[PATCH] mmc: limit the number of retries when analyse tuples failed

[PATCH] mmc: limit the number of retries when analyse tuples failed

[Fengnan Chang]

when analyse tuples failed, may enter an endless loop，so limit the number of retries.

Signed-off-by: Fengnan Chang <fengnanchang@gmail.com>
---
 drivers/mmc/core/sdio_cis.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/mmc/core/sdio_cis.c b/drivers/mmc/core/sdio_cis.c
index dcb3dee59fa5..a3f0c3cc0c2c 100644
--- a/drivers/mmc/core/sdio_cis.c
+++ b/drivers/mmc/core/sdio_cis.c
@@ -266,6 +266,7 @@ static int sdio_read_cis(struct mmc_card *card, struct sdio_func *func)
 
 	do {
 		unsigned char tpl_code, tpl_link;
+		int  tries = 100;
 
 		ret = mmc_io_rw_direct(card, 0, 0, ptr++, 0, &tpl_code);
 		if (ret)
@@ -318,6 +319,9 @@ static int sdio_read_cis(struct mmc_card *card, struct sdio_func *func)
 			prev = &this->next;
 
 			if (ret == -ENOENT) {
+				tries--;
+				if (!tries)
+					break;
 				/* warn about unknown tuples */
 				pr_warn_ratelimited("%s: queuing unknown"
 				       " CIS tuple 0x%02x (%u bytes)\n",
-- 
2.25.1

Re: [PATCH] mmc: limit the number of retries when analyse tuples failed

[Ulf Hansson]

On Tue, 12 Jan 2021 at 15:18, Fengnan Chang <fengnanchang@gmail.com> wrote:
>
> when analyse tuples failed, may enter an endless loop，so limit the number of retries.
>
> Signed-off-by: Fengnan Chang <fengnanchang@gmail.com>
> ---
>  drivers/mmc/core/sdio_cis.c | 4 ++++
>  1 file changed, 4 insertions(+)
>
> diff --git a/drivers/mmc/core/sdio_cis.c b/drivers/mmc/core/sdio_cis.c
> index dcb3dee59fa5..a3f0c3cc0c2c 100644
> --- a/drivers/mmc/core/sdio_cis.c
> +++ b/drivers/mmc/core/sdio_cis.c
> @@ -266,6 +266,7 @@ static int sdio_read_cis(struct mmc_card *card, struct sdio_func *func)
>
>         do {
>                 unsigned char tpl_code, tpl_link;
> +               int  tries = 100;
>

From a general point of view I agree, we should set a limit to avoid
looping forever. Although, I am wondering why exactly is 100 a good
value for retries?

Did you encounter this problem and tried to limit the loop to 100 to
fix the problem - or is the fix done based on pure code analysis?

An option to limit the loop with retries is to bail out when a timer
elapses. Perhaps that is better suited for this case or what do you
think?

>                 ret = mmc_io_rw_direct(card, 0, 0, ptr++, 0, &tpl_code);
>                 if (ret)
> @@ -318,6 +319,9 @@ static int sdio_read_cis(struct mmc_card *card, struct sdio_func *func)
>                         prev = &this->next;
>
>                         if (ret == -ENOENT) {
> +                               tries--;
> +                               if (!tries)
> +                                       break;
>                                 /* warn about unknown tuples */
>                                 pr_warn_ratelimited("%s: queuing unknown"
>                                        " CIS tuple 0x%02x (%u bytes)\n",
> --
> 2.25.1
>

Kind regards
Uffe