Bug 1121980 - Regression for vgdb
Summary: Regression for vgdb
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: Fedora
Classification: Fedora
Component: valgrind
Version: rawhide
Hardware: x86_64
OS: Linux
unspecified
unspecified
Target Milestone: ---
Assignee: Mark Wielaard
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2014-07-22 09:19 UTC by Jan Kratochvil
Modified: 2014-11-01 16:53 UTC (History)
11 users (show)

Fixed In Version: valgrind-3.10.0-4.fc21
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2014-11-01 16:53:55 UTC


Attachments (Terms of Use)

Description Jan Kratochvil 2014-07-22 09:19:17 UTC
Description of problem:
I got a nightly GDB testsuite regression:
-PASS: gdb.base/valgrind-infcall.exp: target remote for vgdb
+FAIL: gdb.base/valgrind-infcall.exp: target remote for vgdb

Version-Release number of selected component (if applicable):
PASS: valgrind-3.9.0-19.svn20140715r14165.fc22.x86_64
FAIL: valgrind-3.9.0-21.svn20140718r14176.fc22.x86_64

How reproducible:
Always.

Steps to Reproduce:
/usr/bin/valgrind --vgdb-error=0 true
+
gdb true
(gdb) target remote | /usr/lib64/valgrind/../../bin/vgdb --pid=42

Actual results:
0x0000000004000d40 in ?? ()

Expected results:
0x0000000004000d40 in _start () from /lib64/ld-linux-x86-64.so.2

Additional info:
It may be also a GDB bug in the end but it is caused by the valgrind version change.  I do not know more but filing it at least as a tracker.

Comment 1 Philippe Waroquiers 2014-07-22 21:29:21 UTC
I looked at the difference in valgrind between revision 14615 and 14176.
It is unclear which change could have caused this.

Also, from what I can see, the above is a stack trace which is produced
by GDB just at startup.
At this stage, the only thing valgrind+vgdb gives to gdb is the
initial program counter.
It is then up to gdb to translate this pc in a function/object/source file.

What is the stack trace given by gdb when running true natively in gdb ?
I.e. doing:
   gdb true
   break _start (or break *0x0000000004000d40?)
   run
   bt

Philipppe

Comment 2 Jan Kratochvil 2014-08-03 17:14:10 UTC
The regression is due to Fedora local patch:
Patch1002: valgrind-3.9.0-ppc64le-functional.patch

Upstream trunk is not affected.

As tested on:
http://pkgs.fedoraproject.org/cgit/valgrind.git/tree/?h=f21&id=da645ddbb5fe9fcac281887b27d29e225ec1ff1a

PASS: svn://svn.valgrind.org/valgrind/trunk@14227
     +svn://svn.valgrind.org/vex/trunk@2911
PASS: valgrind-3.9.0-svn20140718r14176.tar.bz2
PASS: valgrind-3.9.0-svn20140718r14176.tar.bz2
     +Patch1: valgrind-3.9.0-cachegrind-improvements.patch
     +Patch2: valgrind-3.9.0-helgrind-race-supp.patch
     +Patch3: valgrind-3.9.0-stat_h.patch
     +Patch4: valgrind-3.9.0-ldso-supp.patch
     +Patch1001: valgrind-3.9.0-ppc64le-initial.patch
FAIL: valgrind-3.9.0-svn20140718r14176.tar.bz2
     +Patch1: valgrind-3.9.0-cachegrind-improvements.patch
     +Patch2: valgrind-3.9.0-helgrind-race-supp.patch
     +Patch3: valgrind-3.9.0-stat_h.patch
     +Patch4: valgrind-3.9.0-ldso-supp.patch
     +Patch1001: valgrind-3.9.0-ppc64le-initial.patch
     +Patch1002: valgrind-3.9.0-ppc64le-functional.patch

Comment 3 Mark Wielaard 2014-08-04 08:45:12 UTC
Thanks for tracking that down. valgrind-3.9.0-ppc64le-functional.patch is part of the ppc64le port:

# ppc64le
# Initial PPC LE support KDE#334384
# PPC LE functional support KDE#334834
# PPC LE testsuite change KDE#334836
# Note these require regeneration of the auto* files in %%build.
Patch1001: valgrind-3.9.0-ppc64le-initial.patch
Patch1002: valgrind-3.9.0-ppc64le-functional.patch
Patch1003: valgrind-3.9.0-ppc64le-test.patch
Patch1004: valgrind-3.9.0-ppc64le-extra.patch

In particular it is the patch attached to bug:
https://bugs.kde.org/show_bug.cgi?id=334834

I don't immediately see anything that should impact gdb integration on none-ppc64le setups. But apparently something does affect non-ppc64le setups in that patch.

Comment 4 Philippe Waroquiers 2014-08-07 23:05:04 UTC
(In reply to Mark Wielaard from comment #3)
> 
> In particular it is the patch attached to bug:
> https://bugs.kde.org/show_bug.cgi?id=334834
> 
> I don't immediately see anything that should impact gdb integration on
> none-ppc64le setups. But apparently something does affect non-ppc64le setups
> in that patch.
It might have been discovered by a code review:
"Fixed a couple of issues found by Julian during his review.  One of the issues
would have caused issues on non PPC64 systems.  Fixes were in file
coregrind/m_ume/elf.c".

So, the last version of the ppc64 patch might solve the issue.

Comment 5 Fedora Update System 2014-08-29 18:49:20 UTC
valgrind-3.9.0-26.svn20140829r14384.fc21 has been submitted as an update for Fedora 21.
https://admin.fedoraproject.org/updates/valgrind-3.9.0-26.svn20140829r14384.fc21

Comment 6 Fedora Update System 2014-08-30 04:47:41 UTC
Package valgrind-3.9.0-26.svn20140829r14384.fc21:
* should fix your issue,
* was pushed to the Fedora 21 testing repository,
* should be available at your local mirror within two days.
Update it with:
# su -c 'yum update --enablerepo=updates-testing valgrind-3.9.0-26.svn20140829r14384.fc21'
as soon as you are able to.
Please go to the following url:
https://admin.fedoraproject.org/updates/FEDORA-2014-9965/valgrind-3.9.0-26.svn20140829r14384.fc21
then log in and leave karma (feedback).

Comment 7 Fedora Update System 2014-10-11 21:14:35 UTC
valgrind-3.10.0-4.fc21 has been submitted as an update for Fedora 21.
https://admin.fedoraproject.org/updates/valgrind-3.10.0-4.fc21

Comment 8 Fedora Update System 2014-11-01 16:53:55 UTC
valgrind-3.10.0-4.fc21 has been pushed to the Fedora 21 stable repository.  If problems still persist, please make note of it in this bug report.


Note You need to log in before you can comment on or make changes to this bug.