This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Using /dev/full in execv
- From: Dag Nygren <dag at newtech dot fi>
- To: Ulrich Drepper <drepper at redhat dot com>
- Cc: Dag Nygren <dag at newtech dot fi>, libc-alpha at sourceware dot org, dag at newtech dot fi
- Date: Wed, 23 Aug 2006 09:56:06 +0300
- Subject: Re: Using /dev/full in execv
> Dag Nygren wrote:
> > My problem here is that glibc shouldn't segfault here but preferably
> > tell me what is going on....
>
> You haven't spent a second looking at the code in question, have you?
> Any error message can be fatal, that's the whole purpose of opening that
> device. There even is a comment in the code explaining that.
OK, checked the code and has one question and one request:
Q. Is doing a forced ABORT really the only thing we can do here?
REQ:
Add a comment like this in the FAQ:
"Strange SIGSGEV:s when forking".
To work as planned glib needs full access (666) to the
/dev/null device and from glibc 2.4 even the /dev/full device. If this
access is
not available, you will see fork aborting (with a SIGSEGV in Linux).
This kind of comment would have saved me quite some time trying
to find out what was going on and I am sure I am not alone
Best
Dag