Please ignore this duplicate patch. My apologies.
Thanks,
Bao
On 6/1/2023 4:27 AM, Bao D. Nguyen wrote:
If the hwq's queue depth is not a multiple of 4s, the
logic used for wrapping around the increase-by-1 sq_head_slot
variable in ufshcd_mcq_sqe_search() will give an incorrect
result because the binary expression of the mask is not
a uniform of all 1s.
Signed-off-by: Bao D. Nguyen <quic_nguyenb@xxxxxxxxxxx>
---
drivers/ufs/core/ufs-mcq.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/drivers/ufs/core/ufs-mcq.c b/drivers/ufs/core/ufs-mcq.c
index 66ac02e..8e5bc88 100644
--- a/drivers/ufs/core/ufs-mcq.c
+++ b/drivers/ufs/core/ufs-mcq.c
@@ -585,7 +585,6 @@ static bool ufshcd_mcq_sqe_search(struct ufs_hba *hba,
{
struct ufshcd_lrb *lrbp = &hba->lrb[task_tag];
struct utp_transfer_req_desc *utrd;
- u32 mask = hwq->max_entries - 1;
__le64 cmd_desc_base_addr;
bool ret = false;
u64 addr, match;
@@ -610,7 +609,10 @@ static bool ufshcd_mcq_sqe_search(struct ufs_hba *hba,
ret = true;
goto out;
}
- sq_head_slot = (sq_head_slot + 1) & mask;
+
+ sq_head_slot++;
+ if (sq_head_slot == hwq->max_entries)
+ sq_head_slot = 0;
}
out:
