Re: 2.4.18-rcx: Dual P3 + VIA + APIC

[Martin J. Bligh]

>> > <4>CPU1<T0:1339376,T1:446448,D:8,S:446460,C:1339380>
>> > <4>checking TSC synchronization across CPUs: passed.
>> > <4>Waiting on wait_init_idle (map = 0x2)
>> > <4>All processors have done init_idle
>> >
>> > I would say this means the TSC skew fix is broken and shooting down
>> > your box. What do you think, Alan?
>>
>> Seems a reasonable guess. However that TSC skew itself may point to other
>> problems. It means one processor started running successfully a little
>> after the other. That might be normal behaviour for that board or might
>> point to  something else
>
> It seems no normal behaviour, I checked several other boards of this type
> and none had a TSC skew (and all work). Purely guessing I would suggest
> two try some other 2 processors to verify the behaviour is really
> processor-independent. Another guess would of course be the MB itself
> being broken to some extent.
>
> Has anybody ever seen a _working_ skew correction? Is this known-to-work
> code?

Yes. Works every time for me (on NUMA-Q), with huge corrections:

checking TSC synchronization across CPUs:
BIOS BUG: CPU#0 improperly initialized, has 6571 usecs TSC skew! FIXED.
BIOS BUG: CPU#1 improperly initialized, has 6571 usecs TSC skew! FIXED.
BIOS BUG: CPU#2 improperly initialized, has 6571 usecs TSC skew! FIXED.
BIOS BUG: CPU#3 improperly initialized, has 6571 usecs TSC skew! FIXED.
BIOS BUG: CPU#4 improperly initialized, has 20664 usecs TSC skew! FIXED.
BIOS BUG: CPU#5 improperly initialized, has 20664 usecs TSC skew! FIXED.
BIOS BUG: CPU#6 improperly initialized, has 20665 usecs TSC skew! FIXED.
BIOS BUG: CPU#7 improperly initialized, has 20664 usecs TSC skew! FIXED.
BIOS BUG: CPU#8 improperly initialized, has -4424 usecs TSC skew! FIXED.
BIOS BUG: CPU#9 improperly initialized, has -4424 usecs TSC skew! FIXED.
BIOS BUG: CPU#10 improperly initialized, has -4424 usecs TSC skew! FIXED.
BIOS BUG: CPU#11 improperly initialized, has -4424 usecs TSC skew! FIXED.
BIOS BUG: CPU#12 improperly initialized, has -22812 usecs TSC skew! FIXED.
BIOS BUG: CPU#13 improperly initialized, has -22812 usecs TSC skew! FIXED.
BIOS BUG: CPU#14 improperly initialized, has -22812 usecs TSC skew! FIXED.
BIOS BUG: CPU#15 improperly initialized, has -22811 usecs TSC skew! FIXED.

I did try disabling it once, which stopped the system booting.
I never looked at it any further.

If you are crashing near the wait_init_idle fix, you might
try Ingo's scheduler patch - it has a different way of
fixing this race condition.

M.