This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-ssa libmudflap long] multithreading fixes, build reorg, testsuite extensions
- From: Eyal Lebedinsky <eyal at eyal dot emu dot id dot au>
- To: "Frank Ch. Eigler" <fche at redhat dot com>
- Cc: gcc at gcc dot gnu dot org
- Date: Sun, 06 Jul 2003 00:11:15 +1000
- Subject: Re: [tree-ssa libmudflap long] multithreading fixes, build reorg, testsuite extensions
- Organization: Eyal at Home
- References: <20030704190945.A5731@redhat.com> <3F0649CF.4DE8626@eyal.emu.id.au> <20030705124604.GA8165@redhat.com>
"Frank Ch. Eigler" wrote:
> Why not just post the diffs as formal patch submissions? You
> have a copyright assignment now - you could do your development
> right in our CVS repository.
Are you saying that I can do commits? I understood that posting to
gcc-patches in the way to go (which I will do once I get some
confidence from a full test run).
Talking about tests... I had my tests running all day, and after
about 10 hours I see this logged:
mf: erroneous reentrancy detected in `__mf_register'
ssasrsv.orig: ../../../gcc/libmudflap/mf-runtime.c:668: __mf_check:
Assertion `rc==0' failed.
The second line is then repeated 1300 times. The program
running then (my main multithreaded server) is not active
anymore (some kind of lockup?).
mf-runtime.c:668 is the first line of __mf_check, a LOCKTH ().
BTW, this error did show up once, very early in the tests, in
a non-critical program. That program is one of the simplest
I have (does the same as "echo `date` $*"). While it does launch
a thread, it proceeds to join it, so there is hardly any
concurrency. It logged a line identical to the first above
and aborted.
--
Eyal Lebedinsky (eyal@eyal.emu.id.au) <http://samba.org/eyal/>