From: "Ulrich Weigand" <uweigand@de.ibm.com>
To: gdb-patches@sourceware.org
Subject: [rfc] Fix removing breakpoint from shared library race
Date: Wed, 13 Aug 2008 20:36:00 -0000 [thread overview]
Message-ID: <200808132034.m7DKYEEK005260@d12av02.megacenter.de.ibm.com> (raw)
Hello,
after a shared library was unloaded, we can no longer insert breakpoints
into its (no longer present) code segment. Therefore, code in breakpoint.c
(disable_breakpoints_in_unloaded_shlib etc.) takes care to disable such
breakpoints.
However, in a multi-threaded application we cannot really guarantee that
we have noticed the shlib unload event at the time breakpoints are to be
inserted or removed. For the insertion case, insert_bp_location therefore
has its own check, and handles unloaded shared libraries appropriately.
When *removing* breakpoints, however, there is no such check. I have a
multi-threaded test case that reproducibly runs into an error when trying
to remove a breakpoint from a shared library that was *just* unloaded.
The patch below fixes this, by simply silently ignoring failures to remove
a breakpoint from a shared library code segment. The breakpoint will be
cleanly disabled once disable_breakpoints_in_unloaded_shlib gets a chance
to run (or at the next attempt to insert it).
Am I missing some reason why we shouldn't get to this point? Otherwise,
this seems a reasonble solution to me ...
Tested on powerpc-linux and powerpc64-linux.
Bye,
UIrich
ChangeLog:
* breakpoint.c (remove_breakpoint): Do not fail if unable to remove
breakpoint from shared library.
diff -urNp gdb-orig/gdb/breakpoint.c gdb-head/gdb/breakpoint.c
--- gdb-orig/gdb/breakpoint.c 2008-08-08 16:42:41.000000000 +0200
+++ gdb-head/gdb/breakpoint.c 2008-08-13 21:56:44.567419172 +0200
@@ -1642,6 +1642,13 @@ remove_breakpoint (struct bp_location *b
val = 0;
}
}
+
+ /* In some cases, we might not be able to remove a breakpoint
+ in a shared library that has already been removed, but we
+ have not yet processed the shlib unload event. */
+ if (val && solib_address (b->address))
+ val = 0;
+
if (val)
return val;
b->inserted = (is == mark_inserted);
--
Dr. Ulrich Weigand
GNU Toolchain for Linux on System z and Cell BE
Ulrich.Weigand@de.ibm.com
next reply other threads:[~2008-08-13 20:36 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-08-13 20:36 Ulrich Weigand [this message]
2008-08-18 11:47 ` Joel Brobecker
2008-08-18 14:15 ` Ulrich Weigand
2008-09-09 20:43 ` Joel Brobecker
2008-09-09 22:32 ` Ulrich Weigand
2008-08-26 17:43 ` Ulrich Weigand
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=200808132034.m7DKYEEK005260@d12av02.megacenter.de.ibm.com \
--to=uweigand@de.ibm.com \
--cc=gdb-patches@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox