[Bug 293382] Dead lock and kernel crash around closefp_impl

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=293382

--- Comment #55 from Konstantin Belousov <kib@FreeBSD.org> ---
(In reply to Kyle Evans from comment #54)
Note that rep;movs and like are very special kind of instructions in x86 ISA.
They are interruptible, and CPU must update %rsi%rdi/%rcx to reflect the
progress.
Of course for short moves interrupts are kept pending, but for large they are
not,
and I believe that the case of the reporter which copies hundreds of KBs, if
not some
MBs, should be.

That said, could you share some details about the case, and how it was solved?
Did the microcode update resolved it?  I lost access to AMD internal docs some
time ago.

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