Re: [PATCH] symfile.c, find_separate_debug_file

["Michael Snyder" <msnyder@sonic.net>]

----- Original Message ----- 
From: "Daniel Jacobowitz" <drow@false.org>
To: "Michael Snyder" <msnyder@sonic.net>
Cc: <gdb-patches@sourceware.org>; "Mark Kettenis" <mark.kettenis@xs4all.nl>
Sent: Friday, August 03, 2007 8:00 AM
Subject: Re: [PATCH] symfile.c, find_separate_debug_file


> On Fri, Aug 03, 2007 at 07:41:56AM -0700, Michael Snyder wrote:
> > Sure.  It generates a string such as
> > "/usr/local/lib/debug//opt/grbx/lib.so",
> > which is then passed to the libc function 'open', which fails to find
the
> > file.
> >
> > When I remove the extra slash, open succeeds.
> >
> > This is glibc 2.5 running under UML in Scratchbox, on Ubuntu.
> > Gdb is x86-cross-x86, not that that's probably relevant.
>
> That is strange.  If you try it outside of UML and scratchbox, I bet
> that won't happen; my money is on a UML bug in the host filesystem
> driver.
>
> I can't remember where in POSIX it is, and I can't find it, but I'm
> pretty sure that behavior is non-conforming.

OK, well I'm interested to find out.  I *think* I saw the same failure
under Ubuntu, but I'll have to confirm when I get back to the office.

Either way, though, there's nothing wrong with taking out an extra
slash, is there?  It doesn't serve any purpose.  Maybe gdb should
work even if the underlying OS isn't Posix compliant.