Return-Path: Message-Id: <20080529125728.607920561@de.ibm.com> References: <20080529125501.196123527@de.ibm.com> Date: Thu, 29 May 2008 14:55:03 +0200 From: Martin Schwidefsky Subject: [patch 02/11] tape: Fix race condition in tape block device driver Content-Disposition: inline; filename=002-tape-block.diff Sender: linux-kernel-owner@vger.kernel.org List-Archive: List-Post: To: linux-kernel@vger.kernel.org, linux-s390@vger.kernel.org Cc: Michael Holzheu List-ID: From: Michael Holzheu Due to incorrect function call sequence it can happen that a tape block request is finished before the request is taken from the block request queue. The following sequence leads to that condition: * tapeblock_start_request() -> start CCW program * Request finishes -> IO interrupt * tapeblock_end_request() * end_that_request_last() If blkdev_dequeue_request() has not been called before end_that_request_last(), a kernel bug is triggered in end_that_request_last() because the request is still queued. To solve that problem blkdev_dequeue_request() has to be called before starting the CCW program. Signed-off-by: Michael Holzheu --- drivers/s390/char/tape_block.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) Index: quilt-2.6/drivers/s390/char/tape_block.c =================================================================== --- quilt-2.6.orig/drivers/s390/char/tape_block.c +++ quilt-2.6/drivers/s390/char/tape_block.c @@ -179,11 +179,11 @@ tapeblock_requeue(struct work_struct *wo tapeblock_end_request(req, -EIO); continue; } + blkdev_dequeue_request(req); + nr_queued++; spin_unlock_irq(&device->blk_data.request_queue_lock); rc = tapeblock_start_request(device, req); spin_lock_irq(&device->blk_data.request_queue_lock); - blkdev_dequeue_request(req); - nr_queued++; } spin_unlock_irq(&device->blk_data.request_queue_lock); atomic_set(&device->blk_data.requeue_scheduled, 0); -- blue skies, Martin. "Reality continues to ruin my life." - Calvin.