I know that sometimes testing a build is not necessarily a precise
indicator of whether a package will work. I do think that it is a
general indicator of success. I also have this view, not supported by
any technical information, that graphics applications are not as
"forgiving" as others. What got my attention in this situation was the
phrase "Segmentation fault." Five MesaLib tests, all in the same
directory, failed with this. Here is the applicable portion of my test log:
> make check-TESTS
> make[4]: Entering directory
> `/usr/src/MesaLib-9.2.3/Mesa-9.2.3/src/gallium/state_trackers/xvmc'
> make[5]: Entering directory
> `/usr/src/MesaLib-9.2.3/Mesa-9.2.3/src/gallium/state_trackers/xvmc'
> ../../../../bin/test-driver: line 95: 24915 Segmentation fault
> "$@" > $log_file 2>&1
> FAIL: tests/test_subpicture
> ../../../../bin/test-driver: line 95: 24914 Segmentation fault
> "$@" > $log_file 2>&1
> FAIL: tests/test_blocks
> ../../../../bin/test-driver: line 95: 24912 Segmentation fault
> "$@" > $log_file 2>&1
> FAIL: tests/test_context
> ../../../../bin/test-driver: line 95: 24916 Segmentation fault
> "$@" > $log_file 2>&1
> FAIL: tests/test_rendering
> ../../../../bin/test-driver: line 95: 24913 Segmentation fault
> "$@" > $log_file 2>&1
> FAIL: tests/test_surface
> make[6]: Entering directory
> `/usr/src/MesaLib-9.2.3/Mesa-9.2.3/src/gallium/state_trackers/xvmc'
> make[6]: Nothing to be done for `all'.
This was followed by the summary:
> Testsuite summary for Mesa 9.2.3
>
> # TOTAL: 5
> # PASS: 0
> # SKIP: 0
> # XFAIL: 0
> # FAIL: 5
> # XPASS: 0
> # ERROR: 0
>
> See src/gallium/state_trackers/xvmc/test-suite.log
> Please report to https://bugs.freedesktop.org/enter_bug.cgi?product=Mesa
The test-suite.log reiterated this excerpt.
After this the test suite continued on and conducted more tests that passed.
I didn't add them, but I think that 50-60 tests got run and all passed
except these 5. It was, as I said, "Segmentation fault" along with all
tests in src/gallium/state_trackers/xvmc/ failing that got my
attention. I couldn't find anything about this test error in my search
for info, but I did find a cvs commit three years ago that said simply,
"These tests are failing." I honestly don't think there's a problem,
but I wanted to present this so that someone who has more experience and
knowledge than I might tell me if this is really OK.
When I tested LLVM just before MesaLib, I got 3 unexpected failures out
of 8619 tests. That borders on a 0.03% failure rate in LLVM as opposed
to about 10% in MesaLib, and I'm not concerned about what happened in LLVM.
If anyone has anything they'd like to offer about this, I would be grateful.
Thanks,
Dan
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
