[issue13817] deadlock in subprocess while running several threads using Popen

Changes by Doug Hellmann : -- nosy: +doughellmann ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail.p

[issue13817] deadlock in subprocess while running several threads using Popen

Charles-François Natali added the comment: Committed. Christoph, thanks for the report. -- resolution: -> fixed stage: commit review -> committed/rejected status: open -> closed ___ Python tracker __

[issue13817] deadlock in subprocess while running several threads using Popen

Roundup Robot added the comment: New changeset c3649173d093 by Charles-François Natali in branch '2.7': Issue #13817: After fork(), reinit the ad-hoc TLS implementation earlier to fix http://hg.python.org/cpython/rev/c3649173d093 New changeset 7b24dd587a7b by Charles-François Natali in branch '

[issue13817] deadlock in subprocess while running several threads using Popen

Changes by Stephen White : -- nosy: +Stephen.White ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail.

[issue13817] deadlock in subprocess while running several threads using Popen

Charles-François Natali added the comment: I thought about this a bit more, and I realized that a slight variation of this bug also affects 2.7, and also older versions (i.e. before _PyGILState_Reinit() was introduced), because any code that gets called from PyOS_AfterFork() and uses the TLS API

[issue13817] deadlock in subprocess while running several threads using Popen

Changes by Charles-François Natali : -- stage: patch review -> commit review ___ Python tracker ___ ___ Python-bugs-list mailing list

[issue13817] deadlock in subprocess while running several threads using Popen

Changes by Ross Lagerwall : -- nosy: +rosslagerwall ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail

[issue13817] deadlock in subprocess while running several threads using Popen

Charles-François Natali added the comment: And here's the test. To sum up: - reinit_tls.diff makes sure to call PyThread_ReInitTLS() at the beginning of PyOS_AfterFork(), so that the TLS API is usable within PyOS_AfterFork() (e.g. in _after_fork()). It would be applied to 3.2 and default, even

[issue13817] deadlock in subprocess while running several threads using Popen

Martijn van Oosterhout added the comment: What a concidence, we had this problem as well and just today tracked it down, and turns out someone filed this bug yesterday. I note the referenced bug (#13156) says that it's been fixed in the release branch but not yet released. I don't suppose the

[issue13817] deadlock in subprocess while running several threads using Popen

Charles-François Natali added the comment: Here's the patch. It's probably possible to add a test for this, however I don't have access to my development machine, so I can't write it now. -- Added file: http://bugs.python.org/file24275/reinit_tls.diff _

[issue13817] deadlock in subprocess while running several threads using Popen

Antoine Pitrou added the comment: I think you forgot to attach the patch :) -- nosy: +pitrou ___ Python tracker ___ ___ Python-bugs-l

[issue13817] deadlock in subprocess while running several threads using Popen

Charles-François Natali added the comment: Here's the backtrace: """ #0 0x003bfb20c9b1 in sem_wait () from /lib64/libpthread.so.0 #1 0x0051a7c3 in PyThread_acquire_lock (lock=0x17db0750, waitflag=1) at Python/thread_pthread.h:321 #2 0x0051a9b4 in find_key (key=1, valu

[issue13817] deadlock in subprocess while running several threads using Popen

Changes by Antoine Pitrou : -- nosy: +neologix ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail.pyth

[issue13817] deadlock in subprocess while running several threads using Popen

New submission from Christoph Glaubitz : Starting several threads, each just starting a subprocess.Popen and use communicate cause a deadlock in subprocess. (see attached file) I can only reproduce this with python 2.7.2, not with any other versions. 2.6.5, 2.6.7 and 3.2.2 are working fine. Al