[1003.1(2004)/Issue 6 0000609]: It is not clear what threads are considered blocked with respect to a call to pthread_cond_signal() or pthread_cond

Austin Group Bug Tracker

2014-08-15 14:16:00 UTC

A NOTE has been added to this issue.
======================================================================
http://austingroupbugs.net/view.php?id=609
======================================================================
Reported By: mmihaylov
Assigned To: ajosey
======================================================================
Project: 1003.1(2004)/Issue 6
Issue ID: 609
Category: System Interfaces
Type: Clarification Requested
Severity: Editorial
Priority: normal
Status: Under Review
Name: Mihail Mihaylov
Organization:
User Reference:
Section: pthread_cond_broadcast, pthread_cond_signal
Page Number: 1043
Line Number: 33043 - 33046
Interp Status: ---
Final Accepted Text:
======================================================================
Date Submitted: 2012-09-20 14:18 UTC
Last Modified: 2014-08-15 14:16 UTC
======================================================================
Summary: It is not clear what threads are considered blocked
with respect to a call to pthread_cond_signal() or pthread_cond_broadcast()
======================================================================

----------------------------------------------------------------------
(0002348) torvald (reporter) - 2014-08-15 14:16
http://austingroupbugs.net/view.php?id=609#c2348
----------------------------------------------------------------------
What is the current status of this? Is there consensus that POSIX wants to
provide the stronger wake-up ordering guarantees? If so, is there still
discussion about how to actually express, or is the final text already in
some document?

For reference, here's how C++ clarified the specification:
http://cplusplus.github.io/LWG/lwg-defects.html#2190
If this matches with POSIX' intent, please let me know. Also, while the
C++ wording can't be used as is for POSIX, I think it specifies the
semantics in a useful way, so maybe POSIX can say something similar.

Thanks.

Issue History
Date Modified Username Field Change
======================================================================
2012-09-20 14:18 mmihaylov New Issue
2012-09-20 14:18 mmihaylov Status New => Under Review
2012-09-20 14:18 mmihaylov Assigned To => ajosey
2012-09-20 14:18 mmihaylov Name => Mihail Mihaylov
2012-09-20 14:18 mmihaylov Section =>
pthread_cond_broadcast, pthread_cond_signal
2012-09-20 14:18 mmihaylov Page Number => 1043
2012-09-20 14:18 mmihaylov Line Number => 33043 - 33046
2012-09-21 14:24 torvald Issue Monitored: torvald
2012-09-21 15:37 torvald Note Added: 0001371
2012-09-21 17:46 dalias Note Added: 0001372
2012-09-21 18:29 mmihaylov Note Added: 0001373
2012-09-21 22:31 dalias Note Added: 0001374
2012-09-25 16:36 siddhesh Note Added: 0001378
2012-09-25 17:05 torvald Note Added: 0001379
2012-10-10 16:17 msbrown Note Added: 0001396
2012-10-10 19:43 mmihaylov Note Added: 0001398
2012-10-11 06:14 mmihaylov Note Edited: 0001398
2012-10-17 16:39 geoffclare Note Added: 0001403
2012-10-17 16:42 geoffclare Note Edited: 0001403
2012-10-18 07:45 Konrad_Schwarz Note Added: 0001404
2012-10-18 12:32 dalias Note Added: 0001405
2012-10-24 17:34 torvald Note Added: 0001408
2012-10-25 23:31 nialldouglas Note Added: 0001409
2012-10-26 00:02 dalias Note Added: 0001410
2012-10-28 00:37 nialldouglas Note Added: 0001411
2012-10-28 00:38 nialldouglas Note Edited: 0001411
2012-10-28 01:11 dalias Note Added: 0001412
2012-10-28 18:33 nialldouglas Note Added: 0001413
2014-08-15 14:16 torvald Note Added: 0002348
======================================================================