Open theotheroracle opened 3 months ago
I assume you are using qemu-user binfmt on x86_64 system to run riscv64 binaries?
Tests udp_multicast_interface
, udp_multicast_interface6
, and udp_no_autobind
are known to fail under QEMU.
To disable them, a switch -DQEMU=ON
should be used (ref).
This switch affects only tests, not the library itself.
Perhaps we need to add this switch to the nix expression.
The test fork_threadpool_queue_work_simple
is not among tests being disabled under QEMU
flag.
Perhaps, it is failing only under qemu-user, but not qemu-system?
Needs to be investigated/reported to the upstream.
I've tried to build this on a native RISC-V machine, not QEMU. None of these tests failed.
However, I've got a new failure on a native machine:
not ok 242 - spawn_exercise_sigchld_issue
# timeout
# Output from process `spawn_exercise_sigchld_issue`: (no output)
This could be caused by slow hardware, or my system being overloaded by other processes, or different nixpkgs revision.
Also needs to be investigated.
Maybe we should add it to toDisable
, like other tests that are timing out on slow machines.
I assume you are using qemu-user binfmt on x86_64 system to run riscv64 binaries?
correct
Not sure if this is related, but I'm getting the same failure when compiling on a native x86-64 7950X3D with nixpkgs.localSystem.gcc.arch = "znver4"
. Build log here, latest 24.05.
Steps To Reproduce
Steps to reproduce the behavior:
Build log
Additional context
Add any other context about the problem here.
Notify maintainers
@marsam @vcunat
Metadata
Please run
nix-shell -p nix-info --run "nix-info -m"
and paste the result.Add a :+1: reaction to issues you find important.