBSD to TurtleBSD
thread stage 1: convert curproc to curthread, embed struct thread in proc.
thread stage 2: convert npxproc to npxthread.
thread stage 3: create independant thread structure, unembed from proc.
thread stage 4: remove curpcb, use td_pcb reference instead. Move the pcb
Oops commit the thread.h file.
remove unused variable.
get rid of (incorrect) gcc warning.
Add missing prototype (fixes warning).
thread stage 5: Separate the inline functions out of sys/buf.h, creating
thread stage 6: Move thread stack management from the proc structure to
thread stage 7: Implement basic LWKTs, use a straight round-robin model for
thread stage 8: add crit_enter(), per-thread cpl handling, fix deferred
rename td_token to td_xtoken to deal with conflict against sys/thread.h
Add kern/lwkt_rwlock.c -- reader/writer locks. Clean up the process exit &
thread stage 10: (note stage 9 was the kern/lwkt_rwlock commit). Cleanup
Add parens for code readability (no functional change)
Finish migrating the cpl into the thread structure.
minor code optimization.
proc->thread stage 1: change kproc_*() API to take and return threads. Note:
Optimize lwkt_rwlock.c a bit
proc->thread stage 2: MAJOR revamping of system calls, ucred, jail API,
proc->thread stage2: post-commit fixes/cleanup
proc->thread stage2: post-commit fixes/cleanup(2)
proc->thread stage3: make time accounting threads based and rework it for
proc->thread stage 3: synchronize ps, top, and libkvm, and add some convenience
proc->thread stage 3.5: Add an IO_CORE flag so coda doesn't have to dig
proc->thread stage 4: rework the VFS and DEVICE subsystems to take thread
proc->thread stage 4: post commit, introduce sys/file2.h. As with other header
proc->thread stage 4: post commit cleanup. Fix minor issues when recompiling
Introduce cratom(), remove crcopy().
proc->thread stage 5: BUF/VFS clearance! Remove the ucred argument from
simple cleanups (removal of ancient macros)
cleanup
cleanup some odd uses of curproc. Remove PHOLD/PRELE around physical I/O
proc->thread stage 6: kernel threads now create processless LWKT threads.
Cleanup lwkt threads a bit, change the exit/reap interlock.
go back to using gd_cpuid instead of gd_cpu.
smp/up collapse stage 1 of 2: Make UP use the globaldata structure the same
Give ps access to a process's thread structure.
smp/up collapse stage 2 of 2: cleanup the globaldata structure, cleanup
format cleanup for readability. Tab out back-slashes.
threaded interrupts 1: Rewrite the ICU interrupt code, splz, and doreti code.
Implement interrupt thread preemption + minor cleanup.
Add 64 bit display output support to sysctl plus convenient macros.
Misc interrupts/LWKT 1/2: interlock the idle thread. Put execution of
Misc interrupts/LWKT 1/2: threaded interrupts 2: Major work on the
Fix a race in sysctl_out_proc() vs copyout() that could crash the kernel.
Add threads to the process-retrieval sysctls so they show up in top, ps, etc.
Sync userland up with the kernel. This primarily adjusts ps, etc to handle
Operations on the user scheduler must be inside a critical section (fixes
For the moment uio_td may sometimes be NULL -> nfsm_request -> nfs_request ->
Enhance debugging (sync before MP work).
Remove pre-ELF underscore prefix and asnames macro hacks.
fix unused variable warning
Add some temporary debugging.
Split the struct vmmeter cnt structure into a global vmstats structure and
properly initialize ncpus for UP
Split the struct vmmeter cnt structure into a global vmstats structure and
Split the struct vmmeter cnt structure into a global vmstats structure and
misc quota proc->thread
Misc interrupts/LWKT 2/2: Fix a reentrancy issue with thread repriortization
Document hardwired indexes for fields.
add gd_other_cpus
Generic MP rollup work.
Integrate the interrupt related operations for /dev/random support
fix a bug in the exit td_switch function, curthread was not always being
Remove td_proc dependancy on cred that is no longer used.
oops, forgot one. Remove another curproc/cred dependancy
MP Implementation 1/2: Get the APIC code working again, sweetly integrate the
MP Implementation 2/4: Implement a poor-man's IPI messaging subsystem,
MP Implmentation 2A/4: Post commit cleanup, fix missing token releases that
The syncer is not a process any more, deal with it as a thread.
MP Implmentation 3/4: MAJOR progress on SMP, full userland MP is now working!
MP Implmentation 3A/4: Cleanup MP lock operations to allow the MP lock to
MP Implmentation 3A/4: Fix stupid bug introduced in last commit.
MP Implmentation 3B/4: Remove Xcpuast and Xforward_irq, replacing them
MP Implmentation 4/4: Final cleanup for this stage. Deal with a race
misc cleanup. Add a case where we don't want an idlethread to HLT (if there
Cleanup hardclock() and statclock(), making them work properly even though
Fix overflow in delta percentage calculation due to the fact that our 32 bit
Make the cpu/stat display work properly again.
account for the time array being in microseconds now, and allow the
partially fix pctcpu and userland rescheduling. We really have to distribute
Forward FAST interrupts to the MP lock holder + minor fixes.
Collapse gd_astpending and gd_reqpri together into gd_reqflags. gd_reqflags
Make the pigs display more meaningful by showing processes which haven't
oops, fix bug in last commit, and adjust the p_slptime check.
Add missing required '*' in indirect jmp (fix assembler warning).
Fix minor buildworld issues, mainly #include file dependancies and fields
Fix minor compile warning.
The comment was wrong, ptmmap *is* used, put it back in (fix crash
GDB changes required for gdb -k kernel /dev/mem. Still selected by
Make the kernel load properly recognize ABS symbols (.SET assembly
Remove an unnecessary cli that was causing 'trap 12 with interrupts disabled'
Nuke huge mbuf macros stage 1/2: Remove massive inline mbuf macros to reduce
Remove the priority part of the priority|flags argument to tsleep(). Only
Nuke huge mbuf macros stage 2/2: Cleanup the MCL*() cluster inlines by
Remove references to the no longer existant PZERO.
zfreei->zfree (there is no zfreei anymore)
This is the initial implmentation of the LWKT messaging infrastructure.
Add externs for *_nonlocked atomic extensions to avoid warning.
Profiling cleanup 1/2: fix crashes (all registers need to be left intact
doreti was improperly clearing the entire gd_reqflags field when, in
Minor cleanups so GENERIC compiles. Fix underscores in assembly and an
Fix a minor compile-time errors when INVARIANTS is not defined.
DEV messaging stage 1/4: Rearrange struct cdevsw and add a message port
DEV messaging stage 1/4: Rearrange struct cdevsw and add a message port
LINT build test. Aggregated source code adjustments to bring most of the
Throw better sanity checks into vfs_hang_addrlist() for argp->ex_addrlen
DEV messaging stage 2/4: In this stage all DEV commands are now being
Remove two unnecessary volatile qualifications.
LINT pass. Cleanup missed proc->thread conversions and get rid of warnings.
LINT cleanup, add a static function back in which I thought wasn't used.
Merge from FreeBSD 2003/07/15 15:49:53 PDT commit to sys/netinet.
2003-07-22 Hiten Pandya <hmp@nxad.com>
Here is an implementation of Limited Transmit (RFC 3042) for DragonFlyBSD.
Add all_sysent target to /usr/src/sys/Makefile to rebuild syscalls.
Preliminary syscall messaging work. Adjust all <syscall>_args structures
Bring RCNG in from 5.x and adjust config files and scripts accordingly.
Fix some stub prototypes (some missed proc->thread conversions).
Have MFS register a device as a VCHR instead of VBLK, fixing a panic.
Fix NULL td crash in net/if.c when detaching a net interface.
libcr copy: Retarget build paths from ../libc to ../libcr and retarget
Syscall messaging work 2: Continue with the implementation of sendsys(),
Regenerate all system calls
Oops, need to update truss's conf files re: new syscall support header
Fix a minor bug in lwkt_init_thread() (the thread was being added to the
Remove thread->td_cpu. thread->td_gd (which points to the globaldata
Performance cleanup. Greatly reduce the number of %fs prefixed globaldata
Fix __asm syntax error from previous commit.
Fix minor bug in last commit, add the required register keyword back in
syscall messaging 2: Change the standard return value storage for system
Fix the msgsys(), semsys(), and shmsys() syscalls which were broken by the
add missing lwkt_msg to manually specified syscall args structure.
fileops messaging stage 1: add port and feature mask to struct fileops and
Remove KKASSERT(p)'s, 'p' (curproc) no longer exists in this functions.
syscall messaging 3: Expand the 'header' that goes in front of the syscall
LINT synchronization, remove remaining manually defined MIN macros.
Turn off CHECK_POINTS debugging.
Clear the BSS earlier in the boot sequence and clean up the comments. This
Fix a pointer bug introduced by syscall messaging. semget() should work
Cleanup remaining tsleep priority issues.
Remove NOSECURE which no longer serves a purpose. Note: FreeBSD also removed
Explicitly use an unsigned index for 'which' in shmsys(), msgsys(), and
Fix a compile time error. Rename MAX to RAINMAX and get rid of all the
kernel tree reorganization stage 1: Major cvs repository work (not logged as
kernel tree reorganization stage 1: Major cvs repository work (not logged as
kernel tree reorganization stage 1: Major cvs repository work (not logged as
kernel tree reorganization stage 1: Major cvs repository work (not logged as
kernel tree reorganization stage 1a: remove -I- for kmod build
Set _THREAD_SAFE preprocessor symbol when the -pthread option is used,
Kernel tree reorganization stage 2: Major cvs repository work.
Fix the buildkernel target.
Add a few missing cratom() calls. In particular the code in kern_exec()
Add the 'wmake' script to /usr/bin and wmake support to Makefile.inc1.
add cvs-site dist (The layout for the dragonflybsd.org web site).
Add src/test and throw in some of the ad-hoc timing and testing programs
Syscall messaging 4: Further expand the kernel-version of the syscall message.
Syscall messaging 4a: LINT build.
Remove FBSDID, move $ rcs id back to a comment field.
Add softlinks so ports like sysutils/ffsrecov will compile. Eventually
Make modules work again part 1: Rename netgraph's Makefile.module to
Make modules work again part 1: wire up the module build for bus/
Make modules work again part 1: linkup emulation/ and change the architecture
Make modules work again part 1: linkup "net" and rename Makefile.module files
Make modules work again part 1: linkup vfs, rename Makefile.module files,
Make modules work again part 1: linkup crypto.
Make modules work again part 1: forgot a file in 'emulation'.
master makefile for netproto modules, of which there is exactly one
Standardize the syscall generation target to 'sysent'.
Make modules work again part 1: hook up 'dev'. Note that not all devices
Fix compile-time error when compiling a profiled GENERIC.
Make modules work again part 2 (final): Link the module build back into the
Add missing Makefile.modules (forgot to add it in the last commit)
The make release process tries to stat/open a non-existant device, which
Linux needs %edx to be 0 on entry. It registers it as an atexit function if
Linux needs %edx to be 0 on entry. It registers it as an atexit function if
Set SYSDIR properly for sys/crypto and deep sys/dev/disk/... drivers by
Silence warnings about _POSIX_C_SOURCE not being defined. This occured when
Fix minor bug in last commit.
Temporarily go back to an absolute path for signal.h's #inclusion of trap.h
Separate system call generation out from the Makefile so kernel and
Modify the header dependancy ordering so the bsd.kmod.mk/bsd.prog.mk
Adjust an #include line to fix a build error.
Reintegrate the module build for usb.ko
Allow a NULL dev to be passed to _devsw(). This should close any remaining
Separate out userland copyin/copyout operations from the core syscall code
Add an alignment feature to vm_map_findspace(). This feature will be used
Remove now unused proc p variables.
Add the 'eatmem' utility. eatmem is a memory stressor.
Add the NO_KMEM_MAP kernel configuration option. This is a temporary option
Remove unnecessary KKASSERT(p) checks for procedures which do not need
Remove additional KKASSERT(p)'s that we do not need (fixes lint)
SLAB ALLOCATOR Stage 1. This brings in a slab allocator written from scratch
A significant number of applications need access to kernel data
Adjust our kvm based utilities to use the _KERNEL_STRUCTURES define
Cleanup some debugging junk and fix a bug in the M_ZERO optimization code.
oops. Forgot a commit.
At Jeffrey Hsu's suggestion (who follows USENIX papers far more closely the
Fix sendfile() bug introduced by the message passing code. The call to
req->r_td can be NULL, remove KKASSERT() and fix check.
cleanup: remove register keyword, ANSIze procedure arguments.
Do a bit of Ansification, add some pmap assertions to catch the
General cleanup, ANSIfication, and documentation. Prep work for the VFS
Adding missing '|' from last commit.
Make bind1() and accept1() non-static for linux emulator use.
Adjust linux emulation calls for bind and listen to use the new broken-out
Make bind1() and accept1() non-static for linux emulator use.
Convert DIRECTIO code to DragonFly.
Add DIRECTIO to LINT
Cleanup some broken #include's for VESA.
Add the relpath utility which /usr/share/mk/bsd.port.mk will use to
Add bsd.dport.mk, which Makefile's in /usr/dports will use instead of
add relpath to the build Makefile
Add UPDATING note on /usr/dports
rename /usr/dports to /usr/dfports. Includes some CVS surgery.
rename dport -> dfport
Add a cvsup tag, cvs-dfports
Cleanup. Remove unused variable.
Change argument from proc to td.
Remove unused thread pointers.
Permanently fix the 'allocating low physmem in early boot' problem which
update version to match FreeBSD after buffer security fix.
Additional comments: ssh may attempt to zero and free the buffer from
Fix a DIAGNOSTIC check in the heavy-weight switch code. A thread's process is
Additional ssh patches relating to the same fatal() cleanup issue. They
Import prebinding code into DragonFly, based on Matthew N. Dodd's
The twe driver requires all requests, including non-data requests, to be
Add notation on the alignment requirement for the twe driver.
namecache work stage 1: namespace cleanups. Add a NAMEI_ prefix to
Clean up thread priority and critical section handling during boot. The
oops, remove some namecache leakage that is not ready for prime time yet.
Cleanup td_wmesg after a tsleep completes for easier debugging.
Apply FreeBSD Security Advisory FreeBSD-SA-03:14.arp. Fix DOS crash due
Do not attempt to access kernel_map in free(). It's a bad idea, and doubly
Fix a number of mp_lock issues. I had outsmarted myself trying to deal with
Allow unlock and non-blocking lock operations from FAST interrupts.
Remove the NO_KMEM_MAP and USE_SLAB_ALLOCATOR kernel options. Temporarily
Remove the NO_KMEM_MAP and USE_SLAB_ALLOCATOR kernel options. Temporarily
Try to generate more debugging information when the critical nesting count
namecache work stage 2: move struct namecache to its own header file and
Hash the link-check data to improve performance on filesystems containing a
Cleanup: get rid of the CNP_NOFOLLOW pseudo-flag. #define 0'd flags are a
Move pst from dev/misc to dev/raid, add a Makefile for 'pst'.
Add pst to the raid Makefile
Move the pst device from dev/misc to dev/raid.
Fix a negative cache entry reaping bug, cache_zap() expects ncp's ref count
Fix a bug in lwkt_trytoken(), it failed to exit its critical section on
Describe the hw.physmem loader variable.
correct a comment.
Fix a number of interrupt related issues.
Addendum: Many thanks for continuing to Galen Sampson for running such an
The CMOV family of instructions do not work across all cpu families. In
The splash_bmp and splash_pcx builds were missing some symbols, add the
Disable background bitmap writes. They appear to cause at least two race
Add _KERNEL_STRUCTURES support for userland use of this header file.
Define _KERNEL_STRUCTURES instead of _KERNEL to get just the
Use the _KERNEL_STRUCTURES define to allow userland to bring in kernel
Sync with FreeBSD-5, fix-up the use of certain flags variables.
namecache work stage 3a: Adjust the VFS APIs to include a namecache pointer
Data reads and writes should not need credentials, and most filesystems
add bsd.dfport.pre.mk and bsd.dfport.post.mk, part of the DragonFly ports
Upgrade isc-dhcp, e.g. dhclient.
Add a -p option to fdisk that allows it to operate on normal files which
document the new fdisk option.
Fix bugs introduced from the last commit. The loadav routine somehow got
More hacks to support DragonFly port overrides. Deal with ports which
Add a splash example / DragonFly BMP
add /usr/share/examples/splash to the mtree, for installworld.
Add bsd.dfport.pre.mk and bsd.dfport.post.mk to the Makefile
Augment falloc() to support thread-only file pointers (with no integer file
Extend NSIG to 64 and introduce a registration function for the checkpointing
Use the one remaining free termios control character slot for a tty
Fix miscellanious kern_fp.c bugs.
Start separating the ucred from NDINIT.
Oops, I gave Kip bad advise. The checkpoint execution code is supposed
Add checkpoint tty signaling support to stty and tcsh. The signal
Make sure PORTSDIR is not in the environment when bootstrapping a
/bin/sh needs to use sys_nsig, not NSIG, when accessing static arrays
NSIG is now 64. Extend the siglist arrays to match.
Update the sys_sig* manual page to include portability suggestions.
Fix an inherited bug in ptrace's PT_IO. The wrong thread was being assigned
Have lwkt_reltoken() return the generation number to facilitate checks
correct the .PATH for access to scvesactl.c
Fix the userland scheduler. When the scheduler releases the P_CURPROC
Remove PUSER entirely. Since p_priority has nothing to do with kernel
Cleanup P_CURPROC and P_CP_RELEASED handling. P_CP_RELEASED prevents the
Fix a uni-processor bug with the last commit... we weren't rescheduling on
Make malloc_type statistics per-cpu, which fixes statistics update races
Fix memory leaks in the namecache code commited so far to stabilize its
Make vmstat -m understand the new per-cpu aggregation for
machine/param.h has to be included outside of _KERNEL for MAXCPU.
Fix races in ihashget that were introduced when I introduced the
Remove a diagnostic message that detected PHOLD() vs exit()/process-reaping
Don't try to call an ipiq function with a UP build when the cpuid doesn't
Hack up param.h even more so sys/socket.h can include portions of it without
Entirely remove the old kernel malloc and kmem_map code. The slab allocator
Fix the INVARIANTS memuse test, the memuse for each cpu must be aggregated
Enhance the fp_*() API. Reorganize the ELF dump code using the fp_*()
Add a fp_vpopen() function to kern_fp.c, and add reserved fields to
Add the 'checkpt' utility to support the process checkpoint module. This
Add the 'checkpt' utility to support the process checkpoint module. This
Additional checkpoint suppor for vmspace info. In particular, the data size
Add the checkpt module to the system tree. Currently this may only be
Do a bit of cleanup and add a bit of debugging to the checkpoint module.
Fix an error message.
Fix a bug introduced by recent INVARIANTS debugging additions, sometimes
Yah yah. Always good to commit the header file needed for the last fix.
A hackish fix to a bug uio_yield().
Fix bug in last commit, flags were not being passed to fo_write() which
Use sys_nsig instead of NSIG so libc-suspplied arrays match up.
Add support for SIGCKPT to csh/tcsh's internal 'kill' command.
MSGF_DONE needs to be cleared when a message is to be handled asynchronously.
Hook i386/vesa up to the module build
Add SIGCKPT support to tcsh's built-in kill.
This is FreeBSD 5.x's code to dump the kernel's identifier through
Honor MODULES_OVERRIDE if defined.
Fix type-o's, minor documentation update.
Interrupt threads could block in free waiting for kernel_map(). Add a
Add a DEBUG_INTERRUPTS option for debugging unexpected (trap 30) interrupts.
Simplify the lazy-release code for P_CURPROC, removing the TDF_RUNQ
Backout last commit, oops! SIGCKPT had already been added and it was redundant.
Fix a bug introduced when I redid the stray interrupt arary. The
Deal with multicast packets in a manner similar to Solaris, RFC 3376, and
Better documentation of the MP lock state for new threads.
Augment the LWKT thread creation APIs to allow a cpu to be specified. This
64 bit address space cleanups which are a prerequisit for future 64 bit
Do a minor update of Groff to deal with the addition of intmax_t and
Do a minor update of getconf to deal with the addition of intmax_t and
Fix a minor compile-time bug introduced in 1.22 when DEBUG_VFS_LOCKS is
Fix the pt_entry_t and pd_entry_t types. They were previously pointers to
Give up trying to do this the clean way and just hack groff to remove
Fix LINT issues with vm_paddr_t
Fix a translation bug in the last commit. The second status translation
Add a missing element related to the dev messaging changes.
Get rid of the intmax_t dependancies and just specify the maximum
add cmpdi2 and ucmpdi2 to conf/files to fix LINT build.
Change ui_ref from an unsigned short to an integer. A 16 bit ref count is
Cleanup the ui*() API in preparation for the addition of variant-symlink
Variant symlink support stage 1/2: Implement support for storing and retrieving
An additional ui*() API cleanup that I missed.
Add the 'varsym' utility which may be used to manipulate system-wide and
Remove named arguments in the varsym prototypes which conflict with not very
Add 3c940 ethernet driver support ('sk' driver) for ASUS K8V (AMD64)
If a panic occurs from a BIOS call (16 bit mode) or VM86 DDB will attempt
Fatal traps were not reporting the correct %cs (code selector) for the SMP
sync from FreeBSD-5.x. No operational changes. Get rid of a .byte macro
Synchronize APMBIOS with FreeBSD-5.x bios.c 1.64. The primary change is
Import a good chunk of the PNP code from FreeBSD-5.x
Hopefully do a better job disassembling code in 16 bit mode.
Document critical section and IPIQ interactions in the message port
Network threading stage 1/3: netisrs are already software interrupts,
Update the systat manual page to reflect the new -ifstat option.
Core integer types header file reorganization stage 1/2: Create and/or modify
Core integer types header file reorganization stage 2/2:
Add a wmakeenv target for use by the wmakeenv script to aid in debugging
Initialize base unconditionally to get rid of a bogus compiler warning.
Add prototypes for __cmpdi2() and __ucmpdi2(), get rid of compiler warnings.
Add a pmap_getport_timeout() call hack that allows rpc.umntall to control
Add the -f 'fast timeout' option to rpc.umntall. Using this option will
Add the -f option to the rpc.umntall call to reduce boot-time stalls when
Fix minor snafu in last commit.
machine/stdarg provides __va* macros, not va* macros, use <stdarg.h> instead
3c940 bug fix update. This patch simplifies IFF_PROMISC checking from
Implement variant symlinks! In order to use variant symlinks you must first
Catch attempts to queue to unregistered ISRs
Fully synchronize sys/boot from FreeBSD-5.x, but add / to the module path
Fully synchronize sys/boot from FreeBSD-5.x, but add / to the module path
Fully synchronize sys/boot from FreeBSD-5.x, but add / to the module path
Add a temporary workaround to prevent a panic due to a truss related problem.
Add buffer bounds checking during check pointing since it is possible for
Document the edx assignment to fds[1]
What happens when you mod a negative number? Mask off the hash value before
The last major syscall separation commit completely broke our lseek() as well
Add the varsym_list() system call and add listing support to the varsym
Add the varsym_list() system call and add listing support to the varsym
Most code that calls vm_object_allocate() assumes that the allocation will
Prep for GCC 3.x kernel compiles, stage 1/2: Remove bad C syntax including
Add Marvell chipset support. Adding the PCI ID is all that was needed.
cleanup extra )'s missed by the __P commit.
The MBWTest program (/tmp/mbw1) attempts to figure out the L1 and L2
Refuse to load dependancies from the root filesystem during early boot when
Bring in the MODULE_DEPEND() and DECLARE_MODULE() macros from FreeBSD-5.x.
MT_TAG mbufs are terrible hacks and cannot be freed. Skip any MT_TAGs
Remove __P() macros from include/
__P removal.
Fix bug in last syscall separation commit, an extra semicolon was causing
Add a make variable listing obsolete header files that need to be removed
Prep for GCC 3.x kernel compiles, stage 2/2: Remove bad __asm embedded-newline
Move bsd.kern.mk from /usr/src/share/mk to /usr/src/sys/conf and
Misc cleanups to take care of GCC3.x warnings. Missing 'U' and 'LL'
Correct bugs introduced in the last syscall separation commit. The
panic() normally tries to sync. Add a sysctl to control the behavior in
Fix a bug in the last fix to a prior locking issue. A vdrop() was
Cleanup the linux exec code to match other recent fixes to the exec code.
Enable Conrad parallel port radio clock support in ntp by default. No reason
Cleanup aux args and 32-byte align the initial user stack pointer. Note that
Fix a missing backslashed in the 1.9 commit.
clarify the solution for typical build snafus in UPDATING.
Backout part of 1.16. It is not necessary to align the stack at this
Correct several bugs. If we fail to add a device be sure to delete its kobj.
Use M_ZERO instead of manually bzero()ing memory allocated with malloc().
Sync TAILQ_FOREACH work from 5.x. The closer we can get this file to 5.x
Turn off CHECK_POINTS in mpboot.s. It was accidently left on which resulted
RC cleanups and minor bug fixes to support the uname change to DragonFly
Add a .makeenv directive to make, which allows variables to be exported to
This file needs sys/cdefs.h for __ extensions and such (similar to what
Change the system name from 'FreeBSD' to 'DragonFly'. We are now officially
Change the system name from 'FreeBSD' to 'DragonFly'. Additional commits
Add syscall4 (/tmp/sc4) perf test. This tests stat() overhead instead of
Redo the 'upgrade' target. Get rid of the old elf stuff and change the
Change the system name from 'FreeBSD' to 'DragonFly'.
Change the system name from 'FreeBSD' to 'DragonFly'.
Adjust nfs module loading to use nfs.ko (4.x style) rather then the
This is a major cleanup of the LWKT message port code. The messaging code
Add a DECLARE_DUMMY_MODULE() so we can get linker_set module names
Add a DECLARE_DUMMY_MODULE for snd_cmi to detect kld/static-kernel conflicts.
Implement an upcall mechanism to support userland LWKT. This mechanism will
Add a big whopping manual page for the upcall syscalls, upc_register and
Add UPC_CRITADD, the value that crit_count must be increased or decreased
Add an upcall performance test program + example context function
Add some comments to the upcall test code.
Tweak the context data a bit and do some code cleanup. Save %edx as well
'Building databases' has 10 seconds worth of sleeps that it doesn't need.
Do some fairly major include file cleanups to further separate kernelland
When looking for files that have already been linked, strip off any
Make the 'bad isr' panics a little more verbose.
NETISR_POLL cannot use isr 0. Use isr 1.
Fix the OID_AUTO collision with static sysctl numbers. This can occur
More LWKT messaging cleanups. Isolate the default port functions by making
Add a new library, libcaps, which is intended to encompass DragonFly-specific
Add /var/caps/root and /var/caps/users. IPC rendezvous services for root
Set the close-on-exec flag for CAPS client descriptors.
#include cleanups for lwkt_msgport.c and lwkt_thread.c... the committed
Cleanup POSIX real-time so the kernel compiles without the P1003_1B
Fix a DOS in rfork(). Disallow kernel-only flags.
Fix bug in last commit (missing ..)
Add the MPIPE subsystem. This subsystem is used for 'pipelining' fixed-size
Add cpdup to /bin (not /usr/bin), so we can start using it during boot
Add the -C option to mount_mfs. This option will automatically copy the
Documentat mount_mfs -C
Add a missing PRELE() when the mfs_mount kernel process exits. Because
Do not require -i0 when -o is used.
When attempting to open a file path do not treat a file that appears as a
Introduce /usr/src/nrelease which begins to implement the new 'live CD'
Fix a minor bug... install cvsup and mkisofs in the correct
Temporary disable the ports checks.. the Makefile has to be runnable from
Add nreleae back in. It's a dummy target designed only to ensure that
Use mkiso -R instead of -r in order to properly record file modes. In
Add a README file, rc.conf, and example fstab. Disable a number of system
After testing the manual installation instructions on a real box make some
Add additional information on cvsup'ing various sources.
Add chmod 1777 for /tmp and make other minor adjustments.
As part of the libcaps threading work a number of routines in lwkt_thread.c
General cleanups as part of the libcaps userland threading work.
Augment the upcall system calls to support a priority mechanism rather then
Major libcaps work to support userland threading. Stage 1/2.
Minor cleanups to sysport. Use ms_msgsize in the sendsys() call.
Modify the upcall code to access the critical count via the current thread
Adjust a comment.
Add additional functionality to the upcall support to allow us to wait for
Major update to libcaps. Implement support for virtual cpus using
Thread testing code for libcaps.
Convert alpm to use the devmethod code.
PCI compat cleanup, part 1. This brings in the LNC and VX drivers
This patch adds a bunch of stuff from FreeBSD5. It consistantly makes
Add strlcpy and strlcat to libkern
Add libc support for propolice. See:
Add -fstack-protector and -fno-stack-protector support to GCC. Note
Add -fstack-protector support for the kernel.
More cleanups to make ports work better.
Do not print out error messages in quiet mode to make shell scriptiong
Add the -o file option to rcorder. This will cause rcorder to generate
Add a -p option to rcorder which generates the PROVIDE keywords for the
Adjust rc.subr to generate system-wide varsyms tracking RCNG operations.
Add /sbin/rcrun and symlinks rcstart, rcstop, rcrestart, rcvar, rclist,
Oops, forgot the rc.subr update
Make a distinction between disabled entries and running entries. Do not
Be smarter about services which are in a disabled state. If the
Add another special return code, 3, which means 'subsystem is not
Minor grammatical fix.
Add an 'enable' and 'disable' target to rcrun, and add softlink shortcuts
use the proper $RC_ variable when returning irrelevancy.
Make savecore return RC_CONFIGURED unconditionally.
Support multicast on the Marvell Yukon Chipset. The GMAC on the Yukon
Fix a memory leak that occurs when an attempt is made to checkpoint
USER_LDT is now required by a number of packages as well as our upcoming
Fix a syscall separation bug in recvfrom() which sometimes dereferenced
Minor syntax cleanup.
Patch to make the P4S8X run in ATA100 mode. This is a compromise, since
Add a p_sched field to accomodate Adam K Kirchhoff's scheduler work. This
The attribution for the last commit was incorrect, it should have been:
Try to catch mbuf cluster mclfree list corruption a little earlier with
Backgroundable NFS mounts can still cause a boot sequence to stall for
FreeBSD-5.x removed the 'read interrupt arrived early' check code, for
Most motherboards these days have at least two USB controllers. Adjust
Synchronize the USB, CAM, and TASKQUEUE subsystems with FreeBSD RELENG_4.
Fix the peripheral list scan code, which was broken when the new linker set
Primarily add a missing lwkt_reltoken() in ntfs_ntput(), plus a little
minor syntax cleanup
Get rid of individual /* $xxx$ */ comments entirely (which was an artifact
Add missing prototypes in last commit, other minor adjustments.
Add support for cam_calc_geometry(), from FreeBSD-5.
Bring in the entire FreeBSD-5 USB infrastructure. As of this commit my
These files have been moved to bus/usb.
Add idle entry halt vs spin statistics counters machdep.cpu_idle_*,
Cleanup a DIAGNOSTIC test so LINT compiles.
Bump the network interface cloning API to what is in 5.x with the following
Get EHCI to compile. This will make USB2.0 work when 'device ehci' is
Rearrange an INVARIANTS overflow test so it works with more immediacy.
Fix a bug: remove an extra crit_enter() in the default waitport/waitmsg
Remove proc related extern's from sys/proc.h that are now static's inside
When you attempt to kldload snd_sb16 and then unload it, the system crashes.
Import the libkern/fnmatch code from FreeBSD-5.
Maintain the fnmatch() API from FreeBSD-5 using an inline, rename the
if_xname support Part 1/2: Convert most of the netif devices to use
if_xname support Part 2/2: Convert remaining netif devices and implement full
if_xname support Part 2b/2: Convert remaining netif devices and implement full
Add the -r option to set the hostname based on a reverse lookup of an IP
bktr depends on bktr_mem. Add a KMODDEPS line.
*** empty log message ***
Compensate for the frequency error that occurs at higher 'hz' settings.
tvtohz() was originally designed for tsleep() and timeout() operations but
note last commit based on evidence supplied by: Paul Herman <pherman@frenchfries.net>
Fix a bug introduced in the last commit. When calculating the delta count
Add necessary critical sections to microtime() and nanotime().
Make the phase synchronization of the hz clock interrupt (I8254 timer0)
Arg3 to kern_fcntl was incorrectly passed as NULL, causing
Minor corrections to the documentation.
ANSIfy procedure arguments.
Add a missing thread pointer to a busdma call that needs it.
Cleanup the vm_map_entry_[k]reserve/[k]release() API. This API is used to
npx_intr() expects an interrupt frame but was given something inbetween an
Fix a long-standing bug in protocol 2 operation. The client will top sending
minor syntax cleanups (non-operational changes).
scrap stderr from the ps output to avoid an annoying warning (due to
Major GCC surgery. Move most compiler-specific files into named
Augment the upgrade_etc target to remove stale compiler-related binaries
Fixup /usr/libexec/gcc2/aout binutils generation. The install targets had
Adjust the upgrade target to remove stale /usr/libexec/aout files now
Handle recursive situations a bit more robustly by adding a second reference
non operational change. vm_map_wire() now takes a flags argument instead
Fix an off-by-one error in fdfree() related to flock/fcntl unlock on
Fix reversed snprintf arguments.
CAPS IPC library stage 1/3: The core CAPS IPC code, providing system calls
CAPS IPC library stage 2/3: Adjust syscalls.master and regenerate our
vm_uiomove() is a VFS_IOOPT related procedure, conditionalize it
Add test programs for the new caps system calls. Temporarily disable the
move the caps_type enum so it is accessible through both user and kernel
Fix a minor malloc leak in ips.
Add and document an example disklabel file for the ISO
Fix a panic if -i is used on an interface which does not have an IP.
Retool the M_* flags to malloc() and the VM_ALLOC_* flags to
Resident executable support stage 1/4: Add kernel bits and syscall support
Resident executable support stage 2/4: userland bits. Augment rtld-elf
Add sys/resident.h for userland syscall prototypes, and give the unregister
Resident executable support stage 3/4: Cleanup rtld-elf and augment and
Resident executable support stage 4/4: remove prebinding support.
Resident executable support stage 4/4: cleanup options. -R and -x id now
It really is supposed to be CAPF_ANYCLIENT.
Follow FreeBSD's lead and remove a test utility which is not under a free
Fix the bandwidth specifier once and for all. Properly distinguish
Use the fliesystem block size instead of BUFSIZ when the client sends files.
Fix some type-o's and add some info on things to try if your initial attempt
Add a 'dd' prior to the first fdisk to cover situations where people are
Get rid of embedded newline continuation entirely and switch to an ANSI
Assume read access for execute requests so we can fill in the read credential
Synch the XL driver with FreeBSD-4.x.
Turn off hardware assisted transmit checksums by default. In buildworld
When creating a new uidinfo structure, check for an allocation race whether
Cleanup the duel-macro specifications in sioreg.h and ns16550.h by having
Prevent killall from killing itself.
Fix a serious bug in the NTPD loopfilter. Basically what happens is that
USB storage devices are standard fare these days, add device atapicam to
Fix type-o's and add a few files to the cleanup list.
Describe variant symlinks in the 'ln' manual page.
func() -> func(void) style.
Try to work-around a DFly-specific crash that can occur in ufs_ihashget()
Fix the DDB 'trace' command. When segment address adjustments were added
Undo some of the previously made changes to deal with cross build issues
Rename Makefile.sub back to Makefile.inc to fix cross builds.
Rename .sub files back to .inc.
Remove the remainder of Makefile.sub handling. Make objformat aware of the
have upgrade / upgrade_etc remove /usr/libexec/gcc2 now as well, it has
Add back a directory path that buildworld needs, remove /usr/libexec/gcc2
oops, undo part of the last commit. cpp, f77 need their gccX frontends in
Set a variable indicating that we support .makeenv, so we can conditionalize
This should hopefully fix current issues with bootstrap buildworlds from
This commit represents a major revamping of the clock interrupt and timebase
This commit represents a major revamping of the clock interrupt and timebase
This commit represents a major revamping of the clock interrupt and timebase
Clean up the sys_nsig verses NSIG code to handle additional fault
Update miscellanious firewire manual pages from FreeBSD-4.x.
Remove limitations on the 'machine' specification.
MAINTAINER lines in Makefile's are no longer applicable, remove them.
Remove genassym and gensetdefs. These programs are obsolete and no longer
remove genassym and gensetdefs in the upgrade_etc target.
Make gcc3 the default for non-i386 architectures, leave gcc2 the default
binutils214 stage 1/4. Bring in the build infrastructure (left untied from
Don't use wildcards in csh expansions to avoid early termination of
binutils214 stage 2/4.
Merge from vendor branch BINUTILS:
Bring GNU binutils-2.14. This commit is an exact copy of the contents of
binutils214 stage 2/4 (continued). remove /usr/bin/gcc{2,3} from
AMD64 infrastructure stage 1/99. This is just a preliminary commit. Many
Update the GCC3 infrastructure Stage 1/2. This commit generates the basic
Undo the xmalloc->malloc optimization FreeBSD made in certain cases
Adjust __FreeBSD__ -> __DragonFly_
Make the 'realclean' target respect KERNCONF.
Add -DSETPWENT_VOID to CFLAGS to work around a __FreeBSD__ conditional in
Adjust osreldate.h to only set __FreeBSD_version if __FreeBSD__ is set,
Fake __FreeBSD__ for various contrib/ code that needs it.
Change __FreeBSD__ -> __DragonFly__
Remove obsolete __FreeBSD_version checks, add __DragonFly__ tests to
Try to clean up more of GCC's xrealloc and xmalloc wrapper mess so
Fake __FreeBSD__ for sendmail contrib use.
This should fix the dependancy loop for sure.
More __FreeBSD__ -> __DragonFly__ translation
Add some basic in-pipeline instruction timing tests. Instruction timings
Add a locked-bus-cycle add to memory test
Fix a DFly buildworld from 4.x issue. Only set HAVE_STDINT_H for
Split the lwkt_token code out of lwkt_thread.c. Give it its own file.
devsw() does not exist in DFly. use dev_dflags() to extract d_flags.
The logical pci busses must attach to the physical pci bridges using the
Convert mbuf M_ flags into malloc M_ flags when calling malloc().
Use a globaldata_t instead of a cpuid in the lwkt_token structure. The
isa_wrongintr() cannot depend on the (void *) unit argument pointing to
activate any tick-delayed software interrupts in the per-cpu hardclock
sys/dev __FreeBSD__ -> __DragonFly__ cleanups.
Change lwkt_send_ipiq() and lwkt_wait_ipiq() to take a globaldata_t instead
Convert __FreeBSD__ tests to __FreeBSD__ and __DragonFly__ tests
Add -D__FreeBSD__ for buildworld (vacation pulls source files from
Use "iq" instead of "ir" for the register constraint. "iq" means 'an
bio ops can be initiated from the buffer cache, td_proc may be NULL here.
Oops. Forgot to renumber the %N's in the macros in the last commit.
Rewrite the IP checksum code. Get rid of all the inline assembly garbage,
Fix forgotten lwkt_send_ipiq() API update. cpuid -> globaldata pointer
C99 specify ary[] instead of ary[0] in structure.
gcc2 doesn't like ary[] inside structures. Add __ARRAY_ZERO to
Add PCI busses to the device list in bus number order to make debug output
Actively manage cam rescan timeouts in the usb subsystem. This does not
Change M_NOWAIT to M_WAITOK. This does not fix any known issues but it
Get rid of some old cruft and add a failsafe for M_WAITOK which guarentees
Create a new machine type, cpumask_t, to represent a mask of cpus, and
Move <machine/in_cksum.h> to <sys/in_cksum.h>. This file is now platform
Remove old in_cksum.c and in_cksum.h (they were moved to netinet and sys
atomic_*_*_nonlocked() inlines are not MP locked. Remove the MPLOCKED
Split the IPIQ messaging out of lwkt_thread.c and move it to its own file,
Cleanup and augment the cpu synchronization API a bit. Embed the maxcount
Install new .mk files (/usr/src/share/mk) as part of the upgrade_etc target.
Make buftimetoken an extern so it is not declared as a common variable.
buftimetoken must be declared in a .c file.
Remove duplicate declarations in preparation for adding -fno-common to
Remove duplicate bioops declaration in preparation for -fno-common
Add a dependant include.
Remove common variable to get ready for -fno-common.
Compile the kernel with -fno-common to guarentee that we do not accidently
Remove common declaration for -fno-common
Remove duplicate declarations for -fno-common
Remove duplicate declarations for -fno-common
Undo part of the last commit. Part of a different patch set leaked into it
Introduce an MI cpu synchronization API, redo the SMP AP startup code,
(followup) remove lockid.
ATAng stage 1: synch ad_attach() and atapi_attach(), including a fix for
ATAng stage 2: sync part of the ata_dma*() API. No operational changes.
ATAng stage 3: sync additional atang from 4.x, mostly non-opertional changes,
ATAng stage 4: sync additional atang from 4.x, all non-operational changes
ATAng stage 5: sync additional function API changes from FBsd-4. We now
ATAng stage 5: sync chipset changes and bug fixes. busdma is not synched yet.
Add experimental (as in hacked) support for the Silicon Image SATA
ATAng stage 6: Comment-only. Many thanks to David Rhodus for generating
Merge vfs/ufs/ufs_disksubr.c into kern/subr_disk.c. The procedures in
RCNG, shutdown ppp connections nicely when told to.
M_NOWAIT work stage 1/999: Fix some boot-time misuses of M_NOWAIT -> M_WAITOK
Implement a pipe KVM cache primarily to reduce unnecessary TLB IPIs between
Synchronize a bunch of things from FreeBSD-5 in preparation for the new
Synchronize a bunch of things from FreeBSD-5 in preparation for the new
Bring in the FreeBSD-5 ACPICA code as a module. Note: not hooked up yet,
Bring in acpica-unix-20031203. As with other contrib imports, this import
Fix a bug in the last commit. 4.x improperly tries to add the children
Include the required machine/bus.h if we do not already have it.
Bring in additional stuff from FreeBSD-5, fixing some issues (fwohci not
INTR_EXCL moved to sys/bus.h, add #include.
Remove static resource_disabled(), the function is now supplied by
Make nexus understand the new INTR_ flags, mainly INTR_ENTROPY.
Newtoken commit. Change the token implementation as follows: (1) Obtaining
Partitions>8: Do not hardwire partition limit at 'h'.
Partitions>8: Leave a whole 512 bytes for the disklabel and squeeze code
Partitions>8: Increase the number of supported partitions from 8 to 16.
Correct bug introduced in last commit.
Simplify LWKT message initialization semantics to reduce API confusion.
Bring libcaps in line with recent LWKT changes.
Additional CAPS IPC work. Add additional system calls to allow a CAPS
Adjust the caps client/server test code to match new CAPS features. The
The sys/xxx2.h files are supposed to be included after all the normal
get rid of thr{1,2,3}, which are obsolete at the moment. Keep the
Initial CAPS IPC structural encoding and decoding support. Note that the
libcaps now compiles ipiq and token in userland, make those files compile
ANSIfy the tsleep() and sched_setup() procedure definitions.
Config cleanup part 1/3: Remove old style C cruft and cleanup some variable
Config cleanup part 2/3: Remove old style C cruft.
Config cleanup part 3/3: Remove the ns() and twisty eq() macros and replace
The "Don't forget to do a ``make depend''" warning no longer serves any
Add -B to display buffer limits instead of current buffer usage.
Increase the default socket buffer for NFS to deal with linux bugs and to
Addendum comment to last commit. Jeffrey Hsu reminded me that kernel writes
ncpus2 must be initialized to 1 in the UP case. ncpus2_mask and ncpus2_shift
Allow the nominal NFS io block size to be set with a sysctl vfs.nfs.nfs_io_size
Explain some of the instruction details in more depth.
ANSIfication, convert K&R procedure declarations and remove 'register'.
dislabel -> 16 partitions work addendum: MAKEDEV now understands and
Minor documentation adjustments.
Allow %e, %E, %f, %g, %G formats to work without producing an error code.
ANSIfication, remove 'register' and 'auto' use, convert K&R procedure decls.
gdb was unable to obtain backtraces of pure threads. Change the 'proc'
Change M_NOWAIT to M_INTWAIT or M_WAITOK. CAM does a mediocre job checking
Use M_INTWAIT and M_WAITOK instead of M_NOWAIT within the USB bus
In an rfork'd or vfork'd situation where multiple processes are sharing
Fix a bunch of NFS races. These races existed in FreeBSD 4.x but are more
The cam_sim structure was being deallocated unconditionally by device
Do some M_WAITOK<->M_INTWAIT cleanups. Code entered from userland, such as
The cam_sim structure was being deallocated unconditionally by device
When deregistering a bus, pending device bus scan timeouts are not deleted
When detaching UMASS, abort all the pipes before detaching the sim. Note
Do not free the old VFS vector when recomputing the vectors. If a module
Code cleanup, remove 'register' and (void) casts. No functional changes.
Move a drain output call to before a dirty block check instead of after
Partial merge from FBsd-5, code to make some PCCARDs work under NEWCARD.
Add a Makefile stub to build pccard as a module.
Add a Makefile stub to build ep as a module.
ANSIfication, K&R cleanups, 'register' removal.
The kernel realloc() does not support M_ZERO, assert the case.
nextsoftcheck (which is a really aweful interrupt interlock hack) needs to
Separate chroot() into kern_chroot(). Rename change_dir() to checkvp_chdir()
main() more typically uses 'char **argv' instead of 'char *argv[]'. Remove
An strlcpy() in the last commit was unconditionally overwriting 'name'
Fix a bug in the recent connectionless commit. When sending a UDP packet
Update the 825xx GigE support. Add a large number of new device id's and
Create /usr/src/test/test ... a dummy directory that new committers can
Merge FreeBSD ifconfig.c rev 1.94, strlcpy() cannot be used if the source
Separate the pieces a bit. The usb.ko module does not build EHCI, and
Add some missing SRCS dependancies to the EHCI module.
Add some USB specific documentation to make life easier for people
Add additional default-generation entries for usb2-5, most new PCs these
Repository copy libkern/{Makefile,iconv*} -> libiconv, and modify the
Adjust the Makefile's to move the iconv files to libiconv, and add it to
Bring in MODULE_VERSION from FreeBSD-5. Even though our kernel doesn't
Fix generation of USB_EVENT_DEVICE_DETACH, which was commented out in
Incorporate NetBSD rev 1.111: Set the device address before reading the
If XL cannot properly attach it tries to detach to clean up. Unfortunately,
Make ALWAYS_MSG the default. This introduces an extra procedural call
Get rid of the obsolete SMP checks in SMBFS.
Add function call overhead tests for (1) direct calls, (2) indirect calls,
Add some additional spaces so the ctl string does not bump the
64-byte-align the test functions so they appear on different cache lines.
Fix p_pctcpu and p_estcpu. When the new systimer stuff was put in the
Fix p_pctcpu and p_estcpu (addendum). Add a ESTCPUFREQ and set it to 10hz.
Add a missing resetpriority() which was causing all newly forked processes
No changes. force commit / update timestamp so ioctl.c is regenerated.
correct a buildworld failure, fix the include file filter to allow the
Undo part of the last commit. OBJFORMAT_PATH controls how the cross
A large number of targets were doing a mkdir -p openssl. A parallel make
grr. fix bug in last commit. Use .ALLSRC instead of .OODATE
This represents a major update to the buildworld subsystem.
buildworld subsystem update addendum. Hopefully fix buildkernel,
Another attempt to fix make -j N issues with this subdirectory.
Add hexdump and kill to the bootstrap list. Correct the tools path used
Add 'route show' to the route command, plus style cleanups.
Make the Destination and Gateway columns wider when printing FQDNs so
Add -w, which prints the full width of the data being represented even if
ANSIfication (procedure args) cleanup.
make -j N support, add a required dependancy.
Fix malloc semantics, M_NOWAIT->M_WAITOK.
agp_nvidia.c was not being linked into the module build.
Fix malloc semantics (M_NOWAIT->M_INTWAIT/M_WAITOK).
Return a low priority for the "hostb%d" catch-all for pci busses, which will
make -j N support, the generated lib_gen.c and nomacros.h files depend
Fix a missing makewhatis related change so buildworld works again.
The NXCC (native C compiler) misnamed OBJFORMATPATH, it neesd to be
Make the .nx/.no native program helper binaries work and add some missing
Protect v_usecount with a critical section for now (we depend on the BGL),
Do some major performance tuning of the userland scheduler.
Import Alan Cox's /usr/src/sys/kern/sys_pipe.c 1.171. This rips out
Add the pipe2 sysperf test. This test issues block writes from parent to
Initialize the pcpu clocks after we've activated the cpu bit in
Generally bring in additional sf_buf improvements from FreeBSD-5. Separate
Bring in a bunch of well tested MPIPE changes. Preallocate a minimum
UDF was not properly cleaning up getblk'd buffers in the face of error
Count vnodes held on the mount list simply by using the
Protect the mntvnode scan for coda with the proper token. Since we do not
Second major scheduler patch. This corrects interactive issues that were
Initial XIO implementation. XIOs represent data through a list of VM pages
Hook XIO up to the kernel build.
Change CAPS over to use XIO instead of the vmspace_copy() junk it was using
Trash the vmspace_copy() hacks that CAPS was previously using. No other
Cleanup libcaps to support recent LWKT changes. Add TDF_SYSTHREAD back
Allow the child priority (receive side of the pipe test) to be specified
Cleanup the forking behavior of the CAPS client test program.
Add missing sf_buf_free()'s.
Get rid of the upper-end malloc() limit for the pipe throughput test.
Implement a convenient gd_cpumask so we don't have to do 1 << gd->gd_cpuid
Fix an unused variable warning (non-operational).
Enhance the pmap_kenter*() API and friends, separating out entries which
Make buildkernel's require a buildworld to be done first, because they
Correct the commented-out example for MODULES_OVERRIDE.
In the sysclock commit I tried to make 'boottime' a fixed value, but it
Fix bugs in xio_copy_*(). We were not using the masked offset when
Create a normal stack frame in generic_bcopy() to aid debugging, so
Cleanup NXENV so it works properly when running buildworld from FreeBSD.
Perl is no longer needed by buildworld/buildkernel.
Setting the date/time does not always properly write-back the RTC, causing
Fix a missing wildcard binding in the recent wildcard binding hash table work.
Quake 3 server (running under linux emulation) was failing with odd '
Undo the last commit. Utility programs which install c++ includes have no
Fix buildworld. Document TOOLS_PREFIX and USRDATA_PREFIX, improve INCLUDEDIR
Bring in FreeBSD 1.2.2.2. Properly unwind the stack when certain
Partial sync from FreeBSD adds some more support and fixes. Also replace a
Remove makewhatis from /usr/bin (it officially resides in /usr/sbin),
per-cpu tcbinfo[]s aren't ready for prime time yet. The tcbinfo is assigned
Export the lwkt_default_*() message port default functions so other
Subsystems which install an so_upcall may themselves call socket functions
Do not reset %gs in signal handlers, some programs depend on it (KDE in
Protect nfs socket locks with a critical section. Recheck rep->r_mrep just
Use hex bit values instead of decimal bit values (non operational change).
General netif malloc() flags cleanup. Use M_INTWAIT or M_WAITOK instead
General bus malloc() flags cleanup, M_NOWAIT -> M_INTWAIT. Note: leave
General ata malloc() flags cleanup. Use M_INTWAIT where appropriate and
Make TCP stats per-cpu.
Enable propolice (stack smashing detector) by default on gcc3.
Make TCP stats per-cpu. (forgot to add new header file)
TCP statistics structure renamed tcpstat -> tcp_stats.
/tmp/motd* files were being left sitting around after a reboot when the
namecache work stage 4:
namecache work stage 4a: Do some minor performance cleanups with negative
Introduce negative (ENOENT) caching for NFS. Before this, an attempt to
ANSIfication and style cleanups. Non operational.
Do some minor critical path performance improvements in the scheduler
Add vfork/exec perf test. exec1 tests static binaries, exec2 tests dynamic
Use the sf_buf facility rather then kmem_alloc_wait/pmap_kenter/kmem_free
Fix the conditional used to determine whether psignal() should be called.
Remove the now obsolete /usr/include/g++. Cleanup after the nawk->awk
ANSIfication/style cleanups (non operational)
The malloc() call in at_fork() needs to use M_WAITOK instead of M_NOWAIT.
get rid of TCP_DISTRIBUTED_TCBINFO, it only added confusion.
res_search only incremented got_servfail for h_errno == TRY_AGAIN *AND*
uio_td might be NULL, do not indirect through uio_td to get to td_proc
Run the exec test for 5 seconds instead of 1 to improve measurement
Implement lwkt_abortmsg() support. This function chases down a message and
Detect when the target process's thread is sitting on a message port and
Change WAIT_FOR_AUTO_NEG_DEFAULT to 0. Do not wait for auto-negotiation to
netisr_queue() needs to reliably allocate the message used to reference the
Use vm_page_hold() instead of vm_page_wire() for exec's mapping of the first
Revamp the initial lwkt_abortmsg() support to normalize the abstraction. Now
When an mpipe was being destroyed, each element in the array was being
Followup commit, redo the way the root file descriptor slop is calculated
Fix a netmsg memory leak in the ARP code. Adjust all ms_cmd function
Fix a race in user_ldt_free() against an interrupt (which attempts to
M_NOWAIT -> M_WAITOK or M_INTWAIT conversions. There is a whole lot of net
Use M_INTWAIT instead of M_NOWAIT in the ip messaging redispatch case to
The temporary message allocated to execute a connect request is not
M_NOWAIT to mostly M_INTWAIT conversions, with a splattering of
Correct type-o in last commit. oops.
More M_NOWAIT -> M_INTWAIT | M_NULLOK conversions, primarily atm and ipsec.
posixlocks resource limit part 1/4: Add support to the login.conf database,
posixlocks resource limit part 2/4: Add support to /usr/bin/limits.
Revamp UPDATING with separate instructions for upgrading from sources
buildiso was assuming a native obj hierarchy when running the make distribute
msync(..., MS_INVALIDATE) will incorrectly remove dirty pages without
Fix a client tail -f vs server-appended file data corruption case by
#ifdef out the PCATCH/CURSIG code for userland (libcaps), it only applies
Support for more video modes: accept mode names like MODE_<NUMBER> where
If the server goes away while the client is trying to copy a message from
Fix a bug noted by David Rhodus and removes minor redundancy.
nextsoftcheck must be a volatile pointer, not a pointer to a volatile.
Bring in the following revs from FreeBS-4:
Revamp the PIPE test a bit. Use a calibration loop to make the test always
Due to pipe buffer chunking the reader side of the pipe was not traversing
Add mem1 and mem2 .... memory copying and zeroing test suites, making it
Rewrite the optimized memcpy/bcopy/bzero support subsystem. Rip out the
Make hash tables one power of 2 larger so they don't (generally) fold
Make SF_BUF_HASH() into an inline routine, sf_buf_hash(), and add an
Remove the (now non existant) i486_bzero assignment for I486_CPU.
Correct a bug in the last FPU optimized bcopy commit. The user FPU state
Fix a race in npxdna(). If an interrupt occurs after we have set npxthread
Fix a race in the FP copy code. If we setup our temporary FP save area
Fix another bug in the recent bcopy revamp. The range checking was
Clear npxthread before setting CR0_TS.
Add bcopyb() back in for the PCVT driver. bcopyb() is explicitly
Commit an update to the pipe code that implements various pipe algorithms.
We must pmap_qremove() pages that we previously pmap_qenter()'d before
(non bug) The hash routines are never called with a size of 1, but make
ipstat needs thread2.h for MP stuff and mycpu.
A memory ordering barrier is needed in crit_exit() ensure that td_pri
ip6_input() must call the IP6_EXTHDR_CHECK() macro with a specified return
Fix an exit-race with ^T. If a process is exiting it may be detached from
Another major mmx/xmm/FP commit. This is a combination of several patches
pmap_qremove() takes a page count, not a byte count. This should fix
Do not trust the third-party ACPI code. Track memory mapping requests
Use M_INTWAIT instead of M_NOWAIT for the rest of the acpica support code.
Document the fact that SYSTIMERS operate without the MP lock.
Followup, fix some missing ODFM->OFDM conversions.
One of the lf_create_range() calls in lf_clearlock() was passing a bogus
Add the filesystem/NFS stress tester program, adapted for BSD by Jordan
Move the fsx filesystem tester program from /usr/src/tools/regression to
Fix IPV6 listen(). It was simply a matter of a missing
Peter Edwards brought up an interesting NFS bug which we both originally
lf_setlock() was not returning the correct error code due to an 'int error'
lf_alloc_range() must initialize the returned structure sufficiently such that
Add an assertion to sys_pipe to cover a possible overrun case and reorder
Followup log-only addendum: It turns out that last commit did not solve
Fix a bug in sys/pipe.c. xio_init_ubuf() might not be able to load up the
Attempting to access a device which has been destroyed, such as a UMASS
sf_buf_free() requires a critical section to safely manipulate the free
Remove DIAGNOSTIC elements that are no longer correct.
Close an interrupt race between vm_page_lookup() and (typically) a
malloc() flags cleanup and fixes. Get rid of M_NOWAIT in places where the
sf_buf_ref() needs a critical section. Note that the function is not
device switch 1/many: Remove d_autoq, add d_clone (where d_autoq was).
INCSLINKS cannot be used to make softlinks within machine/ because
Fix another race in ^T. ttyprintf() can block, during which time the
Fix more ^T panics. calcru() and p_comm also need p_thread checks. Just
Bring in a better seeding and random number generation algorithm from
Device layer rollup commit.
Using the new contrib rev mechanism, bring cvs up to 1.12.8. Add patch files
Make the primary PQ_ macros available to modules by creating the pageq
Cleanup compile issues with the recent dev_t changes.
