public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug libgomp/42829]  New: TLS detection in ./configure is wrong.
@ 2010-01-21 15:41 jos dot de_laender at telenet dot be
  2010-01-21 15:42 ` [Bug libgomp/42829] " jos dot de_laender at telenet dot be
                   ` (5 more replies)
  0 siblings, 6 replies; 7+ messages in thread
From: jos dot de_laender at telenet dot be @ 2010-01-21 15:41 UTC (permalink / raw)
  To: gcc-bugs

In attachment, you find the code that is used inside the configure m4 macro for
detecting the presence of TLS.

This check is unreliable.

When compiled with -O2 (as done in configure) the result might be that the
compiler reports same adress due to optimization and decides there is no TLS.

When compiled with -O0, it yields the correct result.

This is an optimization artefact. Please extend the test with the two lines I
put in comment. This way the test succeeds correctly, even after optimization.


-- 
           Summary: TLS detection in ./configure is wrong.
           Product: gcc
           Version: 4.4.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libgomp
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: jos dot de_laender at telenet dot be
  GCC host triplet: i686-pc-mingw32
GCC target triplet: mingw64 - see description.


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42829


^ permalink raw reply	[flat|nested] 7+ messages in thread

* [Bug libgomp/42829] TLS detection in ./configure is wrong.
  2010-01-21 15:41 [Bug libgomp/42829] New: TLS detection in ./configure is wrong jos dot de_laender at telenet dot be
@ 2010-01-21 15:42 ` jos dot de_laender at telenet dot be
  2010-01-21 19:49 ` pinskia at gcc dot gnu dot org
                   ` (4 subsequent siblings)
  5 siblings, 0 replies; 7+ messages in thread
From: jos dot de_laender at telenet dot be @ 2010-01-21 15:42 UTC (permalink / raw)
  To: gcc-bugs



------- Comment #1 from jos dot de_laender at telenet dot be  2010-01-21 15:42 -------
Created an attachment (id=19681)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=19681&action=view)
The test.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42829


^ permalink raw reply	[flat|nested] 7+ messages in thread

* [Bug libgomp/42829] TLS detection in ./configure is wrong.
  2010-01-21 15:41 [Bug libgomp/42829] New: TLS detection in ./configure is wrong jos dot de_laender at telenet dot be
  2010-01-21 15:42 ` [Bug libgomp/42829] " jos dot de_laender at telenet dot be
@ 2010-01-21 19:49 ` pinskia at gcc dot gnu dot org
  2010-01-21 20:37 ` jakub at gcc dot gnu dot org
                   ` (3 subsequent siblings)
  5 siblings, 0 replies; 7+ messages in thread
From: pinskia at gcc dot gnu dot org @ 2010-01-21 19:49 UTC (permalink / raw)
  To: gcc-bugs



------- Comment #2 from pinskia at gcc dot gnu dot org  2010-01-21 19:48 -------
Oh this test should pass no matter what really as TLS is emulated on mingw via
functions which call pthreads.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42829


^ permalink raw reply	[flat|nested] 7+ messages in thread

* [Bug libgomp/42829] TLS detection in ./configure is wrong.
  2010-01-21 15:41 [Bug libgomp/42829] New: TLS detection in ./configure is wrong jos dot de_laender at telenet dot be
  2010-01-21 15:42 ` [Bug libgomp/42829] " jos dot de_laender at telenet dot be
  2010-01-21 19:49 ` pinskia at gcc dot gnu dot org
@ 2010-01-21 20:37 ` jakub at gcc dot gnu dot org
  2010-01-21 22:14 ` jos dot de_laender at telenet dot be
                   ` (2 subsequent siblings)
  5 siblings, 0 replies; 7+ messages in thread
From: jakub at gcc dot gnu dot org @ 2010-01-21 20:37 UTC (permalink / raw)
  To: gcc-bugs



