On 25 August 2015 at 14:06, Ulf Hansson [off-list ref] wrote:

On 14 August 2015 at 09:30,  [off-list ref] wrote:

quoted

From: Yong Li <redacted>

Signed-off-by: Yong Li <redacted>
---
 drivers/mmc/card/block.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/mmc/card/block.c b/drivers/mmc/card/block.c
index 452782b..d9e3c45 100644
--- a/drivers/mmc/card/block.c
+++ b/drivers/mmc/card/block.c

@@ -1366,7 +1366,8 @@ static void mmc_blk_rw_rq_prep(struct mmc_queue_req *mqrq,
        bool do_rel_wr = ((req->cmd_flags & REQ_FUA) ||
                          (req->cmd_flags & REQ_META)) &&
                (rq_data_dir(req) == WRITE) &&
-               (md->flags & MMC_BLK_REL_WR);
+               (md->flags & MMC_BLK_REL_WR) &&
+               !(card->quirks & MMC_QUIRK_BLK_NO_CMD23);

Further down in mmc_blk_rw_rq_prep() we check for
MMC_QUIRK_BLK_NO_CMD23. That check becomes redundant after this
change, please remove that check as a part of this patch as well.

quoted

        memset(brq, 0, sizeof(struct mmc_blk_request));
        brq->mrq.cmd = &brq->cmd;
--
2.1.0

Please ignore my previous answer. MMC_QUIRK_BLK_NO_CMD23 is intended
to indicate to the mmc block layer whether using CMD23 for regular
block IO request could have a performance impact for some cards. If
that's the case we don't use it - except when using reliable write!

So this patch is just plain wrong.

Kind regards
Uffe