Cleanup compile issues with the recent dev_t changes.
Cleanup pass. No operational changes.
Get rid of obsolete PQ_ options.
Cleanup pass. No operational changes.
Get rid of VM_WAIT and VM_WAITPFAULT crud, replace with calls to
Pass the proper mask/match arguments to cdevsw_add() so the various sound
The extended mbr handling code was improperly using the b_dev from a
ANSIfication and cleanup. No operational changes.
Increment v_opencount before potentially blocking in dopen to avoid
Cleanup cdevsw and dev reference counting ops to fix a warning that occurs
ANSIfication and general cleanup. No operational changes.
count_udev() was being called with the wrong argument.
preload_delete_name() needs to use the same path munging that the other
Cleanup warnings. No operational changes.
Cleanup warnings. No operational changes. Note that the original ACPICA
Correct a bug in the last commit, udev must be assigned from vp->v_udev, not
ANSIfication and cleanup. No functional changes.
Mask bits properly for pte_prot() in case it is called with additional
Bring in the fictitious page wiring bug fixes from FreeBSD-5. Make additional
Attach bind-9.2.4rc4 to the base system. Rip out bind-8 binaries and add
Initialize the FP unit earlier in the AP boot sequence. This solves
Add lwkt_setcpu_self(), a function which migrates the current thread to
Make sysctl_kern_proc() iterate through available cpus to retrieve the
Two unused arguments were recently removed from pmap_init() without
In the root fs search use the correct unit number when checking for
When doing a restart, sleep for 0.1 seconds after the kill to avoid racing
Fix the path to named's pid file.
Since RNDC does not yet support 'restart', use RCNG to restart named
Add RLIMIT_POSIXLOCKS support to csh/tcsh.
Add a pfil_has_hooks() inline to shortcut calls to pfil_run_hooks(),
ANSIfication. No operational changes.
VM86 calls on some BIOSs, apparently mainly VESA calls, use 8254 timers that
ANSIfication and cleanup. No functional changes.
Clean up some misuses of bp->b_dev after a strategy function has completed
ms_cmd has changed to a union, update the test code.
ANSIfication. No functional changes.
Use MPIPE instead of (the really aweful improper use of) mbuf_alloc() for
TEST
Remove dtom() in unused (#ifndef notdef'd) code. Well, just remove the whole
Use MPIPE instead of the really hackish use of m_get() and mtod()/dtom()
Use MPIPE instead of misusing m_getclr() (i.e. m_get) for NETNS's
Add MPF_INT (for mpipe_init()), which allows one to specify that an MPIPE
Use a normal malloc() for PCB allocations instead of the really aweful mbuf
Remove dtom(). dtom() is no longer supported (precursor requirement for
async syscall work: The async syscall code got dated by recent LWKT
Allow dup_sockaddr() to block, otherwise the code becomes non-deterministic.
Rearrange the kern_getcwd() procedure to return the base of the string
M_NOWAIT is just wrong in the init code. The allocation must succeed.
Add the MSFBUF API. MSFBUFs are like SFBUFs but they manage ephermal
Deal with revoke()'d descriptors. The underlying vnode is vgone()'d, which
The usbcom's device was being released too early in the close sequence.
Remove the canwait argument to dup_sockaddr(). Callers of dup_sockaddr()
Add the prototype for the recently added XIO API call xio_init().
Back out the last change. Normal 'make' builds in the source tree are
Remove dtom() from comment.
Add in_pcbinfo_init() to encapsulate basic structural setup (right now just
Recent accept() changes started depending on the protosw->pr_mport field
Additional listhead->pcblisthead and marker support for netinet6.
Convert netproto/ns to the pr_usrreqs structure. This is untested work
ulimit.h needed to be added to the include file list for installation.
Fix improper vm_object ref counting in procfs that was introduced in the last
Both 'ps' and the loadav calculations got broken by thread sleeps, which
Fix a bug in the reply port path related to aborts. Aborted messages are
ANSIfy the remaining K&R functions. Remove two bzero-after-malloc's in
start_forked_proc() should not be called when rfork() is called without
Fix a race with the clearing of p->p_session->s_ttyvp. NULL the pointer
Clean up GCC3.3, rip out all the weird search paths it adds and fix a long
Make sure gcc_local.c and any left over patch mess is cleaned up by make clean,
Don't set prefetch mode on VIA chips, it causes problems on newer chips and
Write a remote configuration utility called 'rconfig'. This initial
chdir into WorkDir before running the downloaded script.
A sample rconfig script which completely wipes and reinstalls dragonfly
Fix an aggregious non-terminated buffer issue and also fix the retry code
Merge from vendor branch GCC:
Bring in a trimmed down gcc-3.4-20040618.
Allow CCVER to be empty, which will cause the compiled-in default compiler
'lrint' is a gcc3.4 builtin and gcc3.4 is unhappy if we use that for anything
Fix a prototype error in gcc-3.3's com.h in order to allow gcc-3.4 to compile
Remove the last vestiges of -DCCVER from 2.95.x contrib so we can remove
Add a missing break after default: for gcc-3.4 support.
Hook gcc 3.4 into the buildworld. Rearrange HOST_CCVER so it becomes the
gcc-3.4 cleanup.
Adding missing break after default: (gcc-3.4).
gcc-3.4 cleanups. Add missing break statements, deal with goto labels,
gcc-3.4 cleanups. missing break / goto labels.
Correct a known panic in the twe driver, M_NOWAIT -> M_INTWAIT work.
General M_NOWAIT -> M_INTWAIT work, except in periodic timeout() routines
Some more M_NOWAIT->M_INTWAIT. Not sure if the sym changes are correct,
Fix -j builds for gcc-3.4. The .nx build in cc_tools was breaking
Fix -j builds for gcc-3.4. The .nx build in cc_tools was breaking
Turn propolice (-fstack-protector) on by default. Propolice rearranges
Unbreak the buildworld by fixing a cc_tools dependancy on cc_prep in
Check for a queued interrupt being dispatched after the ATA driver
Fix an improper DELAY in the ata tag code (but nobody should be using
Do a partial synch from FreeBSD-5 of the NVIDIA and NFORCE ATA setup code.
Hack in the code from FreeBSD-5 to set the timings for NVIDIA/AMD chipsets.
Fix a broken vrele() in the session tty exit code.
The route table treats sockaddr data as opaque, which means that the unused
Backout 1.19. It prevents some machines from booting (READ timeout).
Get rid of the PFIL_HOOKS option, integrate pfil in the system permanently.
Bring in 1.33 from FreeBSD-5:
Convert the M4'd btxldr.s to a preprocessed btxldr.S (taken from FreeBSD-5)
Implement duel-console mode and make it the default. Adjust the boot
Remove a vn == vp case that was breaking out of the namecache lookup loop
Stage 1/999: Rewrite boot0 to relocate to higher memory (beyond 64K). This
Add a signature for bootn0 so boot0cfg doesn't complain about it, even
If Preposterous BIOS basemem is detected, instead of truncating to 640K
Cleanup a comment.
When a kernel-created thread exits, properly remove it from gd_tdallq and
Some general cleanups and use M_ZERO instead of bzero() in one case
Add some early serial console debugging support to the loader. This is all
Unhook bios_howmem (a simple reporting variable), it isn't ready yet.
Add a short unconditional sleep in the lockf retry path to try to
Reorganize the subdirectories into an include/ subdirectory so the
ndp was using the old contrib/tcpdump instead of contrib/tcpdump-3.8.3
Enhance lockf's debugging with macros. Print the originating process in
Move include path work for named related utilities.
Remove 28.cvs for now (the files involved do not exist in the latest contrib)
Update contrib/cvs to contrib/cvs-1.12.8, but note that this Makefile
Bring boot2 and the loader into line with our new dual-console support.
Do not generate the 'Block size restricts cylinder groups to BLAH' warning
Do not try to chflags() a symbolic link when copying an underlying filesytem
Do a bit of cleanup and enable the SIO FIFO (1655x) to reduce latencies
Fix a minor bug in the auto-console selection (handle the -m mute option
Be ultra conservative for now, do not try to initialize the FIFO.
More missed named fixups related to the include directory move.
Bring in YONETANI Tomokazu's acpi-update-2.patch (27-May-2004), a major
Add note to cpu_idle_hook (which is currently asserted so the code doesn't
Show a more complete listing of interrupt sources (do not weed out sources
Make the VR device backdown to emergency polling if the interrupt appears
Implement livelock detection for threaded interrupts and automatically
The schednetisr() routine is supposed to be MP and interrupt safe, but wasn't
Properly probe for the serial port. If the serial port is unmapped or
Addendum: it should be noted that boot2 also probes for a valid serial
Turn the getty on on ttyd0 by default so a CDBOOT will run all the way through
The acpica-unix-20040527 download from intel seems to like to use upper
Fix three bugs in the livelock code. Fix a minor range error in an
Increase PCCARD_CIS_SIZE from 1024 to 4096 as per FreeBSD-5. Add range
Update the README file with useful information about ACPI.
Undo one of the recent optimizations I made (only running the handlers
Change the version string to RC1
Fix a snafu in the last commit. In the normal non-polling case interrupts
Add an explanation for the use of CCVER in /etc/make.conf, and recommend
Make sure a serial port exists by determining whether it is possible to drain
Add AGP support for the i852GM, i855GM, and i865G.
When booting from CD, check cd1a and acd1a after cd0a and acd0a, allowing
Get rid of some debugging printf's that had been accidently committed.
minor cleanups / no functional changes.
Bring in acpica-20040527 from intel. See: http://developer.intel.com/technology/iapc/acpi/downloads/acpica-unix-20040527.tar.gz
acpica5 update part 1/3: Implement support for acpica-unix-20040527.
acpica5 update part 2/3: Fix a bug introduced in the original acpica5
acpica5 update part 3/3: Bring the usr.sbin/acpi tools into the base system
Cleanup conditionals on exit, remove unnecessary free()'s prior to error exit.
Fix a bug in the -i code where an existing interface might not be located.
More optarg and declaration cleanups.
Update newvers.sh to RC2
Don't allow a step size of 0 (leading to an infinite loop). Have
Be more verbose when printing information on transfer phase errors.
Unconditionally reset ATAPI-CD devices during boot. Brian's notes on this:
In ip_mport() (IP packet demux code), check the minimum length requirement
Bring in a bunch of updates from NetBSD:
Bring EHCI up-to-date with NetBSD. The most serious fixes are 1.53, 1.55,
Julian Elischer posted an interesting proof-of-concept to freebsd-current
ugenbuf is associated with the 'ugen' device, not the 'ugenbuf' device.
Give the newly created ugenbuf.c the standard DragonFly copyright.
Give ugenbuf the standard DragonFly copyright.
IPS was using malloc flags of 0 (which is no longer allowed). The helper
Add a missing '$' to the FreeBSD cvs tags.
Fix some issues with the pccard shutdown path (during reboot and halt).
There was a mountlist race in getnewvnode() whereby the system could block
(installer support). We are going to have a special 'installer' login
Import the new nrelease Makefile packaging and root template infrastructure
The temporary 'installer' user is not supposed to have a password.
The release password databases must be regenerated after installing the
The password file the installer writes to the HD should not have an
Update the version to 1.0-RELEASE.
Update the README to include a description of the installer and add some
Give the dfui packages a version number (1.0). Implement serial console
Resynch the pristine ttys for the installer with the base system ttys (put
Adjust the copyright to the new official DragonFly copyright.
Update dfuibe_installer to 1.0.1 to fix a series slice corruption issue.
Update release to 1.0A
Note that Jeff indicated to me that Jonathan Lemon gave his permission to
Update all my personal copyrights to the Dragonfly Standard Copyright.
Fix a URL displayed in an advisory.
Minor documentation update to clarify the effect of the vfs.usermount sysctl.
Compensate sockstat for the CPU column that netstat now adds.
Don't let packets with DF set sneak by through the hardware-assisted
Fix two serious bugs in the IP demux code. First, if ip_mport() m_pullup()'s
Bring the twiddle fix in from FreeBSD5:1.11. This rev also added a
Brute force a register save/restore for int 0x13 (disk I/O) and 0x10 (putchar)
Sync bootn0 with recent boot0 fixes.
Consolidate most constant memory addresses in bootasm.h part1/2. Convert
Consolidate most constant memory addresses in bootasm.h part2/2:
Correct a bug in NXCCFLAGS generation.
MAJOR BOOT CODE REVAMP / 30 hour+ hacking session (50 if you include the
SCSI CD devices require 'cd0c' to be specified instead of 'cd0a', while
Unconditionally print startup 8254 and TSC calibrations.
ata-raid associates raw ata disk devices to record the raid setup and checks
Add a dire warning about PCI_ENABLE_IO_MODES.
Fix a device pager leak for the case where the page already exists in the
Replace the perl man filter with a sed man filter, fixing manual page
Update the userland scheduler. Fix scheduler interactions which were
Adjust gd_vme_avail after ensuring that sufficient entries exist rather
Have DDBs 'ps' command display additional scheduler-related paramters
Make fstat() account for pending direct-write data when run on a pipe.
Move usched_debug out of the INVARIANTS conditional. Make it unconditional.
Add 'propolice' to the version string version_local.c sed patchup.
Boot1 tries to clear boot2's BSS. It makes several assumptions that are
Patch out tcsh's use of 'exp2', which is math-reserved in gcc-3.4.
rename exp() to expx() to avoid conflict with gcc-3.4 built-in.
rename functions that clash with reserved math procedures to avoid gcc3.4
udev2dev() can return NODEV now, make sure it doesn't crash autoconf's
Change the default syslogd flags from -s to -ss, which prevents a network
(From Alan):
Output a more descriptive error message when AGP can't bind memory.
Stage 1/many: mbuf/cluster accounting rewrite and mbuf allocator rewrite.
Implement a kernel strdup() function (API synch with FreeBSD).
Add a stack-size argument to the LWKT threading code so threads can be
Add LWKT convenience functions lwkt_getpri() and lwkt_getpri_self().
Move kthread_create() from lwkt_thread.c to kern_kthread.c. Add a new
add the 'y' and 'Y' options to ps, and add the 'iac' keyword. The 'y'
Remove a recently added incorrect assertion. I was assuming that
Add a test-and-set and release poll function. This is really just a hack
Work to allow pure threads to issue VFS lookups: Only check p->p_ucred
Work to allow pure threads to issue VFS lookups: fp_open() uses proc0's
Work to allow pure threads to issue VFS lookups: (untested/experimental)
Sync the IFM_MAKEMODE() macro from FreeBSD-5.
Bring in NDIS emulation support from FreeBSD-5. NDIS is a Windows device
Generally speaking modules should unconditionally enable things like
NDIS_INFO -> NDIS_LOCK_INFO
Fix more __stdcall issues. Move the __stdcall into the function typedefs
Provide some basic instructions on how to create an NDIS wireless driver
Add the ndiscvt utility from FreeBSD-5, which is used to compile windows
Bring in the latest pkg_install sources from FreeBSD-5.
Change sendfile() to use the new m_ext callback scheme for cleaning up after
Get rid of mb_map. Retool the mbuf and mbuf cluster allocator to use
Since mbufs are no longer limited by an mb_map the kern.ipc.nmbufs and
Fix buggaboos that prevented ACPI_DEBUG from working.
taskqueue_create() should use M_INTWAIT rather then M_NOWAIT.
Add a global, clocks_running, which tells us when timeout/ticks based clocks
Rip out the badly designed softint-based taskqueue used by ACPI for callbacks.
Make doubly sure that timer2 is not used for speaker operation.
tcp_input()'s DELAY_ACK() code checks to see if the delayed ack timer is
The TCP stack is notified every time userland reads from the TCP socket
The obj hierarchy must be built before the ssh-etc target can be run
Merge from vendor branch CVS:
Bring cvs-1.12.9 into the CVS repository
Upgrade our CVS build from 1.12.8 to 1.12.9 to fix a number of pserver
make the __asm for the pushfl fakery __volatile.
GCC3.4's (default) unit-at-a-time optimization is incompatible with -mrtd.
eventhandler_register() M_NOWAIT->M_INTWAIT.
Add an event handler to adjust the cpu throttle state automatically when
Remove the unconditional timer_restore in the bios call path, it is
Do not hack a #define __FreeBSD_version if __FreeBSD__ does not exist,
Improve compatibility with older FreeBSD-4.x systems when cross-building
Add bzip2 to the bootstrap tools list. The compat libs (if enabled in
ffs_dirpref() calculates dirsize = (fs->fs_avgfilesize * fs->fs_avgfpdir).
The SF64-PCR card has no sound support but stupidly uses the same PCI id
Merge FreeBSD ip.c/1.101, commit message:
Synchronize syslogd with FreeBSD. Primarily syslogd.c/1.129. This primarily
Bring in FreeBSD mount.c/1.58, original commit message:
PPP stupidly hardwires some flag constants that it 'steals' from the mbuf
Discard accepted and pending connections after we detach the listen socket
Add a state to sanity check tcp_close() to make sure it is not called
Get rid of the NO_TCSH make.conf variable. We do not support removing
add 'read1', a program that tests reading one byte at a time from a file.
The base/count bounds checking was insufficient, leading to a kernel memory
Close a kernel mem disclosure bug in linprocfs. The uio_offset was not
Oops, undo portions of the last commit, some extra work got committed that
Have make upgrade remove two stale 80211 header files that can mess
VFS messaging/interfacing work stage 1/99. This stage replaces the old
Add a missing uio_td assignment (that unionfs needs).
Add an installer-fetchpkgs target and other related stuff to reduce the
VFS messaging/interfacing work stage 2/99. This stage retools the vnode ops
Properly record and print 64 bit file sizes, do not truncate the file size
Add some robustness to the error-requeue code. FreeBSD-5's (new) ata driver
Test commits after machine upgrade.
more testing
more testing 3
sigh
feh
feh2
*** empty log message ***
test
yet more testing
more testing
*** empty log message ***
*** empty log message ***
bleh
*** empty log message ***
*** empty log message ***
test
[test format strings for cvs server config]
Properly free the temporary sf_buf in uiomove_fromphys() if a copyin/copyout
Make the buildkernel and nativekernel targets completely wipe and regenerate
The -D__FreeBSD__ must be -D__FreeBSD__=4 or sendmail will not be properly
Add the vop_ops for ntfs before obtaining the root vnode(s) rather then
Remove the advertising clause where possible as per the directive from
Include language describing the prefered method for recognizing authors as
Add the standard DragonFly copyright with attribution to the author (which is
Add missing extension return value for __byte_swap32_var() in the case
The VFS work has made vnode_if.awk obsolete.
Fix a SFBUF memory leak in sendfile(). We were not properly tracking
M_EXT_CLUSTER was not being properly inherited in m_copym(), m_copypacket(),
doingdirectory is really a boolean, use an int rather then ino_t and cast
Minor cleanups, no operational changes other then to add an error message if
Minor cleanups.
Minor cleanups.
Minor cleanups. Also, change various exit(10) codes to exit(1).
Document the unorthordox use of getopt().
Check for a mkstemps() failure, generate a proper warning if the fopen()
Minor cleanups. Document -j in usage.
Cleanup various type-o's in comments.
Output an error message if the open fails.
Correct a mistake in the last commit that caused usage() to seg-fault,
Get rid of dfly/fbsd4/fbsd5 checks for the ntohl() return type. We are
VFS messaging/interfacing work stage 3/99: Bring in the journaling
Turn off the getty on ttyd0 by default to avoid certain machines from
Merge FreeBSD/1.212 and FreeBSD/1.213. These only appear to have an
Bring in FreeBSD/1.214 - UC Regent's advertising clause removal per
Bring in FreeBSD/1.218.
TCPS_CLOSED is no longer 0 in DragonFly. Because ipfilter was assuming
Sync VESA support with FreeBSD-CURRENT, adding support for cards that
Since ip_input() truncates the packet to ip->ip_len prior to entering the
VFS messaging/interfacing work stage 4/99. This stage goes a long ways
Fix handling of the recycling of vnodes from a failed hash collision.
Fix a bug that was causing a 'lockmgr: draining against myself' panic when
Minor style cleanups. Start a list of obsolete functions that should no
Add the -8 option to finger for 8-bit output (from NetBSD).
pkill does not compile cleanly, remove it from the build.
Remove some unnecessary extern's of libc functions.
With the 'cu' program removed from the system (long ago), enable 'cu'
Reenable pkill. There wasn't actually a problem with it, it turned out to
Add VESA mode support for syscons. The vesa.ko module must be loaded and
Fix a bug in sillyrename handling in nfs_inactive(). The code was improperly
Correct a bug introduced in a recent commit. This fixes touch -t.
Misc syntax cleanup.
Minor non-functional syntax changes. Remove (void) casts, and remove
Replace all occurences of strcpy by the safe strlcpy where needed.
Avoid WARNS=2 error by renaming the 'print' variable to 'printerr' to avoid
Make timed WARNS 2 compatible.
Remove #ifdef sgi and related conditionals. Remove improper libc prototypes.
Remove function casts to (void).
Set OPENSSH_USE_POSIX_THREADS to 1 to work around a privilege separation
Add a keyboard preference to kbd_init_struct(). When a wildcard keyboard is
Use the new keyboard preference feature.
Leave the USB keyboard registered with the keyboard subsystem even if the
NULL out p_stats when detaching the underlying thread stack, since the
The mount options matching code was incorrectly testing for string
Temporary hack to remove historical schg protection on 'cu' to avoid
With vnode locking now mandatory a number of bugs have cropped up in the
Implement a convenient lwkt_initport_null_rport() call which initializes
Fix a badly written preprocessor macro.
Make sio.S aware of when -mrtd is or is not being used, so it can implement
scalb() takes two doubles, not (double, int) (though there appear to be
Clean up struct session hold/rele management. The tty half-closed support
ANSIfication/cleanup, no functional changes.
Give the MP fields in the thread structure useful names for UP builds so
Add the 'M' status flag to indicate those proceses or threads which
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
When a umass storage device is unplugged there might be pending requests
Improve error reporting when the cdevsw code detects problems.
Fix cdevsw_remove() warnings related to the removal of mass media (e.g.
Don't complain when a cdevsw with non-zero refs is being removed if it still
'vidcontrol show' was broken by the better vesa support commit. Fix it.
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
Remove a redundant bzero (which also specified the wrong length in anycase
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_* [ Adding missing callout_init()'s ].
Move all the softclock/callout initialization out of MD and into MI. Get rid
Remove a superfluous (and incorrect due to the recent callout changes) bzero()
Unbreak the SCSI drivers. Move the callout_init() from xpt_get_ccb() to
Oops CALLOUT_DID_INIT had the same flags value as CALLOUT_MPSAFE, causing
*** empty log message ***
missing log message for last commit: Rearrange the mbuf clearing code in
Fix a race on SMP systems. Since we might block while obtaining the MP
Make the freeing free mbuf assertion a bit more verbose.
Fix a number of races. First, retain PG_BUSY through a vm_page_remove(),
Add a missing free to a failure case (non critical).
Fix a type-o / syntax error in the last commit.
Add missing callout_init().
Joerg pointed out that callout_init is called by the SYSINIT. Reorganize
Fix the ncpu check for 'ps' so it does not display the 'M' flag on UP
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_* [structural fields appeared not to be used,
timeout/untimeout ==> callout_*
malloc() M_NOWAIT -> M_WAITOK, plus remove bzero's in favor of M_ZERO.
timeout/untimeout ==> callout_*
Add DragonFly-stable-supfile and cleanup the comments in it and
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
cleanup syntax a bit. non functional changes.
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*
timeout/untimeout ==> callout_*. NDIS hides the callout in a windoz
Remove timeout() and untimeout() plus all related support. Not having to
testing
Quick cleanup in preparation for a more complete cleanup.
Add a missing #include and add a missing semicolon.
Do not unconditionally fork() after accept(). accept() can return -1 due
if_pfsync.h needs pfvar.h for struct pf_addr. The ioctl header collection
Use the appropriate #defines instead of hard-coding constants.
Here are (BSD licensed) replacements for bc and dc:
M_NOWAIT -> M_INTWAIT/M_WAITOK. Plus fix a NULL free() that can occur if
M_NOWAIT -> M_WAITOK
Hook the new bc and dc into the tree. Unhook the old gnu bc/dc and add
Fixup the location of openssl / fix release builds.
Remove unused variable.
Add the -L and -l options to install. -L allows an alternative /etc directory
Add support for the LinkSys EG1032 GigE PCI card, and add support for
The inode size must be unsigned-extended to a quad, not sign-extended, so
Fix a bug where DragonFly's nat was closing TCP connections every 10 minutes.
Only assume a numeric group id if the ENTIRE string is numeric.
VFS messaging/interfacing work stage 5/99. Start work on the new
Document the additional vendors applicable to the 'sk' driver.
Get rid of some conditionalized code which the pmap invalidation API took
The wrong vendor-id was used when adding Adaptec ServeRAID Adapter support.
VFS messaging/interfacing work stage 5/99. Start work on the new
Clarify an element in the BUGS section related to the sticky bit.
VFS messaging/interfacing work stage 5b/99. More cleanups, remove the
Minor cleanups to TIMER_USE_1 (no real fixes or anything).
VFS messaging/interfacing work stage 6/99. Populate and maintain the
Attempt to make the boot code operate in a more deterministic fashion.
VFS messaging/interfacing work stage 7/99. BEGIN DESTABILIZATION!
VFS messaging/interfacing work stage 7a/99: Firm up stage 7 a bit by
Remove ATTY.
Unsetting an undefined variable or function is not an error.
Enforce directory creation ordering for subdirectories to fix a make -j
VFS messaging/interfacing work stage 7b/99: More firming up of stage 7.
Make patch's -b option match the old patch's -b option, at least for now,
Add a debugging utility which dumps that kernel's namecache topology.
Only include opt_vmpage.h if _KERNEL is defined.
VFS messaging/interfacing work stage 7c/99: More firming up of stage 7.
VFS messaging/interfacing work stage 7d/99: More firming up of stage 7.
VFS messaging/interfacing work stage 7e/99: More firming up of stage 7.
Update ncptrace.c to handle DragonFly_Stable and HEAD.
Add the -I option to rm. This option asks for confirmation just once if
Do not run getty on ttyd0 by default. The installer's version of ttys already
alias rm to rm -I by default in interactive shells. rm -I is a safer, more
Last commit inspired by: Giorgos Keramidas <keramida@freebsd.org>, who
If we get an EOF in check2(), terminate the yes/no question loop rather
VFS messaging/interfacing work stage 7e/99: More firming up of stage 7.
Add vnodeinfo - a program which scans each mount's vnode list and dumps
VFS messaging/interfacing work stage 8a/99: Sync other filesystems to stage 7
Fix a bug in the tty clist code. The clist code was only protecting itself
The last commit was improperly documented as being VFS related. It was
VFS messaging/interfacing work stage 7f/99: More firming up of stage 7.
VFS messaging/interfacing work stage 7g/99: More firming up of stage 7.
Fix the -N and M options for ncptrace.
VFS messaging/interfacing work stage 7h/99: More firming up of stage 7.
Conditionalize the flag->string conversions so pstat compiles for both the
Add conditionals so vnodeinfo compiles for both old and new api kernels.
Add the standard DragonFly copyright.
Bring in a fix from NetBSD for hid_report_size(). This fixes a detection
Fix a USB stuttering key issue.
Add a missing agp_generic_detach() call if a bad initial aperture size
Make a DFly buildworld work on a FreeBSD-4.x system again by fixing
VFS messaging/interfacing work stage 8/99: Major reworking of the vnode
Try to close an occassional VM page related panic that is believed to occur
test commits@ list 1/2
test commit@ list 2/2
Make a chdir failure fatal rather then just a warning, otherwise pax could
Forgot to add for last commit:
Avoid redefined symbol warning when libcaps uses thread.h with its own
Do not use non-blocking malloc()'s in the busdma support code. A lot of
Initialize the 'kernel' environment variable from loader.4th
When pxebooted loader is detected not only change the kernel name, but
Add /boot/defaults/loader-bootp.conf, a separate default configuration
Final cleanup. After giving up on trying to avoid having two loader*.conf
Remove the 'ether' module dependancy, it is not defined anywhere and will
Add the "nv" interface, and enable it in the bootp case since the netif/nv
The forth code is a real mess, things are repeated all over the place.
Augment vnodeinfo to retrieve and display the number of resident pages in
*** empty log message ***
Add devices da4-da15 and ad4-ad7 to MAKEDEVs 'all' for convenience. Plug-in
Create a softlink from /kernel to /kernel.BOOTP on the CD. Mount
Implement hotchar support for ucom.
Fix a bug where sc_ctlp() is improperly called when the packet is passed up
Adjust rm's usage and manual page synopsis to include the new -I option.
oops, forgot a mention. Last commit:
MFC FreeBSD if_em.c 1.48 and 1.49.
Fix bugs in the limiting code for negative-hit namecache entries. The system
Fix an assertion, vgone() now requires that the vnode be VX locked and refd.
Old API compatibility: The directory vnode passed to VOP_LOOKUP() must be
The last commit failed to adjust the alignment like it said it did.
Fix a boot panic with the amd device. We inherited some busdma code from
Change the default handling for kernels built with debugging info (DEBUG=-g).
