Hi Eric,
> -----Original Message-----
> From: Eric Biggers <ebiggers@xxxxxxxxxx>
> Sent: 12 August 2020 05:59
> To: Alim Akhtar <alim.akhtar@xxxxxxxxxxx>
> Cc: robh@xxxxxxxxxx; devicetree@xxxxxxxxxxxxxxx;
linux-scsi@xxxxxxxxxxxxxxx;
> krzk@xxxxxxxxxx; avri.altman@xxxxxxx; martin.petersen@xxxxxxxxxx;
> kwmad.kim@xxxxxxxxxxx; stanley.chu@xxxxxxxxxxxx;
> cang@xxxxxxxxxxxxxx; linux-samsung-soc@xxxxxxxxxxxxxxx; linux-arm-
> kernel@xxxxxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; kishon@xxxxxx
> Subject: Re: [RESEND PATCH v10 04/10] scsi: ufs: introduce
> UFSHCD_QUIRK_PRDT_BYTE_GRAN quirk
>
> Hi Alim,
>
> On Sat, Jun 13, 2020 at 08:17:00AM +0530, Alim Akhtar wrote:
> > Some UFS host controllers like Exynos uses granularities of PRDT
> > length and offset as bytes, whereas others uses actual segment count.
> >
> > Reviewed-by: Avri Altman <avri.altman@xxxxxxx>
> > Signed-off-by: Kiwoong Kim <kwmad.kim@xxxxxxxxxxx>
> > Signed-off-by: Alim Akhtar <alim.akhtar@xxxxxxxxxxx>
> > ---
> > drivers/scsi/ufs/ufshcd.c | 30 +++++++++++++++++++++++-------
> > drivers/scsi/ufs/ufshcd.h | 6 ++++++
> > 2 files changed, 29 insertions(+), 7 deletions(-)
> >
> > diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> > index ee30ed6cc805..ba093d0d0942 100644
> > --- a/drivers/scsi/ufs/ufshcd.c
> > +++ b/drivers/scsi/ufs/ufshcd.c
> > @@ -2151,8 +2151,14 @@ static int ufshcd_map_sg(struct ufs_hba *hba,
> struct ufshcd_lrb *lrbp)
> > return sg_segments;
> >
> > if (sg_segments) {
> > - lrbp->utr_descriptor_ptr->prd_table_length =
> > - cpu_to_le16((u16)sg_segments);
> > +
> > + if (hba->quirks & UFSHCD_QUIRK_PRDT_BYTE_GRAN)
> > + lrbp->utr_descriptor_ptr->prd_table_length =
> > + cpu_to_le16((sg_segments *
> > + sizeof(struct ufshcd_sg_entry)));
> > + else
> > + lrbp->utr_descriptor_ptr->prd_table_length =
> > + cpu_to_le16((u16) (sg_segments));
> >
> > prd_table = (struct ufshcd_sg_entry *)lrbp->ucd_prdt_ptr;
> >
> > @@ -3500,11 +3506,21 @@ static void
> ufshcd_host_memory_configure(struct ufs_hba *hba)
> >
> cpu_to_le32(upper_32_bits(cmd_desc_element_addr));
> >
> > /* Response upiu and prdt offset should be in double words
*/
> > - utrdlp[i].response_upiu_offset =
> > - cpu_to_le16(response_offset >> 2);
> > - utrdlp[i].prd_table_offset = cpu_to_le16(prdt_offset >> 2);
> > - utrdlp[i].response_upiu_length =
> > - cpu_to_le16(ALIGNED_UPIU_SIZE >> 2);
> > + if (hba->quirks & UFSHCD_QUIRK_PRDT_BYTE_GRAN) {
> > + utrdlp[i].response_upiu_offset =
> > + cpu_to_le16(response_offset);
> > + utrdlp[i].prd_table_offset =
> > + cpu_to_le16(prdt_offset);
> > + utrdlp[i].response_upiu_length =
> > + cpu_to_le16(ALIGNED_UPIU_SIZE);
> > + } else {
> > + utrdlp[i].response_upiu_offset =
> > + cpu_to_le16(response_offset >> 2);
> > + utrdlp[i].prd_table_offset =
> > + cpu_to_le16(prdt_offset >> 2);
> > + utrdlp[i].response_upiu_length =
> > + cpu_to_le16(ALIGNED_UPIU_SIZE >> 2);
> > + }
> >
> > ufshcd_init_lrb(hba, &hba->lrb[i], i);
> > }
>
> Isn't this patch missing an update to ufshcd_print_trs()? It uses
> ->prd_table_length as the number of segments, not the number of bytes.
>
prd_table_length will be populated before it reaches ufshcd_print_trs()
based on UFSHCD_QUIRK_PRDT_BYTE_GRAN.
> - Eric
