Ah, great - thanks for the details & for looking into the flakiness to help improve buildbot quality and confidence!
On Tue, Jan 25, 2022, 6:40 AM David Spickett david.spickett@linaro.org wrote:
That libl deadlock test has been flaky for us across many bots recently, you just had bad timing there and there is a green build later. (scudo resident memory size is another test like this). I'll make sure these are investigated properly.
What I did see was a debug related profiling test failing "Linux/instrprof-debug-info-correlate.c" https://lab.llvm.org/buildbot/#/builders/176/builds/1325. This was fixed by https://github.com/llvm/llvm-project/commit/38e16e1cebb891ad47b85727bf46f7da... and our 2 stage SVE bots are just lagging behind.
Otherwise we're all green on our bots. Thanks for following up and we'll root out what's causing the flaky tests.
Thanks, David Spickett.
On Tue, 25 Jan 2022 at 00:39, David Blaikie dblaikie@gmail.com wrote:
Seems like my change to make Clang default to DWARFv5 might've caused a buildbot failure on your build worker here: https://lab.llvm.org/buildbot/#/builders/185/builds/1295
But I seem to be able to run this test successfully locally on my Linux machine - so I'm wondering if you can offer any help diagnosing the issue showing up on your builder/worker? _______________________________________________ linaro-toolchain mailing list -- linaro-toolchain@lists.linaro.org To unsubscribe send an email to linaro-toolchain-leave@lists.linaro.org