[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: second try, defunct processes



On Tue, Feb 15, 2000 at 03:16:37PM +0100, Grendel wrote:
> * Aaron Solochek said:
> > kernel 2.2.13, sparcstation 5 (170mhz, 128mb), gcc 2.95.2.  
> is the kernel compiled with that compiler? Can you tell that the processes
> that go defunct are ones that fork during their life span? If so, then that
> could be a problem with the kernel. After a parent forks, the kernel will
> try to notify the parent about the fact that a life of its child(ren) has
> ended by sending a signal. If this signal isn't received by the parent (e.g.
> by using waitpid or wait4) the child process will go zombie (<defunct>). If
> the application is coded properly and it still happens, it might mean that
> the kernel fails to deliver the SIGCHLD signal to the parent for some
> reason. It might be caused by the kernel compiled with 2.95.x gcc.
> Generally, it's safer to use gcc 2.7.2.3 to compile your kernels.

That would be true except this is sparc, and 2.7.2.3 can create some buggy
code for sparcs.

This sounds simply like you need to use a newer kernel, and most
preferrably, the vger cvs. I suggest getting 2.2.14 source, and applying
the patches from the kernel-image-2.2.14-sparc sources in woody. You could
also get the cvs directly (currently 2.2.15-pre7) using this:

cvs -d :pserver:anonymous@cvs.on.openprojects.net:/cvs/linux login
<blank pass>
cvs -d :pserver:anonymous@cvs.on.openprojects.net:/cvs/linux co -r linux_2_2 linux

You can then keep this updated by doing this command in the top directory
of the tree:

cvs -z9 up -dP

Good luck.

-- 
 -----------=======-=-======-=========-----------=====------------=-=------
/  Ben Collins  --  ...on that fantastic voyage...  --  Debian GNU/Linux   \
`     bcollins@debian.org  --  bcollins@openldap.org  --  bmc@visi.net     '
 `---=========------=======-------------=-=-----=-===-======-------=--=---'


Reply to: