This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [commit] delete corefile.c:close_exec_file
- From: Joel Brobecker <brobecker at adacore dot com>
- To: Tom Tromey <tromey at redhat dot com>
- Cc: gdb-patches at sourceware dot org
- Date: Thu, 5 Jan 2012 07:03:25 +0400
- Subject: Re: [commit] delete corefile.c:close_exec_file
- References: <m3zkeu9868.fsf@fleche.redhat.com> <20111218063648.GF21915@adacore.com> <m3hb0sv5ec.fsf@fleche.redhat.com> <20120104151730.GA2730@adacore.com> <m38vln9u8g.fsf@fleche.redhat.com>
> Yes, I think it does, but there are still a lot of bfd_cache_close_all
> calls to contend with. Maybe these hid the fact that close_exec_file
> did nothing.
Ah, yes, of course. These, I know we're responsible for some of them.
And there is at least one that we tried to contribute, wasn't quite
right, and then fell on the side. I think. But we should be able to
find the history of each one.
> I imagine at least some of these calls are needed on some platforms.
> I haven't tried to test this yet.
Yes, I am pretty sure they are, unfortunately.
> Some of the calls seem a bit weird. E.g., the one in
> symbol_file_add_with_addrs_or_offsets, from this thread:
>
> http://sourceware.org/ml/gdb-patches/2004-07/msg00439.html
I reconnected the thread in the archives (a real shame that
the threading does not work across months...):
[RFA] win32: bfd_cache_close after kill
http://www.sourceware.org/ml/gdb-patches/2004-05/msg00678.html
http://www.sourceware.org/ml/gdb-patches/2004-05/msg00756.html
http://www.sourceware.org/ml/gdb-patches/2004-06/msg00008.html
http://www.sourceware.org/ml/gdb-patches/2004-06/msg00512.html
In the process of doing so, I found:
[commit] Close the executable after it exits
http://sourceware.org/ml/gdb-patches/2008-04/msg00706.html
(from DanielJ)
[patch] release handle on object files after program exits
http://sourceware.org/ml/gdb-patches/2009-04/msg00074.html
(from yours truly)
I think it should be reasonably easy to find the history of each
of them should we need to. Based on the few messages I glanced at,
it seems that they deal with unlocking the exe after program exits
or gets killed by GDB. The main platform involved is Windows, with
HP-UX being mentioned as well...
--
Joel