Skip to contentSkip to conversation

Upcoming oops in lockdep

DISCUSSION Arjan van de Ven 2 participants 3 messages Aug 23, 2008 – Aug 25, 2008

lore
Share & tools
lore
#0 lore raw

in kerneloops.org a crash (slab poison) in mark_lock is coming up rapidly:

http://www.kerneloops.org/searchweek.php?search=mark_lock

and despite a few earlier hits on the same function it seems to be something introduced in 2.6.27-rc... (the older ones aren't slab poison)

Signature · 3 lines
If you want to reach me at my work email, use [email protected]
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org
#1 lore raw
Quoted 7 lines · Arjan van de Ven

in kerneloops.org a crash (slab poison) in mark_lock is coming up rapidly:

http://www.kerneloops.org/searchweek.php?search=mark_lock

and despite a few earlier hits on the same function it seems to be something introduced in 2.6.27-rc... (the older ones aren't slab poison)

infradeads.org hehe

can this crash be reproduced without lockdep? Normally when I see crashes like this its crashing because the lock itself is stomped on (use after free, for example), and lockdep happens to be the one to touch it first (the lockdep state is by far the largest bit of the spinlock data too).

#2 lore raw
Quoted 5 lines · Peter Zijlstra
Quoted 4 lines · Arjan van de Ven

in kerneloops.org a crash (slab poison) in mark_lock is coming up rapidly:

http://www.kerneloops.org/searchweek.php?search=mark_lock

Also, would be marvelous if you can at least put one of these bug reports in the email, saves me from having to grope for a web browser thingy.

Quoted 10 lines
Quoted 2 lines

and despite a few earlier hits on the same function it seems to be something introduced in 2.6.27-rc... (the older ones aren't slab poison)

infradeads.org hehe

can this crash be reproduced without lockdep? Normally when I see crashes like this its crashing because the lock itself is stomped on (use after free, for example), and lockdep happens to be the one to touch it first (the lockdep state is by far the largest bit of the spinlock data too).