[PATCH v2] mmc: mmci: manage MMC_CAP_NEED_RSP_BUSY for stm32 variant

[Ulf Hansson]

On Thu, 25 Feb 2021 at 15:55, Yann Gautier <yann.gautier at foss.st.com> wrote:
>
> To properly manage commands awaiting R1B responses, the capability
> MMC_CAP_NEED_RSP_BUSY is enabled in mmci driver, for stm32 variant.
> The issue was seen on STM32MP157C-EV1 board, with an erase command,
> with secure erase argument, letting the card stuck, possibly waiting
> for 4 hours before timeout.
>
> Fixes: 94fe2580a2f3 ("mmc: core: Enable erase/discard/trim support for all mmc hosts")
>
> Signed-off-by: Yann Gautier <yann.gautier at foss.st.com>
> ---
> This is somehow a v2 for patch [1].
> Changes:
> - Only apply MMC_CAP_NEED_RSP_BUSY to stm32 variant
> - Cap the used timeout written to MMCIDATATIMER (when using
> MMC_CAP_NEED_RSP_BUSY, cmd->busy_timeout may be greater than
> host->max_busy_timeout)
>
>  [1] https://patchwork.kernel.org/project/linux-mmc/patch/20210204120547.15381-2-yann.gautier@foss.st.com/
>
>  drivers/mmc/host/mmci.c             | 8 +++++++-
>  drivers/mmc/host/mmci_stm32_sdmmc.c | 1 +
>  2 files changed, 8 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/mmc/host/mmci.c b/drivers/mmc/host/mmci.c
> index 17dbc81c221e..89e0e9ccfb71 100644
> --- a/drivers/mmc/host/mmci.c
> +++ b/drivers/mmc/host/mmci.c
> @@ -1242,7 +1242,13 @@ mmci_start_command(struct mmci_host *host, struct mmc_command *cmd, u32 c)
>                 if (!cmd->busy_timeout)
>                         cmd->busy_timeout = 10 * MSEC_PER_SEC;
>
> -               clks = (unsigned long long)cmd->busy_timeout * host->cclk;
> +               if (host->mmc->caps & MMC_CAP_NEED_RSP_BUSY &&
> +                   host->mmc->max_busy_timeout &&
> +                   cmd->busy_timeout > host->mmc->max_busy_timeout)

We are already within "if (host->variant->busy_timeout ....", a few
lines above, which means this can be simplified into:

if (cmd->busy_timeout > host->mmc->max_busy_timeout)

> +                       clks = (unsigned long long)host->mmc->max_busy_timeout * host->cclk;
> +               else
> +                       clks = (unsigned long long)cmd->busy_timeout * host->cclk;
> +
>                 do_div(clks, MSEC_PER_SEC);
>                 writel_relaxed(clks, host->base + MMCIDATATIMER);
>         }
> diff --git a/drivers/mmc/host/mmci_stm32_sdmmc.c b/drivers/mmc/host/mmci_stm32_sdmmc.c
> index 51db30acf4dc..2ad577618324 100644
> --- a/drivers/mmc/host/mmci_stm32_sdmmc.c
> +++ b/drivers/mmc/host/mmci_stm32_sdmmc.c
> @@ -522,6 +522,7 @@ void sdmmc_variant_init(struct mmci_host *host)
>
>         host->ops = &sdmmc_variant_ops;
>         host->pwr_reg = readl_relaxed(host->base + MMCIPOWER);
> +       host->mmc->caps |= MMC_CAP_NEED_RSP_BUSY;

To make it more clear that this is for variants having the
->busy_timeout flag set, I suggest to move this into mmci_probe().

>
>         base_dlyb = devm_of_iomap(mmc_dev(host->mmc), np, 1, NULL);
>         if (IS_ERR(base_dlyb))
> --
> 2.17.1
>

Well, I decided to help out a bit. I have amend the patch according to
the above and extended the commit message with some valuable
information, based upon our earlier discussions.

Patch is applied at my fixes branch with a stable tag, please have a
look, test and shout at me if there is something that looks wrong!

Thanks and kind regards
Uffe