Set the file modes for those rc files which are unconditionally replaced by
Fix a final bug in the vfs cache cleaning code. An incorrect assertion was
Make an existing vnode lock assertion a bit more verbose.
The old lookup() API is extremely complex. Even though it will be ripped out
This is a really nice rewrite of rc.firewall that cleans it up and adds
Fix bugs in the last commit. Some islink checks and a fd->fdval change was
Generate the BRANCH field from the current tag, use CURRENT if the current
Synchronize a few libi386 issues from FreeBSD. Fix a bounce buffer bug,
Add a default for X_WINDOW_SYSTEM in the dfports override case.
Synchronize bug fixes from FreeBSD/RELENG_4.
Fix a compiler warning by pre-declaring struct vnode;
Remove the vfs page replacement optimization and its ENABLE_VFS_IOOPT option.
Document bus_dmamem_alloc() a bit more.
Fix bugs in the vm_map_entry reservation and zalloc code. This code is a bit
Fix a seg-fault if -l is used without a /dev prefix, e.g. cu -l cuaa0. Of
Add a section to UPDATING describing the users and groups that might have
Fix a NULL pointer dereference panic that occurs when the TCP protocol
null_revoke() needs to return 0.
Remove unused variable.
unmount was not removing the negative hits associated with a mount point.
Conditionalize _CTYPE_SW* for bootstrap purposes.
The last fix wasn't good enough. This one causes the SWIDTH lines to be
The PRId64 check was wrong, causing bootstrapping failures on
Bring in Jeff Wheelhouse's CLOG / circular log file support for syslogd,
Fix 'route add -host <target> -interface <interface_name>. This was
Remove incorrect cache_purge() calls in *_rmdir() (OLD API). These could
Note last commit: changes were made primarily to avoid global variable name
Fix a possible remote DOS against pppd, described in detail at
Save and restore the 'version' counter file when doing a full buildkernel
Add humanize_number(3) and split the trimdomain(3) function out from
Bring in some more PCI ID's from FreeBSD.
Give makewhatis an absolute path to make upgrading etc from single-user
Fix a Trident DMA limitation. Symptom as reported by Brock was that his
Remove an assertion that used to double-check the cred passed into vn_open().
Correct a typo and a mdoc(7) style issue.
The last commit created a memory leak because 'buf' is static. Fix that,
Add tip(1)'s emulation of cu(1) to tip's manual page.
Sync make.1 with the rest of the make source.
Do not explicitly set PCIM_CMD_SERRESPEN or PCIM_CMD_PERRESPEN. This was
Correct a softupdates bug, an ir_savebp buffer was not being properly
vfs_object_create() was being called too early on devvp in the FFS mount
Fix another minor bug-a-boo inherited from 4.x sources, the wrong indirdep
Create a non-blocking version of BUF_REFCNT() called BUF_REFCNTNB() to be
Add lockcountnb() - a non-blocking version of lockcount() to be used only
vmpageinfo is a program which runs through the vm_page_array and the
The min() and max() macros in sys/libkern.h are typed u_int and thus do not
Fix a very serious bug in contigmalloc() which we inherited from FreeBSD-4.x.
The 'start = 1' change wasn't needed, revert that one little bit back to
VFS messaging/interfacing work stage 9/99: VFS 'NEW' API WORK.
Correct minor tinderbox -DDEBUG error.
Add a 'preupgrade' target which creates any missing users and groups
Temporarily remove union and nullfs from the vfs build list, they are
Default vfs.fastdev to 1 for wider testing, so the vnode bypass for device
Forced commit, correct comment for last commit. 1.25 has nothing to do with
Add a missing com_unlock() to the serial port drain test code. This code
Remove various forms of NULL, and cleanup types. This is a partial sync from
The strings package passes back a pointer via brk_string() to internal
efree()->free(). remove #define efree (which was defined to free).
Do some deregisterization.
Partial sync from FreeBSD/jmallet:
Diff reduction for great justice against NetBSD, cast to unsigned char when
Make the DEBUGF() macro portable by (ugh) adding a Debug() function, which
Convert make(1) to use ANSI style function declarations. Variable
Possibly expand the variable name's embedded variables before using it, as
Split var.c into var.c and var_modify.c and move all the modification funcs
Split var.c into var.c and var_modify.c and move all the modification funcs
Make both arguments to str_concat() const char *'s and remove STR_DOFREE
Spelling corrections.
Add support for the ATI Radeon 9600 XT and XT_S.
* Restore indentation to tabs.
Partial sync from FreeBSD, add dummy syscalls for extended attribte
direct sysargs to /dev/null for the emualtion system call sets to avoid a
Continuing synchronization from FreeBSD.
Fix an inverted conditional which could lead to nameBuf being truncated in
Temporarily change the net.inet.tcp.sack default from 1 to 0 after confirmed
Fix a bug in the checking of malloc()'s return value. It turns out to have
Cave in and remove NULL checks for M_WAITOK mallocs. DragonFly's M_WAITOK
Follow NOFSCHG if defined. (It's needed to be able to run make installworld
Remove unused junk from the slab allocator.
Do a cleanup pass on the mbuf allocator. Reorder the mmbfree cache tests
Bring in various fixes from FreeBSD:
Clear the NOCORE flag on any text mappings that the RTLD modifies due to
Lots of bug fixes to the checkpointing code. The big fix is that you can
Cleanup some dangling issues with cache_inval(). A lot of hard work went
Re-enable SACK by default. Jeff fixed the corruption issue reported by
Give init the ability to chroot to a directory based on kernel environment
Fix a conditional. sdl was not unconditionally being checked for NULL.
Add code to the BIOS VM86 emulator to detect writes to the 8254. If a
sendfile() was seriously broken. It was calling vm_page_free() without
Fix a number of SMP issues.
Cleanup the 'cache_lock: blocked on..' warning message. Fix a minor
Add support for adjusting the interrupt throttling rate via
The fp argument to vn_open() is optional but the code wasn't treating it
Fix the PRI field to not display bogus process priorities for pure
Fix a bug in chown, chmod, and chflags. When the setfflags(), setffown(),
Document the mandatory use of vget() prior to modifying vnode/inode ops.
There is enough demand for Kip Macy's checkpointing code to warrent
Add the sys_checkpoint(2) manual page and expand checkpt(1).
Remove checkpt/ from Makefile.modules. checkpt is not integrated into the
Remove the checkpoint module. checkpointing is now integrated into the kernel
patch-4.10: Clean some includes and remove ifdef __STDC__, -Wall cleanup,
o Print a warning when we are given two scripts for one target.
o Pacify ``make -f /dev/null -V FOO''.
o Fix proto type
o Give make(1) the ability to use KQUEUE to wait for worker
Cleanup some ESTALE issues on the client when files are replaced on
Fix a bug in ESTALE handling for NFS. If we get ESTALE in vn_open() we
Fix the boottime calculation when the time of day is set in absolute terms.
Unlock the namecache record when traversing a mount point, then relock and
test
Bring in FreeBSD/1.206 by Alan Cox, bde@, and tegge@:
Bring in elements from the FreeBSD usbdevs that the DFly usbdevs does not
vm_page_free_*() now requires the page to be busied, fix a case in
Do not reinitialize the translation mode if reattaching to an existing
Document vmpageinfo.c
Fix a diagnostic check related to new VOP_INACTIVE semantics.
Fix field for recent radix.h cleanups.
Remove bool and boolean_t typedefs from header files where they don't belong,
Correct a bug where incoming connections do not properly initialize the
'bool' is a really bad name for a variable, rename it.
Fix the keyboard lockup problem. There were two big clues: The first was a
VFS messaging/interfacing work stage 10/99:
Synchronize usbdevs with NetBSD and regenerate.
Cleanup missing and duplicate defines from the last commit, rename an
Temporarily allow recursion on locks to deal with a double lock in the
The dos slice scanner was incorrectly including extended partition entries
There seems to be a race during shutdown where ifa->ifa_addr can become
Correct conditional which would always make kcore_open() fail and return
Move the CCVER override for the release build from a make command line
Remove bogus DIAGNOSTIC code that checked if the process was SZOMB or SRUN
Add missing ..
Fix format %lx->%x for ntohl conversions in diagnostic warning printfs.
Fix duplicate script warnings byt remove a duplicate .include of bsd.subdir.mk.
Fix printf format specifier from %lx->%x for ntohl argument.
Remove duplicate _EXTRADEPEND entry. bsd.lib.mk already has one. It is
Remove duplicate .include of bsd.subdir.mk. It is already indirectly
The ../Makefile.inc chain may be .included early (it is also included by
Clean up "`cache' might be used uninitialized" warnings. These come from
Move the doscmd: depdendancy to after the .include so it does not
Correct the make target the user is told to use to fetch required
Remove confusing comment.
Make sure that cn_flags is properly updated to account for side effects
The grouplist variable made local in the last commit was not being NULLed
Do not loop 3 times in the interrupt processing code, it's an unnecessary
Remove now-unused loop_cnt variable.
Add a missing initialization for the error variable which resulted in netstat
Add the -P flag which displays more PCB information (in particular, TCP).
Make sure ntfs_lookup() has the correct side effects by ensuring that
Try to be a bit smarter when closing down a ULPT device.
Silence a compiler warning by adding parenthesis.
Journaling layer work. Add a new system call, mountctl, which will be used
Properly recognize a YUKON device's on-board ram.
Fix to make jumbo frames work properly.
Do not allow '.' or '..' to be specified as the last path component when
Add support for tail -f'ing multiple files. part 1/2.
Improve the printing of ==> filename <==. Do not print the same filename
Fix a range check bug in lseek()
Journaling layer work.
Rename the proc pointer p to pp to clarify its use in portions of the tty
nvp may fall through as NULL, check it before vput()ing.
Get rid of dead non-DragonFly code.
Journaling layer work. Lock down the journaling data format and most
WARNS 6, cleanup compiler warnings, de-register, staticize, etc.
Restore b_data prior to calling relpbuf(). This isn't really necessary but
Journaling layer work. Add shims and skeleton code for most of the
WARNS 2->6 and a minor code readability cleanup.
Make sure the temporary .c file generated from the .y file is properly
Fix a kernel crash that occurs when the SMB protocol stack is used. The
Merge of FreeBSD rev. 1.36+1.37 of ip_nat.c. Conditionalize declarations
Correct two bugs that may result in incorrect CBCP response for
While removing a memory leak, rev 1.32 introduced a
Do not specify the -B option when executing the sub-make. In the BSD
Fix one of probably several smbfs issues. smbfs is improperly tracking
Fix a memory leak in regex.
This is the initial skeleton for the new mountctl utility. Manual page,
Add support for retrieving the journal status via mountctl. Increase some
Revamp the argument format a bit, add basic suppot for creating, deleting,
Add /sbin/mountctl to the build.
When re-connecting an already connected datagram socket be sure to clean
Oops, undo accidental commit. The last commit was not related to the
proc0 is still used by e.g. smbfs to fork off a kernel thread and certain
Do not early terminate if ^C is hit just as a valid job is returned by
Followup note last commit: FreeBSD PR/66242, FreeBSD/1.68 originally
Add syscall primitives for generic userland accessible sleep/wakeup
Add missing kern_umtx.c to sys/conf/files.
Add umtx.c, a simple utility which implements userland mutexes using
Minor correction in umtx_*() calls, the mutex pointer should point to
Minor correction in umtx_*() calls, the mutex pointer should point to
falloc() was not returning an error code on failure.
When a PCMCIA networking card is removed the IF code may free() the network
Add some descriptive comments.
Change lockcount->lockcountnb in an assertion.
Properly vget() vnodes that the syncer intends to VOP_FSYNC(), rather then
Replace the cache-point linear search algorithm for VM map entries with
The vnode reclamation code contains a race whereby a blocking condition may
Repo-copy vinumparser.c and vinumutil.c from /usr/src/sys/dev/raid/vinum
Disable hardware checksum support by default, it produces packet corruption.
Tell the user more explicitly what port needs to be installed to get the
Mount points use a special empty namecache entry to transition from one
Do not leave VCTTYISOPEN set if our attempt to open /dev/tty fails, otherwise
Fix the virtual 'status' file for procfs. The wrong length was being used,
getblk() has an old crufty API in which the logical block size is not a
Redo argv processing to better conform to standards. A NULL argv is no
Add a sysctl to control 8254 bios overwrite warnings. Default is on.
Take advantage of our new namecache topology to generate the actual paths
Remove _THREAD_SAFE depenendancies. Create weakly associated stubs for
encap_getarg() was not properly loading the pointer argument associated
Fix bug in last commit that broke 'df'. 'sfsp' is now a structural pointer
Count time spent in interrupts in the overall runtime calculation so ps
Fix an invariant test that tries to catch locked tokens being left on the
Add a intrmask_t pointer to register_int() and register_swi(), and make
Fix type-o in last commit.
The last commit broke ttyname(), which broke ssh -1. Fix that, plus
Minor adustments to avoid a signed/unsigned comparison compiler warning
Fix a hard-to-find bugaboo in the struct file list sysctl. The code was
Generate more useful -v information on the console during device attach.
m_clalloc() was improperly assuming that an mcl malloc would always succeed
Annotate the class byte with a class name in the bootverbose pci "found->"
One of the last things the system does before it tries to mount root is
Revert the last device_print_child() change, it was too confusing to
The last commit was not sufficient. Rework the code a bit to make it
gdb-6 uses /dev/kmem exclusively for kernel addresses when gdb'ing a live
Add support for pure kernel thread stack frames. Pure kernel threads do not
Up the initial interrupt configuration hook delay to 20 seconds before
Fix a case that can prevent the vnlru_proc vnode recycler from operating.
Fix a cache_resolve() vs cache_inval() race which can result in a livelock.
Greatly reduce the size of ISOFS's inode hash table. CDs and DVDs are small
Add support for the Intel 82562ET/EZ/GT/GZ (ICH6/ICH6R) Pro/100 VE Ethernet.
rcorder tries real hard to free things while processing the list but this
Track the last read and last write timestamp at the device level and modify
Add a manual page for the umtx_*() system calls.
Implement TLS support, tls manual pages, and link the umtx and tls manual
Implement TLS support, tls manual pages, and link the umtx and tls manual
Add system call prototypes for userland.
Improve the contigmalloc() memory allocator. Fix a starting index bug,
Add DFly copyright to cache.c. This file was originally my creation in
Have the getroot script chdir into /etc/namedb itself instead of relying
Fix an issue that the iwi driver seems to hit, that of routing socket
Rewrite the loops which extract the interpreter name and arguments out
Add signal mask save/restore to the checkpoint code. Reorder the file
Fix a firewall rule ordering problem, the 'OPEN' firewall mode was
Allow the #! command line to be up to PAGE_SIZE long, rather then
Journaling layer work. Generate output for path names, creds, and vattr,
Clean up the XIO API and structure. XIO no longer tries to 'track' partial
PHOLD is a throwback to FreeBSD that we don't actually need here. This
off_t is a signed value. The last commit caused the kernel to fail with
Also, do not use M_NOWAIT for an allocation that must succeed.
This is a major revamping of our MSFBUF API. MSFBUFs are used to map
Pass the memfifo option to the kernel.
Make msf_buf_kva() return the correct base address. It was not including
msf_buf_kva() now returns a char * rather then a vm_offset_t.
Journaling layer work. Write the actual data associated with a VOP_WRITE
Remove some of the nastier debugging printfs.
msf_buf_alloc() can be called with a NULL vm_page_t when e.g. the first
Fix an incorrect pointer in a journal_build_pad() that led to a panic.
Add a simple msf_buf_bytes() inline which returns the number of bytes
Fix a number of alignment and offset bugs that were corrupting the
Replace references to TurtleBSD with DragonFlyBSD. Turtle was one of the
Bring in the IWI driver from FreeBSD and merge in ALTQ support.
Do some WARNS 6 cleanups. Add __unused, add a few type casts, and
Bring in the IWI driver from FreeBSD and merge in ALTQ support.
Bring in Damien's IPW driver.
Change NOWAITs into WAITOKs. In FreeBSD NOWAIT allocations seem to be
First cut of the jscan utility. This will become the core utility for
Cleanup the debug output, fix a few data conversion issues.
Sync with FreeBSD/1.103. In the softupdates case for ffs_truncate() we
Add a couple more PCI IDs for the Intel ICH5 ATA100, ICH6 SATA150,
pipe->pipe_buffer.out was not being reset to 0 when switching from direct
Fix a serious bug in cache_inval_vp()'s TAILQ iteration through v_namecache.
Add a function that returns an approximate time_t for realtime for
Fix a couple of NFS client side caching snafus. First, we have to update
Implement CLOCK_MONOTONIC using getnanouptime(), which in DragonFly is
Bring in the minix editor 'mined', in its original form except for necessary
Bring in PicoBSD changes from the FreeBSD Attic.
Make ^L redraw the screen, like people expect.
Support XTERM function key escape sequences.
Add 'mined' to the bin build and add a minimal manual page for it.
flesh out a tiny bit so people at least know what the help key is.
Bring in some CAM bug fixes from FreeBSD.
src/games WARNS 6 cleanups.
Fix a bugaboo in the last commit. We cannot safely modify n_mtime based
Fix a potential security hole by outputing non-printable characters related
Close a possible security hole by using strvis() when displaying td_comm.
Clean up a number of caching edge cases in NFS, rework the code to be
Ignore additional stty control characters like ^Y which interfere with
Add RCNG support for setting the negative attribute cache timeout, make
Add a sysctl "net.inet.tcp.aggregate_acks" which controls the tcp ack
Add Windoz autorun support to the CD to automatically pop up a browser with a
Fix a crash in the firewire / DCONS code that occurs when the crom data
Pick up changes in rev 1.8 of src/sys/dev/ic/mpt_netbsd.c from NetBSD.
Rewrite a good chunk of MAKEDEV, commonizing paths, cleaning up device
Fix a minor bug in pass*) generation. 'units' was not properly optioned.
Do not allow the journaling descriptor to be a regular file on the same
Start working on the full-duplex journaling feature, where the target can
Reverse the order of dragonfly and FreeBSD since it says right after
Allow an info.size field of -1 to indicate that the descriptor should map
Because destroy_all_dev() checks the mask/match against the device's si_udev,
Correct the th_off check against ip_len. The check in ip_demux occurs
Additional note to last commit. GCC-3.4 improperly generates a warning
Remove an assertion in bundirty() that requires the buffer to not be on
Have the server complain if the NFS rpc stream from a client gets corrupted,
Fix a server-side TCP NFS mount stream sequencing problem which could result
Display additional information about a vnode's object.
Sync with the kernel to make ncptrace work again. CINV_PARENT no longer
Fix a bug where the main system clock stops operating. We were using a field
Do a better job distributing RPC requests on the NFS server side. This
The old ntpdate displayed the clock error by default. Change rdate to do
Temporarily back out the last change due to time zone display issues
loginfo/commitinfo test
test 2
Add another syscall test, and test cvs commits after loginfo/commitinfo/config
Don't hide the library build's ar/ranlib lines, or shared library link
Cleanup and retool portions of the TLS support and make sure that
Move the setting of sopt->sopt_dir down into the lower level sogetopt()
The vnode recycler was not handling a hardlinking case that is capable
WARNS=6 cleanup.
General WARNS cleanups: staticize functions. Adjust strdup()
WARNS cleanup of array initializer.
WARNS=6 cleanup.
WARNS=6 cleanup.
Bring in some work from FreeBSD:
A cache invalidation race that generates a warning was not properly
WARNS?=6 cleanup for fstat.
Add the 'webstress' utility. This program will fork a number of child
A program which dumps the wildcard hash table for the TCP protocol for
Fix a bug in the distributed PCB wildcardhash code for TCP. For the SMP
Apply same bug fix as last commit to IPV6.
Automatically calculate the kernel revision for release tags by extracting
Make sure neither the boot loader nor the kernel is compiled with the
Add cases for the nvidia nForce3 PRO and PRO S1. Who knows how well it
Add some minor in-code documentation.
libstand is used by the boot loader, make sure the stack protector is
Correct an NFS bug related to ftruncate() operations. When the client
Include fsid information in the mount point dump.
Update the various DragonFly supfiles, rename some of them to make their
add ports/polish
The cdrom MNT_ROOTFS check was interfering with the NFS export handling,
Add a hardwired dhcpd.conf for the installer's netboot server support.
Sync up to the latest installer packages (with netboot server capabilities!).
The release is 1.2, bump the HEAD of the tree to 1.3 for post-release
Bump HEAD's official release #define to 130000.
Clearly state when ACPI is overriding APM's device entries.
Optimize lwkt_send_ipiq() - the IPI based inter-cpu messaging routine.
Only bump the switch_count counter when lwkt_switch() actually switches
Zero the whole structure, not just the name, so we don't trip up on this
Implement Red-Black trees for the vnode clean/dirty buffer lists.
Fix a bug in the last commit. A FIFO calculation was incorrect (3/2 instead
staticize lwkt_reqtoken_remote().
CVS was seg-faulting on systems with hostnames greater then 34 characters
Abstract out the routines which manipulate the mountlist.
NULL-out two stack-declared variables that were incorrectly assumed to
Get rid of VPLACEMARKER and retool vmntvnodescan() to use a secondary
Make the kernel malloc() and free() MP safe by pushing the BGL past the
Document the rwlock routines as being MP safe.
Add some convenient targets for cvs maintanance by the cvs meister.
Make access to basetime MP safe and interrupt-race safe by using a simple
Fix the structural type for kern.boottime and kern.basetime. Fix basetime
Initial commit for the DragonFly home-made ntpd client. Why? Because
Correct some packet ntohs/htons issues. Fix a documentation bug.
Cleanup compiler warnings. Generate cleaner debug output. Change the
Fix an overflow in the delta time calculation. A double as incorrectly
Allows 16 samples with a correllation >= 0.96 in addition to the
Change client_check() to calculate the best offset and the best frequency
Implement a variable polling rate capability.
Don't issue a frequency correction if it is the same as the last
Implement a course offset adjustment for large time steps.
Add information on interrupt preemptions by the current thread on any given
Make the trace code a bit smarter. It now looks around for something
Add config file support, daemon() backgrounding, pidfile,
Redo the way the thread priority is displayed. Split out the critical
Fix an SMP bug. The SMP startup code waits for the APs to finish
Minor rearrangement of an mpcount. This isn't actually a bug because the
On MP systems, malloc's using R_NOWAIT are not supposed to block, so don't
Add -n (not for real) option. The program goes through motions, but
Do not try to collect offset data if a prior offset correction is still
Correct a bug in the last two commits. The time_second global was not
Implement -s/-S (do quick course adjustment on startup).
Add a manual page, clean up the options display, and link dntpd into the
Clean up the manual page, correct spelling, add additional information.
Final forced commit to correct log comments. char's are *signed*, not
Report on segmentation violations (from testing file mmap) rather then
A program which sets up a TLS segment and then loops with random sleeps
Add a binary library compatibility infrastructure. Library updates are
The library compat Makefile needs 'tail'.
If a process does not block between setting up %gs and fork()ing, a
Adjust the upgrade target to remove libraries from /usr/lib that exist in
incore() is used to detect logical block number collisions, and other
Initial commit of the generation utility and configuration files for the
Give the DEVELOPMENT branch a subversion so we can do build/installworld
Do not allow installworld to proceed if the currently running kernel
Major TLS cleanups.
Document additional work in last commit. Bumped library to ld-elf.so.2
Add NEWBUS infrastructure for interrupt enablement and disablement. This
Have the EM device call the new interrupt enablement and disablement
Do better range checking on the LDT. FreeBSD-SA-05:07.ldt
Get rid of bus_{disable,enable}_intr(), it wasn't generic enough for
Get rid of the bad hack that was doing network polling from the trap code.
(add missing file related to last commit)
Rewrite the polling code. Instead of trying to do fancy polling enablement
Fix a bugaboo in the last commit. We can't move to a generic mux yet
Change ifconfig 'poll' to ifconfig 'polling', the same keyword that it
Get rid of IFCAP_POLLING for now.
Properly conditionalize a call to ether_poll_deregister via
Fix a race in the serializer's use of atomic_intr_cond_enter(). When
Remove some debugging that crept int othe last commit.
Fix a bug in the serializer's race detection code. It is possible for
Properly initialize the serializer by calling lwkt_serialize_init().
Implement a new cputimer infrastructure to allow us to support different
More cleanups, add the API implementation to select the system clock.
Disable the ability to change the system clock with a sysctl. More
Use the ACPI timer as the system clock if possible. This should free up
Add a simple API tha allows the interrupt timer to efficiently convert
Fix a recursive clock_lock() on SMP systems which was deadlocking on boot.
Fix a bugaboo in the last commit. The wildcard patterns were not accounting
Only do the OS version requirements check if DESTDIR is non-existant,
Clean up type-o's.
Remove spl*() calls from the bus/ infrastructure, replacing them with
Conditionalize an #include so libcam can build the file.
Remove spl*() calls from the atm code, replacing them with critical sections.
Remove spl*() calls from the crypto device driver, replacing them with
Conditionalize thread2.h so /usr/src/usr.bin/fstat can include vm_object.h.
Remove variable names from procedure declarations so userland doesn't
Fix a warning by conditionalizing a forward declaration.
After some thought, replace the splhigh()/spl0() combination in swapout()
Remove spl*() calls from the netproto/atm driver, replacing them with
Remove spl*() calls from netinet, replacing them with critical sections.
Make -I only apply to rm's run in the foreground. Silently discard it if
Remove spl*() calls from net/i4b, replacing them with critical sections.
Remove spl*() calls from i386, replacing them with critical sections.
cpu_mb2() needs to load %eax with $0 before running cpuid.
Add a sysctl, debug.use_malloc_pattern, that explicitly initializes data
Augment the panic when attempting to switch from a FAST interrupt to include
vm_contig_pg_free() must busy the page before freeing it in the case
Replace cpu_mb1() and cpu_mb2() with cpu_mfence(), cpu_lfence(), cpu_sfence(),
The acpi module was failing to load due to exposed crit_*() functions. Add
Remove spl*() calls from kern, replacing them with critical sections.
Add some missing #include's from the last commit.
Get rid of an unused variable due to the last commit.
Reorder code in m_chtype() to properly decrement the mbtype stats. Before
Temporary hack to fix interrupt race when decrementing a shared
Handle the case where the version file might be empty, which can occur
Rollup mbuf/objcache fixes.
Add a missing lwkt_reltoken() in the NULL return path. Do not count NULL
Fix a bug in the mbstats accounting. m_mbufs was being decremented for
Attempt to avoid a livelocked USB interrupt during boot by not enabling
Another terrible hack to leave interrupts disabled until the USB bus
Replace SPLs with critical sections in the opencrypto code.
spl->critical section conversion.
spl->critical section conversion.
spl->critical section conversion.
spl->critical section conversion.
spl->critical section conversion
spl->critical section conversion.
spl->critical section conversion, plus remove some macros which are now
spl->critical section conversion.
spl->critical section conversions.
spl->critical section conversion.
Add missing #include <thread2.h> to support the critical section calls.
Move sys/buf2.h and sys/thread2.h into the #ifdef _KERNEL section.
Fix mismatched crit_*() pair.
Add additional sanity checks, remove unused arguments to vm_page_startup().
Remove illegal parens from #ifndef.
The callout timer init code was using { SI_SUB_CPU , SI_ORDER_FIRST }.
Rip out bad spl manipulating junk from mpt and clean it up.
spl->critical section conversion.
spl->critical section conversion. Also fixes a missed spl in DGM.
spl->critical section conversion.
spl->critical section conversion.
Remove all remaining SPL code. Replace the mtd_cpl field in the machine
When cleaning an mbuf header for reinsertion into the objcache, make sure
Abstract out the location of an m_tag's data by adding a m_tag_data() inline.
Fix two bugs in the LWKT token code.
Add more magic numbers for the token code.
Fix a bug in the physmap[] array limit calculation and rewrite portions of
Add a missing crit_exit(). The code path in question only got executed
Fix a serious SMP bug. The RWLOCK support used by dev/raid/aac,
Tokens are recursive in the context of the same thread. This also means
Add a DEBUG_TOKENS option which causes token operations to be logged to
Introduce an ultra-simple, non-overlapping, int-aligned bcopy called bcopyi().
Reimplement the kernel tracepoint facility. The new implementation is
Use the KTR facility to trace token operations.
Add a caller backtrace feature (enabled by default), which records part of
Use KTR's built-in call chain recording rather then hacking it up ourselves.
Include a bitmap of allocated entries when built with INVARIANTS. I
Add KTR support to the slab allocator. Track malloc's, free's, oversized
Correct a missing macro element for the NON-KTR case.
Correct KTR masks for memory logging.
Rewrite a good chunk of the ktrdump utility to work with the new DragonFly
Add additional sanity checks to IPIQ processing, do some cleanups,
Add KTR support to the IPIQ code.
Have ktrdump run 'nm' on the kernel execfile and translate the caller1,2
Bump fd_lastfile, freefile, and refcnt to 32 bit ints. Also bump cmask
File descriptor cleanup stage 2, remove the separate arrays for file
Synchronize the fstat program with recent file descriptor cleanups.
