>
> (note btw the drivers fix the effects of the problem not the bug).
> Loic Prylli posted a possible real bug fix (I agree with his diagnosis
> Im not yet totally happy his fix is the right one in every case)
>
Hello,
Indeed, the fix I have proposed was not meant to be the definitive
solution: actually I first tried to remove even the acknowledgement to
the 8259, because I thought the 8259 was not involved for IPI, but in
this case kernel was not able to boot. So I tried leaving the acknowledgement,
only removing the cache_ax and irq mask manipulations. And that
worked. But overall, the fix itself was more a random test and try,
than a real fix (I have yet to read one day the datasheet that
describe the IO-APIC features).
Regards,
Loic
