This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
[PATCH 0/3 v4] Demangler crash handler
- From: Gary Benson <gbenson at redhat dot com>
- To: gdb-patches at sourceware dot org
- Cc: Andrew Burgess <aburgess at broadcom dot com>, Doug Evans <xdje42 at gmail dot com>, Eli Zaretskii <eliz at gnu dot org>, Florian Weimer <fw at deneb dot enyo dot de>, Mark Kettenis <mark dot kettenis at xs4all dot nl>, Pedro Alves <palves at redhat dot com>, Tom Tromey <tromey at redhat dot com>
- Date: Thu, 5 Jun 2014 14:01:40 +0100
- Subject: [PATCH 0/3 v4] Demangler crash handler
- Authentication-results: sourceware.org; auth=none
Hi all,
This patch is an updated version of the demangler crash handler I
posted yesterday. The main changes from the previous version are:
1) All #ifdef SIGSEGV conditionals have been removed.
2) If the user set "ulimit -c 0" then no core file will be
created and a warning will be printed. This mirrors what
internal_error and internal_warning currently do.
3) A separate signal stack is created for the SIGSEGV handler
to allow it to function correctly if the normal process
stack overflows. This signal stack is currently only used
by the SIGSEGV handler--all other signal handlers use the
normal process stack as before.
Doug requested that I change the patch to emit warnings for every
demangler crash, not just the first. I've not done this, my reason
being that subsequent failures could have been caused by the first,
by memory corruption or some leftover state: they could be bogus,
and could lead to us chasing bugs that don't exist. I prefer this
way, but I'm not hung up on it and if it's a blocker for Doug or
anyone else I will add the extra warnings.
I've split the patches as follows:
1/3 - adds a new category of internal problem for demangler
warnings. This patch is unchanged from the previous
version (PATCH 1/2 v3):
https://sourceware.org/ml/gdb-patches/2014-06/msg00142.html
2/3 - refactors and exposes the core-dumping functions in utils.c.
This is a completely new patch.
3/3 - the crash catcher itself. This patch differs from the
previous version by the removal of the #ifdef SIGSEGV
conditionals, the addition of a check to see if a core
dump should be performed, and the creation of a separate
stack to allow the signal handler to function when the
normal stack is exhausted.
I would push all three patches as one commit. The news file entries
for the commit would be:
* New options
maint set catch-demangler-crashes (on|off)
maint show catch-demangler-crashes
Control whether GDB should attempt to catch crashes in the
symbol name demangler.
maint set demangler-warning quit (yes|no|ask)
maint show demangler-warning quit
Control whether GDB should exit if it catches a crash in the
symbol name demangler.
* New commands
maint demangler-warning
Cause GDB to call the internal function demangler_warning and
hence behave as though an internal error in the demangler has
been detected.
Is this ok to commit?
Thanks,
Gary
--
http://gbenson.net/