diff options
| author | Stefan Haberland <stefan.haberland@de.ibm.com> | 2009-10-06 10:34:15 +0200 |
|---|---|---|
| committer | Martin Schwidefsky <sky@mschwide.boeblingen.de.ibm.com> | 2009-10-06 10:35:11 +0200 |
| commit | 6fca97a958bc3c67566aa91eafc6a5be2e66d6b3 (patch) | |
| tree | 9e5ca7c9c3d14c45f469feb58cab15212cee9d6b /drivers/s390/block/dasd.c | |
| parent | af9d2ff9afaae8040dbf09238b2579f92c93579e (diff) | |
[S390] dasd: fix race condition in resume code
There is a race while re-reading the device characteristics. After
cleaning the memory area a cqr is build which reads the device
characteristics. This may take a rather long time and the device
characteristics structure is zero during this. Now it could be
possible that the block tasklet starts working and a new cqr will be
build. The build_cp command refers to the device characteristics
structure and this may lead into a divide by zero exception.
Fix this by re-reading the device characteristics into a temporary
structur and copy the data to the original structure. Also take the
ccwdev_lock.
Signed-off-by: Stefan Haberland <stefan.haberland@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Diffstat (limited to 'drivers/s390/block/dasd.c')
| -rw-r--r-- | drivers/s390/block/dasd.c | 5 |
1 files changed, 3 insertions, 2 deletions
diff --git a/drivers/s390/block/dasd.c b/drivers/s390/block/dasd.c index dad0449475b..53b8c255360 100644 --- a/drivers/s390/block/dasd.c +++ b/drivers/s390/block/dasd.c @@ -2508,8 +2508,6 @@ int dasd_generic_restore_device(struct ccw_device *cdev) device->stopped &= ~DASD_UNRESUMED_PM; dasd_schedule_device_bh(device); - if (device->block) - dasd_schedule_block_bh(device->block); if (device->discipline->restore) rc = device->discipline->restore(device); @@ -2520,6 +2518,9 @@ int dasd_generic_restore_device(struct ccw_device *cdev) */ device->stopped |= DASD_UNRESUMED_PM; + if (device->block) + dasd_schedule_block_bh(device->block); + dasd_put_device(device); return 0; } |