| Summary: | media-libs/mesa-23.0.3-r1:0: 4 test failures on hppa | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Rolf Eike Beer <eike> |
| Component: | Current packages | Assignee: | Gentoo X packagers <x11> |
| Status: | CONFIRMED --- | ||
| Severity: | normal | Keywords: | TESTFAILURE |
| Priority: | Normal | ||
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| URL: | https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26978 | ||
| See Also: |
https://bugs.gentoo.org/show_bug.cgi?id=915660 https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26978 https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26991 https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26964 |
||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: |
build.log
test logs |
||
|
Description
Rolf Eike Beer
2023-06-08 20:53:43 UTC
Created attachment 863589 [details]
test logs
There is also some weird binary data in the build log, I don't know if that is an unrelated problem or related. > src/gbm/libgbm.so.1.0.0: unknown symbol exported: _GLOBAL_OFFSET_TABLE_
Looks like I need to add this to the Mesa symbol checking scripts.
This failure: > [----------] 3 tests from half_to_float_test > [ RUN ] half_to_float_test.nan_test > ../mesa-23.0.3/src/util/tests/half_float_test.cpp:61: Failure > Value of: issignaling(TEST_NAN) > Actual: true > Expected: false > [ FAILED ] half_to_float_test.nan_test (0 ms) is apparently because hppa uses a different signalling NaN? At least the test author appears to be aware of it: > /* Sanity test our test values */ > TEST(half_to_float_test, nan_test) > { > EXPECT_TRUE(isinf(TEST_POS_INF)); > EXPECT_TRUE(isinf(TEST_NEG_INF)); > > EXPECT_TRUE(isnan(TEST_NAN)); > /* Make sure that our 32-bit float nan test value we're using is a > * non-signaling NaN. The sign of the bit for signaling was apparently > * different on some old processors (PA-RISC, MIPS?). This test value should > * cover Intel, ARM, and PPC, for sure. > */ > EXPECT_FALSE(issignaling(TEST_NAN)); > } I'll have to do some more research. https://en.wikipedia.org/wiki/NaN#:~:text=Two%20separate%20kinds%20of%20NaNs,to%20basic%20floating%2Dpoint%20arithmetic. says > in NaNs generated by the PA-RISC and old MIPS processors, the signaling/quiet > bit is zero if the NaN is quiet, and non-zero if the NaN is signaling. Thus, > on these processors, the bit represents an is_signaling flag. Right, so on MIPS it's even more complex because there are multiple possibilities. Thanks MIPS. I honestly don't know what the "correct" thing here even is. The hppa CPU represents sNaN/qNaN in one way, but this code is almost always used for converting values to/from half-float for use in shader code... which is going to run on a GPU that almost certainly uses the IEEE-754-2008 convention for NaNs (i.e., not what hppa does). I think we should disable this test on hppa/old-mips, but I don't know how to check or whether there is a macro that tells us which convention is in use... Please give https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26978 https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26991 a try. The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=93b9d9dc97403326a984a056615295a163e259fa commit 93b9d9dc97403326a984a056615295a163e259fa Author: Matt Turner <mattst88@gentoo.org> AuthorDate: 2024-01-10 21:34:47 +0000 Commit: Matt Turner <mattst88@gentoo.org> CommitDate: 2024-01-11 04:41:51 +0000 media-libs/mesa: Version bump to 23.3.3 Bug: https://bugs.gentoo.org/908079 Closes: https://bugs.gentoo.org/921297 Signed-off-by: Matt Turner <mattst88@gentoo.org> media-libs/mesa/Manifest | 1 + media-libs/mesa/files/23.3.3-big-endian.patch | 84 ++++ ...3-symbols-check-Add-_GLOBAL_OFFSET_TABLE_.patch | 31 ++ media-libs/mesa/mesa-23.3.3.ebuild | 450 +++++++++++++++++++++ 4 files changed, 566 insertions(+) |
