Re: [PATCH]: update tls and lock tests, gl_cond_t WIN32 implementation

Yoann Vandoorselaere wrote: > I just run the test suite under wine and it run correctly. Please apply. Applied. > > Anyway, for reference, here's the same thing again as an attachment. > > Thanks, this solved the problem! I guess some line wrapped around in the > original patch. But as I said,

Re: [PATCH]: update tls and lock tests, gl_cond_t WIN32 implementation

Hi Bruno, Le jeudi 16 octobre 2008 à 12:27 +0200, Bruno Haible a écrit : > > The patch won't apply to the current GIT repository: > > > > patching file lib/glthread/cond.h > > patching file lib/glthread/cond.c > > Hunk #2 FAILED at 235. > > Hunk #3 FAILED at 306. > > Hunk #4 FAILED at 422. > > Hu

Re: [PATCH]: update tls and lock tests, gl_cond_t WIN32 implementation

Hi Yoann, > The patch won't apply to the current GIT repository: > > patching file lib/glthread/cond.h > patching file lib/glthread/cond.c > Hunk #2 FAILED at 235. > Hunk #3 FAILED at 306. > Hunk #4 FAILED at 422. > Hunk #5 FAILED at 439. > 4 out of 6 hunks FAILED -- saving rejects to file > lib/

Re: [PATCH]: update tls and lock tests, gl_cond_t WIN32 implementation

Hi Bruno, Le jeudi 16 octobre 2008 à 02:43 +0200, Bruno Haible a écrit : [...] > Find here an adaption of the "one event per thread" idea, and the use of > linked lists rather than array to keep the removal an O(1) operation. > It passes the tests, and should be scalable. OK to apply? > > I thin

Re: [PATCH]: update tls and lock tests, gl_cond_t WIN32 implementation

Hi Yoann, When looking in detail at your implementation of condition variables for Win32, I find it has two major flaws: - Already mentioned in [1] is the fact that it can busy-loop. Namely, when a broadcast is done, all threads are woken up and loop until the event is reset, which happ

Re: [PATCH]: update tls and lock tests, gl_cond_t WIN32 implementation

Hi Yoann, Thanks for the rapid testing. > Attached is a patch that fixes some compilation problem due to the > latest cond and select changes. > > Additionally, it seems the cond timedwait test are now failing under > WIN32, although I did not attempt to trace the problem yet. Your fix for the

Re: [PATCH]: update tls and lock tests, gl_cond_t WIN32 implementation

Hi Bruno, Le dimanche 12 octobre 2008 à 00:16 +0200, Bruno Haible a écrit : > You wrote on 2008-09-18 > > and 2008-10-02 > : > > > provide a working implement

Re: [PATCH]: update tls and lock tests, gl_cond_t WIN32 implementation

Yoann Vandoorselaere wrote: > > during this conversion an integer overflow could occur in at least 2 > > places > > (1. result->tv_sec = x->tv_sec - y->tv_sec > > 2. (res.tv_sec * 1000) + (res.tv_usec / 1000)). > > You might want to push your #1 changes up to the GLIBC team: the > ti

Re: [PATCH]: update tls and lock tests, gl_cond_t WIN32 implementation

Hi Bruno, Le dimanche 12 octobre 2008 à 00:16 +0200, Bruno Haible a écrit : > You wrote on 2008-09-18 > > and 2008-10-02 > : > > > provide a working implement

Re: [PATCH]: update tls and lock tests, gl_cond_t WIN32 implementation

Hi Yoann, You wrote on 2008-09-18 and 2008-10-02 : > provide a working implementation of gl_cond_t for WIN32 platform > (based on http://www.cs.wustl.edu/~sch

Re: [PATCH]: update tls and lock tests, gl_cond_t WIN32 implementation

Le mercredi 01 octobre 2008 à 03:40 +0200, Bruno Haible a écrit : > Yoann Vandoorselaere wrote on 2008-09-18: > > > > - Add the missing cond-tests module. > > Applied as follows. I > - removed ENABLE_LOCKING, > - made use of gl_cond_define_initialized and gl_lock_define_initialized > (beca

Re: [PATCH]: update tls and lock tests, gl_cond_t WIN32 implementation

Yoann Vandoorselaere wrote on 2008-09-18: > > > - Add the missing cond-tests module. Applied as follows. I - removed ENABLE_LOCKING, - made use of gl_cond_define_initialized and gl_lock_define_initialized (because the macros gl_cond_initializer and gl_lock_initializer are not public),

Re: [PATCH]: update tls and lock tests, gl_cond_t WIN32 implementation

Yoann Vandoorselaere wrote on 2008-09-18: > Here is a new version of the patches, contain more dependencies fixes, Yup, I noticed these as well: - 'errno' is needed as a dependency because of ETIMEDOUT, - 'time' is needed as a dependency for 'struct timespec'. Also, let me make the definition

Re: [PATCH]: update tls and lock tests, gl_cond_t WIN32 implementation

Le lundi 15 septembre 2008 à 14:12 +0200, Yoann Vandoorselaere a écrit : > Le dimanche 14 septembre 2008 à 20:46 +0200, Yoann Vandoorselaere a > écrit : > > Hi, > > > > Attached is a patch implementing the following glthread fixes: > > > > - Correct pthread_atfork() detection. > > > > - Add the