[Bug libgomp/65385] [libgomp] omp task untied test case fails
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65385 Jack Howarth changed: What|Removed |Added CC||howarth.at.gcc at gmail dot com --- Comment #5 from Jack Howarth --- FYI, this test is now supported in llvm's openmp support in clang trunk as of... r266853 | abataev | 2016-04-20 00:01:36 -0400 (Wed, 20 Apr 2016) | 5 lines [OPENMP] Codegen for untied tasks. If the untied clause is present on a task construct, any thread in the team can resume the task region after a suspension. Patch adds proper codegen for untied tasks. It now produces the expected results of... #Tested Directive t ct ot oct omp_task_untied 100 100 100 100 on x86_64-apple-darwin15 for -m32 and -m64.
[Bug libgomp/65385] [libgomp] omp task untied test case fails
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65385 --- Comment #4 from Jakub Jelinek --- IMHO not, I view the untied clause purely as an optimization hint (and libgomp parses it, but ignores it).
[Bug libgomp/65385] [libgomp] omp task untied test case fails
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65385 Sebastian Huber changed: What|Removed |Added Status|UNCONFIRMED |RESOLVED Resolution|--- |WONTFIX --- Comment #3 from Sebastian Huber --- I will forward this to the test suite authors. Is it then possible to test the task untied at all?
[Bug libgomp/65385] [libgomp] omp task untied test case fails
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65385 --- Comment #2 from Jakub Jelinek --- That test is completely bogus. The spec doesn't require untied tasks to change threads at any point, it is strictly a "may" case. So the test is testing something not required by the standard.
[Bug libgomp/65385] [libgomp] omp task untied test case fails
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65385 --- Comment #1 from Sebastian Huber --- Created attachment 35008 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35008&action=edit Test program.
