scsi: qla2xxx: Fix DMA unmap leak
authorHimanshu Madhani <hmadhani@marvell.com>
Fri, 26 Jul 2019 16:07:26 +0000 (09:07 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sat, 21 Dec 2019 09:42:16 +0000 (10:42 +0100)
[ Upstream commit 5d328de64d89400dcf9911125844d8adc0db697f ]

With debug kernel we see following wanings indicating memory leak.

[28809.523959] WARNING: CPU: 3 PID: 6790 at lib/dma-debug.c:978
dma_debug_device_change+0x166/0x1d0
[28809.523964] pci 0000:0c:00.6: DMA-API: device driver has pending DMA
allocations while released from device [count=5]
[28809.523964] One of leaked entries details: [device
address=0x00000002aefe4000] [size=8208 bytes] [mapped with DMA_BIDIRECTIONAL]
[mapped as coherent]

Fix this by unmapping DMA memory.

Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Himanshu Madhani <hmadhani@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
drivers/scsi/qla2xxx/qla_bsg.c

index 33e4dce..e7db95d 100644 (file)
@@ -336,6 +336,8 @@ qla2x00_process_els(struct fc_bsg_job *bsg_job)
                dma_map_sg(&ha->pdev->dev, bsg_job->request_payload.sg_list,
                bsg_job->request_payload.sg_cnt, DMA_TO_DEVICE);
        if (!req_sg_cnt) {
+               dma_unmap_sg(&ha->pdev->dev, bsg_job->request_payload.sg_list,
+                   bsg_job->request_payload.sg_cnt, DMA_TO_DEVICE);
                rval = -ENOMEM;
                goto done_free_fcport;
        }
@@ -343,6 +345,8 @@ qla2x00_process_els(struct fc_bsg_job *bsg_job)
        rsp_sg_cnt = dma_map_sg(&ha->pdev->dev, bsg_job->reply_payload.sg_list,
                bsg_job->reply_payload.sg_cnt, DMA_FROM_DEVICE);
         if (!rsp_sg_cnt) {
+               dma_unmap_sg(&ha->pdev->dev, bsg_job->reply_payload.sg_list,
+                   bsg_job->reply_payload.sg_cnt, DMA_FROM_DEVICE);
                rval = -ENOMEM;
                goto done_free_fcport;
        }