This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [patch] Fix watchpoint-at-epilogue 7.0 regression (s390x)
- From: Joel Brobecker <brobecker at adacore dot com>
- To: Jan Kratochvil <jan dot kratochvil at redhat dot com>
- Cc: gdb-patches at sourceware dot org
- Date: Sat, 26 Dec 2009 09:35:55 +0400
- Subject: Re: [patch] Fix watchpoint-at-epilogue 7.0 regression (s390x)
- References: <20091225142031.GA5442@host0.dyn.jankratochvil.net>
> gdb/
> 2009-12-25 Jan Kratochvil <jan.kratochvil@redhat.com>
>
> * breakpoint.c (watchpoint_check): Check the call
> gdbarch_in_function_epilogue_p before calling frame_find_by_id.
> Extend the comment.
This part looks OK to me.
On a more general side, I wondered for a while if frame_find_by_id
should be modified to never raise an exception (and return NULL instead).
But decided that this was probably a bad idea. Using the same kind
of example as provided by Jan, if our program was to stop inside
the epilogue a function called by the block where the expression
is valid, we might end up deleting the watchpoint prematurely.
> gdb/testsuite/
> 2009-12-25 Jan Kratochvil <jan.kratochvil@redhat.com>
>
> * gdb.base/watchpoint-cond-gone.exp, gdb.base/watchpoint-cond-gone.c,
> gdb.base/watchpoint-cond-gone-stripped.c: New.
OK, with a few changes:
- The copyright notices needs to be GPL v3;
- One nit, see below.
> + gdb_suppress_entire_file "Testcase compile failed, so all tests in this file will automatically fail."
We are now trying to get rid of gdb_suppress_entire_file. Can you just
"return" instead?
BTW: Thank you for using some of the convenience routines. This reminded
me that I wanted to start writing a GDB testcase writing cookbook. Your
testcase had a lot of stuff that I wanted to put in there, so I just
finally started it! Will send an email to gdb@ when I'm done...
--
Joel