This is the mail archive of the
libc-help@sourceware.org
mailing list for the glibc project.
Re: dead-lock in glibc
- From: Joël Krähemann <jkraehemann at gmail dot com>
- To: "Carlos O'Donell" <carlos at systemhalted dot org>
- Cc: "libc-help at sourceware dot org" <libc-help at sourceware dot org>, Torvald Riegel <triegel at redhat dot com>
- Date: Fri, 17 Mar 2017 13:58:51 +0100
- Subject: Re: dead-lock in glibc
- Authentication-results: sourceware.org; auth=none
- References: <CA+Owze40Onq_uZs2wOjY=O5Xv3D75Ce_b7Sf5qEjMZ-bAnW_wA@mail.gmail.com> <CAE2sS1gXkrLAZf2o54QSkE_fqFMrSd987nP=QYRe=GQEdq26_w@mail.gmail.com> <CA+Owze6vtqJ4jURD2H4fouw5izePVaQ9iun2LCLQ+HqwVvkvWw@mail.gmail.com> <CAE2sS1iF1ua0w9379zm-nMToTxQfVJfTxa78uMgs6z=LEqy5GA@mail.gmail.com> <CA+Owze6nxNpB+FWAQfu_6duy0wEA1n+K5mY1C6KAZEO1-dn4eQ@mail.gmail.com> <CAE2sS1hWPORM=dvAoQti_eHb4NcKX81sdevL3fgm2gZ1vwOuLQ@mail.gmail.com>
- Reply-to: jkraehemann-guest at users dot alioth dot debian dot org
Hi Carlos
I wanted to extend the test-suite of my application anyway. May be
we are able to reproduce the issue.
I'm going to tell you when the test is ready.
Bests,
Joël
On Thu, Mar 16, 2017 at 3:28 PM, Carlos O'Donell
<carlos@systemhalted.org> wrote:
> On Thu, Mar 16, 2017 at 2:30 AM, Joël Krähemann <jkraehemann@gmail.com> wrote:
>> Thank you for the hints. If you need additional information please let me know.
>
> What would be invaluable would be a program that demonstrates, using
> the APIs, the same behaviour outside of the interactive environment
> you were using. I know that's a tall order, but usually until someone
> drives the problem complexity down a bit we can't easily look at the
> issue.
>
> In some areas we have made progress to simplify this, for example we
> have an API tracer for the malloc family of functions that allows
> users to trace an API sequence and replay it in a simulator to attempt
> to reproduce a failure. When we extend our tracing to enough APIs we
> might be able to one day just get a "API trace" from you and replay it
> to observe the same behaviour on our end.
>
> Thanks for the feedback.
>
> Cheers,
> Carlos.