This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH] Make only user-specified executable filenames sticky
- From: Joel Brobecker <brobecker at adacore dot com>
- To: Gary Benson <gbenson at redhat dot com>
- Cc: Don Breazeal <donb at codesourcery dot com>, "gdb-patches at sourceware dot org" <gdb-patches at sourceware dot org>
- Date: Fri, 17 Jul 2015 14:48:32 -0700
- Subject: Re: [PATCH] Make only user-specified executable filenames sticky
- Authentication-results: sourceware.org; auth=none
- References: <20150505151448 dot GA1417 at blade dot nx> <1430907977-30605-1-git-send-email-gbenson at redhat dot com> <5550E357 dot 9040209 at codesourcery dot com> <20150605093724 dot GA26604 at blade dot nx> <5571B828 dot 1080208 at codesourcery dot com> <20150703111418 dot GB15448 at blade dot nx>
On Fri, Jul 03, 2015 at 12:14:18PM +0100, Gary Benson wrote:
> Don Breazeal wrote:
> > On 6/5/2015 2:37 AM, Gary Benson wrote:
> > > Don Breazeal wrote:
> > > > On 5/6/2015 3:26 AM, Gary Benson wrote:
> > > > > In GDB some executable files are supplied by the user
> > > > > (e.g. using a "file" command) and some are determined by GDB
> > > > > (e.g. while processing an "attach" command). GDB will not
> > > > > attempt to determine a filename if one has been set. This
> > > > > causes problems if you attach to one process and then attach
> > > > > to another: GDB will not attempt to discover the main
> > > > > executable on the second attach. If the two processes have
> > > > > different main executable files then the symbols will now be
> > > > > wrong.
> > > > >
> > > > > This commit updates GDB to keep track of which executable
> > > > > filenames were supplied by the user. When GDB might attempt
> > > > > to determine an executable filename and one is already set,
> > > > > filenames determined by GDB may be overridden but
> > > > > user-supplied filenames will not.
> > > >
> > > > How does this interact with follow-exec-mode? If
> > > > follow-exec-mode is 'new' and the program execs, then 'run' will
> > > > use the original executable file. But if follow-exec-mode is
> > > > 'same' and the program execs, then 'run' will use the executable
> > > > file that was active after the exec call.
Hi Gary,
> > > > In the follow-exec-mode == 'same' instance, is the assumption
> > > > that the exec'd executable file takes on the same
> > > > 'user-supplied' attribute as the initial executable, since it is
> > > > using the original inferior?
> > > >
> > > > If so, is there a scenario where:
> > > > * the user supplies the exec file name
> > > > * the program execs, so the exec file name is now different
> > > > * then the user tries to do an attach (without an exec file name)
> > > > to a process running the original exec file, and gets the wrong
> > > > exec file name?
> > >
> > > I'm not sure. Where would I need to look to check this out?
> > > (Where is the bit that updates the filename after exec?)
> >
> > I think it goes like this: infrun.c:follow_exec calls
> > exec.c:exec_file_attach, which updates the name of the executable.
>
> Ah, there is exactly the scenario you describe Don, good call.
> Joel, I can fix v3 of this patch to zero exec_file_is_user_supplied
> before the exec_file_attach in follow_exec.
>
> I'm not convinced this patch will not introduce new bugs, the whole
> handling of how/where executable and/or symbol files get changed
> seems... haphazard :) That's mainly why I'd put this patch on the
> back burner.
>
> I'm on the fence as to whether this should be committed, so I'll
> defer to you Joel. If you say commit I'll re-test and push.
Now that the branch has been cut, I'm thinking it would be a good
to start getting the ball rolling again. WDYT?
--
Joel