2008-02-19 14:47:12

by Martin Schwidefsky

[permalink] [raw]
Subject: [patch 05/13] dasd: fix locking in __dasd_device_process_final_queue

From: Stefan Weinhuber <[email protected]>

After setting the status of the cqr and releasing the lock for the
block cqr queue, we call the cqr callback function, which will usually
just trigger the dasd_block_tasklet. But when the tasklet is already
running the cqr might be processed before we invoke the callback
function. In rare cases the callback pointer may already be invalid
by the time we want to call it, which will result in a panic.
Solution: Call the callback function first and then release the lock.

Signed-off-by: Stefan Weinhuber <[email protected]>
Signed-off-by: Martin Schwidefsky <[email protected]>
---

drivers/s390/block/dasd.c | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)

Index: quilt-2.6/drivers/s390/block/dasd.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd.c
+++ quilt-2.6/drivers/s390/block/dasd.c
@@ -1149,12 +1149,14 @@ static void __dasd_device_process_final_
{
struct list_head *l, *n;
struct dasd_ccw_req *cqr;
+ struct dasd_block *block;

list_for_each_safe(l, n, final_queue) {
cqr = list_entry(l, struct dasd_ccw_req, devlist);
list_del_init(&cqr->devlist);
- if (cqr->block)
- spin_lock_bh(&cqr->block->queue_lock);
+ block = cqr->block;
+ if (block)
+ spin_lock_bh(&block->queue_lock);
switch (cqr->status) {
case DASD_CQR_SUCCESS:
cqr->status = DASD_CQR_DONE;
@@ -1172,15 +1174,13 @@ static void __dasd_device_process_final_
cqr, cqr->status);
BUG();
}
- if (cqr->block)
- spin_unlock_bh(&cqr->block->queue_lock);
if (cqr->callback != NULL)
(cqr->callback)(cqr, cqr->callback_data);
+ if (block)
+ spin_unlock_bh(&block->queue_lock);
}
}

-
-
/*
* Take a look at the first request on the ccw queue and check
* if it reached its expire time. If so, terminate the IO.

--
blue skies,
Martin.

"Reality continues to ruin my life." - Calvin.