DragonFly bugs List (threaded) for 2006-09
Re: 2 panics: assertion: p->p_numposixlocks == 0 in exit1
:Hope this is the right place to post this:
:Had 2 kernel panics:
:assertion: p->p_numposixlocks == 0 in exit1
:Prior to that I had several messages (Don't know if this is helpful or not)
:Sep 5 23:52:56 mercury kernel: swap_pager: out of swap space
:Sep 5 23:52:58 mercury kernel: swap_pager_getswapspace: failed
:Sep 5 23:52:58 mercury kernel: pid 76579 (httpd), uid 80, was killed: out of swap space
:Sep 5 23:53:05 mercury kernel: swap_pager: out of swap space
:Thought it kind of odd, since I didn't have that many things running, the
:machine wasn't thrashing or anything.
That can really only happen if the machine actually does run out
of swap space. It sounds like one of the httpd's ate all available
:Debugger(c0524821,c05eaea0,c0511d95,cb262bf0,0) at Debugger+0x44
:panic(c0511d95,c0523c57,c04f334f,cb262c98,cb258594) at panic+0xae
:exit1(d,c02a7c21,cb262c50,c02a7c21,c0af8e80) at exit1+0x219
:expand_name(cb258420,d,c0af8e80,cb262c98,0) at expand_name
:postsig(d,cb262c90,1,ffffffff,1000) at postsig+0xdf
:userret(cb258594,cb262d48,0,cb258420,0) at userret+0x107
:syscall2(2f,2f,2f,10,0) at syscall2+0x29d
:Xint0x80_syscall() at Xint0x80_syscall+0x2a
:The second panic, I did exactly this:
:logged in using ssh.
:su'd to root
:mounted an nfs partition /usr/pkgsrc
:(Can't recall if I started apache first or second)
:Then in /usr/pkgsrc/mail/procmail
:echo 'bmake package ; bmake install ; bmake distclean' | batch
:A short time later it paniced. (probably when the above commands were
What version of DragonFly are you running? There are known bugs
prior to the 1.6 release that can result in that panic but they
should have been fixed in 1.6.
:I'll try to go straight to nfs mounting and bmake to see if apache
:was related in any way.
:I set up dumpdev on the swap partition, however, I don't understand how
:this could be useful to anyone? (I'd imagine no one wants 128M file
:of my swap partition..)
It was almost certainly the bmake. procmail does a bunch of locking
tests during the build and that probably triggered the bug.
If the DragonFly is < 1.6 then upgrade. If its >= 1.6 then definitely
try to get a kernel core out of it.
p.s. I'm leaving today for a few days, I'll be back Friday evening.