This is the mail archive of the
gdb-testers@sourceware.org
mailing list for the GDB project.
Re: diffgdbdaymail 20141002
- From: Jan Kratochvil <jan dot kratochvil at redhat dot com>
- To: Pedro Alves <palves at redhat dot com>
- Cc: gdb-testers at sourceware dot org
- Date: Fri, 3 Oct 2014 20:28:34 +0200
- Subject: Re: diffgdbdaymail 20141002
- Authentication-results: sourceware.org; auth=none
- References: <201410021312 dot s92DCARu006338 at host1 dot jankratochvil dot net> <542DEC71 dot 3070904 at redhat dot com>
On Fri, 03 Oct 2014 02:23:13 +0200, Pedro Alves wrote:
> > --- 20141001Build-gdbgit-gdbserver-f21/fedora-21-x86_64/out/gdb-m64.sum 2014-10-01 12:14:18.075234612 +0200
> > +++ 20141002Build-gdbgit-gdbserver-f21/fedora-21-x86_64/out/gdb-m64.sum 2014-10-02 14:50:47.782072676 +0200
> >
> > #gdb.base/random-signal.exp
> > Running gdb/testsuite/gdb.base/random-signal.exp ...
> > PASS: gdb.base/random-signal.exp: set can-use-hw-watchpoints 0
> > PASS: gdb.base/random-signal.exp: watch v
> > PASS: gdb.base/random-signal.exp: continue
> > -PASS: gdb.base/random-signal.exp: stop with control-c
> > +FAIL: gdb.base/random-signal.exp: stop with control-c (timeout)
>
> Hmm. I'm wondering whether this is still related to b57bacecd5f3684cd9f58b0da0e2caccbb6a546d
> (Fix non-stop regressions caused by "breakpoints always-inserted off" changes).
This is a long-term fuzzy result.
I have checked now the directory
20141002Build-gdbgit-gdbserver-f21/fedora-21-x86_64/ has GIT commit
2278c276a887b12b85ae30d63c446bf45a3bfd9f (therefore already containing the
commit b57bacecd5f3684cd9f58b0da0e2caccbb6a546d you ask about).
That is I think b57bacecd5f3684cd9f58b0da0e2caccbb6a546d is unrelated to this
"FAIL: gdb.base/random-signal.exp" fuzzy result.
I have checked that in my local results from 2013-06-20 when this testcase was
checked in
commit 427cd150eed8c0dd4f0d0a1105448b4ebe36da6d
Author: Tom Tromey <tromey@redhat.com>
Date: Tue Jun 18 19:57:49 2013 +0000
this FAIL race was already happening. Therefore I have added it to the
already rich set of ignored patterns:
http://git.jankratochvil.net/?p=nethome.git;a=blob;f=bin/diffgdb;h=b44c2c09d7ed3ab19bb584aaebf24939a090a9eb;hb=HEAD
> It'd be great if the git hash that was tested was included in the report mail.
> Would that be possible?
I have implemented it now as it was very simple;
--- 20141002Build-gdbgit-gdbserver-rawhide//fedora-rawhide-x86_64/out/gdb-m32.sum 2014-10-02 15:11:00.763184923 +0200
--- GIT commit 2278c276a887b12b85ae30d63c446bf45a3bfd9f
+++ 20141003Build-gdbgit-gdbserver-rawhide//fedora-rawhide-x86_64/out/gdb-m32.sum 2014-10-03 09:10:22.507555435 +0200
+++ GIT commit 665bd7cfef61fc6afb7968920b590cccd258c6a9
but I no longer want to invest much into these scripts as they should be done
some other way. I started them as my private aid for my former RHEL GDB
maintenance only.
Even this addition I find only very rarely useful - these scripts are run
nightly (=per day), not per commit. Therefore one has to always run some
"git bisect" after any suspicious is found anyway.
Thanks,
Jan