Re: [PATCH v11 6/6] mmc: core: Add support for resetting an eMMC at system resume

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] mmc: core: unchecked _mmc_hw_reset() error and premature active state in mmc_resume()
--

commit f67582c90bab341460117ecbfa06f572b7755129
Author: Ulf Hansson <ulf.hansson@oss.qualcomm.com>

mmc: core: Add support for resetting an eMMC at system resume

This commit adds support for the reset-card-at-resume DT property by setting
MMC_CAP2_RESET_AT_RESUME and MMC_PM_KEEP_POWER, and calling _mmc_hw_reset()
during mmc_resume().

> diff --git a/drivers/mmc/core/mmc.c b/drivers/mmc/core/mmc.c
> index c183afbed30c8..793767c5159f2 100644
> --- a/drivers/mmc/core/mmc.c
> +++ b/drivers/mmc/core/mmc.c
> @@ -2320,6 +2320,14 @@ static int mmc_resume(struct mmc_host *host)
>  	}
>  
>  	pm_runtime_enable(&host->card->dev);
> +
> +	if (mmc_card_keep_power(host) &&
> +	    host->caps2 & MMC_CAP2_RESET_AT_RESUME) {
> +		mmc_claim_host(host);
> +		_mmc_hw_reset(host);
> +		mmc_release_host(host);
> +	}

[Severity: Medium]
Should mmc_resume() check the return value of _mmc_hw_reset(host) and avoid
marking the card active before the reset succeeds?

Right now, when mmc_card_keep_power(host) is true, mmc_resume() unconditionally
marks the card runtime-active and clears the suspended state first:

drivers/mmc/core/mmc.c:mmc_resume() {
	if (mmc_card_keep_power(host)) {
		pm_runtime_set_active(&host->card->dev);
		mmc_card_clr_suspended(host->card);
	}
	pm_runtime_enable(&host->card->dev);
	...
}

After that, _mmc_hw_reset(host) resets or power-cycles the card, resets the
host IOS state to initial values via mmc_set_initial_state(host), and calls
mmc_init_card(host, card->ocr, card).

If mmc_init_card() fails (for example due to a command timeout, voltage or
timing switch failure, or tuning error), mmc_resume() discards the error and
returns 0 to mmc_bus_resume(). That suppresses the resume failure warning in
mmc_bus_resume() and leaves the card marked runtime-active and unsuspended
while the hardware remains in an uninitialized reset state, preventing a
subsequent runtime resume via _mmc_resume() from re-initializing the card.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261030181034.3811225-1-kamal.dasu@broadcom.com?part=6