[1/2] mmc: jz4740: Don't change parent clock rate for some SoCs

Message ID 20221108045300.2084671-2-lis8215@gmail.com
State New
Headers
Series mmc: jz4740: Don't change parent clock rate for some SoCs |

Commit Message

Siarhei Volkau Nov. 8, 2022, 4:52 a.m. UTC
  Some SoCs have one clock divider for all MMC units, thus changing one
affects others as well. This leads to random hangs and memory
corruptions, observed on the JZ4755 based device with two MMC slots
used at the same time.

List of SoCs affected includes: JZ4725b, JZ4755, JZ4760 and JZ4760b.
However, the MMC driver doesn't distinguish JZ4760 and JZ4770
which shall remain its behavior. For the JZ4755 is sufficient to
use JZ4725b's binding. JZ4750 is outside of the patch.

The MMC core has its own clock divisor, rather coarse but suitable well,
and it shall keep the role of tuning clock for the MMC host in that
case.

Signed-off-by: Siarhei Volkau <lis8215@gmail.com>
---
 drivers/mmc/host/jz4740_mmc.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
  

Comments

Ulf Hansson Nov. 18, 2022, 8:45 a.m. UTC | #1
On Tue, 8 Nov 2022 at 05:53, Siarhei Volkau <lis8215@gmail.com> wrote:
>
> Some SoCs have one clock divider for all MMC units, thus changing one
> affects others as well. This leads to random hangs and memory
> corruptions, observed on the JZ4755 based device with two MMC slots
> used at the same time.

Urgh, that sounds like broken HW to me.

The MMC blocks could share a parent clock (that would need a fixed
rate for it to be applied), assuming there is a separate gate/divider
available per block. But there isn't'?

>
> List of SoCs affected includes: JZ4725b, JZ4755, JZ4760 and JZ4760b.
> However, the MMC driver doesn't distinguish JZ4760 and JZ4770
> which shall remain its behavior. For the JZ4755 is sufficient to
> use JZ4725b's binding. JZ4750 is outside of the patch.
>
> The MMC core has its own clock divisor, rather coarse but suitable well,
> and it shall keep the role of tuning clock for the MMC host in that
> case.

The mmc core doesn't have a clock divisor, but it does control the bus
clock frequency through the ->set_ios() host ops. It needs to do that,
to be able to conform to the (e)MMC, SD and SDIO specifications.

Can you please try to elaborate on the above, so I can better
understand your point?

>
> Signed-off-by: Siarhei Volkau <lis8215@gmail.com>

Kind regards
Uffe

> ---
>  drivers/mmc/host/jz4740_mmc.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/mmc/host/jz4740_mmc.c b/drivers/mmc/host/jz4740_mmc.c
> index dc2db9c18..d390ff31d 100644
> --- a/drivers/mmc/host/jz4740_mmc.c
> +++ b/drivers/mmc/host/jz4740_mmc.c
> @@ -114,6 +114,7 @@ enum jz4740_mmc_version {
>         JZ_MMC_JZ4740,
>         JZ_MMC_JZ4725B,
>         JZ_MMC_JZ4760,
> +       JZ_MMC_JZ4770,
>         JZ_MMC_JZ4780,
>         JZ_MMC_X1000,
>  };
> @@ -887,7 +888,13 @@ static int jz4740_mmc_set_clock_rate(struct jz4740_mmc_host *host, int rate)
>         int real_rate;
>
>         jz4740_mmc_clock_disable(host);
> -       clk_set_rate(host->clk, host->mmc->f_max);
> +
> +       /*
> +        * Changing rate on these SoCs affects other MMC units too.
> +        * Make sure the rate is configured properly by the CGU driver.
> +        */
> +       if (host->version != JZ_MMC_JZ4725B && host->version != JZ_MMC_JZ4760)
> +               clk_set_rate(host->clk, host->mmc->f_max);
>
>         real_rate = clk_get_rate(host->clk);
>
> @@ -992,6 +999,7 @@ static const struct of_device_id jz4740_mmc_of_match[] = {
>         { .compatible = "ingenic,jz4740-mmc", .data = (void *) JZ_MMC_JZ4740 },
>         { .compatible = "ingenic,jz4725b-mmc", .data = (void *)JZ_MMC_JZ4725B },
>         { .compatible = "ingenic,jz4760-mmc", .data = (void *) JZ_MMC_JZ4760 },
> +       { .compatible = "ingenic,jz4770-mmc", .data = (void *) JZ_MMC_JZ4770 },
>         { .compatible = "ingenic,jz4775-mmc", .data = (void *) JZ_MMC_JZ4780 },
>         { .compatible = "ingenic,jz4780-mmc", .data = (void *) JZ_MMC_JZ4780 },
>         { .compatible = "ingenic,x1000-mmc", .data = (void *) JZ_MMC_X1000 },
> --
> 2.36.1
>
  
Paul Cercueil Nov. 18, 2022, 9:27 a.m. UTC | #2
Hi,

(Ingenic SoCs maintainer here)

Le ven. 18 nov. 2022 à 09:45:48 +0100, Ulf Hansson 
<ulf.hansson@linaro.org> a écrit :
> On Tue, 8 Nov 2022 at 05:53, Siarhei Volkau <lis8215@gmail.com> wrote:
>> 
>>  Some SoCs have one clock divider for all MMC units, thus changing 
>> one
>>  affects others as well. This leads to random hangs and memory
>>  corruptions, observed on the JZ4755 based device with two MMC slots
>>  used at the same time.
> 
> Urgh, that sounds like broken HW to me.
> 
> The MMC blocks could share a parent clock (that would need a fixed
> rate for it to be applied), assuming there is a separate gate/divider
> available per block. But there isn't'?

They do share a parent clock and have separate gates, and each MMC IP 
block has an internal divider for the bus frequency derived from that 
shared clock.

>> 
>>  List of SoCs affected includes: JZ4725b, JZ4755, JZ4760 and JZ4760b.
>>  However, the MMC driver doesn't distinguish JZ4760 and JZ4770
>>  which shall remain its behavior. For the JZ4755 is sufficient to
>>  use JZ4725b's binding. JZ4750 is outside of the patch.
>> 
>>  The MMC core has its own clock divisor, rather coarse but suitable 
>> well,
>>  and it shall keep the role of tuning clock for the MMC host in that
>>  case.
> 
> The mmc core doesn't have a clock divisor, but it does control the bus
> clock frequency through the ->set_ios() host ops. It needs to do that,
> to be able to conform to the (e)MMC, SD and SDIO specifications.
> 
> Can you please try to elaborate on the above, so I can better
> understand your point?

Yes, I don't really understand the patch, TBH.

The "clk_set_rate" call will only set the shared clock to the *maximum* 
clock frequency (host->mmc->f_max) which should be the exact same 
across all MMC IPs.

So it doesn't matter if it's set 3 times by 3 different instances of 
the IP, as long as they all request the same value.

Besides, I know for a fact that the mainline driver works fine on the 
JZ4760(B) and JZ4725B.

Finally... even if it was correct, this change would break 
compatibility with old Device Tree files.

Cheers,
-Paul

>> 
>>  Signed-off-by: Siarhei Volkau <lis8215@gmail.com>
> 
> Kind regards
> Uffe
> 
>>  ---
>>   drivers/mmc/host/jz4740_mmc.c | 10 +++++++++-
>>   1 file changed, 9 insertions(+), 1 deletion(-)
>> 
>>  diff --git a/drivers/mmc/host/jz4740_mmc.c 
>> b/drivers/mmc/host/jz4740_mmc.c
>>  index dc2db9c18..d390ff31d 100644
>>  --- a/drivers/mmc/host/jz4740_mmc.c
>>  +++ b/drivers/mmc/host/jz4740_mmc.c
>>  @@ -114,6 +114,7 @@ enum jz4740_mmc_version {
>>          JZ_MMC_JZ4740,
>>          JZ_MMC_JZ4725B,
>>          JZ_MMC_JZ4760,
>>  +       JZ_MMC_JZ4770,
>>          JZ_MMC_JZ4780,
>>          JZ_MMC_X1000,
>>   };
>>  @@ -887,7 +888,13 @@ static int jz4740_mmc_set_clock_rate(struct 
>> jz4740_mmc_host *host, int rate)
>>          int real_rate;
>> 
>>          jz4740_mmc_clock_disable(host);
>>  -       clk_set_rate(host->clk, host->mmc->f_max);
>>  +
>>  +       /*
>>  +        * Changing rate on these SoCs affects other MMC units too.
>>  +        * Make sure the rate is configured properly by the CGU 
>> driver.
>>  +        */
>>  +       if (host->version != JZ_MMC_JZ4725B && host->version != 
>> JZ_MMC_JZ4760)
>>  +               clk_set_rate(host->clk, host->mmc->f_max);
>> 
>>          real_rate = clk_get_rate(host->clk);
>> 
>>  @@ -992,6 +999,7 @@ static const struct of_device_id 
>> jz4740_mmc_of_match[] = {
>>          { .compatible = "ingenic,jz4740-mmc", .data = (void *) 
>> JZ_MMC_JZ4740 },
>>          { .compatible = "ingenic,jz4725b-mmc", .data = (void 
>> *)JZ_MMC_JZ4725B },
>>          { .compatible = "ingenic,jz4760-mmc", .data = (void *) 
>> JZ_MMC_JZ4760 },
>>  +       { .compatible = "ingenic,jz4770-mmc", .data = (void *) 
>> JZ_MMC_JZ4770 },
>>          { .compatible = "ingenic,jz4775-mmc", .data = (void *) 
>> JZ_MMC_JZ4780 },
>>          { .compatible = "ingenic,jz4780-mmc", .data = (void *) 
>> JZ_MMC_JZ4780 },
>>          { .compatible = "ingenic,x1000-mmc", .data = (void *) 
>> JZ_MMC_X1000 },
>>  --
>>  2.36.1
>>
  
Siarhei Volkau Nov. 18, 2022, 9:51 a.m. UTC | #3
пт, 18 нояб. 2022 г. в 12:27, Paul Cercueil <paul@crapouillou.net>:
>
> Hi,
>
> (Ingenic SoCs maintainer here)
>
> Le ven. 18 nov. 2022 à 09:45:48 +0100, Ulf Hansson
> <ulf.hansson@linaro.org> a écrit :
> > On Tue, 8 Nov 2022 at 05:53, Siarhei Volkau <lis8215@gmail.com> wrote:
> >>
> >>  Some SoCs have one clock divider for all MMC units, thus changing
> >> one
> >>  affects others as well. This leads to random hangs and memory
> >>  corruptions, observed on the JZ4755 based device with two MMC slots
> >>  used at the same time.
> >
> > Urgh, that sounds like broken HW to me.
> >
> > The MMC blocks could share a parent clock (that would need a fixed
> > rate for it to be applied), assuming there is a separate gate/divider
> > available per block. But there isn't'?
>
> They do share a parent clock and have separate gates, and each MMC IP
> block has an internal divider for the bus frequency derived from that
> shared clock.
>
> >>
> >>  List of SoCs affected includes: JZ4725b, JZ4755, JZ4760 and JZ4760b.
> >>  However, the MMC driver doesn't distinguish JZ4760 and JZ4770
> >>  which shall remain its behavior. For the JZ4755 is sufficient to
> >>  use JZ4725b's binding. JZ4750 is outside of the patch.
> >>
> >>  The MMC core has its own clock divisor, rather coarse but suitable
> >> well,
> >>  and it shall keep the role of tuning clock for the MMC host in that
> >>  case.
> >
> > The mmc core doesn't have a clock divisor, but it does control the bus
> > clock frequency through the ->set_ios() host ops. It needs to do that,
> > to be able to conform to the (e)MMC, SD and SDIO specifications.
> >
> > Can you please try to elaborate on the above, so I can better
> > understand your point?
>
> Yes, I don't really understand the patch, TBH.
>
> The "clk_set_rate" call will only set the shared clock to the *maximum*
> clock frequency (host->mmc->f_max) which should be the exact same
> across all MMC IPs.

That's the case I need different "f_max" for my HW, for some reason
internal slot can't do a full rate (48MHz) but the external can, the same
card used for checking. So I want to set 24M for mmc0, and 48M for mmc1
with respect to hardware limitation.

>
> So it doesn't matter if it's set 3 times by 3 different instances of
> the IP, as long as they all request the same value.
>
> Besides, I know for a fact that the mainline driver works fine on the
> JZ4760(B) and JZ4725B.
>
> Finally... even if it was correct, this change would break
> compatibility with old Device Tree files.
>
> Cheers,
> -Paul
>
> >>
> >>  Signed-off-by: Siarhei Volkau <lis8215@gmail.com>
> >
> > Kind regards
> > Uffe
> >
> >>  ---
> >>   drivers/mmc/host/jz4740_mmc.c | 10 +++++++++-
> >>   1 file changed, 9 insertions(+), 1 deletion(-)
> >>
> >>  diff --git a/drivers/mmc/host/jz4740_mmc.c
> >> b/drivers/mmc/host/jz4740_mmc.c
> >>  index dc2db9c18..d390ff31d 100644
> >>  --- a/drivers/mmc/host/jz4740_mmc.c
> >>  +++ b/drivers/mmc/host/jz4740_mmc.c
> >>  @@ -114,6 +114,7 @@ enum jz4740_mmc_version {
> >>          JZ_MMC_JZ4740,
> >>          JZ_MMC_JZ4725B,
> >>          JZ_MMC_JZ4760,
> >>  +       JZ_MMC_JZ4770,
> >>          JZ_MMC_JZ4780,
> >>          JZ_MMC_X1000,
> >>   };
> >>  @@ -887,7 +888,13 @@ static int jz4740_mmc_set_clock_rate(struct
> >> jz4740_mmc_host *host, int rate)
> >>          int real_rate;
> >>
> >>          jz4740_mmc_clock_disable(host);
> >>  -       clk_set_rate(host->clk, host->mmc->f_max);
> >>  +
> >>  +       /*
> >>  +        * Changing rate on these SoCs affects other MMC units too.
> >>  +        * Make sure the rate is configured properly by the CGU
> >> driver.
> >>  +        */
> >>  +       if (host->version != JZ_MMC_JZ4725B && host->version !=
> >> JZ_MMC_JZ4760)
> >>  +               clk_set_rate(host->clk, host->mmc->f_max);
> >>
> >>          real_rate = clk_get_rate(host->clk);
> >>
> >>  @@ -992,6 +999,7 @@ static const struct of_device_id
> >> jz4740_mmc_of_match[] = {
> >>          { .compatible = "ingenic,jz4740-mmc", .data = (void *)
> >> JZ_MMC_JZ4740 },
> >>          { .compatible = "ingenic,jz4725b-mmc", .data = (void
> >> *)JZ_MMC_JZ4725B },
> >>          { .compatible = "ingenic,jz4760-mmc", .data = (void *)
> >> JZ_MMC_JZ4760 },
> >>  +       { .compatible = "ingenic,jz4770-mmc", .data = (void *)
> >> JZ_MMC_JZ4770 },
> >>          { .compatible = "ingenic,jz4775-mmc", .data = (void *)
> >> JZ_MMC_JZ4780 },
> >>          { .compatible = "ingenic,jz4780-mmc", .data = (void *)
> >> JZ_MMC_JZ4780 },
> >>          { .compatible = "ingenic,x1000-mmc", .data = (void *)
> >> JZ_MMC_X1000 },
> >>  --
> >>  2.36.1
> >>
>
>
  
Ulf Hansson Nov. 18, 2022, 10:06 a.m. UTC | #4
On Fri, 18 Nov 2022 at 10:52, Siarhei Volkau <lis8215@gmail.com> wrote:
>
> пт, 18 нояб. 2022 г. в 12:27, Paul Cercueil <paul@crapouillou.net>:
> >
> > Hi,
> >
> > (Ingenic SoCs maintainer here)
> >
> > Le ven. 18 nov. 2022 à 09:45:48 +0100, Ulf Hansson
> > <ulf.hansson@linaro.org> a écrit :
> > > On Tue, 8 Nov 2022 at 05:53, Siarhei Volkau <lis8215@gmail.com> wrote:
> > >>
> > >>  Some SoCs have one clock divider for all MMC units, thus changing
> > >> one
> > >>  affects others as well. This leads to random hangs and memory
> > >>  corruptions, observed on the JZ4755 based device with two MMC slots
> > >>  used at the same time.
> > >
> > > Urgh, that sounds like broken HW to me.
> > >
> > > The MMC blocks could share a parent clock (that would need a fixed
> > > rate for it to be applied), assuming there is a separate gate/divider
> > > available per block. But there isn't'?
> >
> > They do share a parent clock and have separate gates, and each MMC IP
> > block has an internal divider for the bus frequency derived from that
> > shared clock.
> >
> > >>
> > >>  List of SoCs affected includes: JZ4725b, JZ4755, JZ4760 and JZ4760b.
> > >>  However, the MMC driver doesn't distinguish JZ4760 and JZ4770
> > >>  which shall remain its behavior. For the JZ4755 is sufficient to
> > >>  use JZ4725b's binding. JZ4750 is outside of the patch.
> > >>
> > >>  The MMC core has its own clock divisor, rather coarse but suitable
> > >> well,
> > >>  and it shall keep the role of tuning clock for the MMC host in that
> > >>  case.
> > >
> > > The mmc core doesn't have a clock divisor, but it does control the bus
> > > clock frequency through the ->set_ios() host ops. It needs to do that,
> > > to be able to conform to the (e)MMC, SD and SDIO specifications.
> > >
> > > Can you please try to elaborate on the above, so I can better
> > > understand your point?
> >
> > Yes, I don't really understand the patch, TBH.
> >
> > The "clk_set_rate" call will only set the shared clock to the *maximum*
> > clock frequency (host->mmc->f_max) which should be the exact same
> > across all MMC IPs.
>
> That's the case I need different "f_max" for my HW, for some reason
> internal slot can't do a full rate (48MHz) but the external can, the same
> card used for checking. So I want to set 24M for mmc0, and 48M for mmc1
> with respect to hardware limitation.

This sounds like a board specific problem, right?

The simple solution would be to use 24M for both hosts, but that would
unnecessarily degrade the speed for the host for the internal slot.

It sounds like we need a new DT binding to describe a capped
max-frequency for the "broken slot". And in case that is available in
the DTS, the mmc->f_max value should be overridden with it, while also
respecting the original f_max value while calling clk_set_rate().

Br
Uffe
  
Siarhei Volkau Nov. 18, 2022, 12:48 p.m. UTC | #5
пт, 18 нояб. 2022 г. в 13:06, Ulf Hansson <ulf.hansson@linaro.org>:
>
> On Fri, 18 Nov 2022 at 10:52, Siarhei Volkau <lis8215@gmail.com> wrote:
> >
> > пт, 18 нояб. 2022 г. в 12:27, Paul Cercueil <paul@crapouillou.net>:
> > >
> > > Hi,
> > >
> > > (Ingenic SoCs maintainer here)
> > >
> > > Le ven. 18 nov. 2022 à 09:45:48 +0100, Ulf Hansson
> > > <ulf.hansson@linaro.org> a écrit :
> > > > On Tue, 8 Nov 2022 at 05:53, Siarhei Volkau <lis8215@gmail.com> wrote:
> > > >>
> > > >>  Some SoCs have one clock divider for all MMC units, thus changing
> > > >> one
> > > >>  affects others as well. This leads to random hangs and memory
> > > >>  corruptions, observed on the JZ4755 based device with two MMC slots
> > > >>  used at the same time.
> > > >
> > > > Urgh, that sounds like broken HW to me.
> > > >
> > > > The MMC blocks could share a parent clock (that would need a fixed
> > > > rate for it to be applied), assuming there is a separate gate/divider
> > > > available per block. But there isn't'?
> > >
> > > They do share a parent clock and have separate gates, and each MMC IP
> > > block has an internal divider for the bus frequency derived from that
> > > shared clock.
> > >
> > > >>
> > > >>  List of SoCs affected includes: JZ4725b, JZ4755, JZ4760 and JZ4760b.
> > > >>  However, the MMC driver doesn't distinguish JZ4760 and JZ4770
> > > >>  which shall remain its behavior. For the JZ4755 is sufficient to
> > > >>  use JZ4725b's binding. JZ4750 is outside of the patch.
> > > >>
> > > >>  The MMC core has its own clock divisor, rather coarse but suitable
> > > >> well,
> > > >>  and it shall keep the role of tuning clock for the MMC host in that
> > > >>  case.
> > > >
> > > > The mmc core doesn't have a clock divisor, but it does control the bus
> > > > clock frequency through the ->set_ios() host ops. It needs to do that,
> > > > to be able to conform to the (e)MMC, SD and SDIO specifications.
> > > >
> > > > Can you please try to elaborate on the above, so I can better
> > > > understand your point?
> > >
> > > Yes, I don't really understand the patch, TBH.
> > >
> > > The "clk_set_rate" call will only set the shared clock to the *maximum*
> > > clock frequency (host->mmc->f_max) which should be the exact same
> > > across all MMC IPs.
> >
> > That's the case I need different "f_max" for my HW, for some reason
> > internal slot can't do a full rate (48MHz) but the external can, the same
> > card used for checking. So I want to set 24M for mmc0, and 48M for mmc1
> > with respect to hardware limitation.
>
> This sounds like a board specific problem, right?

Exactly.

>
> The simple solution would be to use 24M for both hosts, but that would
> unnecessarily degrade the speed for the host for the internal slot.
>
Indeed.

> It sounds like we need a new DT binding to describe a capped
> max-frequency for the "broken slot". And in case that is available in
> the DTS, the mmc->f_max value should be overridden with it, while also
> respecting the original f_max value while calling clk_set_rate().
>
> Br
> Uffe

I think it is unnecessary, "max-frequency" is enough and clear to use here,
it just needs to be tuned independently for each controller. However, the
controllers have hidden dependency on each other. Break this dependency
that's the subject of the patchset.

Siarhei
  
Paul Cercueil Nov. 18, 2022, 1:04 p.m. UTC | #6
Hi Siarhei,

Le ven. 18 nov. 2022 à 12:51:54 +0300, Siarhei Volkau 
<lis8215@gmail.com> a écrit :
> пт, 18 нояб. 2022 г. в 12:27, Paul Cercueil 
> <paul@crapouillou.net>:
>> 
>>  Hi,
>> 
>>  (Ingenic SoCs maintainer here)
>> 
>>  Le ven. 18 nov. 2022 à 09:45:48 +0100, Ulf Hansson
>>  <ulf.hansson@linaro.org> a écrit :
>>  > On Tue, 8 Nov 2022 at 05:53, Siarhei Volkau <lis8215@gmail.com> 
>> wrote:
>>  >>
>>  >>  Some SoCs have one clock divider for all MMC units, thus 
>> changing
>>  >> one
>>  >>  affects others as well. This leads to random hangs and memory
>>  >>  corruptions, observed on the JZ4755 based device with two MMC 
>> slots
>>  >>  used at the same time.
>>  >
>>  > Urgh, that sounds like broken HW to me.
>>  >
>>  > The MMC blocks could share a parent clock (that would need a fixed
>>  > rate for it to be applied), assuming there is a separate 
>> gate/divider
>>  > available per block. But there isn't'?
>> 
>>  They do share a parent clock and have separate gates, and each MMC 
>> IP
>>  block has an internal divider for the bus frequency derived from 
>> that
>>  shared clock.
>> 
>>  >>
>>  >>  List of SoCs affected includes: JZ4725b, JZ4755, JZ4760 and 
>> JZ4760b.
>>  >>  However, the MMC driver doesn't distinguish JZ4760 and JZ4770
>>  >>  which shall remain its behavior. For the JZ4755 is sufficient to
>>  >>  use JZ4725b's binding. JZ4750 is outside of the patch.
>>  >>
>>  >>  The MMC core has its own clock divisor, rather coarse but 
>> suitable
>>  >> well,
>>  >>  and it shall keep the role of tuning clock for the MMC host in 
>> that
>>  >>  case.
>>  >
>>  > The mmc core doesn't have a clock divisor, but it does control 
>> the bus
>>  > clock frequency through the ->set_ios() host ops. It needs to do 
>> that,
>>  > to be able to conform to the (e)MMC, SD and SDIO specifications.
>>  >
>>  > Can you please try to elaborate on the above, so I can better
>>  > understand your point?
>> 
>>  Yes, I don't really understand the patch, TBH.
>> 
>>  The "clk_set_rate" call will only set the shared clock to the 
>> *maximum*
>>  clock frequency (host->mmc->f_max) which should be the exact same
>>  across all MMC IPs.
> 
> That's the case I need different "f_max" for my HW, for some reason
> internal slot can't do a full rate (48MHz) but the external can, the 
> same
> card used for checking. So I want to set 24M for mmc0, and 48M for 
> mmc1
> with respect to hardware limitation.

The JZ4760B programming manual states that the controller is "fully 
compatible with the SD Memory Card Specification 2.0". In that 
specification, the bus speed is max. 25 MHz.

The programming manual also says: "In data transfer mode, the MSC 
controller can operate card with clock rate fpp (0 ~ 25Mhz)."

So the max-frequency really should be 25 MHz.

Cheers,
-Paul

>> 
>>  So it doesn't matter if it's set 3 times by 3 different instances of
>>  the IP, as long as they all request the same value.
>> 
>>  Besides, I know for a fact that the mainline driver works fine on 
>> the
>>  JZ4760(B) and JZ4725B.
>> 
>>  Finally... even if it was correct, this change would break
>>  compatibility with old Device Tree files.
>> 
>>  Cheers,
>>  -Paul
>> 
>>  >>
>>  >>  Signed-off-by: Siarhei Volkau <lis8215@gmail.com>
>>  >
>>  > Kind regards
>>  > Uffe
>>  >
>>  >>  ---
>>  >>   drivers/mmc/host/jz4740_mmc.c | 10 +++++++++-
>>  >>   1 file changed, 9 insertions(+), 1 deletion(-)
>>  >>
>>  >>  diff --git a/drivers/mmc/host/jz4740_mmc.c
>>  >> b/drivers/mmc/host/jz4740_mmc.c
>>  >>  index dc2db9c18..d390ff31d 100644
>>  >>  --- a/drivers/mmc/host/jz4740_mmc.c
>>  >>  +++ b/drivers/mmc/host/jz4740_mmc.c
>>  >>  @@ -114,6 +114,7 @@ enum jz4740_mmc_version {
>>  >>          JZ_MMC_JZ4740,
>>  >>          JZ_MMC_JZ4725B,
>>  >>          JZ_MMC_JZ4760,
>>  >>  +       JZ_MMC_JZ4770,
>>  >>          JZ_MMC_JZ4780,
>>  >>          JZ_MMC_X1000,
>>  >>   };
>>  >>  @@ -887,7 +888,13 @@ static int jz4740_mmc_set_clock_rate(struct
>>  >> jz4740_mmc_host *host, int rate)
>>  >>          int real_rate;
>>  >>
>>  >>          jz4740_mmc_clock_disable(host);
>>  >>  -       clk_set_rate(host->clk, host->mmc->f_max);
>>  >>  +
>>  >>  +       /*
>>  >>  +        * Changing rate on these SoCs affects other MMC units 
>> too.
>>  >>  +        * Make sure the rate is configured properly by the CGU
>>  >> driver.
>>  >>  +        */
>>  >>  +       if (host->version != JZ_MMC_JZ4725B && host->version !=
>>  >> JZ_MMC_JZ4760)
>>  >>  +               clk_set_rate(host->clk, host->mmc->f_max);
>>  >>
>>  >>          real_rate = clk_get_rate(host->clk);
>>  >>
>>  >>  @@ -992,6 +999,7 @@ static const struct of_device_id
>>  >> jz4740_mmc_of_match[] = {
>>  >>          { .compatible = "ingenic,jz4740-mmc", .data = (void *)
>>  >> JZ_MMC_JZ4740 },
>>  >>          { .compatible = "ingenic,jz4725b-mmc", .data = (void
>>  >> *)JZ_MMC_JZ4725B },
>>  >>          { .compatible = "ingenic,jz4760-mmc", .data = (void *)
>>  >> JZ_MMC_JZ4760 },
>>  >>  +       { .compatible = "ingenic,jz4770-mmc", .data = (void *)
>>  >> JZ_MMC_JZ4770 },
>>  >>          { .compatible = "ingenic,jz4775-mmc", .data = (void *)
>>  >> JZ_MMC_JZ4780 },
>>  >>          { .compatible = "ingenic,jz4780-mmc", .data = (void *)
>>  >> JZ_MMC_JZ4780 },
>>  >>          { .compatible = "ingenic,x1000-mmc", .data = (void *)
>>  >> JZ_MMC_X1000 },
>>  >>  --
>>  >>  2.36.1
>>  >>
>> 
>>
  
Paul Cercueil Nov. 18, 2022, 1:14 p.m. UTC | #7
Le ven. 18 nov. 2022 à 13:04:30 +0000, Paul Cercueil 
<paul@crapouillou.net> a écrit :
> Hi Siarhei,
> 
> Le ven. 18 nov. 2022 à 12:51:54 +0300, Siarhei Volkau 
> <lis8215@gmail.com> a écrit :
>> пт, 18 нояб. 2022 г. в 12:27, Paul Cercueil 
>> <paul@crapouillou.net>:
>>> 
>>>  Hi,
>>> 
>>>  (Ingenic SoCs maintainer here)
>>> 
>>>  Le ven. 18 nov. 2022 à 09:45:48 +0100, Ulf Hansson
>>>  <ulf.hansson@linaro.org> a écrit :
>>>  > On Tue, 8 Nov 2022 at 05:53, Siarhei Volkau <lis8215@gmail.com> 
>>> wrote:
>>>  >>
>>>  >>  Some SoCs have one clock divider for all MMC units, thus 
>>> changing
>>>  >> one
>>>  >>  affects others as well. This leads to random hangs and memory
>>>  >>  corruptions, observed on the JZ4755 based device with two MMC 
>>> slots
>>>  >>  used at the same time.
>>>  >
>>>  > Urgh, that sounds like broken HW to me.
>>>  >
>>>  > The MMC blocks could share a parent clock (that would need a 
>>> fixed
>>>  > rate for it to be applied), assuming there is a separate 
>>> gate/divider
>>>  > available per block. But there isn't'?
>>> 
>>>  They do share a parent clock and have separate gates, and each MMC 
>>> IP
>>>  block has an internal divider for the bus frequency derived from 
>>> that
>>>  shared clock.
>>> 
>>>  >>
>>>  >>  List of SoCs affected includes: JZ4725b, JZ4755, JZ4760 and 
>>> JZ4760b.
>>>  >>  However, the MMC driver doesn't distinguish JZ4760 and JZ4770
>>>  >>  which shall remain its behavior. For the JZ4755 is sufficient 
>>> to
>>>  >>  use JZ4725b's binding. JZ4750 is outside of the patch.
>>>  >>
>>>  >>  The MMC core has its own clock divisor, rather coarse but 
>>> suitable
>>>  >> well,
>>>  >>  and it shall keep the role of tuning clock for the MMC host in 
>>> that
>>>  >>  case.
>>>  >
>>>  > The mmc core doesn't have a clock divisor, but it does control 
>>> the bus
>>>  > clock frequency through the ->set_ios() host ops. It needs to do 
>>> that,
>>>  > to be able to conform to the (e)MMC, SD and SDIO specifications.
>>>  >
>>>  > Can you please try to elaborate on the above, so I can better
>>>  > understand your point?
>>> 
>>>  Yes, I don't really understand the patch, TBH.
>>> 
>>>  The "clk_set_rate" call will only set the shared clock to the 
>>> *maximum*
>>>  clock frequency (host->mmc->f_max) which should be the exact same
>>>  across all MMC IPs.
>> 
>> That's the case I need different "f_max" for my HW, for some reason
>> internal slot can't do a full rate (48MHz) but the external can, the 
>> same
>> card used for checking. So I want to set 24M for mmc0, and 48M for 
>> mmc1
>> with respect to hardware limitation.
> 
> The JZ4760B programming manual states that the controller is "fully 
> compatible with the SD Memory Card Specification 2.0". In that 
> specification, the bus speed is max. 25 MHz.
> 
> The programming manual also says: "In data transfer mode, the MSC 
> controller can operate card with clock rate fpp (0 ~ 25Mhz)."
> 
> So the max-frequency really should be 25 MHz.

Nevermind. I read wrong, at least the SD spec. (the quote of the 
programming manual is still concerning though).

It's rated for 25 MB/s, so 50 MHz on 4 lanes.

Cheers,
-Paul

> 
>>> 
>>>  So it doesn't matter if it's set 3 times by 3 different instances 
>>> of
>>>  the IP, as long as they all request the same value.
>>> 
>>>  Besides, I know for a fact that the mainline driver works fine on 
>>> the
>>>  JZ4760(B) and JZ4725B.
>>> 
>>>  Finally... even if it was correct, this change would break
>>>  compatibility with old Device Tree files.
>>> 
>>>  Cheers,
>>>  -Paul
>>> 
>>>  >>
>>>  >>  Signed-off-by: Siarhei Volkau <lis8215@gmail.com>
>>>  >
>>>  > Kind regards
>>>  > Uffe
>>>  >
>>>  >>  ---
>>>  >>   drivers/mmc/host/jz4740_mmc.c | 10 +++++++++-
>>>  >>   1 file changed, 9 insertions(+), 1 deletion(-)
>>>  >>
>>>  >>  diff --git a/drivers/mmc/host/jz4740_mmc.c
>>>  >> b/drivers/mmc/host/jz4740_mmc.c
>>>  >>  index dc2db9c18..d390ff31d 100644
>>>  >>  --- a/drivers/mmc/host/jz4740_mmc.c
>>>  >>  +++ b/drivers/mmc/host/jz4740_mmc.c
>>>  >>  @@ -114,6 +114,7 @@ enum jz4740_mmc_version {
>>>  >>          JZ_MMC_JZ4740,
>>>  >>          JZ_MMC_JZ4725B,
>>>  >>          JZ_MMC_JZ4760,
>>>  >>  +       JZ_MMC_JZ4770,
>>>  >>          JZ_MMC_JZ4780,
>>>  >>          JZ_MMC_X1000,
>>>  >>   };
>>>  >>  @@ -887,7 +888,13 @@ static int 
>>> jz4740_mmc_set_clock_rate(struct
>>>  >> jz4740_mmc_host *host, int rate)
>>>  >>          int real_rate;
>>>  >>
>>>  >>          jz4740_mmc_clock_disable(host);
>>>  >>  -       clk_set_rate(host->clk, host->mmc->f_max);
>>>  >>  +
>>>  >>  +       /*
>>>  >>  +        * Changing rate on these SoCs affects other MMC units 
>>> too.
>>>  >>  +        * Make sure the rate is configured properly by the CGU
>>>  >> driver.
>>>  >>  +        */
>>>  >>  +       if (host->version != JZ_MMC_JZ4725B && host->version !=
>>>  >> JZ_MMC_JZ4760)
>>>  >>  +               clk_set_rate(host->clk, host->mmc->f_max);
>>>  >>
>>>  >>          real_rate = clk_get_rate(host->clk);
>>>  >>
>>>  >>  @@ -992,6 +999,7 @@ static const struct of_device_id
>>>  >> jz4740_mmc_of_match[] = {
>>>  >>          { .compatible = "ingenic,jz4740-mmc", .data = (void *)
>>>  >> JZ_MMC_JZ4740 },
>>>  >>          { .compatible = "ingenic,jz4725b-mmc", .data = (void
>>>  >> *)JZ_MMC_JZ4725B },
>>>  >>          { .compatible = "ingenic,jz4760-mmc", .data = (void *)
>>>  >> JZ_MMC_JZ4760 },
>>>  >>  +       { .compatible = "ingenic,jz4770-mmc", .data = (void *)
>>>  >> JZ_MMC_JZ4770 },
>>>  >>          { .compatible = "ingenic,jz4775-mmc", .data = (void *)
>>>  >> JZ_MMC_JZ4780 },
>>>  >>          { .compatible = "ingenic,jz4780-mmc", .data = (void *)
>>>  >> JZ_MMC_JZ4780 },
>>>  >>          { .compatible = "ingenic,x1000-mmc", .data = (void *)
>>>  >> JZ_MMC_X1000 },
>>>  >>  --
>>>  >>  2.36.1
>>>  >>
>>> 
>>> 
>
  
