| 1 |
pageexec@××××××××.hu wrote: |
| 2 |
> On 5 Nov 2007 at 11:59, Brian Kroth wrote: |
| 3 |
> |
| 4 |
>> Actually, it looks like ipmi watchdog caused it. |
| 5 |
> |
| 6 |
> that would probably be a hard-stuck kernel that was detected |
| 7 |
> by the watchdog. |
| 8 |
|
| 9 |
Yeah, I was just confused since nothing showed up in the logs, but I |
| 10 |
guess hard panics like that wouldn't. |
| 11 |
|
| 12 |
> |
| 13 |
>> pagexec: were those pax messages any help to you? |
| 14 |
> |
| 15 |
> well, sort of if 'making me ever more confused' counts as help ;-). |
| 16 |
> the thing is, the bad pages all seem to be part of an otherwise |
| 17 |
> non-executable mapping, meaning they can't possibly be part of |
| 18 |
> vma mirroring (since only executable regions are mirrored), and |
| 19 |
> my code doesn't (well, shouldn't) change any behaviour there. so |
| 20 |
> i'm still thinking... |
| 21 |
> |
| 22 |
|
| 23 |
Well, I could always try a vanilla kernel to see if I can reproduce |
| 24 |
these problems there. Still working on getting my test vm setup. BTW, |
| 25 |
the other machine that I saw the message on was also a vm, but the |
| 26 |
original was a physical machine. |
Archives