Synchronize libkcore with recent file descriptor cleanups.
Synchronize the ipfilter contrib code with recent file descriptor cleanups.
Randomize the initial stack pointer for a user process. Introduce a
Document cleanvar_enable in rc.conf.5 and document the purge code
The recent file descriptor work is significant enough to deserve a
Fix a race between fork() and ^Z. If the ^Z is handled just as the forked
Do a quick cleanup pass on the userland scheduler and move resetpriority()
Repo-copy kern_switch.c to usched_4bsd.c, remove kern_switch.c, and point
Remove unused variables (from prior spl->critical section conversion)
Associate a userland scheduler control structure with every process and
Cleanup indentation, no operational changes.
Move remaining scheduler-specific functions into the usched abstraction.
Move more scheduler-specific defines from various places into usched_bsd4.c
* Remove a procedural layer in the scheduler clock code by having
Print out additional information for a magic number failure assertion.
Allow the CTYPE macros to be disabled, forcing procedure calls to be used
Remove an assertion that does not compile due to a lack of a KERNLOAD
The recent commit to propogate kernel options to modules enabled the
Yet more scheduler work. Revamp the batch detection heuristic and fix a few
Re-commit usched_bsd4.c (losing the history) to try to fix a repository
Do not abort the entire core dump if VOP_VPTOFH() fails. VPTOFH is not
The pipe code was not properly handling kernel space writes. Such writes
Fix a few issues in the kernel-side journal.
The size of a nesting record may not be known (due to the virtual stream
Major continuing work on jscan, the userland backend for the journaling
* Fix a number of alignment errors that was causing garbage to be parsed.
Remove some debugging printfs and fix a bug where libc's fread() returns
Generate the correct referential data when journaling hardlinks.
Add support for mirroring symlinks and hardlinks.
Revert the last commit until a better solution can be found, it breaks
Add another argument to fp_read() to tell the kernel to read the entire
Implement the full-duplex ack protocol. refurbish some of the memory
Bring mountctl up-to-date with memory fifo statistics structural changes.
Add an option and test implementation for the full-duplex ack protocol.
Work around a ctype bug when displaying printable characters in the
It is not acceptable to index the array out of bounds if an illegal index
Adjust the inline to take a pointer to a constant array to avoid a
Add missing m_freem() in BPF if the mbuf exceeds the interface mtu.
Make shutdown() a fileops operation rather then a socket operation.
Add journaling restart support, required to produce a robust journaling
Check for a free-after-send case and panic if detected.
For now just use lwkt_gettoken() rather then a trytoken/gettoken combination.
Add some missing crit_exit()'s. The original code just assumed that the
Move a mistaken crit_exit() into a crit_enter(), which was panicing the
Reorder the INVARIANTS test in crit_enter() to occur prior to modifying
* Fix a bug that could cause dc_stop() to try to m_freem() something that's
KTR_MALLOC should be KTR_MEMORY
When a usb mass storage device is removed the related CAM SIM structure is
DELAY() is a spin loop, we can't use it any more because shutdown
Add KTR support for usb_mem to trace usb-related allocations.
Fix numerous extremely serious bugs in OHCI's iso-synchronous code. I'm
Fix a MP lock race. The MP locking state can change when lwkt_chktokens()
If a fatal kernel trap occurs from an IPI or FAST interrupt on a cpu not
Interlock panics that occur on multiple cpus before the first cpu is able to
Limit switch-from-interrupt warnings to once per thread to avoid an endless
Add some additinal targets to allow elements of a buildworld to be
Add some debugging code to catch any dirty inodes which are destroyed
Add some conditionalized debugging 'PANIC_DEBUG', to allow us to panic a
When a cpu is stopped due to a panic or the debugger, it can be in virtually
Support disablement of chflags in a jail, part 1/2.
Only compile in lwkt_smp_stopped() on SMP builds.
Additional work to try to make panics operate better on SMP systems.
Fix a critical bug in the IPI messaging code, effecting SMP systems. In
Fix a sockbuf race. Currently the m_free*() path can block, due to
Do not compile the kernel with the stack protector. I've decided to tolerate
Support disablement of chflags in a jail, part 2/2. This actually isn't
Add a missing FREE_LOCK() call.
Add missing crit_exit().
Add a new kernel compile debugging option, DEBUG_CRIT_SECTIONS. This fairly
Add a missing crit_exit().
If multiple processes are being traced and some other process has a write
There is a case when B_VMIO is clear where a buffer can be placed on the
Stephan believes that this patch, just committed to FreeBSD, may fix
Bump the development branch sub-version from 1.3.2 to 1.3.3, indicating
Add a new system config directive called "nonoptional" which specifies
Port a major reworking of the way IPS driver commands are managed from
Fix a bug in the last commit. When using the UFS dirent directly,
Add a sanity check for the length of the file name to vop_write_dirent().
Fix a race in rename when relocking the source namecache entry. Since we
Fix an inode bitmap scanning bug. Due to an error in the length adjustment
UFS sometimes reports: 'ufs_rename: fvp == tvp (can't happen)'. The case
Convert RANDOM_IP_ID into a sysctl.
Have vidcontrol set the video history size based on a new rc.conf variable,
Instead of resetting the video buffer's history size to the system
Add a TTY_PATH_MAX limit, set to 256, and reduce the size of the ttyname
Reduce the buffer size for the threaded version of ttyname() to TTY_PATH_MAX.
Filesystem journaling. Reorganize the journal scan for the mountpoint to
When writing UNDO records, only try to output the file contents for VREG
Implement FSMID. Use one of the spare 64 bit fields in the stat structure
Reduce critical section warnings for AHC when critical section debugging
Fix a serious bug in cache_inefficient_scan() related to its use of
Only include thread2.h for kernel builds (its macros are used by vm_page.h's
Add a typedef ufs1_ino_t to represent inodes for UFS1 filesystems.
Dump and restore need to use a UFS-centric inode representation. Convert
Bump the development sub-version to 1.3.5.
Require HEAD users to upgrade to 1.3.5 before running installworld, due
Fix a deadlock in ffs_balloc(). This function was incorrectly obtaining a
Use a typedef that already conveniently exists instead of anonymous
Get rid of smp_rendezvous() and all associated support circuitry. Move
Make sure the vnode is unlocked across a dev_dclose() call, otherwise a
Add a missing crit_exit(), fixing a panic. Attempt to continue with the
Print out a little more information on the PXE boot configuration.
Remove old #if 0'd sections of code, add a few comments, and report a bit
Compile up both the NFS and the TFTP version of the PXE loader and
Give the kernel a native NFS mount rpc capability for mounting NFS roots by
Cleanup the module build and conditionalize a goto label.
Revert the very last commit to ehci.c (1.12). It was locking the system
Merge the following revs from NetBSD (in an attempt to bring new material
1.101 Update code comments.
Rework and expand the algorithms in JSCAN, part 1/2. Implement a new
Rework and expand the algorithms in JSCAN, part 2/?.
* Generally change NOXXX to NO_XXX, similar to work done in FreeBSD.
Rework and expand the algorithms in JSCAN, part 3/?.
Document a special case for Journaling PAD records. PAD records have to
Rework and expand the algorithms in JSCAN, part 4/?.
Rework and expand the algorithms in JSCAN, part 5/?.
Slightly reorganize the transaction data. Instead of placing the REDO data
Rework and expand the algorithms in JSCAN, part 6/?.
Syntax cleanup, add a code comment, add a newline in a bootverbose
Fix isa_wrongintr. The APIC vector was being directly assigned to a C
DragonFly's malloc only guarentees X alignment when X is a power of 2,
Fix a serializer bug. The SMP serializer would in some cases fail to
Fix a token bug. A preempting interrupt thread blocking on a token cannot
Cleanup minor syntax/informational issues.
Rename all the functions and structures for the old VOP namespace API
Cleanup a couple of serious issues with vinum.
Bump the development sub-version to 6, covering the following major changes
Add -d /cvs to tag update targets.
Bump __Dragonfly_version to 130006 (+ slip preview on the file)
Reserve the same amount of space for the spinlock structure whether we are
Split spinlock.h into spinlock.h and spinlock2.h so we can embed spinlocks in
Add an option, -y, which displays the 64 bit FSMID for a file or directory.
Add an argument to vfs_add_vnodeops() to specify VVF_* flags for the vop_ops
With the new FSMID flag scheme we can optimize the update of the chain by
Using the ACPI idle hook while shutting down ACPI during a halt or reboot
Fix the infinite-watchdog timeout problem. the pending_txs count was not
Re-initialize the interrupt mask on ACPI wakeup. This seems to
Because recalculate is only called once or twice for long sleeps,
Do not attempt to modify read-only-mounted filesystems in ufs_inactive().
Add a sysctl, kern.unprivileged_read_msgbuf (defaults to enabled) which if
Implement sysctls to restrict a user's ability to hardlink files owned by
Allow the target safety check to be overridden. May be necessary in certain
Ensure that FP registers are not used for integer code.
Update subvers to 7 so we can sync up the Preview tag prior to Simon
Bump subversion in param.h
Add missing crit_exit(). If VR fails to attach the system will assert
Attempt to add generic GigE support to MII. If this creates issues we will
Properly serialize access in the NV ethernet driver and attempt to fix
Remove the INTR_TYPE_* flags. The interrupt type is no longer used to
Move a bunch of per-interrupt-thread variables from static storage to
Major cleanup of the interrupt registration subsystem.
Use sysctl's instead of KVM to access the interrupt name and count list.
Fix a bug where fsetfd() was not returning an error when it failed,
Move the polling systimer initialization code out of kern_clock.c and into
Fix a bad panic check from the last commit in register_randintr().
FreeBSD commit message:
Bring another softupdates fix in from FreeBSD, FreeBSD commit message:
Implement an emergency interrupt polling feature. When enabled, most
Display all IOAPIC pin assignments when bootverbose is specified, not
MPTable fixup for Shuttle XPC with an AMD Athlon X2 - there is no entry
Only resynchronize the RTC on shutdown if we had previously loaded it and
Do a run through of the fragment allocation and freeing code, documenting
Oops, fix the polling enable name, it's supposed to be kern.polling.enable,
Only check GiGE related flags in the generic code when MIIF_IS_1000X is set
Add an mii_flags field to the attach arguments, to make it easier to create
Pass mii_attach_args to mii_softc_init() rather then initializing the softc's
Add two checks for potential buffer cache races.
Document kern.emergency_intr_{enable,freq} in loader.8.
Temporarily work around a race in the kernel. The kernel does a sanity check
Avoid a recursive kernel fault and subsequent double fault if the VM fault
Add a missing BUF_UNLOCK in the last commit.
Remove the dummy IPI messaging routines for UP builds and properly
Temporary hack until corecode can fix it. There is a p_rtprio and also
Redo the interrupt livelock code. Simplify the frequency calculations
Fix a long-standing bug in the livelock code. An interrupt thread normally
Fix a bug in the ppbus code where an interrupt cookie might be torn down
Assert that the vnode is locked when modifying an inode inside ffs_balloc.
Add a redundant double-check in ffs_reallocblks and assert that the number
Increase the MSGBUF_SIZE from 32K to 64K, boot verbose messages don't
Cleanup some of the newbus infrastructure.
Add another parameter to BUS_ADD_CHILD to allow children to inherit
An exclusive lock on the vnode is required when running vm_object_page_clean(),
Move the freebsd package system from /usr/sbin to /usr/freebsd_pkg/sbin
add an acpi_enabled() function, invert the "pci" logic and require that
Bump config to 400022. Added an 'arch' softlink. 'machine' goes into
ICU/APIC cleanup part 1/many.
ICU/APIC cleanup part 2/many.
ICU/APIC cleanup part 2/many.
ICU/APIC cleanup part 3/many.
ICU/APIC cleanup part 4/many.
ICU/APIC cleanup part 5/many.
ICU/APIC cleanup part 6/many.
ICU/APIC cleanup part 7/many.
Make rndcontrol use the new ioctl so it can properly list interrupt sources
Reimplement IPI forwarding of FAST interrupts to the cpu owning the BGL
ICU/APIC cleanup part 8/many.
Add more documentation for the APIC registers and rename some of the
De-confuse the IO APIC mapping code by creating a helper procedure to
ICU/APIC cleanup part 9/many.
ICU/APIC cleanup part 10/many. Be a lot more careful programming the IO APIC.
Display warnings for any configured IO APIC pins that do not actually exist.
Fix a bug in the last commit. The wrong argument was being passed to
We are already serialized when nv_ospackettx() is called, so it must
Allow 'options SMP' *WITHOUT* 'options APIC_IO'. That is, an ability to
Fix a symbol not found problem by not including madt.c in the ACPI module.
Switch to the BSP when doing a normal shutdown. ACPI can't power the machine
Fix the cpu the softclock thread(s) are created on. The softclock threads
Fix a comment. The slave is connected to IRQ 2 on the first 8259, not
Mark our fine-grained interrupt timer as being INTR_MPSAFE, because it is.
The 'picmode' variable was mis-named. The MPTable is actually simply
ICU/APIC cleanup part 11/many.
Solve the continuous stream of spurious IRQ 7's that occur on machines
Make sure that the apic error, performance counter, and timer local
When operating in SMP+ICU mode, try to disconnect the 8259 from the cpu
Clean up the CPU_AMD64X2_INTR_SPAM option to check the cpu_id and provide
Document CPU_AMD64X2_INTR_SPAM in LINT.
Minor manual adjustment to add debug.acpi.enabled.
Adjust the globaldata initialization code to accomodate globaldata
Make tsleep/wakeup MP SAFE part 1/2.
Turn around the spinlock code to reduce the chance of programmer error.
Revert part of the last commit. We aren't ready for the per-cpu _wakeup
USB mouse fix for certain mice, such as the Logitech LX700. Do not assume
Fix a broken array lookup in the old 4.3 BSD mmap compatibility code
Make tsleep/wakeup() MP SAFE for kernel threads and get us closer to
Temporarily check for and correct a race in getnewbuf() that exists due
Continue work on our pluggable scheduler abstraction. Implement a system
If a /dev/<disk> device node is fsynced at the same time the related
Remove ancient interrupt handler hacks that are no longer needed.
Do a better job formatting vmstat -i output. Output entries that look like
Convert the lockmgr interlock from a token to a spinlock. This fixes a
Protect allproc scans with PHOLD/PRELE, in particular to support the
Fix a bug in the last commit. The proc pointer can be NULL at the
Remove inthand_add() and inthand_remove(). Instead, register_int() and
Add a thread flag, TDF_MPSAFE, which is used during thread creation to
Add a sysctl and tunable kern.intr_mpsafe which allows threaded interrupts
Start working on making the printf() path MPSAFE, because it isn't at the
Add a sysctl and tunable kern.syscall_mpsafe which allows system calls
Followup to last commit, cleanup some SMP/UP conditionals.
Fix the design of ifq_dequeue/altq_dequeue by adding an mbuf pointer and
Wrap psignal() and a few other places that require the MP lock when
Add a sysctl and tunable kern.trap_mpsafe which allows some traps to run
Remove unused label.
Document the fact that the vm86 instruction emulator is MPSAFE.
Do not try to set up hardware vectors for software interrupts.
Consolidate the initialization of td_mpcount into lwkt_init_thread().
Add a lwkt_serialize_try() API function.
Fix some minor bugs in lwkt_serialize_handler*() which upcoming code will
Add the ips driver to GENERIC.
Assert that he mbuf type is correct rather then blinding setting m_type.
Fix a mbuf statistics bug.
tcp_syncache.cache_limit is a per-cpu limit, reserve enough space for all
Make all network interrupt service routines MPSAFE part 1/3.
Properly serialize IPW.
ipw (is the only driver that) needs a working interrupt to perform
Jumbo mbuf mangement's extbuf callbacks must be MPSAFE. Use a serializer
Get rid of the p_stat SZOMB state. p_stat can now only be SIDL, SSLEEP, or
tsleep_interlock() must be called prior to testing the serializer lock
The primary mbuf cluster management code needs to be made MPSAFE since
For MPSAFE syscall operation, CURSIG->__cursig->issignal() may be called
Do not hold the ifnet serializer when entering tsleep() in the PPP TUN
Change the initial path from /bin:/usr/bin to /bin:/usr/bin:/sbin:/usr/sbin
cred may be NULL due to a prior error code. crhold() handles NULL creds,
Fix a bug in the big tsleep/wakeup cleanup commit. When stopping a
Fix a process exit/wait race. The wait*() code was making a faulty test
Reduce SCSI_DELAY in GENERIC from 15 seconds to 5 seconds.
Fix a bogus proc0 test that is no longer accurate. This should allow the
Add support for DLink 528(T) Gigabit cards.
Add /usr/pkg/[s]bin to /bin/sh's default path and login.conf's default path.
Require pkgsrc to be installed, include pkgsrc bootstrap binaries and mk.conf
Update more default cshrc/profiles with new paths.
Fix a bug in our RB_SCAN calls. A critical section is required to
Sync up misc work on the currently inactive syscall mapping library.
Add an option -b which dumps the clean/dirty buffer cache RB trees for
Document the nfs dirent conversion code. No functional changes.
doreti and splz were improperly requiring that the MP lock be held in order
The new lockmgr() function requires spinlocks, not tokens. Take this
Synchronize the TSC between all cpus on startup and provide a sysctl,
Enhance ktrdump to generate relative timestamps in fractional microseconds,
Add KTR_TESTLOG and debug.ktr.testlogcnt, which issues three ktrlog() calls
Add additional KTR lines to allow us to characterize the total overhead
Clean up more spinlock conversion issues and fix related panics.
SB_NOINTR must be set for the socket to prevent nbssn_recv() from hard
Fix the directory scan code for SMBFS. It was losing track of the directory
Beef up error reporting for a particular assertion to try to track down a
Do not ldconfig -m /usr/lib/gcc2 if there is no /usr/lib/gcc2 directory.
Fix UP build issues. Move tsc_offsets[] from mp_machdep.c to kern_ktr.c,
Make the KTR test logging work for UP as well as SMP.
Add KTR logging for tsleep entry/exit and wakeup entry/exit.
Add KTR logging for IF_EM to measure interrupt overhead and packet
Don't display the file and line by default.
Fix another interesting bug. td_threadq is shared by the LWKT scheduler,
Add a feature and a sysctl (debug.ktr.testipicnt) to test inter-cpu
Remove the 'all' target to fix buildworld, and fix a
Move tsc_offsets[] to a new home to unbreak kernel builds, again.
pfnStop() seems to take a flags argument which as far as I can tell from
Properly integrate the now mandatory serializer into the WI network driver.
Add KTR logging to the core tcp protocol loop.
Fix a number of panic conditions for network cardbus devices by implementing
Add a feature that allows a correction factor to be applied to attempt
Clean up some minor typos in comments.
Fix numerous nrelease build issues related to the pkgsrc conversion.
run ./bootstrap in a chroot so it configures the correct paths in the
Cleanup minor typeos.
Bump us to 1.3.8 in preparation for pre-release tagging.
By the time list_net_interfaces() is called in /etc/rc.d/netif, clone_up()
1.4 Release branched, HEAD is now 1.5.
Fix the installer_quickrel target. Also fix the pkgsrc bootstrap, not
A ^Z signals the whole process group, causing the parent process (vipw) to
Fix a type-o that was causing the wrong mbuf's csum_data to be adjusted
After much hair pulling the problem with dual BGE interfaces not coming up
Make wait*() behave the same as it did before we moved TSTOP handling
Add a 2 second delay after configuring interfaces before continuing.
Add a cvsup example to track the 1.4 release.
Give up trying to port ezm3, add a cvsup binary bootstrap package to
Do not require a .sh extension for scripts in local_startup dirs,
ether_input() no longer allows the MAC header to be passed separately,
The stat and ino_t changes were not intended to break dump/restore
Do not call ether_ifdetach() with the serializer held in IF_TAP.
Weed out files with .sample, .orig, or .dist extensions. Proceduralize
Preliminary ndis cleanup. The serializer has taken over the functionality
Finish fixing NDIS serialization. Wrap detach in the serializer and remove
Add a target that will update the 1.4-release slip tag.
Mostly fix nullfs. There are still namespace race issues between
Clean up unmount() by removing the vnode resolution requirement. Just
bx is supposed to point to twiddle_chars, not contain the first element
Correct sizeof(pointer) bugs that should have been sizeof(*pointer)
Add mk.conf to the ISO and have the installer install it in /etc
Use the DragonFly contrib patch system to correct improper sizeof(pointer)
Add the '-c cpu' option to arp, netstat, and route, to allow the route
Switch the type and how argument declarations around to match what the
The random number generator was not generating sufficient entropy by
Add a missing #include <sys/lock.h> to fix a UP kernel build problem.
Lobotomize libcaps so it compiles again and can be used by the code
Bring in the parallel route table code and clean up ARP. The
Remove the VBWAIT flag test, it will soon go away.
Remove serialization calls that are no longer correct, fixing a panic
Make warn() a weak reference.
Bring in a bunch of malloc features from OpenBSD and fundamentally change
Get rid of seriously out of date example code.
Properly assert the state of the MP lock in the async syscall message
Make the entire BUF/BIO system BIO-centric instead of BUF-centric. Vnode
Fix a bunch of race cases in the NFS callout timer code, in the handling
Reformulate some code which was #if 0'd out in the last patch. When
Reduce the default NFSv3 access cache timeout from 60 seconds to 10 seconds.
bioops.io_start() was being called in a situation where the buffer could
Change the server side NFS write gather delay from 10ms to 20ms. This
nvi's "memory use after free" bug exists in following call path:
Do not set the pcb_ext field until the private TSS has been completely
Properly check for buffered data in ukbd_check_char(). This fixes issues
A thread may be freed from a different cpu then it was assigned to,
dvp must be unlocked prior to issuing a VOP operation to avoid obtaining
Pass LK_PCATCH instead of trying to store tsleep flags in the lock
buftimespinlock is utterly useless since the spinlock is released
vfs_bio_awrite() was unconditionally locking a buffer without checking
Add additional red-black tree functions for fast numeric field lookups.
Unlock vnodes prior to issuing VOP_NREMOVE to accomodate filesystem
Replace the global buffer cache hash table with a per-vnode red-black tree.
Roll 1.5.1 for the slip tag
Remove two incorrect serializer calls in the NDIS code.
Struct buf's cannot simply be bcopy'd any more due to linkages in the
cluster_read() was very dangerously issuing a blind BMAP for a buffer
Add missing block number assignment in ATA raid mirroring code. The
Change KKASSERT() to not embed #exp in the control string. Instead pass
cache_fromdvp() uses a recursive algorithm to resolve disconnected
Implement a VM load heuristic. sysctl vm.vm_load will return an indication
Cleanup the copyright in this file. Easy since its my code.
Add an option to add a slight time skew to the execution of scripts to
Fix a serious bug in the olddelta microseconds calculation returned by
Prevent the driver from reinitializing the card when it's already running.
strnstr() was testing one byte beyond the specified length in certain
Make a slight adjustment to the last commit. Change saved_ncp to saved_dvp
Backout the rest of 1.29. There are a number of issues with the other
Bump 1.5.2 for the preview tag, synchronized to just before the
Add missing commit for the VM load heuristic and page allocation rate
Major BUF/BIO work commit. Make I/O BIO-centric and specify the disk or
Don't just assume that the directory offset supplied by the user is
Fix numerous translation problems in msdosfs, related to the recent BUF/BIO
Sync fstat up to the changes made in msdosfsmount.h.
Sanitize status message.
Add options to allow the dump offset or system memory size (for the purposes
Undo the last commit. At the moment we require access to the structure
Add the initialization of blockoff back in, the variable is still used.
Clean up the extended lookup features in the red-black tree code.
Change *_pager_allocate() to take off_t instead of vm_ooffset_t. The
Remove NQNFS support. The mechanisms are too crude to co-exist with
Correct some minor bugs in the last patch to fix kernel compilation.
Add a RB_PREV() function which returns the previous node in a red-black
Add PCI IDs for Intel's ICH7 and ICH7M ATA/SATA hardware, used in
Remove VOP_GETVOBJECT, VOP_DESTROYVOBJECT, and VOP_CREATEVOBJECT. Rearrange
A VM object is now required for vnode-based buffer cache ops. This
NFS needs to instantiate a backing VM object for the vnode to read a symlink.
ffs_truncate(), called from, truncate(), remove(), rmdir(), rename-overwrite,
Fix a race condition between nlookup and vnode reclamation. Even though the
Use the vnode v_opencount and v_writecount universally. They were previously
Clone cd9660_blkatoff() into a new procedure, cd9660_devblkatoff(), which
Require that *ALL* vnode-based buffer cache ops be backed by a VM object.
Give the MFS pseudo block device vnode a VM object, as is now required
ufs_dirempty() issues I/O on the directory vnode and needs to make sure
A floating point fault (instead of DNA fault) can occur when the TS bit
Document the use of SDT_SYS386IGT vs SDT_SYS386TGT when setting up the
A number of structures related to UFS and QUOTAS have changed name.
A number of structures related to UFS and QUOTAS have changed name.
A number of structures related to UFS and QUOTAS have changed name.
ufs_readdir() can be called from NFS with the vnode being opened, create
Because multiple opens of /dev/tty only issue one actual open to the
Properly calculate the ronly flag at unmount time.
Transplant all the UFS ops that EXT2 used to call into the EXT2 tree and
Unconditionally initialize a VM object for a directory vnode. Continue
Synchronize vinitvmio() calls from UFS to EXT2FS.
Due to continuing issues with VOP_READ/VOP_WRITE ops being called without
Get rid of bogus 'pushing active' reports. Initialize a VM object for VREG
Followup last commit, fix missing argument to vinitvmio().
Fix one place where the superblock was being read (and written) at the
Remove debugging printfs.
Calculate the correct buffer size when reading a symlink via NFS.
in_ifadown() was only cleaning up the route table on the originating cpu,
/dev/random was almost always returning 0 bytes. This was due to several
Note: the previous rev's CVS comment was messed up due to an editor snafu.
NTFS sometimes splits the initialization of a new vnode into two parts.
If a process forks while being scanned, a non-zero p_lock will be inherited
Supply version of wakeup() which only operate on threads blocked on the
Fix a livelock in the objcache blocking code. PCATCH was being improperly
vop_stdopen() must be called when a fifo_open fails in order to then be
Fix an edge case where objects can be returned to a per-cpu cache while
Always guarentee at least one space between two network addresses.
Conditionalize a lwkt_send_ipiq2() to fix the UP build.
Fix a bug in the pkg_add -n tests where pkg_add was incorrectly reporting
Recent bug fixes make this worthy for testing, update to 1.5.3 and slip the
Generate a host-unreachable failure rather then a crash if the MTU is too
Bring in some small changes from FreeBSD.
Minor typing cleanups for aicasm.
Add spin_uninit() to provide symmetry with spin_init().
Misc sysperf cleanups. Add another mutex tester to test the xchgl
Get rid of unused arg.
Get rid of LK_PCATCH in the agp lockmgr() calls. AGP ignores the return
Remove unused code label.
Get rid of LK_DRAIN in dounmount(). LK_DRAIN locks are not SMP friendly and
LK_DRAIN locks are no longer obtained on vnodes, rip out the check.
Run the lockmgr() call independant of the KASSERT() in smb_co_init().
Get rid of LK_DRAIN and LK_INTERLOCK interactions. Recode interlocks when
Get rid of LK_DRAIN, rely on nc_lwant to interlock lock races against
Remove all remaining support for LK_DRAIN lockmgr locks. LK_DRAIN was a
Remove remaining uses of the lockmgr LK_INTERLOCK flag.
Remove the now unused interlock argument to the lockmgr() procedure.
Remove LK_REENABLE (related to the LK_DRAIN removal).
The nticks calculation is still broken. Sometimes the delta systimer
Fix an incorrect header length comparison for IPSEC AH. Add required
B_INVAFTERWRITE is no longer used, remove it.
If softupdates or some other entity re-dirties a buffer, make sure
Call vnode_pager_setsize() before BALLOC rather than after.
vfsync() is not in the business of removing buffers beyond the file EOF.
Add a memory wrap check to kernacc to try to reduce instances of a bogus
Rename KVM_READ() to kread() and make it a real procedure. Also incorporate
Generate unique identifiers for simulated FSMIDs so any errors appear to
Separate the MD5 code into its own module.
Get rid of the weird FSMID update path in the vnode and namecache code.
Add the preadv() and pwritev() systems and regenerate.
Fix the range checking for all read and write system calls. Fix the
Get rid of libcr, the idea(s) behind it are not really applicable anymore
Move most references to the buffer cache array (buf[]) to kern/vfs_bio.c.
Fix a bug in the POSIX locking code. The system could lose track of
Remove the buffer cache's B_PHYS flag. This flag was originally used as
Get rid of the remaining buffer background bitmap code. It's been turned
Remove non-existant variable from debugging message.
Get rid of pbgetvp() and pbrelvp(). Instead fold the B_PAGING flag directly
Bring in some fixes from NetBSD:
Bring in SHA256 support from FreeBSD.
Replace the the buffer cache's B_READ, B_WRITE, B_FORMAT, and B_FREEBUF
Remove b_xflags. Fold BX_VNCLEAN and BX_VNDIRTY into b_flags as
The pbuf subsystem now initializes b_kvabase and b_kvasize at startup and
Remove buf->b_saveaddr, assert that vmapbuf() is only called on pbuf's. Pass
Plug xform memory leaks. Don't re-initialize an xform for an SA that
m_cat() may free the mbuf on 2nd arg, so m_pkthdr manipulation
32bit from 64bit value fixup. Fix typo.
Fix fencepost error causing creation of 0-length mbufs when
more strict sanity check for ESP tail. [From KAME]
32bit from 64bit value fixup.
Supply a stack pointer for a pure thread context so backtrace works.
Plug memory leak in umass.
- Add workarounds for dropped interrupts on VIA and ATI controllers.
The wrong pointer was being used to calculate the page offset, leading
Fix a bug in close(). When a descriptor is closed, all process leaders
Fix an information disclosure issue on AMD cpus. The x87 debug registers,
Fix a biodone/AR_WAIT case. b_cmd was not getting set to BUF_CMD_DONE,
Add a missing ohci_waitintr() call that allows polled operation of
Minor cleanup, plus initialize a few additional fields in the proc
Remove the accounting argument from lf_create_range() and lf_destroy_range().
Invert a mistaken test. Set b_resid to 0 if B_ERROR is not set.
Document the handling of a file holes in ufs_strategy() and clean up
- Clarify the definitions of b_bufsize, b_bcount, and b_resid.
Block devices generally truncate the size of I/O requests which go past EOF.
Cleanup procedure prototypes, get rid of extra spaces in pointer decls.
Remove VOP_BWRITE(). This function provided a way for a VFS to override
Remove the thread pointer argument to lockmgr(). All lockmgr() ops use the
Simplify vn_lock(), VOP_LOCK(), and VOP_UNLOCK() by removing the thread_t
Remove the thread_t argument from vfs_busy() and vfs_unbusy(). Passing a
The thread/proc pointer argument in the VFS subsystem originally existed
Add some ifioctl() td -> ucred changes that were missed.
The fdrop() procedure no longer needs a thread argument, remove it.
Remove the thread_t argument from nfs_rslock() and nfs_rsunlock().
Remove the thread argument from ffs_flushfiles(), ffs_mountfs(),
Remove the thread argument from ext2_quotaoff(), ext2_flushfiles(),
Remove the thread argument from all mount->vfs_* function vectors,
Fix a null pointer indirection, the VM fault rate limiting code only
We have to use pmap_extract here, pmap_kextract will choke if the page
We have to use pmap_extract() here. pmap_kextract() will choke on a missing
We have to use pmap_extract() here. If we lose a race against page
Recode the streamid selector. The streamid was faked before. Do it for
lockmgr_kernproc() wasn't checking whether the lockholder as already
Remove the internal F_FLOCK flag. Either F_POSIX or F_FLOCK must be set,
Add a little program that allows one to test posix range locks.
Rewrite the POSIX locking code. It was becomming impossible to track
Split kern/vfs_journal.c. Leave the low level journal support code in
Recognize the cpu ident for additional VIA processors.
Fix three bugs in the last commit and document special cases. Tighten
UMAPFS has been disabled (and non-working) for a long time. Scrap it
Most of the fields in vnodeop_desc have been unused for a while now.
Remove mount_umap.
cbb_probe() assumes that the subclass field is unique. This patch further
Remove vnode lock assertions that are no longer used. Remove the
Attempt to interlock races between the buffer cache and VM backing store
Remove the (unused) copy-on-write support for a vnode's VM object. This
Pass the process (p) instead of the vnode (p->p_tracep) to the kernel tracing
The ktracing code was not properly matching up VOP_OPEN and VOP_CLOSE calls.
Oops, last commit was slightly premature. Fix a bug-a-boo and remove
Add another mutex tester for Jeff's spinlock code w/ the refcount
p_tracep -> p_tracenode, tracking changes made in recent commits.
p_tracep -> p_tracenode, tracking changes made in recent commits.
Replace the LWKT token code's passive management of token ownership with
Make spinlocks panic-friendly.
Remove the last vestiges of UF_MAPPED. All the removed code was already
Consolidate the file descriptor destruction code used when a newly created
Convert most manual accesses to filedesc->fd_files[] into the appropriate
Recent lwkt_token work broke UP builds. Fix the token code to operate
I'm growing tired of having to add #include lines for header files that
Embed the netmsg in the mbuf itself rather than allocating one for
Fix a build issue with libnetgraph. net/bpf.h does not need to include
Remove so_gencnt and so_gen_t. The generation counter is not used any more.
Remove the (unmaintained for 10+ years) svr4 and ibcs2 emulation code.
A little script that runs through all the header files and checks that
Clean up more #include files. Create an internal __boolean_t so two or
Only _KERNEL code can optimize based on SMP vs UP. User code must always
Implement a much faster spinlock.
Give struct filedesc and struct file a spinlock, and do some initial
Do a major cleanup of the file descriptor handling code in preparation for
Fix a minor bug in fdcopy() in the last commit, Consolidate the
Sync to head. Add a verbose option to vmpageinfo which dumps all the
The pageout daemon does not usually page out pages it considers active.
Move all the resource limit handling code into a new file, kern/kern_plimit.c.
spinlock more of the file descriptor code. No appreciable difference in
Start consolidating process related code into kern_proc.c. Implement
Move the code that inserts a new process into the allproc list into its
Fix issues with an incorrectly initialized buffer when formatting a floppy.
When a vnode is vgone()'d its v_ops is replaced with dead_vnode_ops.
Modifying lk_flags during lock reinitialization requires a spinlock.
Adjust pamp_growkernel(), elf_brand_inuse(), and ktrace() to use
Convert almost all of the remaining manual traversals of the allproc
Fix several buffer cache issues related to B_NOCACHE.
More MP work.
* Make falloc() MPSAFE. filehead (the file list) and nfiles are now
Add #include <sys/lock.h> where needed to support get_mplock().
* Fix a number of cases where too much kernel memory might be allocated to
Remove FFS function hooks used by UFS. Simply make direct calls from ufs
Add a read-ahead version of ffs_blkatoff() called ffs_blkatoff_ra(). This
Implement msleep(). This function is similar to the FreeBSD msleep() except
Greatly reduce the MP locking that occurs in closef(), and remove
Clear the new VMAYHAVELOCKS flag when after an unlock we determine that
Mark various forms of read() and write() MPSAFE. Note that the MP lock is
Get rid -y/-Y (sort by interactive measure). The interactive measure has
Further isolate the user process scheduler data by moving more variables
Clean up compiler warnings when KTR is enabled but KTR_ALL is not.
Remove conditional memory allocation based on KTR_ALL. Allocate memory
Add two KTR (kernel trace) options: KTR_GIANT_CONTENTION and
Shortcut two common spinlock situations and don't bother KTR logging them.
Fix numerous bugs in the BSD4 scheduler introduced in recent commits.
gd_tdallq is not protected by the BGL any more, it can only be manipulated
Use the MP friendly objcache instead of zalloc to allocate temporary
If the scheduler clock cannot call bsd4_resetpriority() due to spinlock
Update the manual page to reflect additional spinlock requirements.
Another update. Clarify that a shared spinlock can be acquired while holding
Since we can only hold one shared spinlock at a time anyway, change the
namecache->nc_refs is no longer protected by the MP lock. Atomic ops must
Remove vnode->v_id. This field used to be used to identify stale namecache
Add an option which dumps the filename from the vnode's namecache link.
Fix a file descriptor leak, add a missing vx_put() after linprocfs
Rename arguments to atomic_cmpset_int() to make their function more obvious.
Fix a bug in the linux emulator's getdents_common() function. The function
Misc cleanup - move another namecache list scan into vfs_cache.c
An inodedep might go away after the bwrite, do not try to access
Fix blocking races in various *_locate() functions within softupdates.
Remove LWKT reader-writer locks (kern/lwkt_rwlock.c). Remove lwkt_wait
Fix a minor bug in the last commit. lwp_cpumask has to be in the LWP copy
Modify kern/makesyscall.sh to prefix all kernel system call procedures
Fix a file descriptor leak in cam_lookup_pass() when the ioctl to
Fix a WARNS=3 gcc warning related to longjmp clobbers, fix a possible use
Remove lwp_cpumask assignment. lwp_cpumask is handled in the bcopy section.
Remove an inappropriate crit_exit() in ehci.c and add a missing crit_exit()
Add an INVARIANTS test in both the trap code and system call code. The
Cleanup crit_*() usage to reduce bogus warnings printed to the console
Some netisr's are just used to wakeup a driver via schednetisr(). The
Add missing crit_exit()
Remove the asynchronous system call interface sendsys/waitsys. It was an
Add an option, DEBUG_PCTRACK, which will record the program counter of
Add a new utility, 'pctrack', which dumps program counter tracking data
Fix namespace pollution.
We shouldn't have to fninit to make the FP unit usable for MMX based copies.
Move selinfo stuff to the separate header sys/selinfo.h. Make sys/select.h
Remove the select_curproc vector from the usched structure. It is used
Add kernel syscall support for explicit blocking and non-blocking I/O
The pread/preadv/pwrite/pwritev system calls have been renamed. Create
Use the _SELECT_DECLARED method to include the select() prototype instead
Add two more system calls, __accept and __connect. The old accept() and
Well, ok, if you are going to turn off writable strings, then the code
Do not set O_NONBLOCK on a threaded program's descriptors any more. Instead,
fcntl(.., F_SETFL, ..) should only do an FIOASYNC ioctl if the FASYNC
Replace the random number generator with an IBAA generator for /dev/random
Replace the random number generator with an IBAA generator for /dev/random
Fix a case where RTP_PRIO_FIFO was not being handled properly. The bug led
/dev/[k]mem was not allowing access to the CPU globaldata area, because it
Swap out FAT12 for NTFS so the boot0 prompt says 'DOS' instead of '??' for
Add a new option -H <path> to cpdup. This option allows cpdup to be used
Do not attempt to read the slice table or disk label when accessing a raw
If we hit the file hardlink limit try to copy the file instead of hardlinking
Add a missing initbufbio() to fix a panic when vinum tries to issue a
Include kernel sources on the release CD.
Use bzip instead of gzip, rename the tar file to make it more obvious that
Disassociate the VM object after calling VOP_INACTIVE instead of before.
Cleanup, no functional changes.
Change the seeder array from a modulo to a logical AND, improving performance
Turn on the new kern.seedenable sysctl when seeding the PRNG.
Update the manual pages for the kernel random number generator.
Correct a problem with the user process scheduler's estimated cpu
A broken pipe error is sometimes reported by zcat if the user quits out
Add missing prototype.
Use pbufs instead of ebufs.
Attempt to fix an occassional panic in pf_purge_expired_states() by
Add a note on where to find the release engineering document.
Bump sub-versions and DragonFly_version in preparation for branching.
Add a new target for cvs administration of the 1.6 slip tag. Add a cvsup
Adjust HEAD version from 1.5 to 1.7.
Add a fairly bad hack to detect ripouts that might occur during a list
Remove several layers in the vnode operations vector init code. Declare
Get rid of the weird coda VOP function arguments and void casts and
Introduce sys/syslink.h, the beginnings of a VOP-compatible RPC-like
Check the the ops mount pointer is not NULL before indirecting through it.
Add code to dump lockf locks associated with a vnode.
Fix a bug where the VMAYHAVELOCKS flag on a vnode may get lost, resulting
Make a few more #define's visible when compiling with _KERNEL_STRUCTURES
Fix a minor bug that prevented compilation.
MASSIVE reorganization of the device operations vector. Change cdevsw
For the moment adjust dd to find the new location of the device type
Why is ip_fil.h trying to declare kernel procedures for userland #include's?
Remove duplicate code line.
Fix an incorrect #ifndef label. Also remove a now unnecessary
Update the syslink structural documentation. Add syslink_msg.h, containing
Get rid of some unused fields in the fileops and adjust the declarations
Get rid of a bogus check that cut the blocked-lock wakeup code a little
Add structures and skeleton code for a new system call called syslink()
Instead of indirectly calling vop_stdlock() and friends, install direct
minor syslink cleanups to get the syslink_read() and syslink_write()
LK_NOPAUSE no longer serves a purpose, scrap it.
Protect the pfshead[] hash table with a token.
VNode sequencing and locking - part 1/4.
VNode sequencing and locking - part 2/4.
VNode sequencing and locking - part 3/4.
Update the X11 path for the default login.conf.
Add a #define that source code can check to determine that the stat
Add some linux compatibility defines, _DIRENT_HAVE_D_NAMLEN and
Add a remote host capability for both the source and target directory
Generate a nice message and make sure the program exits if we lose a
Add a postscript printer filter example using ghostscript for a
Remove the coda fs. It hasn't worked in a long time.
Fix a memory leak and greatly reduce the memory allocated when remembering
Properly update the mtime for directories.
VNode sequencing and locking - part 4/4 - subpart 1 of many.
Bring in the initial cut of the Cache Coherency Management System module.
Fix a case where a spinlock was not being released.
Add skeleton procedures for the vmspace_*() series of system calls which
Rename functions to avoid conflicts with libc.
Rename functions to avoid conflicts with libc.
Rename functions to avoid conflicts with libc.
Rename functions to avoid conflicts with libc.
Rename functions to avoid conflicts with libc.
Rename functions to avoid conflicts with libc.
Split extern in6* declarations for libc vs the kernel.
Create 'k' versions of the kernel malloc API.
Rename malloc->kmalloc, free->kfree, and realloc->krealloc. Pass 1
Rename malloc->kmalloc, free->kfree, and realloc->krealloc. Pass 2
Make KMALLOC_ONLY the default, remove compatibility shims for the
Remove KMALLOC_ONLY from LINT
Move the code that eats certain PNP IDs into a ISA bus-specific file.
Clean up module build failures when compiling a kernel without PCI.
Fix malloc macros for dual-use header file.
Attempt to fix a vnode<->namecache deadlock in NFS's handling of stale
Get rid of a struct device naming conflict.
Rename struct specinfo into struct cdev. Add a new typedef 'cdev_t' for cdev
Rename the kernel NODEV to NOCDEV to avoid conflicts with the userland NODEV.
Change the kernel dev_t, representing a pointer to a specinfo structure,
Reserve upcall IDs 0-31 for system use.
Move flag(s) representing the type of vm_map_entry into its own vm_maptype_t
MAP_VPAGETABLE support part 1/3.
MAP_VPAGETABLE support part 2/3.
MAP_VPAGETABLE support part 3/3.
More cleanups + fix a bug when taking a write fault on a mapping that uses
Clean up some #include's that shouldn't have been in there. Unbreak
Collapse some bits of repetitive code into their own procedures and
Fix a bug in sysctl()'s handling of user data. You can't wire 0 bytes
Fix a bug in sendmsg() and two compatibility versions of sendmsg().
Fix a bug when '-f -H' is used and the target already exists. cpdup was
Bump the version number reported by cpdup from 1.06 to 1.07
Add a README file with some helpful porting hints. I'd rather do this in
Commit a comprehensive file describing how to do incremental backups along
Make some adjustments to low level madvise/mcontrol/mmap support code to
Make some adjustments to low level madvise/mcontrol/mmap support code to
Move an assertion in the bridge code so it only gets hit if the bridge
Disallow writes to filesystems mounted read-only via NULLFS. In this case
Set f_ncp in the struct file unconditionally. Previously we only set it
Remove the last bits of code that stored mount point linkages in vnodes.
Check that namecache references to the mount point are no longer present
Fix a bug in the script softlink code. The softlinks were not being
Add an option that causes cpdup to skip CHR or BLK devices. This option
Recent dev_t work confused sysctl. Adjust the reported type to udev_t
Fix a compile error when DDB is not defined. db_print_backtrace() is
Try to clean up any remaining filesystem references when rebooting. Clean
Fix an off-by-one error. Track #1 is index 0 in the TOC buffer.
Fix a bug in the device intercept code used by /dev/console. The
Follow up to kern_conf.c 1.16. We can't just ignore the ops comparison, it
Fix a bug where mmap()ing a revoked descriptor caused a kernel panic on a
Do not temporarily set signals to SIG_IGN when polling whether the parent
Add Marc's monthly statistics script to DragonFly's base dist. These
Fix a long-standing bug inherited from FreeBSD. It is possible for a
Add two more vmspace_*() system calls to read and write a vmspace. These
NULLFS was not releasing a reference to the root of the underlying
Correct a compiler warning from the last commit.
Add a device that attaches to the memory controller. If ECC is enabled in
Greatly reduce memory requirements when fsck'ing filesystems with lots
Remove inode free race warning messages. These were originally added to
Add a ton of infrastructure for VKERNEL support. Add code for intercepting
Reformulate the way the kernel updates the PMAPs in the system when adding
Reorganize the way machine architectures are handled. Consolidate the
Reorganize the way machine architectures are handled. Consolidate the
Add advice if a kernel config file cannot be found to remind people that
memset must be a real procedure rather then an indirect pointer because
Fix paths to arch/i386, related to the recent architecture topology changes.
Get rid of the indirect function pointer for bzero(). We haven't used it
Bump the config version. Add a 'cpu_arch' directive that allows the
Further normalize the _XXX_H_ symbols used to conditionalize header file
Further normalize the _XXX_H_ symbols used to conditionalize header file
Purge the IFQ when associating a new altq. Packets that have already been
Do a major clean-up of the BUSDMA architecture. A large number of
Fix a stack overflow due to recursion. When the namecache must invalidate
Major namecache work primarily to support NULLFS.
Adjust fstat to properly traverse mount points when constructing a
Sync our rm -P option with OpenBSD - if the file has a hardlink count
test 4
test 5
Major kernel build infrastructure changes, part 1/2 (sys).
Major kernel build infrastructure changes, part 2/2 (user).
Misc cleanups and CVS surgery. Move a number of header and source files
Remove system dependancies on <machine/ipl.h>. Only architecture files
Move <machine/ccbque.h> to <sys/ccbque.h>. ccbque.h is not a
Move <machine/dvcfg.h> to the one device that actually uses it, remove
More Machine-dependant/Machine-independant code and header file separation.
Move the Maxmem extern to systm.h
Get the MI portions of VKERNEL to build, start linking it against libc.
bmake uses /usr/share/mk/sys.mk, so we cannot require that MACHINE_CPU be
Add a missing #undef to properly compile atomic.c functions into the
Adjust for symbol name changes.
buildworld depends on hw.machine exported from the kernel being correct.
Enable the building of boot0cfg for pc32.
Check for subdirectories for both the platform architecture and the
unresolve the vnode associated with the namecache entry for a mount point
Fictitious VM pages must remain structurally stable after free.
Use spinlocks instead of tokens to interlock the objcache depot.
Allow M_ZERO to be specified when using simple object cache setups which
Replace the global VM page hash table with a per-VM-object RB tree. No
Check an additional special pattern to detect dangerously dedicated mode.
Add a manual page outlining the rules for committers.
Make int bootverbose and int cold declarations machine independant.
Misc vkernel work.
Add a generic interrupt controller type that the virtual kernel build can use.
Generate forwarding header files to mimic /usr/include -> /usr/include/sys
Fix a NULL pointer dereference introduced in the previous commit.
For the moment conditionally remove the declaration of certain libc
Local variables that were improperly named 'errno' must be renamed so as
Use ${.CURDIR} to get the correct path to the root skeleton directory.
Document MADV_SETMAP and MAP_VPAGETABLE. These functions support virtualized
Add another ICH PCI ID.
Fix umct and add F5U409 USB Serial Adaptor.
rename sscanf -> ksscanf
Pass NULL to objcache_create() to indicate that null_ctor() and/or null_dtor()
Repo copy machine/pc32/i386/mem.c to kern/kern_memio.c and separate out
Add a prototype for the new mcontrol() system call.
Rename kvprintf -> kvcprintf (call-back version)
Add 'k' versions for printf, sprintf, and snprintf. kprintf, ksprintf, and
Undo some renames that don't apply to the boot code (linked against libstand).
Add IFF_MONITOR support.
Rename sprintf -> ksprintf
Remove unused procedures and declarations.
Continue fleshing out VKERNEL support. Initialize the per-cpu globaldata
Rename virtual_avail to virtual_start, so name follows function.
Remove unused defines.
Make a chunk of low level initialization code for proc0 and thread0 machine
Make certain libc prototypes / system calls visible to kernel builds
Introduce globals: KvaStart, KvaEnd, and KvaSize. Used by the kernel
Make kernel_map, buffer_map, clean_map, exec_map, and pager_map direct
Correct a conditional used to detect a panic situation. The index was off by
zbootinit() was being called with too few pv_entry's on machines with small
Fix compilation error when building without INET6. Fix a number of minor
Fix manual page references to omshell.
None of the patches in dhclient/client were being applied. Add the patches
Remove an old debugging kprintf.
Try to locate any instances where pmap_enter*() is called with a kernel
Get most of the VKERNEL pmap handling code in.
Remove pmap_kernel() (which just returned a pointer to kernel_pmap), and
Move dumplo from MD to kern/kern_shutdown.c
Continue fleshing out the VKERNEL.
Rename system calls, removing a "sys_" prefix that turned out not to be
Move uiomove_fromphyhs() source from MD to MI.
Remove fuswintr() and suswintr(), they were never implemented and it was a
Repo-move machine/pc32/i386/i386-gdbstub.c to cpu/i386/misc/i386-gdbstub.c.
GDB stubs were only being compiled on systems with serial I/O installed,
The last commit was incomplete, correct.
Conditionalize all of the subdirectories in dev so we skip them for
Conditionalize the existance of a gdt[] array for the in-kernel disassembler.
Conditionalize the existance of a gdt[] array for the in-kernel disassembler.
Rename errno to error to avoid conflict with errno.h
Remove all physio_proc_*() calls. They were all NOPs anyhow and used
VKERNEL work, deal with remaining undefined symbols.
Remove the hack.So hack for virtual kernels.
Remove old debugging printf.
Use Maxmem instead of physmem. physmem is used only within pc32
Initialize thread0.td_gd prior to calling various gdinit functions, because
Offset KernelPTD and KernelPTA so we can directly translate a kernel virtual
If no memory image file is specified, locate or create one by running
Add a new procedure, vm_fault_page(), which does all actions related to
Assign proc0 a dummy frame to bootstrap vm_fork().
Fix a conflict with libc's killpg().
Fix compiler warnings
Signal handlers usually inherit %gs. Make them inherit %fs as well. This
Fix symbol conflict with falloc()
define _KERNEL_VIRTUAL if not defined to hack-fix conflicts with normal
Make libc prototypes available for kernel builds.
Allow certain cpufunc.h inlines to be overridden by virtual kernel builds.
Use itimers to implement the virtual kernel's SYSTIMER backend.
Add a virtual disk device for virtual kernels to boot from.
Add support for a root disk device file.
The stack frame available from a signal to user mode stores the fault address
Handle page faults within the virtual kernel process itself (what would be
Set rootdevnames[0] to automatically boot from ufs:vd0a when a root disk
Make the vmspace_*() system call prototypes available to (virtual) kernel
Implement nearly all the remaining items required to allow the virtual kernel
Modify the trapframe sigcontext, ucontext, etc. Add %gs to the trapframe
The signal return code was trying to access user mode addresses
When removing a page directory from a page map, the KVA representing
Add missing header file.
Make more libc prototypes available to _KERNEL builds.
Make libc prototypes available to kernel builds
Implement vm_fault_page_quick(), which will soon be replacing
Rewrite vmapbuf() to use vm_fault_page_quick() instead of vm_fault_quick().
Use CBREAK mode for the console.
Add the virtual kernel's virtual disk device to the fray (vd*).
Rename the following special extended I/O system calls. Only libc, libc_r,
Add a missing pmap_enter() in vm_fault_page(). If a write fault does a COW
Disable terminal control characters while the virtual kernel is running,
Use our interrupt infrastructure to handle the clock interrupt, but
Fix two incorrect sigblock() calls.
A virtual kernel running an emulated process context must pop back into
A virtual kernel running another virtual kernel running an emulated process
Name the virtual kernel disk device 'vkd' instead of 'vd'.
Get floating point working in virtual kernels. Add a feature that allows
Fix collision with variable named 'errno'.
Have vectorctl silently succeed to remove a run-time warning.
Tell the real kernel not to sync the file that backs a virtual kernel's
Make the size of the pmap structure the same for both pc32 and vkernel
Open the root disk with O_DIRECT. We do not want both the real kernel and
Properly block SIGALARM and disable interrupts (i.e. signals) in init
Fix a bug vm_fault_page(). PG_MAPPED was not getting set, causing the
When removing a page directory, tell the real kernel to invalidate the
Major pmap update. Note also that this commit temporarily nerfs performance
Replace remaining uses of vm_fault_quick() with vm_fault_page_quick().
Remove unused SWI's.
Add missing link options to export global symbols to the _DYNAMIC section,
Fix a number of places where the kernel assumed it could directly access
Misc cleanups.
Adjust the gdb patch to account for the new register structure.
Set kernel_vm_end to virtual_start instead of virtual_end so it can be used
Add missing bzero() during low boot after malloc().
Add single-user mode boot option (-s).
Fix the recently committed (and described) page writability nerf. The real
Pull in a few bits from FreeBSD. Add a structure size field and floating
Include the VN device by default.
Allow VKERNEL builds to build certain non-hardware disk modules as well.
The FP subsystem might not work properly when a vkernel is run inside
Implement a new signal delivery mechanism, SA_MAILBOX. If specified the
Link up the interrupt frame to the systimer API. Use PGEX_U to indicate
Rename type to avoid conflict with 'kqueue' symbol.
Add KQUEUE support to the TAP interface.
Add O_ASYNC (SIGIO) support to kqueue(). Also add F_SETOWN support.
Add kqueue based async I/O support to the virtual kernel. Convert VKE to
Pass an interrupt frame to kqueue-based interrupts. Modify the console
Close two holes in the pmap code. The page table self mapping does not
cputimer_intr_reload() - prevent a negatively indexed or too-small a reload
We want the virtual kernel to be default-secure. Disable writes to kernel
Implement getcontext(), setcontext(), makecontext(), and swapcontext().
Add missing -I path when compiling assembly files.
Rename /usr/src/sys/machine to /usr/src/sys/platform. Give the platform
Fix license issue by removing -lreadline. The programs don't reference
Remove the advertising clause from vinum with permission from Greg Lehey,
Fix a crash related to the NPX (floating point) interrupt. The interrupt
checkdirs() was being passed the wrong mount point, resulting in a panic
Poor attempt to tracking the stack frame through a trap. Adjust for
Stop init before ripping its filesystem references out in order to
Fix the incorrect addition of a leading '/' in file paths in the journaling
Setup for 1.8 release - bump HEAD to 1.7.1 and synchronize preview tag.
Setup for 1.8 release - add new targets to Makefile to udpate the slip tag
Setup for 1.8 release - Adjust HEAD to 1.9.0
Fix generation of the mount path for "/" when a process is chrooted into
Remove gobsd from the list.
Implement -D
smbfs was not guarenteeing a NULL return vnode on error. This is required
Add note on using 'handle SIGSEGV noprint' when gdb'ing a virtual kernel.
Make sure all string buffers passed from userland are terminated before
Initial syslink system call documentation and overview.
Minor syntax cleanup
Generate a warning if a wired page is encountered on a queue during a free
Try to catch double-free bugs in the ACPI code. For now generate a warning
Add syslink.2 to the install list.
syslink work - Implement code for a reformulated system call, giving the
Update the syslink documentation. A number of major design changes have
Reformulate the syslink_msg structure a bit. Instead of trying to create
Update the syslink documentation. This is still a work in progress. The
Cleanup and reformulate some of the comments.
Add IP_MINTTL socket option - used to set the minimum acceptable TTL a
Add subr_alist.c. This is a bitmap allocator that works very similarly to
Remove ldconfig_paths_aout, it is no longer used.
Use SHUT_RD instead of a hardcoded value of 0 in calls to shutdown().
Kernel virtual memory must be mapped on a segment address boundary. Try
Bring in the skeleton infrastructure and manual page for the new syslink
Probably the last change to the syslink() system call. Allow a generic
Don't allow snd_nxt to be set to a value less then snd_una when restoring
Allocations of size greater then the radix were not returning the correct
We have a few generation sets for Red-Black trees that implement RLOOKUP
Clean up the so_pru_soreceive() API a bit to make it easier to read
Make 'last mounted on' reporting by fsck work again. Add a new option
Convert all pr_usrreqs structure initializations to the .name = data format.
Just throw all the main arguments for syslink() into syslink_info and
Give the sockbuf structure its own header file and supporting source file.
Sync netstat up to the sockbuf changes.
sbappendcontrol() was improperly setting sb_lastmbuf, creating a possible
IPV6 type 0 route headers are a flawed design, insecure by default, and
Move syslink_desc to sys/syslink_rpc.h so kernel code does not need
Fix various paths in rc.d/diskless and friends.
Add a generally accessible cpu_pause() inline for spin loops.
Implement ncpus_fit and ncpus_fit_mask. Admittedly not the best name.
