mmc: block: do not assign mq_rq when aborting command
authorLinus Walleij <linus.walleij@linaro.org>
Tue, 24 Jan 2017 10:17:54 +0000 (11:17 +0100)
committerUlf Hansson <ulf.hansson@linaro.org>
Mon, 13 Feb 2017 12:20:38 +0000 (13:20 +0100)
The code in mmc_blk_issue_rq_rq() aborts a command if the request
is not properly aligned on large sectors. As part of the path
jumping out, it assigns the local variable mq_rq reflecting
a MMC queue request to the current MMC queue request, which is
confusing since the variable is not used after this jump.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
drivers/mmc/core/block.c

index b60d1fb3a07ae92002cd34430a66eed3fb393914..13e6fe060f26b7ceef5850ad01ee4f3ecb32bac6 100644 (file)
@@ -1649,7 +1649,6 @@ static int mmc_blk_issue_rw_rq(struct mmc_queue *mq, struct request *rqc)
                                !IS_ALIGNED(blk_rq_sectors(rqc), 8)) {
                                pr_err("%s: Transfer size is not 4KB sector size aligned\n",
                                        rqc->rq_disk->disk_name);
-                               mq_rq = mq->mqrq_cur;
                                req = rqc;
                                rqc = NULL;
                                goto cmd_abort;