On 13/02/2019 11:42, Hannes Reinecke wrote:
The change to use dma_set_mask_and_coherent() incorrectly made a second call with the 32 bit DMA mask value when the call with the 64 bit DMA mask value succeeded. This resulted in FC connections failing due to corrupted data buffers, and various other SCSI/FCP I/O errors.
Fixes: e4db40e7a1a2 ("scsi: hisi_sas: use dma_set_mask_and_coherent") Cc: stable@vger.kernel.org Suggested-by: Ewan D. Milne emilne@redhat.com Signed-off-by: Hannes Reinecke hare@suse.com
drivers/scsi/hisi_sas/hisi_sas_main.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/drivers/scsi/hisi_sas/hisi_sas_main.c b/drivers/scsi/hisi_sas/hisi_sas_main.c index eed7fc5b3389..bc17fa0d8375 100644 --- a/drivers/scsi/hisi_sas/hisi_sas_main.c +++ b/drivers/scsi/hisi_sas/hisi_sas_main.c @@ -2323,6 +2323,7 @@ static struct Scsi_Host *hisi_sas_shost_alloc(struct platform_device *pdev, struct Scsi_Host *shost; struct hisi_hba *hisi_hba; struct device *dev = &pdev->dev;
int error;
shost = scsi_host_alloc(hw->sht, sizeof(*hisi_hba)); if (!shost) {
@@ -2343,8 +2344,11 @@ static struct Scsi_Host *hisi_sas_shost_alloc(struct platform_device *pdev, if (hisi_sas_get_fw_info(hisi_hba) < 0) goto err_out;
- if (dma_set_mask_and_coherent(dev, DMA_BIT_MASK(64)) &&
dma_set_mask_and_coherent(dev, DMA_BIT_MASK(32))) {
Aren't we supposed to be fixing hisi_sas_v3_probe()? We have a path for platform device probe and a path for pci device probe. You're changing the platform device probe, which is currently ok.
I meant to unify all this. Now would be a good time, as it's not a fix. Can do it later.
Thanks, John
- error = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(64));
- if (error)
error = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(32));- if (error) { dev_err(dev, "No usable DMA addressing method\n"); goto err_out; }