-
Notifications
You must be signed in to change notification settings - Fork 1.5k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[macOS] gperftools
2.15: some tests fail
#1477
Comments
cmake failing tests is an existing ticket. #1422 dupe ? |
@alk I can re-run from makefiles, but at least on Intel with v. 2.10 I was getting nearly identical failures. So unless makefiles tests were broken in 2.10 but fixed since then, while CMake tests we broken but not fixed, results should be similar. I will try from makefiles and update on results. UPD. By the way, with v. 2.10 I was getting a freeze with makefiles and CMake alike, while with v. 2.15 there is no freeze with CMake. |
This is on 14.2.1/arm64 with makefiles build of 2.15 release:
I will try on PowerPC, and also re-run with CMake on Sonoma, since turned out I used a slightly earlier commit here, not the actual release. |
14.2.1/arm64, CMake build of 2.15 release:
So yes, while results for 2.15 release are marginally better over the original results from commit 8987d08 (I did not notice it was set to use a commit), CMake build has substantially more broken tests. |
10.6 powerpc (32-bit), makefiles build of 2.15 release (CMake results above are also for 2.15 release):
Worse than on Sonoma, much better than with CMake. @alk If I could provide any additional info which will help, please tell me. |
Yes, we have pprof issue(s) on osex. #1292 (comment) |
@alk So we got the following failures on
vs
Is it possible to say based on logs what may cause a few extra tests fail on |
So looked at failures log (from autotools build) comment #1477 (comment) above. First is stacktrace issue. I am not sure we actually want to do ppc frame-pointer based stacktrace capturing methods on ppc/osx. But it is up to you what you want. (OSX has it's own backtrace implementation which seems to work) Second is lack of MallocHook_GetCallerStackTrace support on this platform. This has become even more broken on OSX recently. And I am actively attempting to replace it (but not 100% sure yet if this is direction we want). It is a bigger change too. And last is pprof issue. You're welcome to help with first and third sub-issues. |
I have ran tests now on 14.2.1 aarch64 with clang and on 10.6 ppc with gcc. There are a number of failures:
14.2.1/aarch64
10.6/ppc
I have also tested v. 2.10 today on 10.6.8 Rosetta and 10.6.8 x86_64, and got similar failures, though there also one test was freezing, so I could not reach to the end. So at least for the most part these failure are not new and not specific to CMake build, though it did seem to me that at least one extra test failed with CMake which passed with makefiles.
@alk Could you please take a look at these?
Logs:
gperftools_2.15_powerpc_tests.log
gperftools_2.15_aarch64_tests.log
The text was updated successfully, but these errors were encountered: