(In reply to Vladyslav Yamkovyi from comment #441)
> (In reply to Owen Swerkstrom from comment #439)
> > This comes up as a Linux problem, but it sure smells like a hardware defect
> > to me.
> Any system that does not provides a software workaround must be affected
> according to published errata, not even a question. I've published one of
> their revision guides in my previous comments. They have no plans on fixing
> this and suggest using software workarounds. We're left on our own.
What is the "this" that AMD have no plans to fix ?
1057 MWAIT or MWAITX Instructions May Fail to Correctly Exit From
the Monitor Event Pending State
1059 In Real Mode or Virtual-8086 Mode MWAIT or MWAITX Instructions May
Fail to Correctly Exit From the Monitor Event Pending State
1109 MWAIT Instruction May Hang a Thread
but I could not find anything else that might be related to the "freeze when idle" problem.
FWIW, here's the full text for the Erratum 1109:
1109 MWAIT Instruction May Hang a Thread
Description: Under a highly specific and detailed set of internal timing conditions, the MWAIT instruction may cause a thread to hang in SMT (Simultaneous Multithreading) Mode.
Potential Effect on System: The system may hang or reset.
Suggested Workaround: System software may contain the workaround for this erratum.
Fix Planned: No fix planned
so there ! I guess "idle=nomwait" is "the workaround" ?
"Typical Current Idle" appears to work for some (including me) but not for everyone. If one or more of these MWAIT errata is the root cause of the "freeze when idle" problem, I wonder why AMD introduced "Typical Current Idle" and how that relates to these MWAIT issues ??
(In reply to Vladyslav Yamkovyi from comment #441)
> (In reply to Owen Swerkstrom from comment #439)
> > This comes up as a Linux problem, but it sure smells like a hardware defect
> > to me.
> Any system that does not provides a software workaround must be affected
> according to published errata, not even a question. I've published one of
> their revision guides in my previous comments. They have no plans on fixing
> this and suggest using software workarounds. We're left on our own.
What is the "this" that AMD have no plans to fix ?
I had a look at <https:/ /www.amd. com/system/ files/TechDocs/ 55449_Fam_ 17h_M_00h- 0Fh_Rev_ Guide.pdf> which you referenced earlier. I found 3 MWAIT issues:
1057 MWAIT or MWAITX Instructions May Fail to Correctly Exit From
the Monitor Event Pending State
1059 In Real Mode or Virtual-8086 Mode MWAIT or MWAITX Instructions May
Fail to Correctly Exit From the Monitor Event Pending State
1109 MWAIT Instruction May Hang a Thread
but I could not find anything else that might be related to the "freeze when idle" problem.
FWIW, here's the full text for the Erratum 1109:
1109 MWAIT Instruction May Hang a Thread
Description: Under a highly specific and detailed set of internal timing
condition s, the MWAIT instruction may cause a thread to
hang in SMT (Simultaneous Multithreading) Mode.
Potential Effect on System: The system may hang or reset.
Suggested Workaround: System software may contain the workaround for
this erratum.
Fix Planned: No fix planned
so there ! I guess "idle=nomwait" is "the workaround" ?
"Typical Current Idle" appears to work for some (including me) but not for everyone. If one or more of these MWAIT errata is the root cause of the "freeze when idle" problem, I wonder why AMD introduced "Typical Current Idle" and how that relates to these MWAIT issues ??