------- Comment #5 from belyshev at depni dot sinp dot msu dot ru 2007-09-21
06:56 -------
Problem still persists, see eg.
http://gcc.gnu.org/ml/gcc-testresults/2007-09/msg00911.html
pass40-frag fails due to small timeout (not every box can create 20K threads in
20 seconds especially under high load of make -j4).
pass37-frag and pass39-frag also fail due to improper use of sched_yield(),
this patch cures problem for me most of the time:
Index: gcc43/libmudflap/testsuite/libmudflap.cth/pass37-frag.c
===================================================================
--- gcc43.orig/libmudflap/testsuite/libmudflap.cth/pass37-frag.c
+++ gcc43/libmudflap/testsuite/libmudflap.cth/pass37-frag.c
@@ -20,7 +20,6 @@ func (void *p)
libmudflap. */
array[x] = i;
}
- sched_yield (); /* sleep (1); */
}
return (NULL);
Index: gcc43/libmudflap/testsuite/libmudflap.cth/pass39-frag.c
===================================================================
--- gcc43.orig/libmudflap/testsuite/libmudflap.cth/pass39-frag.c
+++ gcc43/libmudflap/testsuite/libmudflap.cth/pass39-frag.c
@@ -20,7 +20,6 @@ func (void *p)
assert (arrays[i] != NULL);
arrays[i][x] = i;
free (arrays[i]);
- sched_yield (); /* sleep (1); */
}
return (NULL);
--
belyshev at depni dot sinp dot msu dot ru changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |fche at redhat dot com,
| |belyshev at depni dot sinp
| |dot msu dot ru
Last reconfirmed|2005-10-11 07:32:46 |2007-09-21 06:56:31
date| |
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20003
