Re: [PATCH] scsi: libsas: Reset num_scatter if libata mark qc as NODATA

[Jolly Shah <jollys@google.com>]

Hi John,


On Thu, Mar 18, 2021 at 9:19 AM John Garry <john.garry@huawei.com> wrote:
>
> On 18/03/2021 00:24, Jolly Shah wrote:
> > Hi John,
> >
> > Thanks for the review.
> >
> >
> > On Wed, Mar 17, 2021 at 4:44 AM John Garry <john.garry@huawei.com> wrote:
> >>
> >> On 16/03/2021 19:39, Jolly Shah wrote:
> >>> When the cache_type for the scsi device is changed, the scsi layer
> >>> issues a MODE_SELECT command. The caching mode details are communicated
> >>> via a request buffer associated with the scsi command with data
> >>> direction set as DMA_TO_DEVICE (scsi_mode_select). When this command
> >>> reaches the libata layer, as a part of generic initial setup, libata
> >>> layer sets up the scatterlist for the command using the scsi command
> >>> (ata_scsi_qc_new). This command is then translated by the libata layer
> >>> into ATA_CMD_SET_FEATURES (ata_scsi_mode_select_xlat). The libata layer
> >>> treats this as a non data command (ata_mselect_caching), since it only
> >>> needs an ata taskfile to pass the caching on/off information to the
> >>> device. It does not need the scatterlist that has been setup, so it does
> >>> not perform dma_map_sg on the scatterlist (ata_qc_issue).
> >>
> >> So if we don't perform the dma_map_sg() on the sgl at this point, then
> >> it seems to me that we should not perform for_each_sg() on it either,
> >> right? That is still what happens in sas_ata_qc_issue() in this case.
> >>
>
> Hi Jolly Shah,
>
> >
> > Yes that's right. To avoid that, I can add elseif block for
> > ATA_PROT_NODATA before for_each_sg() is performed. Currently there's
> > existing code block for ATA_PROT_NODATA after for_each_sg()  is
> > performed,
> > reused that to reset num_scatter. Please suggest.
> >
>
> How about just combine the 2x if-else statements into 1x if-elif-else
> statement, like:
>
>
> if (ata_is_atapi(qc->tf.protocol)) {
>         memcpy(task->ata_task.atapi_packet, qc->cdb, qc->dev->cdb_len);
>         task->total_xfer_len = qc->nbytes;
>         task->num_scatter = qc->n_elem;
>         task->data_dir = qc->dma_dir;
> } else if (qc->tf.protocol == ATA_PROT_NODATA) {
>         task->data_dir = DMA_NONE;
> } else {
>         for_each_sg(qc->sg, sg, qc->n_elem, si)
>                 xfer += sg_dma_len(sg);
>
>         task->total_xfer_len = xfer;
>         task->num_scatter = si;
>         task->data_dir = qc->dma_dir;
> }
>
Updated in v2.

> >>> Unfortunately,
> >>> when this command reaches the libsas layer(sas_ata_qc_issue), libsas
> >>> layer sees it as a non data command with a scatterlist. It cannot
> >>> extract the correct dma length, since the scatterlist has not been
> >>> mapped with dma_map_sg for a DMA operation. When this partially
> >>> constructed SAS task reaches pm80xx LLDD, it results in below warning.
> >>>
> >>> "pm80xx_chip_sata_req 6058: The sg list address
> >>> start_addr=0x0000000000000000 data_len=0x0end_addr_high=0xffffffff
> >>> end_addr_low=0xffffffff has crossed 4G boundary"
> >>>
> >>> This patch assigns appropriate value to  num_sectors for ata non data
> >>> commands.
> >>>
> >>> Signed-off-by: Jolly Shah <jollys@google.com>
> >>> ---
> >>>    drivers/scsi/libsas/sas_ata.c | 6 ++++--
> >>>    1 file changed, 4 insertions(+), 2 deletions(-)
> >>>
> >>> diff --git a/drivers/scsi/libsas/sas_ata.c b/drivers/scsi/libsas/sas_ata.c
> >>> index 024e5a550759..94ec08cebbaa 100644
> >>> --- a/drivers/scsi/libsas/sas_ata.c
> >>> +++ b/drivers/scsi/libsas/sas_ata.c
> >>> @@ -209,10 +209,12 @@ static unsigned int sas_ata_qc_issue(struct ata_queued_cmd *qc)
> >>>                task->num_scatter = si;
> >>>        }
> >>>
> >>> -     if (qc->tf.protocol == ATA_PROT_NODATA)
> >>> +     if (qc->tf.protocol == ATA_PROT_NODATA) {
> >>>                task->data_dir = DMA_NONE;
> >>> -     else
> >>> +             task->num_scatter = 0;
> >>
> >> task->num_scatter has already been set in this function. Best not set it
> >> twice.
> >>
> >
> > Sure. Please suggest if I should update patch to above suggested
> > approach. That will avoid setting num_scatter twice.
> >
>
> Thanks,
> John
>
> BTW, could we add a fixes tag?

Added in v2.

Thanks,
Jolly