When <sys/user.h> is included, it MUST be included first because it sets
Add a shortcut function, objcache_create_mbacked(), which is more complex
Implement SYSREF - structural reference counting, allocation, and sysid
When <sys/user.h> is included, it MUST be included first because it sets
Use the __boolean_t defined in machine/stdint.h instead of the
Remove unneeded references to sys/syslink.h. Get syslink_desc from
* Use SYSREF for vmspace structures. This replaces the vmspace structure's
Store the frequency and cputimer used to initialize a periodic systimer.
Revamp SYSINIT ordering. Relabel sysinit IDs (SI_* in sys/kernel.h) to
EST's module was being installed before the module list got initialized,
Reorder cpu interrupt enablement, do it in the code that drops
It is possible for spurious interrupt(s) to be posted to an AP cpu
Make the mountroot> prompt a bit more user-friendly.
More cleanups, do not allow backspacing beyond the start of the line.
Implement kern.do_async_attach. default disabled. To enable add
Move clock registration from before SMP startup to after. APIC_IO builds
Add missing crit_exit();
ata_boot_attach() is no longer used, #if 0 it out.
Document the interrupt moderation timer and the fact that even though
Reduce the livelock limit from 50Khz to 40Khz.
When thread0 is initialized it must also be LWKT scheduled or LWKT will
Add a new system call, lwp_rtprio(), and regenerate system calls.
Make libthread_xu use the new lwp_rtprio() system call, mainly taken from
Followup commit - fix a bug in the last commit.
pci_get_resource_list() was returning an illegal pointer instead of NULL
Update the vget, vput, vref, vrele, vhold, and vdrop documentation
Update vnode.9, correct spelling.
The bus_get_resource_list DEVMETHOD is primarily used to by PCI devices
Use SYSREF to reference count struct vnode. v_usecount is now
Update for vnode changes.
Update vnodeinfo to handle the recent vnode changes.
Changes to consdev - low level kernel console initialization.
Add fields to the ktrace header to allow kdump to also display the TID
Replace NOCDEV with NULL. NOCDEV was ((void *)-1) and as inherited
Give the device major / minor numbers their own separate 32 bit fields
Synchronize libkvm etc, with recent kernel major/minor device
Fix a bug where multiple mounts on the same mount point cause the
Fix the location of Make.tags.inc
Add the ID for the Realtek ALC862 codec to the hda sound driver.
Remove variables that are no longer used due to the previous commit.
Fix a vnode recyclement namecache<->vnode deadlock introduced with recent
Ooops, cache_inval_vp_nonblock() was being called too late, after the
Make the kern.ipc.nmbclusters and kern.ipc.nmbufs sysctls read-only.
Remove old unused cruft.
Remove the ancient diskpart program.
Start untangling the disklabel from various bits of code with the goal of
Continue untangling the disklabel. Have most disk device drivers fill out
Continue untangling the disklabel. Use the generic disk_info structure
Continue untangling the disklabel. Reorganize struct partinfo and the
Add a new command, /sbin/diskinfo, which uses the revamped DIOCGPART
The normal ATA driver is capable of handling 48 bit block addressing, but
* The diskslice abstraction now stores offsets/sizes as 64 bit quantities.
Support 64 bit file sizes and 64 bit sector numbers.
Continue untangling the disklabel. Add sector index reservation fields
Port 16 byte SCSI command support from FreeBSD. This adds support for
Add dev_drefs() - return the number of references on a cdev_t
Remove the roll-your-own disklabel from CCD. Use the kernel disk manager
Remove the roll-your-own disklabel from the ATA CD driver. Use the
Remove support for mcd and scd - these were old proprietary ISA cdrom
Remove libdisk from the build.
Synchronize the NATA kernel build.
Use DIOCGPART instead of DIOCGDINFO to remove references to the disklabel
Remove #include <sys/disklabel.h> from various source files which no longer
Remove libdisk from the Makefile.
Add getdisktabbyname() to libc. This will soon replace getdiskbyname().
Fix buildworld, getdiskbyname() has moved to <disktab.h>
Add back PCI_MAP_FIXUP, it was mistakenly removed.
Remove the NATA config file generation rules. Add a rule to the check:
use_mcd.h, use_scd.h no longer exist.
Finish moving boot/i386 to boot/pc32 (it was left half done), remove
Cleanup shutdown(2) usage and make it consistent. The implementation in rsh
Update # comments and documentation for disktab(5).
Continue untangling the disklabel.
Use the WHOLE_DISK_SLICE instead of the compatibility slice to
Implement raw extensions for WHOLE_DISK_SLICE device accesses for acd0.
Temporary hack until we can get rid of the disklabel dependancies.
Make savecore work wagain by using the new 64 bit dumplo (_dumplo64).
Continue untangling the disklabel. Clean up dumpdev handling by having
Keep the ds_skip_* fields in struct diskslice properly synchronized.
Fix a one-character allocated string buffer overflow that was introduced
Fix device recognition, /dev/vn0 now uses WHOLE_SLICE_PART, not partition 'c'.
Include geometry data in DIOCGPART so fdisk can use it instead of trying
Disklabel operations are no longer legal on the raw disk, use DIOCGPART to
Quickly update UPDATING with 1.8 -> 1.9+ documentation.
Cleanup diskerr() output a bit - don't say it was trying to write when
When a traditional bsd disklabel is present, try to reserve SBSIZE bytes
Remove all dependancies newfs had on accessing the disklabel. Use
More disklabel disentanglement - use DIOCGPART instead of DIOCGDLABEL.
Remove the fragment size, block size, and cpg fields from the disklabel
Continue untangling the disklabel.
Revert sc1 to testing getuid() like it says it does in the output.
Use 0x%08x for all minor numbers, not just those > 255.
Fix a bug in recent commits. When creating a virgin disk label for devices
Handle disklabels with the disk management layer instead of rolling our own
Remove DIOCWLABEL operation. Doing it destroyed the purpose of having
Remove unused define.
Simplify the lwkt_msg structure by removing two unused fields and a number
* Greatly reduce the complexity of the LWKT messaging and port abstraction.
The dump device must be opened before ioctls can be performed on it.
Fix the kinfo run/sleep state for pure kernel threads. This effects /bin/ps
LWKT message ports contain a number of function pointers which abstract
Properly detect interruptable LWKT sleeps and display as 'S' instead of 'D'.
Do an even better job discerning between blocked threads and blocked
Add lwkt_sleep() to formalize a shortcut numerous bits of code have been
Update documentation.
Add a -c file option to the vkernel to specify CD images. The first -c or -r
Add the kernel support function allocsysid().
The proper root device for a vkernel fs is vkd0s0a, not vkd0a.
Syslink API work - greatly simplify the syslink_msg structure. Reimplement
From within a virtual kernel, make /sbin/shutdown and /sbin/halt actually
Open the root image O_EXLOCK|O_NONBLOCK and exit with an error message
Add some syslink debugging programs.
Remove unnecessary initialization and fix a GCC-4.x run-time linking issue
Rename private to privdata and class to srclass to avoid conflicts with
Adjust M_NOWAIT to M_WAITOK or M_INTWAIT as appropriate.
Add flsl() for the NATA driver.
Add flsl() for the NATA driver.
Add a missing header file dependancy.
Merge all the FreeBSD work done since our initial import of NATA, except
Bring in 1.343 from FreeBSD. FreeBSD commit message:
Synchronize to FreeBSD 1.35 - just adds a #define for ATA_SMART_CMD.
Add a timings line for UDMA6 in two places. The drivers in question may or
M_NOWAIT can only be used in a driver where a failed memory allocation is
Implement boundary and maximum segment size handling in bus_dmamap_load().
Set the IDE DMA start bit as a separate I/O write from the DMA port
Part 1/2: Add a sanity check to the NATA interrupt code to assert that
Go to bed before the sun comes up.
Catch up a bit with FreeBSD netgraph by replacing *LEN constants with
Do any crash dump operation before the shutdown_post_sync event handler
Add a field to the keyboard abstraction structure that allows the USB
When compiling a kernel with all ktr logging (KTR_ALL), do not auto-enable
Add polling support to BGE.
Use the slab cache for PAGE_SIZE and PAGE_SIZE*2 sized allocations. This
Add KTR logging for SMP page table syncing ops.
Implement vm_fault_object_page(). This function returns a held VM page
Entirely remove exec_map from the kernel. Use the new vm_fault_object_page()
Get rid of some broken _KERNEL_VIRTUAL hacks.
Remove unused variables after last commit.
Remove the last source of SMP TLB invalidations in the critical code path
Formalize the object sleep/wakeup code when waiting on a dead VM object and
Increase the tsleep/wakeup hash table size and augment the KTR logging a bit.
Add a new option to ktrdump (-l) which causes it to loop awaiting new data
Change the -a option to be a bit friendlier. Have it print sorted,
Create an upgrade target for MAKEDEV. This target will attempt to
Remove the temporary NATA kernel build config file. Change GENERIC to
This patch allows umct (USB<->RS232 adapter) to write to devices that do
Move initialization of a vnode's various red-black trees to the CTOR function
Dont poll PS/2 mouse interrupts, it can cause the mouse to get jumpy.
Refuse to poll a detached sim.
dev_dopen() can be called multiple times with only one dev_dclose() when
Expand the diskslice->ds_openmask from 8 bits to 256 bits to cover all
Do not destroy the device queue, it is needed by the peripheral code
xpt_bus_deregister() never returns 0, don't test for it.
When getnewvnode() is called the vnode's v_type defaults to VNON. Syncer
Add vfs.nfs.pos_cache_timeout to complement vfs.nfs.neg_cache_timeout.
Import the kernel GPT and UUID header files from FreeBSD, and bring in
Bring uuidgen(3) into libc and implement the uuidgen() system call.
Implement an opaque function, if_getanyethermac(), which retrieves MAC
Regenerate system callsa (add uuidgen()).
Update all sound code to use the snd_*() locking abstraction and sndlock_t.
Correct mistake in last commit.
Create the USB task queues before creating the event thread to avoid
Bring the gpt labeling program in from FreeBSD.
Backout the last commit, it's creating panics.
Implement (non-bootable) GPT support. If a PMBR partition type is detected
Add subr_diskgpt.c to the platform conf files.
Add subr_diskgpt.c - oops. part of the GPT commit.
Bring in the uuidgen utility from FreeBSD.
Have UFS set the vnode type to VBAD instead of VNON so it gets cleaned
Add two new UUID helper functions to libc, uuid_name_lookup() and
Lines in /etc/uuids[.local] beginning with '#' are considered comments.
Change the location of the files to /etc/defaults/uuids and /etc/uuids.
Fix mistake in last commit, the file locations were not changed properly.
Augment RB tree macros even more, allowing for static declarations,
Adjust gpt to use the new UUID lookup functions via /etc/[defaults/]uuids.
Create defaults/uuids and adjust the build to copy the file to /etc/defaults
Make indexes start at 0, not 1, so they match the GPT partition numbers.
Fix compiler warning (embedded /*)
Fix an overflow in the GPT code, I wasn't allocating enough slice structures.
Disable per-channel interrupt sources before enabling the master interrupt
Implement SIGTERM handling. When a SIGTERM is received by a VKERNEL, it
More syslink messaging work. Now basically done except for the I/O 'DMA'
Disklabel separation work - Generally shift all disklabel-specific
Disklabel separation work - more.
Move all the code related to handling the current 32 bit disklabel
Improve the error message for gpt add a little.
Implement non-booting support for the DragonFly 64 bit disklabel:
Make some adjustments to clean up structural field names. Add type and
Correct a couple of uuid retention issues. Output the storage uuid for
Rename d_obj_uuid to d_stor_uuid to conform to the naming convention being
Add the -p pidfile option to the vkernel.
Refuse to label media that is too large to handle a 32 bit disklabel
The fstype was not being properly tested for a CCD uuid.
Do not blindly allow the block count to overflow. Restrict newfs filesystem
Correct a bug in the -S truncation mode where the mode was not being passed
Fix an issue with positive namecache timeouts. Locked children often
Fix rts_input() which is the only procedure which calls raw_input(). As
Recode the state machine to make it a bit less confusing. Collapse the
Adjust debug output so columns line up better.
Create a default dntpd.conf file for DragonFly using three pool.ntp.org
Add a new option (-i) that allows the insane deviation value to be set, and
Repo-copy numerous files from sys/emulation/posix4 to sys/sys and sys/kern
A file descriptor of -1 is legal when accessing journal status. Just allow
Implement jscan -o. Take the patch from Steve and add some additional
Fix a bug-a-boo, the type uuid was being printed instead of the storage
Clarify cpu localization requirements when using callout_stop() and
Get out-of-band DMA buffers working for user<->user syslinks. This
Add a new flag, XIOF_VMLINEAR, which requires that the buffer being mapped
Add O_MAPONREAD (not yet implemented). This will have the semantics of
Clean up syslink a bit and add an abstraction that will eventually allow
This is a simple little syslink test program which ping-pongs a 64K
Implement struct lwp->lwp_vmspace. Leave p_vmspace intact. This allows
Flag the checkpoint descriptor so on restore we can identify it and use the
Add MLINKS for checkpoint.1, because most point looking for information
Update the documentation for sys_checkpoint().
Move the P_WEXIT check from lwpsignal() to kern_kill(). That is, disallow
Try to avoid accidental foot shooting by not allowing a virtual kernel
A signal is sent to a particular LWP must be delivered to that LWP and never
More multi-threaded support for virtualization. Move the save context
Bring in all of Joe Talbott's SMP virtual kernel work to date, which makes
Conditionalize SMP bits for non-SMP builds.
Use dladdr() to obtain symbol names when possible and try to dump the
Also credit lots of help from Aggelos Economopoulos <aoiko@cc.ece.ntua.gr>
Do not allow umtx_sleep() to restart on a restartable signal. We want to
Implement an architecture function cpu_mplock_contested() which is
Clean up a kprintf() that was missing a newline.
Only use the symbol returned by dladdr() if its address is <= the
Copy a junk file from pc32 needed for <time.h>
sched_ithd() must be called from within a critical section.
The kernel perfmon support (options PERFMON) was trying to initialize its
The real-kernel madvise and mcontrol system calls handle SMP interactions
Add an option (-n ncpus) to specify the number of cpus a virtual kernel
Increase SMP_MAXCPU to 31. Can't do 32 (boo hoo!) because spinlocks need
sigwinch has to run with the big giant lock so use the DragonFly
Exhaust the virtual kernel network interface even if we cannot allocate
Put a timeout on the umtx_sleep() in the idle loop and add conditional
Because the objcache caches up to two magazines on each cpu some pretty
Make the virtual kernel's systimer work with SMP builds. Have it
Give virtual kernels access to sysctl() prototypes and clean up warnings.
A virtual kernel can cause a vm_page's hold_count to exceed 32768. Make
Implement an architecture call for contended spinlocks so the vkernel can
The vkernel's copyin/copyout implementation is not MP safe, acquire and
Add usched_mastermask - a master cpu mask specifying which cpus user
If more then 2 virtual cpus are present, dedicate one to handle I/O
Fix a number of races in the controlling terminal open/close code.
Fix an issue which arises with the TAP interface when the highly
Add a section on how to build the world inside a virtual kernel.
Try to catch double-replies a little earlier so we get a more meaningful
Fix an insufficient test of the message flags when determining whether
Update the CCD and disklabel documentation to reflect changes in
Do not synchronously waitmsg in the unix domain socket's custom putport
Use I/O size limits in underlying devices to govern I/O chunk
Add SHUTDOWN_PRI_DRIVER and move all driver shutdown functions from
Add an sc_maxiosize field which the ccd driver now needs.
Fix LWP support on exec. exec now properly kills all LWPs.
Be more verbose in the bad-opencount assertion.
Synchronous libarchive to 2.2.4 from FreeBSD, including fixes related to
Merge from vendor branch LIBARCHIVE:
Synchronous libarchive to 2.2.4 from FreeBSD, including fixes related to
Update to include info on last update.
Reparse device specifications. The slice is no longer optional.
Clean up the ioctl switch and add support for DIOCGPART which is
Rename the new cvsup bootstrap kit so make nrelease knows a new one
Temporarily reenable writing to the label area for backwards compatibility.
Update the installer to dfuibe_installer-1.1.7nb1. Add logic to remove
Be a little more verbose when reporting unmount errors.
Remove the requirement that calls to vn_strategy() be limited to the
Temporarily hack around an interrupt race against device detach by
Properly initialize next_cpu to 0 when '-l map' is used for the default
The disk layer must not inherit the D_TRACKCLOSE flag from the underlying
Do not loop forever doing 0-sized I/Os if si_iosize_max is 0. Instead
When CAM creates the disk layer, set the underlying raw device's
Incorporate the device DMA limitations into the request transfer size
Add assertions for 0-length transfers and panic if one is attempted.
SCSI CD devices do not support raw mode accesses (yet). This fixes
dssetmask() was being called too early, causing the disk layer to believe
Remove a duplicate SUBDIR +=
Clarify two usage cases for umtx.2
Synchronize the syslink manual page with the system call.
Minor checkpt usage() patch.
Bump DragonFly_version and create a subvers-DEVELOPMENT file for HEAD for
Add logic to allow the first hog partition to specify a base offset. This
release engineering: Add a slip tag for 1.10 and add an example cvsup
Release engineering: Update version information in HEAD to reflect 1.11
Update to 9.3.4-P1 with key id generation security fix.
Merge from vendor branch BIND:
Turn syscall_mpsafe and trap_mpsafe on by default. This causes system calls
Include a virtual kernel in the release along with GENERIC.
Add an ordering field to the interrupt config hook structure and adjust
Fix missing history->hist adjustments from libreadline->libedit commit.
Add infrastructure to locate a disk device by name by scanning the disk
When an inode collision occurs a partially initialized vnode will be
Introduce krateprintf(), a rate-controlled kprintf(), and the related
Rate-limit residual I/O warnings from the VM system that occur when a
Allow the compatibility slice (s0) to be specified.
vrecycle() is typically called from a VFS's inactive function, which
Detect the case where rename()ing over a file that has been unlinked but
Fix vinum. Vinum illegally uses device ops such as dev_dopen(),
Fix a coding mistake when dequeueing memory disk BIOs.
I missed this file. We are running 9.3.4-P1 now, not 9.3.4.
The LWP run/sleep state does not always match the actual state of the
The distribution installs a Makefile in /usr with easy-to-use targets to
Breakout a target for preview as well, and use the slip tag for the
Update some packages supplied by corecode. The new bootstrap is a binary
Do not require AHC_PCI_CONFIG to be defined.
Adjust the installer to properly install /usr/Makefile.
Fix build errors when MSDOSFS_DEBUG is defined.
Fix a mbuf leak that was introduced in April. In April I made a change
Add the MBUF_DEBUG option. This is a fairly invasive option that should
Explicitly set a large receive buffer for datagram sockets to give syslog
nfe_init() can be called as a side effect of certain ioctl operations with
Introduce two delays in nfe_stop().
Add support for a new revision of the RealTek 8168B/8111B called SPIN3.
The 1's complement checksum in m->m_pkthdr.csum_data was not being properly
It is possible to catch a LWP while it is being created or destroyed,
Port FreeBSD/pf_norm.c 1.18 - fix 1's complement carry for csum_data when
Oops. Correct attribution for the last commit - 1's complement csum_data
Add another fix to the 1's complement checksum. A second carry does not
Make m_mclfree() MP safe by fixing a N->0 dereferencing race. The spinlock
The cvs checkout commands were improperly specified the -d option.
Change the ordering of the zombie test for ^T to avoid a NULL pointer
Fix the fstab installation example. vkd0a -> vkd0s0a.
Explicitly extract the sector size for the underlying media. This solves
Replace the huge mess that was vnode_pager_generic_getpages() with a much
Add xio_init_pages(), which builds an XIO based on an array of vm_page_t's.
Remove the vpp (returned underlying device vnode) argument from VOP_BMAP().
The new VOP_N*() (namespace) operations pass a pointer to a namecache
Add additional functionality to the syslink implementation. Give the
Part 1/many USERFS support. Bring in the initial userfs infrastructure.
Fix for amd geode cs5536 companion (ehci) panic. Also fix a word-reversed
Fix pci bus detection on certain motherboards. Fixes bus detect on
Add '-H', 'nlwp', and 'tid' options to ps(1) to display some LWP data (inspired
Signals have to be blocked when creating our LWPs or a LWP may receive a
Convert the lwp list into a red-black tree. This greatly reduces the
Clean up the kvm process code. This is only used when trying to get a
Bring CARP into the tree. CARP = Common Address Redundancy Protocol, which
Deprecate 'installer_*' targets. If used a warning is generated and the
Fix the root device selection to match the manual page. Before it was always
Fix another ^T race related to catching a LWP just as it is being created,
Add a prototype and wrapper for lockuninit() to complement spin_uninit().
getpages/putpages fixup part 1 - Add support for UIO_NOCOPY VOP_WRITEs to
Do not try to dump UIO_NOCOPY writes to the journal. There's nothing
Temporary hack up a fix for msdos which avoids a deadlock between the VM
Add missing xfer->state assignment.
Add missing if_softc assignment, allowing pppX interfaces to
Add geode companion support to ohci probe code
part2 ppX fix for altq/pf.
Add T2300 cpu support to EST.
Fix a bugaboo in the last commit. Pages are hard-busied for getpages,
Fix a misleading error message when the device specified at the mountroot
Add vop_stdgetpages() and vop_stdputpages() and replace those filesystem
Fix a bug in vnode_pager_generic_getpages(). This function was improperly
Add a MNTK_ flag to the mount structure allowing a VFS to specify that
Bring in FreeBSD/1.177 - fix a bug in a call made to useracc(). This
Force an over-the-wire transaction when resolving the root of an NFS mount
Change the virtual kernel's default hz to 20, because the kqueue timers we
Set si_iosize_max to silence run-time warnings.
Remove a bogus assertion. in_vm86call may have been set by some unrelated
General userfs fleshing out work. Abstract out construction and
kern_access() had the same bug kern_stat() had with regards to a
Fix bugs in the handling of CIDR specifications such as 'route add 128/8
Fix a bootstrapping issue with the change of the default gcc to 4.1.
alist debug mode - report unrounded byte usage.
Interrupt thread preemption was switching in threads with held tokens
Add a note that dntpd can be started at boot time even if the network
Add an install location for the vkernel binary, add a second on how to
Add a uuid for the "DragonFly HAMMER" filesystem type.
libiconv was declaring a base kobj_class structure instead of an extended
Remove the /usr/lib/crt* files. These files have been located in
Allow the crc32.c module to be used in userland or kernel code.
Indicate that alist_free() calls do not have to supply power-of-2 aligned
Add a typedef for uuid_t for kernel compiles. One already existed for
Adjust RB_PROTOTYPEX to match RB_GENERATE_XLOOKUP. These declare a Red-Black
Add syslink_vfs.h for userfs, defining the syslink element infrastructure
Initial commit of mount_hammer - basic working skeleton for testing.
Initial commit of newfs_hammer - basic working skeleton for testing.
Primary header file infrastructure and A-list implementation for the
Adjust the description of HAMMER's storage limitations. I have rearranged
Clean up the structural organization. Separate out A-lists and make
Give the A-list code the ability to do a forward or reverse allocation
Add volume, super-cluster, cluster, and buffer abstractions to provide
Fix a race between exit and kinfo_proc. proc->p_pgrp and the related
Fix more NULL structural dereferences that can occur when a process is in
Reactivate a vnode after associated it with deadfs after a forced unmount.
HAMMER part 1/many. This is a clear-my-plate commit.
Add a HAMMER kernel build option, add a VFS type for HAMMER, add a file
Synchronize newfs_hammer with recent changes.
Correct a bug in the lockf code. F_NOEND was not being properly set.
Properly set the buf_type in the volume, super-cluster, and cluster headers.
A delete_tid of 0 indicates a record which has not yet been deleted and
HAMMER 2/many - core mount and unmount code now works, the B-Tree search
Add a PHOLD/PRELE sequence around a sysctl_out to fix a race against
Modify struct vattr:
Break-out the standard UNIX uid/gid tests for VOP_ACCESS into a helper file.
Convert the global 'bioops' into per-mount bio_ops. For now we also have
When the quotacheck has not been run the quota code may have to
Add regetblk() - reacquire a buffer lock. The buffer must be B_LOCKED or
Silence an annoying compiler warning.
HAMMER part 2/many.
Add bio_ops->io_checkread and io_checkwrite - a read and write pre-check
Correct bug in last commit.
Remove i386 support.
Separate ssb_lock() and ssb_unlock() into its own header file and reimplement
HAMMER 3/many - more core infrastructure.
HAMMER 4/many - more core infrastructure
Update the documentation for getdirentries(2). Describe issues with using
Add a helper function vop_helper_setattr_flags() modeled off of UFS's
Adjust getdirentries() to allow basep to be NULL. Use off_t for the loff
Make necessary changes to readdir/getdirentries to support HAMMER. HAMMER
Clean up some missing 32->64 bit cookie conversions. Adjust the NFS server
Add vop_helper_create_uid() - roughly taken from UFS. Figure out special
Break out the scan info structure's support routines so external code
HAMMER 5/many - in-memory cache and more vnops.
Fix loc_seek - using lseek to acquire the directory cookie.
Replace the very predictable 'random' IP sequence number generator with
Remove debugging printfs.
Drop into DDB if the vkernel hits a floting point exception (SIGFPE).
Catch vkernel divide-by-0 traps a bit earlier so they are reported properly.
HAMMER 6/many - memory->disk flush, single-cluster sync to disk, more vnops.
HAMMER 7/many - deletions, overwrites, B-Tree work.
Initialize idx_ldata - a forward iterator for allocating large (16K) data
Make fixes to the A-list initialization and properly allocate records
HAMMER 8/many - A-list, B-Tree fixes. As-of queries
HAMMER 9/many - btree removal cases, mount nohistory
Bring the getent(1) program in from FreeBSD and link it into the build.
Bring in Matthias Schmidt's nice little pkg_search script.
Add clarifying comments for LWP_WSTOP and LWP_WEXIT.
Fix a 'panic: vm_page_cache: caching a dirty page' assertion. Even though
Add missing sys/proc.h
Fix krateprintf(). The frequency was improperly being multiplied by hz
Update pkg_search to download and use the pkg_summary file, allowing
Synchronize most of the remaining FreeBSD changes for Elf64 typedefs.
Install hammer includes in /usr/include/vfs/hammer.
Fix bug in as-of mount date specification.
Save and restore the FP context in the signal stack frame.
HAMMER 10/many - synchronize miscellaneous work.
Properly set the mc_fpformat field in the ucontext so libc_r knows which FP
Use the mc_fpformat field to determine the correct FP save/restore
FP registers are now saved and restored by the kernel, remove the
HAMMER 11/many - initial spike commit.
HAMMER 12/many - add VOPs for symlinks, device, and fifo support.
HAMMER 12/many - buffer cache sync, buffer cache interactions, misc fixes.
HAMMER 13/many - Stabilization commit
HAMMER 13B/many - addendum to 13.
Add the 'hammer' utility. This is going to be a catch-all for various
HAMMER 14/many - historical access cleanup, itimes, bug fixes.
UFS vnodes must have VM objects before they can be truncated. This is
Back out the last commit, it asserts in the getblk code due to the vnode
HAMMER 15/many - user utility infrastructure, refactor alists, misc
fill_kinfo_proc() may be asked to load information on a zombied process,
HAMMER 16/many - Recovery infrastructure, misc bug fixes
HAMMER 16B/many: Fix data overwrite case.
Fix buffer cache deadlocks by splitting dirty buffers into two categories:
HAMMER 17/many: Refactor IO backend, clean up buffer cache deadlocks.
HAMMER 18/many: Stabilization pass
HAMMER 18B/many: Stabilization pass
Attempt to fix an interrupt recursion which can occur in specific
HAMMER 19/Many - Cleanup pass, cluster recovery code.
HAMMER 20A/many: B-Tree lookup API cleanup, B-Tree changes.
HAMMER utilities:
HAMMER 20B/many: New spike topology, simplify the B-Tree code.
HAMMER utilities:
HAMMER 21/many: B-Tree node locking finalization.
Add hammer_recover.c for kernel builds w/ HAMMER.
Fix an issue with cache_rename(). This procedure previously copied a
Fix time conversion bugs in the stamp command.
HAMMER 22/many: Recovery and B-Tree work.
HAMMER utilities:
HAMMER 23/many: Recovery, B-Tree, spike, I/O work.
HAMMER utilities: Features and sync with VFS.
HAMMER 24/many: Clean up edge cases
HAMMER utilities: Add a verbose (-v) option.
HAMMER 25/many: get fsx (filesystem test) working, cleanup pass
HAMMER 24B/many: Edge cases, cleanups
HAMMER utilities: synchronize newfs_hammer.
Conditionalize the illegal MXCSR tests on SSE support. Machines that did
Address a potential weakness in IBAA. The generator needs to be warmed up
Make sure scb->lastfound is NULLed out when it matches the entry being
HAMMER 25/many: Add an ioctl API for HAMMER.
HAMMER utilities: Add the 'prune' and 'history' commands.
