mmc: core: Avoid hanging to claim host for mmc via some nested calls

As the block layer, since the conversion to blkmq, claims the host using a
context, a following nested call to mmc_claim_host(), which isn't using a
context, may hang.

Calling mmc_interrupt_hpi() and mmc_read_bkops_status() via the mmc block
layer, may suffer from this problem, as these functions are calling
mmc_claim|release_host().

Let's fix the problem by removing the calls to mmc_claim|release_host()
from the above mentioned functions and instead make the callers responsible
of claiming/releasing the host. As a matter of fact, the existing callers
already deals with it.

Fixes: 81196976ed94 ("mmc: block: Add blk-mq support")
Reported-by: Dmitry Osipenko <digetx@gmail.com>
Suggested-by: Adrian Hunter <adrian.hunter@intel.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Shawn Lin <shawn.lin@rock-chips.com>

diff --git a/drivers/mmc/core/mmc_ops.c b/drivers/mmc/core/mmc_ops.c
index 908e4db03535b038d548563188c0a11ac1ce4761..42d6aa89a48a95ad82bc60445f73971ac48730d2 100644 (file)
--- a/drivers/mmc/core/mmc_ops.c
+++ b/drivers/mmc/core/mmc_ops.c
@@ -848,7 +848,6 @@ int mmc_interrupt_hpi(struct mmc_card *card)
                return 1;
        }
 
-       mmc_claim_host(card->host);
        err = mmc_send_status(card, &status);
        if (err) {
                pr_err("%s: Get card status fail\n", mmc_hostname(card->host));
@@ -890,7 +889,6 @@ int mmc_interrupt_hpi(struct mmc_card *card)
        } while (!err);
 
 out:
-       mmc_release_host(card->host);
        return err;
 }
 
@@ -932,9 +930,7 @@ static int mmc_read_bkops_status(struct mmc_card *card)
        int err;
        u8 *ext_csd;
 
-       mmc_claim_host(card->host);
        err = mmc_get_ext_csd(card, &ext_csd);
-       mmc_release_host(card->host);
        if (err)
                return err;