Re: cvs commit: src/sys/ddb db_ps.c src/sys/kern init_main.c kern_exit.c kern_fork.c kern_resource.c kern_sig.c sys_generic.c src/sys/platform/pc32/i386 pmap.c src/sys/platform/vkernel/platform pmap.c src/sys/sys proc.h tree.h src/sys/vm vm_vmspace.c

["Simon 'corecode' Schubert" <corecode@xxxxxxxxxxxx>]

Matthew Dillon wrote:
>     There are several places where iterations are replaced with a nearly
>     direct lookups.  None of these places scaled well to a linear list,
>     and almost all of them are in critical paths.  rtprio, direct LWP

I think rtprio only for other threads setting the prio for a thread, not if the thread changes its own prio.

>     signaling (such as the vkernel uses to send an IPI to another cpu),

Yes, that's bad.

>     selrecord(), selwakeup(), and TID selection for newly forked threads.
>
>     selrecord() was O(nthread*nfd).  Now it is O(nfd) like it was originally.
>     selwakeup() was O(ioreadyrate*nthread*(nthread again if collisions occur)).
>     Now it is O(ioreadyrate*(nthread if collisions occur)).

The problem is that sel*() is anyways O(nproc), because it only stores the PID and not the proc pointer.  So that is definitely an area for optimization.

cheers
 simon

--
Serve - BSD     +++  RENT this banner advert  +++    ASCII Ribbon   /"\
Work - Mac      +++  space for low €€€ NOW!1  +++      Campaign     \ /
Party Enjoy Relax   |   http://dragonflybsd.org      Against  HTML   \
Dude 2c 2 the max   !   http://golden-apple.biz       Mail + News   / \