| Message ID | 20240327024545.138351-1-ricky_wu@realtek.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | mmc: core: resume not check card exist before powerup | expand |
On Wed, 27 Mar 2024 at 03:46, Ricky Wu <ricky_wu@realtek.com> wrote: > > _mmc_sd_resume > add get_cd before call powerup, make sure the card exist Please elaborate more on what problem you are trying to solve, the above doesn't make much sense to me, sorry. Yes, the card may be removed while the system is suspended. Then, even if ->get_cd() indicates that there is no card inserted in the SD-card-slot, we may still have the card being registered - and then we need to unregister it. That said, we need to call mmc_power_up() in order to try to communicate with the card again. At least that is what the mmc_rescan() work assumes when it runs the bus_ops->detect() callback at some point later in time. > > Signed-off-by: Ricky Wu <ricky_wu@realtek.com> > --- > drivers/mmc/core/sd.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/mmc/core/sd.c b/drivers/mmc/core/sd.c > index 1c8148cdda50..35e036672cfb 100644 > --- a/drivers/mmc/core/sd.c > +++ b/drivers/mmc/core/sd.c > @@ -1750,6 +1750,9 @@ static int _mmc_sd_resume(struct mmc_host *host) > if (!mmc_card_suspended(host->card)) > goto out; > > + if (host->ops->get_cd && !host->ops->get_cd(host)) > + goto out; > + > mmc_power_up(host, host->card->ocr); > err = mmc_sd_init_card(host, host->card->ocr, host->card); > mmc_card_clr_suspended(host->card); > -- > 2.25.1 > Kind regards Uffe
> On Wed, 27 Mar 2024 at 03:46, Ricky Wu <ricky_wu@realtek.com> wrote: > > > > _mmc_sd_resume > > add get_cd before call powerup, make sure the card exist > > Please elaborate more on what problem you are trying to solve, the > above doesn't make much sense to me, sorry. > > Yes, the card may be removed while the system is suspended. Then, even > if ->get_cd() indicates that there is no card inserted in the > SD-card-slot, we may still have the card being registered - and then > we need to unregister it. > That said, we need to call mmc_power_up() in order to try to > communicate with the card again. At least that is what the > mmc_rescan() work assumes when it runs the bus_ops->detect() callback > at some point later in time. > We saw the power up in a short time from waveform when removing the card, So we saw mmc_sd_resume call the power up did not check the card exist. We think this the short time power up maybe cause OCP if no card exist. And this power up we think is unnecessary when no card exist. > > > > Signed-off-by: Ricky Wu <ricky_wu@realtek.com> > > --- > > drivers/mmc/core/sd.c | 3 +++ > > 1 file changed, 3 insertions(+) > > > > diff --git a/drivers/mmc/core/sd.c b/drivers/mmc/core/sd.c > > index 1c8148cdda50..35e036672cfb 100644 > > --- a/drivers/mmc/core/sd.c > > +++ b/drivers/mmc/core/sd.c > > @@ -1750,6 +1750,9 @@ static int _mmc_sd_resume(struct mmc_host > *host) > > if (!mmc_card_suspended(host->card)) > > goto out; > > > > + if (host->ops->get_cd && !host->ops->get_cd(host)) > > + goto out; > > + > > mmc_power_up(host, host->card->ocr); > > err = mmc_sd_init_card(host, host->card->ocr, host->card); > > mmc_card_clr_suspended(host->card); > > -- > > 2.25.1 > > > > Kind regards > Uffe
Soft reminder If this solution, you still have concert on sd.c. can we just do this check card flow in mmc host power up call_back? > > > > > > _mmc_sd_resume > > > add get_cd before call powerup, make sure the card exist > > > > Please elaborate more on what problem you are trying to solve, the > > above doesn't make much sense to me, sorry. > > > > Yes, the card may be removed while the system is suspended. Then, even > > if ->get_cd() indicates that there is no card inserted in the > > SD-card-slot, we may still have the card being registered - and then > > we need to unregister it. > > That said, we need to call mmc_power_up() in order to try to > > communicate with the card again. At least that is what the > > mmc_rescan() work assumes when it runs the bus_ops->detect() callback > > at some point later in time. > > > > We saw the power up in a short time from waveform when removing the card, > So we saw mmc_sd_resume call the power up did not check the card exist. > > We think this the short time power up maybe cause OCP if no card exist. > And this power up we think is unnecessary when no card exist. > > > > > > > > Signed-off-by: Ricky Wu <ricky_wu@realtek.com> > > > --- > > > drivers/mmc/core/sd.c | 3 +++ > > > 1 file changed, 3 insertions(+) > > > > > > diff --git a/drivers/mmc/core/sd.c b/drivers/mmc/core/sd.c index > > > 1c8148cdda50..35e036672cfb 100644 > > > --- a/drivers/mmc/core/sd.c > > > +++ b/drivers/mmc/core/sd.c > > > @@ -1750,6 +1750,9 @@ static int _mmc_sd_resume(struct mmc_host > > *host) > > > if (!mmc_card_suspended(host->card)) > > > goto out; > > > > > > + if (host->ops->get_cd && !host->ops->get_cd(host)) > > > + goto out; > > > + > > > mmc_power_up(host, host->card->ocr); > > > err = mmc_sd_init_card(host, host->card->ocr, host->card); > > > mmc_card_clr_suspended(host->card); > > > -- > > > 2.25.1 > > > > > > > Kind regards > > Uffe
On Fri, 19 Apr 2024 at 08:53, Ricky WU <ricky_wu@realtek.com> wrote: > > Soft reminder > > If this solution, you still have concert on sd.c. > can we just do this check card flow in mmc host power up call_back? > > > > > > > > > _mmc_sd_resume > > > > add get_cd before call powerup, make sure the card exist > > > > > > Please elaborate more on what problem you are trying to solve, the > > > above doesn't make much sense to me, sorry. > > > > > > Yes, the card may be removed while the system is suspended. Then, even > > > if ->get_cd() indicates that there is no card inserted in the > > > SD-card-slot, we may still have the card being registered - and then > > > we need to unregister it. > > > That said, we need to call mmc_power_up() in order to try to > > > communicate with the card again. At least that is what the > > > mmc_rescan() work assumes when it runs the bus_ops->detect() callback > > > at some point later in time. > > > > > > > We saw the power up in a short time from waveform when removing the card, > > So we saw mmc_sd_resume call the power up did not check the card exist. Well, that doesn't really tell me if that is a problem or not? Doesn't the first command sent to the SD card (that has been removed) fail if there is no card inserted anyway? > > > > We think this the short time power up maybe cause OCP if no card exist. OCP? > > And this power up we think is unnecessary when no card exist. It may not be necessary, but in what way does it hurt? > > > > > > > > > > > > Signed-off-by: Ricky Wu <ricky_wu@realtek.com> > > > > --- > > > > drivers/mmc/core/sd.c | 3 +++ > > > > 1 file changed, 3 insertions(+) > > > > > > > > diff --git a/drivers/mmc/core/sd.c b/drivers/mmc/core/sd.c index > > > > 1c8148cdda50..35e036672cfb 100644 > > > > --- a/drivers/mmc/core/sd.c > > > > +++ b/drivers/mmc/core/sd.c > > > > @@ -1750,6 +1750,9 @@ static int _mmc_sd_resume(struct mmc_host > > > *host) > > > > if (!mmc_card_suspended(host->card)) > > > > goto out; > > > > > > > > + if (host->ops->get_cd && !host->ops->get_cd(host)) > > > > + goto out; > > > > + > > > > mmc_power_up(host, host->card->ocr); > > > > err = mmc_sd_init_card(host, host->card->ocr, host->card); > > > > mmc_card_clr_suspended(host->card); > > > > -- Kind regards Uffe
diff --git a/drivers/mmc/core/sd.c b/drivers/mmc/core/sd.c index 1c8148cdda50..35e036672cfb 100644 --- a/drivers/mmc/core/sd.c +++ b/drivers/mmc/core/sd.c @@ -1750,6 +1750,9 @@ static int _mmc_sd_resume(struct mmc_host *host) if (!mmc_card_suspended(host->card)) goto out; + if (host->ops->get_cd && !host->ops->get_cd(host)) + goto out; + mmc_power_up(host, host->card->ocr); err = mmc_sd_init_card(host, host->card->ocr, host->card); mmc_card_clr_suspended(host->card);
_mmc_sd_resume add get_cd before call powerup, make sure the card exist Signed-off-by: Ricky Wu <ricky_wu@realtek.com> --- drivers/mmc/core/sd.c | 3 +++ 1 file changed, 3 insertions(+)