[Zephyr] Reduce CI disk usage and quarantine rv32 FPU test#1578
[Zephyr] Reduce CI disk usage and quarantine rv32 FPU test#1578quic-areg wants to merge 1 commit into
Conversation
We can sometimes hit "No space left on device" mid-setup since the runs sit right at the runner disk limit. Trim usage in zephyr-run.yml (shared by PR and nightly): - west update: drop 30 vendor HALs no QEMU board we build uses (~4-5 GB saved). - twister: pass -M pass to delete passing tests' build dirs as they run. Also quarantine kernel.fpu_sharing.generic.riscv32, which forces hard-float rv32 that cpullvm picolibc has no multilib for. The PR workflow should pass after this change. Signed-off-by: quic-areg <[email protected]>
|
Jonathon Penix (@jonathonpenix) can you comment on the rv32 FPU / CPULLVM toolchain ? |
Jonathon Penix (jonathonpenix)
left a comment
Jonathon Penix (jonathonpenix)
left a comment
There was a problem hiding this comment.
Not that familiar with these options, but seems reasonable to me.
Out of curiosity, how often are you hitting these issues?
We only run 1x per week in cpullvm but I don't think we've hit space issues recently
Oops, left a comment but just to respond this directly: there's already one in cpullvm's mainline, should be in 23.x when that is out |
Almost never. I think I've only seen it once. |
3 participants
We can sometimes hit "No space left on device" mid-setup since the runs sit right at the runner disk limit. Trim usage in zephyr-run.yml (shared by PR and nightly):
Also quarantine kernel.fpu_sharing.generic.riscv32, which forces hard-float rv32 that cpullvm picolibc has no multilib for.
The PR workflow should pass for all targets after this change.