Re: kern/38165 (pipe(2) failure, then uvm_fault(0x*, 0, 1) -> 0xe (kernel: supervisor trap page fault, code=0))

[Matthew Mondor <mm_lists%pulsar-zone.net@localhost>]

The following reply was made to PR kern/38165; it has been noted by GNATS.

From: Matthew Mondor <mm_lists%pulsar-zone.net@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: kern/38165 (pipe(2) failure, then uvm_fault(0x*, 0, 1) -> 0xe
 (kernel: supervisor trap page fault, code=0))
Date: Fri, 27 Nov 2009 22:01:34 -0500

 On Fri, 27 Nov 2009 17:25:02 +0000 (UTC)
 "Greg A. Woods" <woods%planix.com@localhost> wrote:
 
 >  Unfortunately I do not have any systems running or in development beyond
 >  netbsd-4 yet.  (Go Git Go! ;-))
 
 On netbsd-5 I'm unable to reproduce the bug using more than a douzen
 consecutive runs of a command containing a hundred piped cat(1)
 (feeding them a few lines followed by an EOF).
 Assuming that this is enough to trigger the bug on netbsd-4 it seems to
 indeed have been fixed since...
 
 I observed various problems if there were enough piped instances to
 cause sh vfork errors, in which case STOPed cat(1) processes would
 remain pending (possibly an sh(1) bug not handling vfork(2) failures
 gracefully?).  After enough of them sh(1) also got stuck into an
 endless RUN loop, but killing that sh process restored the system to a
 normal state.
 
 Thanks,
 -- 
 Matt