test_exec failed child failed 1 test_exec failed Test defer-taskrun.t failed with ret 1 Running test double-poll-crash.t 0 sec Running test drop-submit.t 0 sec Running test eeed8b54e0df.t cqe error: -95 Test eeed8b54e0df.t failed with ret 1 Running test empty-eownerdead.t 0 sec ------------------------------------------------------------------- This is an unstable amd64 chroot image at a tinderbox (==build bot) name: 17.1_no_multilib-j4_test-20230218-170028 ------------------------------------------------------------------- gcc-config -l: [1] x86_64-pc-linux-gnu-12 * clang/llvm (if any): /usr/lib/llvm/15 15.0.7 Python 3.10.10 Available Ruby profiles: [1] ruby27 (with Rubygems) [2] ruby30 (with Rubygems) * php cli (if any): HEAD of ::gentoo commit 866111267e8579bb016ab2ca1f0a910f9c8b5130 Author: Repository mirror & CI <repomirrorci@gentoo.org> Date: Sun Feb 19 01:02:00 2023 +0000 2023-02-19 01:02:00 UTC emerge -qpvO sys-libs/liburing [ebuild N ] sys-libs/liburing-2.3-r4 USE="test -examples -static-libs"
Created attachment 852856 [details] emerge-info.txt
Created attachment 852858 [details] emerge-history.txt
Created attachment 852860 [details] environment
Created attachment 852862 [details] etc.portage.tar.bz2
Created attachment 852864 [details] logs.tar.bz2
Created attachment 852866 [details] sys-libs:liburing-2.3-r4:20230219-015442.log
Created attachment 852868 [details] temp.tar.bz2
In Toralf's CI fail the same tests as in ago's CI, but a few extra: ago: <500f9fbadef8.t> <cq-overflow.t> <eeed8b54e0df.t> <file-verify.t> <io-cancel.t> <iopoll.t> <read-write.t> <ringbuf-read.t> <sq-poll-dup.t> <sq-poll-share.t> tor: <500f9fbadef8.t> <cq-overflow.t> <defer-taskrun.t> <eeed8b54e0df.t> <file-verify.t> <io-cancel.t> <iopoll.t> <read-write.t> <ringbuf-read.t> <sq-poll-dup.t> <sq-poll-share.t> <xattr.t> Again, for me, all tests pass. I first assumed that it may be due the kernel version, but Toralf is using 6.1 which is pretty recent. The locked-in-memory size ulimit was in older io_uring versions a limiting factory, but IIRC io_uring's accounting changed, so this should also not be the issue here.
