[SCSI] 3w-9xxx: fix iommu_iova leak
authorJames Bottomley <JBottomley@Parallels.com>
Sun, 18 Sep 2011 14:56:20 +0000 (18:56 +0400)
committerJames Bottomley <JBottomley@Parallels.com>
Mon, 26 Sep 2011 14:28:58 +0000 (09:28 -0500)
Following reports on the list, it looks like the 3e-9xxx driver will leak dma
mappings every time we get a transient queueing error back from the card.
This is because it maps the sg list in the routine that sends the command, but
doesn't unmap again in the transient failure path (even though the command is
sent back to the block layer).  Fix by unmapping before returning the status.

Reported-by: Chris Boot <bootc@bootc.net>
Tested-by: Chris Boot <bootc@bootc.net>
Acked-by: Adam Radford <aradford@gmail.com>
Cc: stable@kernel.org
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
drivers/scsi/3w-9xxx.c

index b7bd5b0cc7aa450e23ebe7f8b0ddc10814b82f05..3868ab2397c6bcb3418d7f2b8663739438e86809 100644 (file)
@@ -1800,10 +1800,12 @@ static int twa_scsi_queue_lck(struct scsi_cmnd *SCpnt, void (*done)(struct scsi_
        switch (retval) {
        case SCSI_MLQUEUE_HOST_BUSY:
                twa_free_request_id(tw_dev, request_id);
+               twa_unmap_scsi_data(tw_dev, request_id);
                break;
        case 1:
                tw_dev->state[request_id] = TW_S_COMPLETED;
                twa_free_request_id(tw_dev, request_id);
+               twa_unmap_scsi_data(tw_dev, request_id);
                SCpnt->result = (DID_ERROR << 16);
                done(SCpnt);
                retval = 0;