This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.
| Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
|---|---|---|
| Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
| Other format: | [Raw text] | |
I was looking into why parallel tests like 25_algorithms/headers/algorithm/algorithm_parallel_mode.ccOk. I think your issue may be a special case of libstdc++/34106.
were failing on IA64 HP-UX and I found lines like:
#pragma message: slow fetch_and_add_64 #pragma message: slow compare_and_swap_64
OK, I looked at compatibility.h and basically understood why I get these messages
Yes. I think it's simply because the relevant search path is missing when the dg-require run. We should fix this indeed, affects a few basic headers/ tests.but then I went to IA64 Linux where I didn't see these messages in order to see how the test passes there and I see:
UNSUPPORTED: 25_algorithms/headers/algorithm/algorithm_parallel_mode.cc
along with lines like:
parallel_mode31631.cc:1:17: error: omp.h: No such file or directory
This seems odd ot me since I am building libgomp with my compiler.
... however, to be sure, let's not mistake the handful tests at issue, with proper testing of the parallel algorithms, which is done via make check-parallel, see:And when I saw the same messages in my x86_64 build, then I got really suspicious.
Is there a known problem on Linux where we aren't testing the parallel
algorithms that we should be testing because check_v3_target_parallel_mode
is saying that we don't support a parallel mode?
| Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
|---|---|---|
| Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |