[Bug 219399] System panics after several hours of 14-threads-compilation orgies using poudriere on AMD Ryzen...

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=219399

--- Comment #214 from Don Lewis <truckman at FreeBSD.org> ---
I'm not familiar enough with the Linux memory map to comment, but I haven't
seen any reports of hang/reboot problems that sound like what what we ran into.
 There have been reports of idle machines having problems that seem to be
resolved by avoiding the deep Cx states.  I don't know if that is a potential
problem for us since our default seems to be:
  hw.acpi.cpu.cx_lowest: C1

It would be nice if the "safe area" was documented.  This doesn't appear to be
an issue with Intel CPUs.  My AMD FX-8320E is happy with the pre-Ryzen location
of our signal trampoline code.  I haven't run my test program on it because I'm
keeping it busy doing the port builds that I'd hoped to be doing on my new
Ryzen box.

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