https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=243225

--- Comment #15 from Terry Kennedy <[email protected]> ---
Back when I reported this, it was extremely timing-dependent. It first showed
up on a system that moved to clang 9 (with no other related changes - r356239
to r356557), presumably due to different optimizations in the newer clang.

Moving to a different slot, disabling hyperthreading, going into BIOS setup,
etc. would all make the bug disappear or reappear.

I'm not surprised that various debug prints give enough extra time to allow
whatever timing bug is in there to settle down.

We're now on clang 19 and many FreeBSD major versions later and people are
still running into this.

I think the best way to deal with this is to get a developer who is familiar
with mrsas and give them remote console access with reset / power control (like
a Dell iDRAC) and let them look at it on a system where it is happening.

I may or may not be able to put such a system together myself - my original
report was from nearly 6 years ago and I'd have to see if I still have the
hardware. But feel free to let me know if I should try.

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to