------- Comment #3 from jakub at gcc dot gnu dot org  2010-01-21 20:37 -------
Yeah, the test is certainly fine, if it sporadically fails, there are issues
with the mingw TLS or thread implementation (or both).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42829


^ permalink raw reply	[flat|nested] 7+ messages in thread

* [Bug libgomp/42829] TLS detection in ./configure is wrong.
  2010-01-21 15:41 [Bug libgomp/42829] New: TLS detection in ./configure is wrong jos dot de_laender at telenet dot be
                   ` (2 preceding siblings ...)
  2010-01-21 20:37 ` jakub at gcc dot gnu dot org
@ 2010-01-21 22:14 ` jos dot de_laender at telenet dot be
  2010-01-23 10:35 ` ktietz at gcc dot gnu dot org
  2010-06-11  8:38 ` iains at gcc dot gnu dot org
  5 siblings, 0 replies; 7+ messages in thread
From: jos dot de_laender at telenet dot be @ 2010-01-21 22:14 UTC (permalink / raw)
  To: gcc-bugs



------- Comment #4 from jos dot de_laender at telenet dot be  2010-01-21 22:14 -------
Your call.
I *don't* think it's fine.
Even if there *is* TLS, it can fail due to the -O2 optimizing.
(it doesn't detect the referring via &).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42829


^ permalink raw reply	[flat|nested] 7+ messages in thread

* [Bug libgomp/42829] TLS detection in ./configure is wrong.
  2010-01-21 15:41 [Bug libgomp/42829] New: TLS detection in ./configure is wrong jos dot de_laender at telenet dot be
                   ` (3 preceding siblings ...)
  2010-01-21 22:14 ` jos dot de_laender at telenet dot be
@ 2010-01-23 10:35 ` ktietz at gcc dot gnu dot org
  2010-06-11  8:38 ` iains at gcc dot gnu dot org
  5 siblings, 0 replies; 7+ messages in thread
From: ktietz at gcc dot gnu dot org @ 2010-01-23 10:35 UTC (permalink / raw)
  To: gcc-bugs



------- Comment #5 from ktietz at gcc dot gnu dot org  2010-01-23 10:35 -------
Yes, I can confirm this issue, but want to investigate where the underlying
issue really comes from. The interesting issue I found is, that if a
TLS-variable isn't assigned, the values of the addresses in different threads
are equal. As soon as one thread initialize the variable in program flow (not
by initializing the global __thread variable), everthing works as expected.

Kai


-- 

ktietz at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ktietz at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42829


^ permalink raw reply	[flat|nested] 7+ messages in thread

* [Bug libgomp/42829] TLS detection in ./configure is wrong.
  2010-01-21 15:41 [Bug libgomp/42829] New: TLS detection in ./configure is wrong jos dot de_laender at telenet dot be
                   ` (4 preceding siblings ...)
  2010-01-23 10:35 ` ktietz at gcc dot gnu dot org
@ 2010-06-11  8:38 ` iains at gcc dot gnu dot org
  5 siblings, 0 replies; 7+ messages in thread
From: iains at gcc dot gnu dot org @ 2010-06-11  8:38 UTC (permalink / raw)
  To: gcc-bugs



------- Comment #6 from iains at gcc dot gnu dot org  2010-06-11 08:37 -------
closing as fixed by r160457 (trunk) and r160526 (4.5 release branch)


-- 

iains at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42829


^ permalink raw reply	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2010-06-11  8:38 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2010-01-21 15:41 [Bug libgomp/42829] New: TLS detection in ./configure is wrong jos dot de_laender at telenet dot be
2010-01-21 15:42 ` [Bug libgomp/42829] " jos dot de_laender at telenet dot be
2010-01-21 19:49 ` pinskia at gcc dot gnu dot org
2010-01-21 20:37 ` jakub at gcc dot gnu dot org
2010-01-21 22:14 ` jos dot de_laender at telenet dot be
2010-01-23 10:35 ` ktietz at gcc dot gnu dot org
2010-06-11  8:38 ` iains at gcc dot gnu dot org

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).