Add a conditional so we don't have to drag in everything when a user
HAMMER 25/many: Pruning code
* Implement a mountctl() op for setting export control on a filesystem.
Fix a compiler warning.
Implement NFS support and export control for HAMMER.
HAMMER 26/many: More NFS support work, rename fixes
Fix some NFS related bugs which cause the mount point's mnt_refs counter
HAMMER 26/many: Misc features.
HAMMER Utilities: Add an 'everything' directive to the prune command. This
HAMMER 27/many: Major surgery - change allocation model
HAMMER 28/many: Implement zoned blockmap
HAMMER 28A/many: Translation and write performance optimizations
Make the Brother HL1240 printer work with ulpt.
Adjust nrelease to a new package set.
Bump to 1.11.1 prior to 1.12 branch and update the preview tag.
Release engineering, Add a slip target to /usr/src/Makefile for 1.12 and
Release Engineering on HEAD.
Oops, drop head's version back one for head (it was set to the release's
Fix an issue where the random number generator's random event injector
Fix mount_nfs to allow hostnames which begin with a digit.
Fix a use-after-free bug in the envelope code just after a port 25 fork.
HAMMER 29/many: Work on the blockmap, implement the freemap.
Require the the core file be owned by the user. Please also see the
HAMMER 30/many: blockmap work.
HAMMER 30A/many: blockmap cleanup
HAMMER 30B/many: Minor bug fix.
HAMMER 30C/many: Fix more TID synchronization issues
HAMMER 31A/many: File data size optimization
HAMMER 31B/many: Fix busy block dev on HAMMER umount
HAMMER 31C/many: Fix livelock in deadlock handling code
Clean up the token code and implement lwkt_token_is_stale(). Users of
Patch additional use-after-free cases.
HAMMER 32/many: Record holes, initial undo API, initial reblocking code
HAMMER utilities: Add the reblock command, adjust newfs_hammer.
HAMMER 32B/many: Reblocking work.
HAMMER 33/many: Expand transaction processing, fix bug in B-Tree
HAMMER 33B/many: Further B-Tree fix.
Improve vkernel support.
A threaded process going into SZOMB may still have active threads which are
HAMMER utilities: feature add.
Remove calls to pmap_clear_modify() in the swap_pager, fixing a kernel panic.
HAMMER 33C/many: features and bug fixes.
We must hold the lwp we are trying to kill to prevent it from being
HAMMER 34/many: Stabilization pass
Miscellanious features and adjustments to cpdup.
Bump cpdup's version to 1.08.
Add a FS_HAMMER id for disklabel.
Fix collision in conf/files, add hammer_signal.c.
Add mount_hammer, newfs_hammer, and the hammer utility to the build.
HAMMER 34B/many: Stabilization pass.
Synchronize various changes from FreeBSD. This is not exhaustive but gets
Rename PCIP_STORAGE_SATA_AHCI to PCIP_STORAGE_SATA_AHCI_1_0
HAMMER 35/many: Stabilization pass, cleanups
HAMMER utilities: automatic sync/sleep
HAMMER utilities: Add -lm for double arithmatic.
HAMMER 35B/many: Stabilization pass, cleanups.
HAMMER 35C/many: Stabilization pass.
Make sure there is no possibility of a cothread trying to access the
HAMMER 36/many: Stabilization pass.
It's frankly long past time that we turn net.inet.tcp.always_keepalive
HAMMER 36B/many: Misc debugging.
Fix a snafu with the last commit. Not all of the new AHCI detection support
Add fairq to altq. Fairq is a fair queueing algorithm with bandwidth
Bring the 'probability' keyword into PF from NetBSD. This feature allows
MFC 1.33/pf.c from NetBSD. Don't apply a window scale to the window
Add parallel transaction support for remote source or target specifications.
More cpdup work.
Properly detach children so we dont have to pthread_join() them. Fixes
Properly mark a transaction has being completed so the slave side of
Update PORTING instructions for linux to support pthreads.
Implement a number of major new features to PF.
pfsync_state doesn't have or need a hash field, the state will be hashed
Fix multiple issues with -p<parallel>, including several data corruption
Fix ktrace for threaded processes. Move the KTRFAC_ACTIVE flag to the LWP
Don't free held clean pages when asked to clean.
Minor optimization to LIST_FOREACH_MUTABLE taken from FreeBSD.
Fix a bug in umtx_sleep(). This function sleeps on the mutex's physical
Fix some issues in libthread_xu's condvar implementation.
Update the documentation for umtx_sleep() and umtx_wakeup().
Add __sreadahead() to help with pkgsrc's devel/m4.
Remove debugging assertion.
Refuse to talk with the remote cpdup if it's version is not compatible.
Finish up cpdup. Bump the protocol version to 2 and refuse to talk to
Add an interlock for certain usb task operations.
Dive the scheduler to implement the yield function. For the moment it just
Pass the current LWP to sigexit() instead of the current process. Remove
Fix some IO sequencing performance issues and reformulate the strategy
HAMMER 37/Many: Add a flush helper thread, clean up some inconsistencies.
Fix a free() race due to a misplaced mutex unlock.
* Remove the SINGLEUSE feature for telldir(), it does not conform to the
Fix two A-list corruption cases.
Fix panics which can occur when killing a threaded program. lwp_exit()
HAMMER 38A/Many: Undo/Synchronization and crash recovery
HAMMER 38B/Many: Undo/Synchronization and crash recovery
HAMMER 38C/Many: Undo/Synchronization and crash recovery
HAMMER 38D/Many: Undo/Synchronization and crash recovery
HAMMER 38E/Many: Undo/Synchronization and crash recovery
HAMMER 38E/Many: Undo/Synchronization and crash recovery
HAMMER utilities: Misc documentation and new options.
HAMMER 38E/Many: Undo/Synchronization and crash recovery
HAMMER 38F/Many: Undo/Synchronization and crash recovery, stabilization pass
Fix some pmap races in pc32 and vkernel, and other vkernel issues.
Minor code reordering and documentation adjustments.
Fix a NULL poiner dereference in the statistics collecting code as
KTR_TESTLOG is a valid kernel option (it enables the KTR ipi performance
Paging and swapping system fixes.
HAMMER 39/Many: Parallel operations optimizations
HAMMER Utilities: zone limit
HAMMER 39B/Many: Cleanup pass
The driver was improperly using kmem_free() instead of pmap_unmapdev(),
Add some assertions when a buffer is reused
Change the SMP wakeup() code to send an IPI to the target cpu's in parallel
Cothreads do not have a globaldata context and cannot handle signals
Add pmap_unmapdev() calls to detach functions for drivers which used
Have vfsync() call buf_checkwrite() on buffers with bioops to determine
HAMMER 40A/Many: Inode/link-count sequencer.
HAMMER 40B/Many: Inode/link-count sequencer cleanup pass.
HAMMER 40C/Many: Inode/link-count sequencer cleanup pass.
Print the path even if we do not understand the filesystem type.
HAMMER 40D/Many: Inode/link-count sequencer cleanup pass.
HAMMER 40E/Many: Inode/link-count sequencer cleanup pass.
HAMMER 40F/Many: Inode/link-count sequencer cleanup pass, UNDO cache.
Correct a bug in seekdir/readdir which could cause the directory entry
Print the 64 bit inode as a 64 bit quantity rather then a 32 bit quantity.
The direct-write pipe code has a bug in it somewhere when the system is
HAMMER 40F/Many: UNDO cleanup & stabilization.
HAMMER Utilities: enhanced show, timeout option
HAMMER 40G/Many: UNDO cleanup & stabilization.
HAMMER 41/Many: Implement CRC checking (WARNING: On-media structures changed)
HAMMER Utilities: Feature add
Only call bwillwrite() for regular file write()s, instead of for all write()s.
Keep track of the number of buffers undgoing IO, and include that number
HAMMER Utilities: Sync with recent changes.
HAMMER 41B/Many: Cleanup.
Remove the SMP_MAXCPU override for vkernels, causing the build to revert
Enable kern.trap_mpsafe and kern.syscall_mpsafe by default for vkernels.
Fix a sizeof() the wrong variable name. The correct variable was the same
Correct comments and minor variable naming and sysctl issues.
Bump base development version to 197700 so it is properly distinct from
Clear the direction flag (CLD) on entry to the kernel, to support future
Recode the resource limit core (struct plimit) to fix a few races and
Fix some lock ordering issues in the pipe code.
Fix a race between the namecache and the vnode recycler. A vnode cannot be
Fix a nasty memory corruption issue which can occur due to the kernel bcopy's
Fix fork/vfork statistics. forks and vforks were being improperly counted
Fix many bugs and issues in the VM system, particularly related to
HAMMER 42/Many: Cleanup.
Sync sysperf with some random stuff, and add a cld instruction tester.
Return EINVAL if a NULL pointer is passed to the mutex routines, instead
Fix a HAMMER assertion which turned out to be a bug in VOP_N*(). Sometimes
HAMMER 42A/Many: Stabilization.
Fix feature logic so changing kern.pipe.dwrite_enable on the fly works
Finish moving the kernel from tsc_freq (32 bits) to tsc_frequency (64 bits).
HAMMER Utilities: Feature add
HAMMER 42B/Many: Stabilization.
HAMMER 42C/Many: Stabilization.
HAMMER Utilities: scan feedback
Fix UP real kernel crash, a vkernel define was being used iproperly
HAMMER Utilities: Features
HAMMER 42D/Many: Stabilization.
HAMMER 42E/Many: Cleanup.
HAMMER Utilities: Cleanup.
HAMMER 43/Many: Remove records from the media format, plus other stuff
HAMMER 43A/Many: Cleanup, bug fixes.
HAMMER 43B/Many: Correct delete-on-disk record bug.
HAMMER Utilities: Misc features and adjustments.
HAMMER 43C/Many: Performance cleanup
HAMMER 44/Many: Stabilization pass, user-guaranteed transaction ids
HAMMER Utilities: Feature add
HAMMER Utilities: Features
HAMMER 45/Many: Stabilization pass, undo sequencing.
Fix str[c]spn off by one error.
The char dummy must be dummy[2] to accomodate a nul terminator when dealing
HAMMER Utilities: Stabilization pass.
Add a sysctl jail.allow_raw_sockets (default to diabled) which allows
Syntax cleanup and also commit a missing piece of the jail_allow_raw_sockets
Change cluster_read() to not block on read-ahead buffers it is unable to
HAMMER 46/Many: Performance pass, media changes, bug fixes.
HAMMER Utilities: Update for HAMMER changes.
Fix a number of core kernel issues related to HAMMER operation.
HAMMER 46B/Many: Stabilization pass
Fix an overflow in the time calculation.
Add a DELAY(500) during the register init phase to give the device some time
Properly track the write-open count when updating a msdos mount from RW to RO,
Add a define for IEEE80211_FC1_PROTECTED.
Bring in fixes for a bug which occurs when the filesystem become fulls.
HAMMER 47/Many: Stabilization pass
Fix a pipelining performance issue due to the way reading from the socket
Calls to DIOCSYNCSLICEINFO were being made with the assumption that
Use a per-bucket mutex to reduce contention and fix a seg-fault from a
Bump version to 1.11.
Fix a very old bug where the root mount was not getting a filesystem syncer
Add vop_helper_chmod() and vop_helper_chown(). These helper functions
HAMMER 48/Many: finish vop_setattr support, ncreate/nmknod/etc, minor bug fixes.
* Implement SOCK_SEQPACKET sockets for local communications. These sockets
Create a new daemon called vknetd. This daemon uses the new SOCK_SEQPACKET
Get rid of an old and terrible hack. Local stream sockets enqueue packets
Add vknetd to the build.
Do not try to set-up the bridge or tap interfaces when connecting to
Add the notty utility, a program I wrote long ago which I should have
Fix socketvar.h inclusion by userland. This is a temporary hack and,
Only test the IP protocol (ip_p) for IP frames.
Implement a new utility called vknet. This utility interconnects the
Generate a semi-random MAC address when connecting to a SOCK_SEQPACKET
HAMMER 49/Many: Enhance pruning code
HAMMER Utilities: Add the 'hammer softprune' command.
HAMMER 49B/Many: Stabilization pass
HAMMER Utilities: Cleanup
HAMMER Utilities: New utility 'undo'.
* Implement new system calls in the kernel: statvfs(), fstatvfs(),
Implement a new system call: getvfsstat(). This system call returns
Clean up statvfs() and related prototypes. Place the prototypes in the
More header file cleanups related statvfs.
Add getmntvinfo() which uses the new getvfsstat() system call.
Use newly available libc and system calls related to statvfs to make df
HAMMER Utilities: Performance adjustments, bug fixes.
HAMMER 50/Many: VFS_STATVFS() support, stabilization.
Even using the objcache we need a one-per-cpu free-thread cache in order
Fix kernel compile warnings.
HAMMER Utilities: Correct vol0_stat_freebigblocks.
Add missing exit(1).
Disallow negative seek positions for regular files, directories, and
Add the UF_NOHISTORY and SF_NOHISTORY chflags flags. The nohistory flag
Report the nohistory, noshistory, and nouhistory flags, and allow them
HAMMER 51/Many: Filesystem full casework, nohistory flag.
HAMMER Utilities: More pre-formatting, cleanup
Do not update f_offset on EINVAL.
HAMMER Utilities: Enhance mount_hammer
HAMMER 52/Many: Read-only mounts and mount upgrades/downgrades.
HAMMER 53A/Many: Read and write performance enhancements, etc.
HAMMER Utilities: Critical bug in newfs_hammer
HAMMER 53B/Many: Complete overhaul of strategy code, reservations, etc
HAMMER 53C/Many: Stabilization
Fix a SMP race in signotify_remote(). LWPHOLD() the lwp being
Add an extern for hidirtybuffers.
HAMMER 53D/Many: Stabilization
Change bwillwrite() to smooth out performance under heavy loads. Blocking
HAMMER 53E/Many: Performance tuning
HAMMER 53F/Many: Fix deadlock.
HAMMER 53G/Many: Performance tuning.
Switch from bioq_insert_tail() to bioqdisksort(). When the kernel is
HAMMER 53H/Many: Performance tuning, bug fixes
HAMMER 54/Many: Performance tuning
HAMMER 54B/Many: Performance tuning.
HAMMER 54C/Many: Performing tuning, bug fixes
Add missing LWPHOLD/LWPRELE in kinfo code.
Reimplement B_AGE. Have it cycle the buffer in the queue twice instead of
LWPHOLD/LWPRELE must be atomic ops because an IPI can call LWPRELE.
HAMMER 54D/Many: Performance tuning.
HAMMER 55: Performance tuning and bug fixes - MEDIA STRUCTURES CHANGED!
HAMMER Utilities: Sync with commit 55 - MEDIA STRUCTURES CHANGED!
Change the namecache lock warning delay from 1 to 5 seconds.
We must process incoming IPI messages when spinning in the thread
HAMMER 56A/Many: Performance tuning - MEDIA STRUCTURES CHANGED!
HAMMER Utilities: sync with 56A
HAMMER 56B/Many: Performance tuning - MEDIA STRUCTURES CHANGED!
HAMMER Utilities: Sync with 56B
Miscellanious performance adjustments to the kernel
HAMMER Utilities: Bug fixes
HAMMER 56C/Many: Performance tuning - MEDIA STRUCTURES CHANGED!
Fix a bug in cluster_read(). An error returned by the call to
HAMMER Utilities: Sync with 56D
HAMMER 56D/Many: Media structure finalization, atime/mtime, etc.
HAMMER 56E/Many: Correct bug in 56D
HAMMER 56F/Many: Stabilization pass
HAMMER 57/Many: Pseudofs support
HAMMER Utilities: Add the 'pseudofs' directive for commit 57
Support S_IFDIR mknod() calls for HAMMER. This is used by the Hammer
HAMMER Utilities: Sync with 58A
HAMMER 58A/Many: Mirroring support part 1
HAMMER Utilities: Remove time/transaction-id conversion directives.
HAMMER 58B/Many: Revamp ioctls, add non-monotonic timestamps, mirroring
HAMMER Utilities: Sync to 58B
HAMMER 59A/Many: Mirroring related work (and one bug fix).
HAMMER Utilities: Sync with 59A
HAMMER Utilities: Add "slave" option to hammer_mount.
Make sure UFS disallows mknod()'s with type VDIR.
Add KTR_HAMMER
Vendor import of netgraph from FreeBSD-current 20080626
Merge from vendor branch NETGRAPH:
Netgraph port from FreeBSD - initial porting work
Add additional atomic ops from FreeBSD.
Add files and options lines for NETGRAPH7
Apply patch supplied in FreeBSD-PR to ata-raid code:
dummy_thr does not have to be committed and pthread_t might not even
Increase the default request timeout from 5 seconds to 10 seconds.
HAMMER 59B/Many: Stabilization pass - fixes for large file issues
Fix a system performance issue created by ata_sort_queue(). This function
Bump the sortq_lost check from 8 to 128, letting the disk optimally read or
Replace the bwillwrite() subsystem to make it more fair to processes.
HAMMER 59C/Many: Stabilization pass - fixes for large file issues
Fix hopefully all possible deadlocks that can occur when mixed block sizes
HAMMER 59D/Many: Sync with buffer cache changes in HEAD.
HAMMER 59E/Many: Stabilization pass - fixes for large file issues
Adjust comments.
Fix an issue where CAM would attempt to illegally get a lockmgr() lock
Fix a NULL pointer dereference when a DDB 'ps' attempts to
HAMMER 59D/Many: Stabilization pass
Fix a buf_daemon performance issue when running on machines with small
HAMMER 59E/Many: Stabilization pass
Add a new helper function, kmalloc_limit(). This function returns
Fix a low-memory deadlock in the VM system which can occur on systems
HAMMER 59F/Many: Stabilization pass
Fix numerous pageout daemon -> buffer cache deadlocks in the main system.
HAMMER 59G/Many: Stabilization pass (low memory issues)
HAMMER 59H/Many: Stabilization pass
HAMMER 59I/Many: Stabilization pass
HAMMER 59J/Many: Features
HAMMER 60A/many: Mirroring work
HAMMER Utilities: Mirroring and pseudo-fs directives
HAMMER 60B/many: Stabilization
HAMMER Utilities: Sync with recent work.
HAMMER Utilities: Stabilization
HAMMER 60C/many: Mirroring
HAMMER 60D/Many: Mirroring, bug fixes
Error out if no volumes are specified instead of core-dumping.
HAMMER 60E/Many: Mirroring, bug fixes
HAMMER Utilities: Sync with 60E
When creating a new HAMMER filesystem also create a PFS record for it,
HAMMER 60F/Many: Mirroring
HAMMER Utilities: Sync with 60F
Rename fid_reserved to fid_ext.
UFS+softupdates can build up thousands of dirty 1K buffers and run out
HAMMER 60G/Many: Mirroring, bug fixes
Add the HAMMER filesystem to GENERIC and VKERNEL.
Cleanup - move a warning so it doesn't spam the screen so much, cleanup
HAMMER 60H/Many: Stabilization pass
HAMMER 60I/Many: Mirroring
HAMMER Utilities: Sync with 60I
HAMMER Utilities: Mirroring work
HAMMER 60J/Many: Mirroring
HAMMER Utilities: Sync with 60J
Add crc32_ext() - allows continuation of a 32 bit crc.
HAMMER 61A/Many: Stabilization
HAMMER 61B/Many: Stabilization
HAMMER 61C/Many: Stabilization
Add a vclean_unlocked() call that allows HAMMER to try to get rid of a
Correct a bug in the last commit.
HAMMER 61D/Many: Mirroring features
HAMMER Utillities: Sync with 61D
HAMMER 61E/Many: Stabilization, Performance
HAMMER Utilities: Cleanup
HAMMER 61F/Many: Stabilization
HAMMER 61F2/Many: Fix bug in last commit
HAMMER 61G/Many: Stabilization of new flush_group code
Use a 64 bit quantity to collect file size data instead of
HAMMER 61E/Many: Stabilization, Performance
Fix an asf core dump.
Kernel support for HAMMER:
HAMMER 61F/Many: Stabilization w/ simultanious pruning and reblocking
HAMMER Utilities: Features
2.0 Release Engineering:
2.0 Release Engineering:
2.0 Release Engineering:
NFS performance fixes.
HAMMER 61E/Many: Features
HAMMER Utilities: Sync with 61E
Fix a bug in vmntvnodescan() revealed by the recent NFS sync fix. The
Fix a bug where mount_nfs would properly parse an IP address, but would
Fix an issue where libthread_xu was not accepting the full priority
HAMMER 61G/Many: Stabilization
HAMMER 61H/Many: Stabilization
Fix a lock leak in nfs_create(), tracked down from a crash dump and
HAMMER 62/Many: Stabilization, performance, and cleanup
Add logic to warn of possible renames, and clearly state when failures may
In DragonFly, gpt partitions look like slices in /dev, and we match the
Code documentation only: Describe B_NOCACHE
Give krateprintf() an initial burst capability if count is set to
Make some adjustments to the buffer cache:
Fix multiple bugs in CAM related devices which go away unexpectedly. This
When dealing with a failed read properly set B_INVAL.
HAMMER 63/Many: IO Error handling features
Detach the SIM when a firewire disk device is disconnected. Leave the
Try to make fwohci work more reliably. Stop printing 'phy int' to the
O_CREAT was being allowed to leak through a read-only NFS export.
HAMMER 64/Many: NFS, cross-device links, synctid
HAMMER 65/Many: PFS cleanups and confusion removal
HAMMER Utilities: Sync with HAMMER 65.
HAMMER Utilities: Sync with HAMMER 65.
Change 'default' to 'English'
Synchronize some of the machine-independant AMD64 bits.
Synchronize some of the machine-independant AMD64 bits.
Change newfs_hammer to reserve a minimum of 100M for the UNDO FIFO. Any
HAMMER commit
An off-by-one malloc size was corrupting the installer's memory,
Adjust the desiredvnodes (kern.maxvnodes) calculation for machines
HAMMER - fix kmalloc exhaustion w/ 3G ram
Pass the correct string to md_mount() when doing a diskless nfs mount.
Add a terrible hack to GPT which allows non-EFI BIOSes to boot from it.
Add a quick entry for the new 'gpt boot' directive.
HAMMER: Mirroring, misc bug fixes
HAMMER Utilities: Streaming mirroring!
HAMMER Utilities: Cleanup
HAMMER: Mirroring work
Cast to unsigned long to match the rest of the expression. This is just
AMD64 work:
Create an #include layer for bus/pci and bus/isa so source files do not
Add amd64 files for the ISA and PCI busses and adjust the header files
HAMMER: fsync blocking fixes
Fix a panic on boot that can occur if you hit keys on the keyboard
Keep UFS compatible on 32 and 64 bit builds by changing the 'time_t' embedded
HAMMER 2.1:01 - Stability
HAMMER 2.0:02 - rmdir, stability
Implement a bounce buffer for physio if the buffer passed from userland
Remove daddr_t dependancies in the swap code. Move swblk_t and add
Don't bump intr_context when running from the softint. Hopefully this
AMD64 Support:
AMD64 Support:
AMD64 Support:
AMD64 Support:
AMD64 Support:
AMD64 Support:
Add memset() to help w/amd64 support.
* Add a flag to track an in-transit socket abort to avoid races when closing
Adjust the mcontext code to match amd64.
Adjust the mcontext code to match amd64.
AMD64: Fix bugs in cerror().
AMD64 - Sync AMD64 support from Jordan Gordeev's svn repository and
Back-out the tls change. The field must be signed or the system will not
Add BUF_CMD_FLUSH support - issue flush command to mass storage device.
HAMMER: Mass storage flush command support
AMD64: Add JG64 config file for testing purposes.
AMD64: Fix the crossworld build.
Flesh out BUF_CMD_FLUSH support.
* Move /kernel to /boot/kernel and /modules to /boot/modules.
Remove any vestiges of the old pam, particularly /etc/pam.conf. pam config
Continue working the abort path. Move SS_ABORTING flag handling inward
Adjust the boot code to boot from either a root with a /boot, or directly
Remove boot0gpt - it isn't ready yet (and may never be).
Bring hunt in from OpenBSD. The best multi-player terminal game ever!
Update the rconfig examples. Correct a bug in auto.sh and add hammer.sh.
Add a reference for /usr/share/examples/rconfig to the rconfig manual page.
Fix a crash on access to TAP's owning thread. The owning thread can go away
Fix an endless recursion and double fault which could occur when accessing
Bump up the size of the boot partition in the boot+HAMMER rconfig
Add missing sleep.
Increase sockbuf send and receive buffers to 57344 bytes. In particular,
Fix issues with the scheduler that were causing unnecessary reschedules
Add a MSGF_NORESCHED feature for lwkt thread-based message ports. The
Fix bug in hammer mirror command when used with a remote source
Augment loader.conf's manual page to describe hw.usb.hack_defer_exploration
Improve code flow for KASSERT and KKASSERT using __predict_false().
Fix an invalidation case that tends to occur under load on NFS servers or
* Implement the ability to export NULLFS mounts via NFS.
Unbreak buildworld a slightly different way. Adjust null.h to not
Add the 'hammer cleanup' command. This is a meta-command which will
If snapshots are disabled and the snapshots directory contains no
* Fix a bug in runcmd() - the argv[] list was not NULL terminated.
HAMMER: Fix a couple of minor non-corrupting bugs in HAMMER.
Make two more changes to the ata request queue sorting code.
Change the autoflush code to autoflush when a flush group reaches a
Fix a double-flush which was occuring for every unlinked inode, resulting
Rename the PFSD structure's prune_path[64] to snapshots[64].
sleep for a shorter period of time when an excessive number of inodes are
Additions to 'hammer pfs-*':
Do not loop forever if narg becomes negative. This can happen when asked
Add vop_stdpathconf and default it so filesystems do not need to declare
Use the new vop_stdpathconf() instead of rolling our own.
Linux emulation adjustments.
Add support for "RealTek 8102EL PCIe 10/100baseTX". Checksum support
Do not return an EINVAL error for certain abort and disconnect cases.
HAMMER Utilities: Adjust 'show' defaults.
Add code to verify the data CRC by default, in addition to the B-Tree
Change hammer_str_to_tid() and its callers to restrict the format of
Add a new utility called 'monitor' which uses kqueue to monitor a
Add KQUEUE support to HAMMER.
Add vfs.nfs.flush_on_hlink and default to off.
Allow an alignment default of 0 to be treated as 1.
Try to do a better job aborting active requests when a usb mass storage
Add kmalloc_raise_limit() - allow a subsystem to raise the allocation
Raise the kmalloc limit for the M_HAMMER_INO pool based on desiredvnodes
The priority mask used to compartmentalize the comparison to determine
Add a new feature to mirror-copy and mirror-write. If the target PFS does
Move the HAMMER option around a little and add ALTQ to the default VKERNEL
Add required range checks prior to kmalloc()ing socket option buffer space.
Fix flags handling for the install program. Properly set and clear flags
* Change hc_remove() to return a -errno if an error occurs instead of -1.
Fix a bootstrapping issue, UF_NOHISTORY may not exist on older systems.
HAMMER utilities:
HAMMER:
HAMMER: Correct minor oops in last commit. Do not allow downgrading.
Allow an error code to be returned in the head.error element of the ioctl.
HAMMER Filesystem changes:
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly into devel

Matthias Schmidt (142):
Hello :)
o Change my personal copyright to the DragonFly copyright
Add 'H' to usage() options. This was missing from the last commit.
Don't randomize fortune dat files during build time. fortune will display
Pass NULL instead of getprogname() to pidfile(). pidfile() will automatically
Test cvs add :)
Enhance the comment
Sync the adduser(8) command with FreeBSD. Summary:
Sync etc/periodic with FreeBSD. Short summary:
Update the man page to reflect the recent sync with FreeBSD.
Remove bogus non-reentrant "temporary" implementation of gethostbyaddr_r()
Remove 3rd clause of Berkely copyright, per letter. Don't rename 4. to 3.
Enclose O_ROOTCRED in _KERNEL_STRUCTURES. This is needed for the upcoming
Update kdump(1) to print more human readable information.
o Add missing dot (.)
Update to get entries for the Sun Grid Engine and iSCSI.
Mention /etc/firmware for firmware(9) image files.
Add pam_nologin(8) to the tree. pam_nologin verifies
o Mention pam_nologin(8) in nologin(5). Only adapt the changes from FreeBSD
Major update to pkg_search(1)
Move the following entries from kern to security
Renamed kern.ps_showallprocs to security.ps_showallprocs
Commit pkg_radd(1) on behalf of corecode@. pkg_radd is a wrapper for
If -m is specified, a newfs(8) command is printed that can be used to
Add pam(3) support for cron(8). cron uses pam to check if the user's account
Sync the passive fingerprinting database with OpenBSD to get support for
Remove reference to the FreeBSD developers handbook. We have a chapter
Sync with FreeBSD. This bings us the -m option to look for a specific
Use the new kldstat -q/-m options instead of "| grep"
Replace home-grown list iteration with methods from sys/queue.h
Add support for network devices found on Intel ICH9 hardware. I have one
Warn the user if he uses -e and procfs(5