| Message ID | 20211009204959.5864-1-kirill.kapranov@compulab.co.il |
|---|---|
| State | Accepted |
| Commit | 44645f87de513468a8c8082d057704df0495d7a4 |
| Delegated to: | Jaehoon Chung |
| Headers | show |
| Series | [1/1] mmc: Fix mmc_switch excessive timeout | expand |
Reviewed-by: Marek Behún <marek.behun@nic.cz> Tested-by: Marek Behún <marek.behun@nic.cz> This fixes a bug introduced in commit 3f4e52fc9d81 https://source.denx.de/u-boot/u-boot/-/commit/3f4e52fc9d81e95ee90f6be7b05ca62fc8c7c1e3 and it is incorrect. It caused the mmc initialization to take 4 seconds (previously it was <0.1s). The code in Linux also uses &&, and the comment says: If the host doesn't support HW polling via the ->card_busy() ops and when it's not allowed to poll by using CMD13, then we need to rely on waiting the stated timeout to be sufficient. See https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/mmc/core/mmc_ops.c?h=v5.15-rc5#n598 Peng, could we get this merged ASAP? :) Marek
diff --git a/drivers/mmc/mmc.c b/drivers/mmc/mmc.c index ba54b19c14..4d9871d69f 100644 --- a/drivers/mmc/mmc.c +++ b/drivers/mmc/mmc.c @@ -819,11 +819,11 @@ static int __mmc_switch(struct mmc *mmc, u8 set, u8 index, u8 value, return ret; /* - * In cases when not allowed to poll by using CMD13 or because we aren't + * In cases when neiter allowed to poll by using CMD13 nor we are * capable of polling by using mmc_wait_dat0, then rely on waiting the * stated timeout to be sufficient. */ - if (ret == -ENOSYS || !send_status) { + if (ret == -ENOSYS && !send_status) { mdelay(timeout_ms); return 0; }
Fix branching to avoid premature falling back on a long timeout instead of continuation of the initialization attempt. Clear of the comment to avoid the ambiguity. Signed-off-by: Kirill Kapranov <kirill.kapranov@compulab.co.il> Cc: Pantelis Antoniou <panto@antoniou-consulting.com> Cc: Ye Li <ye.li@nxp.com> --- drivers/mmc/mmc.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)