Skip to contentSkip to conversation

more cfq spinlock badness

DISCUSSION Dave Jones 3 participants 6 messages Jan 31, 2006 – Feb 3, 2006 1 message not in the archive

lore
Share & tools
lore

Dave Jones

author
#0 lore raw

Not seen this break for a while, but I just hit it again in 2.6.16rc1-git4.

Dave

BUG: spinlock bad magic on CPU#0, pdflush/1128
 lock: ffff81003a219000, .magic: 00000000, .owner: <none>/-1, .owner_cpu: 0

Call Trace: <ffffffff80206edc>{spin_bug+177} <ffffffff80207045>{_raw_spin_lock+25}
       <ffffffff801fea4a>{cfq_exit_single_io_context+85} <ffffffff801ff9a6>{cfq_exit_io_context+24}
       <ffffffff801f79b0>{exit_io_context+137} <ffffffff80135fbc>{do_exit+182}
       <ffffffff8010ba49>{child_rip+15} <ffffffff80146087>{keventd_create_kthread+0}
       <ffffffff8014629c>{kthread+0} <ffffffff8010ba3a>{child_rip+0}
Kernel panic - not syncing: bad locking
#1 lore raw
Quoted 4 lines · Dave Jones

Not seen this break for a while, but I just hit it again in 2.6.16rc1-git4.

Dave

Quoted 9 lines
BUG: spinlock bad magic on CPU#0, pdflush/1128
 lock: ffff81003a219000, .magic: 00000000, .owner: <none>/-1, .owner_cpu: 0

Call Trace: <ffffffff80206edc>{spin_bug+177} <ffffffff80207045>{_raw_spin_lock+25}
       <ffffffff801fea4a>{cfq_exit_single_io_context+85} <ffffffff801ff9a6>{cfq_exit_io_context+24}
       <ffffffff801f79b0>{exit_io_context+137} <ffffffff80135fbc>{do_exit+182}
       <ffffffff8010ba49>{child_rip+15} <ffffffff80146087>{keventd_create_kthread+0}
       <ffffffff8014629c>{kthread+0} <ffffffff8010ba3a>{child_rip+0}
Kernel panic - not syncing: bad locking

Again, which devices have you used? Did it happen at shutdown, or? Did the ub bug get fixed, if you are using that? The bug above has in the past always been able to be explained by a driver destroying a structure embedding the queue lock before the queue is dead.

Signature · 1 line
Jens Axboe
author
#2 lore raw
Quoted 8 lines · Jens Axboe
Quoted 7 lines · Dave Jones

Not seen this break for a while, but I just hit it again in 2.6.16rc1-git4.

Dave

BUG: spinlock bad magic on CPU#0, pdflush/1128
 lock: ffff81003a219000, .magic: 00000000, .owner: <none>/-1, .owner_cpu: 0
Quoted 8 lines
Quoted 6 lines
Call Trace: <ffffffff80206edc>{spin_bug+177} <ffffffff80207045>{_raw_spin_lock+25}
       <ffffffff801fea4a>{cfq_exit_single_io_context+85} <ffffffff801ff9a6>{cfq_exit_io_context+24}
       <ffffffff801f79b0>{exit_io_context+137} <ffffffff80135fbc>{do_exit+182}
       <ffffffff8010ba49>{child_rip+15} <ffffffff80146087>{keventd_create_kthread+0}
       <ffffffff8014629c>{kthread+0} <ffffffff8010ba3a>{child_rip+0}
Kernel panic - not syncing: bad locking

Again, which devices have you used?

nothing special (Ie, no usb bits, just the onboard ata_piix SATA)

Quoted 1 line

Did it happen at shutdown, or?

whislt starting up a bunch of gnome panel applets.

Quoted 1 line

Did the ub bug get fixed

yes

Quoted 3 lines

if you are using that? The bug above has in the
past always been able to be explained by a driver destroying a structure embedding the queue lock before the queue is dead.

as there were no ub devices plugged in at the time, I think Pete is off the hook for this one.

Dave

#3 lore raw
Quoted 26 lines · Dave Jones
Quoted 8 lines · Jens Axboe
Quoted 7 lines · Dave Jones

Not seen this break for a while, but I just hit it again in 2.6.16rc1-git4.

Dave

BUG: spinlock bad magic on CPU#0, pdflush/1128
 lock: ffff81003a219000, .magic: 00000000, .owner: <none>/-1, .owner_cpu: 0
Quoted 8 lines
Quoted 6 lines
Call Trace: <ffffffff80206edc>{spin_bug+177} <ffffffff80207045>{_raw_spin_lock+25}
       <ffffffff801fea4a>{cfq_exit_single_io_context+85} <ffffffff801ff9a6>{cfq_exit_io_context+24}
       <ffffffff801f79b0>{exit_io_context+137} <ffffffff80135fbc>{do_exit+182}
       <ffffffff8010ba49>{child_rip+15} <ffffffff80146087>{keventd_create_kthread+0}
       <ffffffff8014629c>{kthread+0} <ffffffff8010ba3a>{child_rip+0}
Kernel panic - not syncing: bad locking

Again, which devices have you used?

nothing special (Ie, no usb bits, just the onboard ata_piix SATA)

Quoted 1 line

Did it happen at shutdown, or?

whislt starting up a bunch of gnome panel applets.

Quoted 1 line

Did the ub bug get fixed

Quoted 8 lines

yes

Quoted 3 lines

if you are using that? The bug above has in the
past always been able to be explained by a driver destroying a structure embedding the queue lock before the queue is dead.

as there were no ub devices plugged in at the time, I think Pete is off the hook for this one.

The ub fix hasn't been merged yet. Just trying to be absolutely certain that you didn't have that loaded?

Signature · 1 line
Jens Axboe
author
#4 lore raw
Quoted 8 lines · Jens Axboe
Quoted 6 lines
Quoted 3 lines

if you are using that? The bug above has in the
past always been able to be explained by a driver destroying a structure embedding the queue lock before the queue is dead.

as there were no ub devices plugged in at the time, I think Pete is off the hook for this one.

The ub fix hasn't been merged yet.

Really? I thought Pete synced up a whole bunch of ub bits including that already.

Quoted 2 lines

Just trying to be absolutely certain
that you didn't have that loaded?

99.9% sure. (It was a few days ago, and I'm already forgetting details).

Dave

A message that is not in the archive (sent off-list or not yet received): [email protected]

#6 lore raw
Quoted 5 lines · Dave Jones
Quoted 3 lines
Quoted 1 line
Quoted 1 line

Not seen this break for a while, but I just hit it again in 2.6.16rc1-git4.

The ub fix hasn't been merged yet.

Really? I thought Pete synced up a whole bunch of ub bits including that already.

I did but the patch sat in Greg's and AKPM's trees for a while, so the first release which contains the CFQ+ub fix is 2.6.16-rc2.

But in any case if no USB devices were connected, ub cannot be in the picture. It's true even if they were connected and disconnected. A queue has to be in use before this happens.

-- Pete