mmc: limit the time of retries when analyse tuples failed
PATCH v2 1/1 mmc: limit the time of retries when analyse tuples failed
when analyse tuples failed, may enter an endless loop,so limit the time of retries.
--- drivers/mmc/core/sdio_cis.c | 3 +++ 1 file changed, 3 insertions(+)
drivers/mmc/core/sdio_cis.c +3 -0
@@ -266,6 +266,7 @@ static int sdio_read_cis(struct mmc_card *card, struct sdio_func *func) do { unsigned char tpl_code, tpl_link; + u64 timeout = get_jiffies_64() + 10 * HZ; ret = mmc_io_rw_direct(card, 0, 0, ptr++, 0, &tpl_code); if (ret)
@@ -318,6 +319,8 @@ static int sdio_read_cis(struct mmc_card *card, struct sdio_func *func) prev = &this->next;if (ret == -ENOENT ) { +if (time_after64 (get_jiffies_64 (), timeout)) +break ;/* warn about unknown tuples */ pr_warn_ratelimited("%s: queuing unknown" " CIS tuple 0x%02x (%u bytes)\n",
--
2.25.1
Quoted 2 lines · Fengnan Chang
when analyse tuples failed, may enter an endless loop,so limit the time of retries.
Since this is fixing a real bug for you, it looks like we should tag this for stable kernels as well, right?
Quoted 11 lines
Signed-off-by Fengnan Chang <[email protected]>--- drivers/mmc/core/sdio_cis.c | 3 +++ 1 file changed, 3 insertions(+)drivers/mmc/core/sdio_cis.c +0
-0
drivers/mmc/core/sdio_cis.c static int sdio_read_cis(struct mmc_card *card, struct sdio_func *func) · line 269
Quoted 4 lines
drivers/mmc/core/sdio_cis.c +1
-0@@ -266,6 +266,7 @@ static int sdio_read_cis(struct mmc_card *card, struct sdio_func *func) do { unsigned char tpl_code, tpl_link; + u64 timeout = get_jiffies_64() + 10 * HZ;
To be consistent with how we do time based polling (see __mmc_poll_for_busy() in mmc_ops.c for example) I would prefer if you use a define for the timeout, rather than a magic value as here.
#define SDIO_READ_CIS_TIMEOUT_MS (10 * 1000) /* 10s */
Additionally, regular jiffies should be sufficient I think. Thus I prefer if you could specify the timeout along the lines like the below:
timeout = jiffies + msecs_to_jiffies(SDIO_READ_CIS_TIMEOUT_MS);
Quoted 13 lines
drivers/mmc/core/sdio_cis.c +2
-1ret =mmc_io_rw_direct (card,0 ,0 , ptr++,0 , &tpl_code);if (ret)@@ -318,6 +319,8 @@ static int sdio_read_cis(struct mmc_card *card, struct sdio_func *func) prev = &this->next;if (ret == -ENOENT ) { +if (time_after64 (get_jiffies_64 (), timeout)) +break ;/* warn about unknown tuples */ pr_warn_ratelimited("%s: queuing unknown" " CIS tuple 0x%02x (%u bytes)\n", --
Quoted 2 lines
2.25.1
Other than that, this looks okay to me.
Kind regards
Uffe
yeah, I think it should tag this for stable kernels as well. I will send a new version later.
Quoted 55 lines · Ulf Hansson <[email protected]> 于2021年1月20日周三 下午11:29写道:
Quoted 2 lines · Fengnan Chang
when analyse tuples failed, may enter an endless loop,so limit the time of retries.
Since this is fixing a real bug for you, it looks like we should tag this for stable kernels as well, right?
Quoted 11 lines
Signed-off-by Fengnan Chang <[email protected]>--- drivers/mmc/core/sdio_cis.c | 3 +++ 1 file changed, 3 insertions(+)drivers/mmc/core/sdio_cis.c +0
-0drivers/mmc/core/sdio_cis.c static int sdio_read_cis(struct mmc_card *card, struct sdio_func *func) · line 269
Quoted 4 lines
drivers/mmc/core/sdio_cis.c +1
-0@@ -266,6 +266,7 @@ static int sdio_read_cis(struct mmc_card *card, struct sdio_func *func) do { unsigned char tpl_code, tpl_link; + u64 timeout = get_jiffies_64() + 10 * HZ;To be consistent with how we do time based polling (see
__mmc_poll_for_busy()inmmc_ops.cfor example) I would prefer if you use a define for the timeout, rather than a magic value as here.#define
SDIO_READ_CIS_TIMEOUT_MS(10 * 1000) /* 10s */Additionally, regular jiffies should be sufficient I think. Thus I prefer if you could specify the timeout along the lines like the below:
timeout = jiffies +msecs_to_jiffies(SDIO_READ_CIS_TIMEOUT_MS);Quoted 13 lines
drivers/mmc/core/sdio_cis.c +2
-1ret =mmc_io_rw_direct (card,0 ,0 , ptr++,0 , &tpl_code);if (ret)@@ -318,6 +319,8 @@ static int sdio_read_cis(struct mmc_card *card, struct sdio_func *func) prev = &this->next;if (ret == -ENOENT ) { +if (time_after64 (get_jiffies_64 (), timeout)) +break ;/* warn about unknown tuples */ pr_warn_ratelimited("%s: queuing unknown" " CIS tuple 0x%02x (%u bytes)\n", --Quoted 2 lines
2.25.1
Other than that, this looks okay to me.
Kind regards
Uffe