Aidan MacDonald Nov. 18, 2022, 2:18 p.m. UTC | #8
Paul Cercueil <paul@crapouillou.net> writes:

> Hi,
>
> (Ingenic SoCs maintainer here)
>
> Le ven. 18 nov. 2022 à 09:45:48 +0100, Ulf Hansson <ulf.hansson@linaro.org> a
> écrit :
>> On Tue, 8 Nov 2022 at 05:53, Siarhei Volkau <lis8215@gmail.com> wrote:
>>>  Some SoCs have one clock divider for all MMC units, thus changing one
>>>  affects others as well. This leads to random hangs and memory
>>>  corruptions, observed on the JZ4755 based device with two MMC slots
>>>  used at the same time.
>> Urgh, that sounds like broken HW to me.
>> The MMC blocks could share a parent clock (that would need a fixed
>> rate for it to be applied), assuming there is a separate gate/divider
>> available per block. But there isn't'?
>
> They do share a parent clock and have separate gates, and each MMC IP block has
> an internal divider for the bus frequency derived from that shared clock.
>
>>>  List of SoCs affected includes: JZ4725b, JZ4755, JZ4760 and JZ4760b.
>>>  However, the MMC driver doesn't distinguish JZ4760 and JZ4770
>>>  which shall remain its behavior. For the JZ4755 is sufficient to
>>>  use JZ4725b's binding. JZ4750 is outside of the patch.
>>>  The MMC core has its own clock divisor, rather coarse but suitable well,
>>>  and it shall keep the role of tuning clock for the MMC host in that
>>>  case.
>> The mmc core doesn't have a clock divisor, but it does control the bus
>> clock frequency through the ->set_ios() host ops. It needs to do that,
>> to be able to conform to the (e)MMC, SD and SDIO specifications.
>> Can you please try to elaborate on the above, so I can better
>> understand your point?
>
> Yes, I don't really understand the patch, TBH.
>
> The "clk_set_rate" call will only set the shared clock to the *maximum* clock
> frequency (host->mmc->f_max) which should be the exact same across all MMC IPs.
>
> So it doesn't matter if it's set 3 times by 3 different instances of the IP, as
> long as they all request the same value.

Isn't the fact that 3 separate driver instances call clk_set_rate()
on a shared clock kind of... problematic? Nor is it documented that
all controllers need to use the same max-frequency.

Given it's a shared clock which can't realistically be controlled by
any of the consumers, I don't think it's unreasonable to assign the
clock frequency from the DT, but it needs to be backward-compatible
with old DTs.

On another note, shouldn't the MMC mux parent be assigned from the DT?
JZ4760 (and most other Ingenic SoCs) have multiple choices and I don't
see the parent being assigned anywhere.

Regards,
Aidan

>
> Besides, I know for a fact that the mainline driver works fine on the JZ4760(B)
> and JZ4725B.
>
> Finally... even if it was correct, this change would break compatibility with
> old Device Tree files.
>
> Cheers,
> -Paul
>
>>>  Signed-off-by: Siarhei Volkau <lis8215@gmail.com>
>> Kind regards
>> Uffe
>>
>>>  ---
>>>   drivers/mmc/host/jz4740_mmc.c | 10 +++++++++-
>>>   1 file changed, 9 insertions(+), 1 deletion(-)
>>>  diff --git a/drivers/mmc/host/jz4740_mmc.c b/drivers/mmc/host/jz4740_mmc.c
>>>  index dc2db9c18..d390ff31d 100644
>>>  --- a/drivers/mmc/host/jz4740_mmc.c
>>>  +++ b/drivers/mmc/host/jz4740_mmc.c
>>>  @@ -114,6 +114,7 @@ enum jz4740_mmc_version {
>>>          JZ_MMC_JZ4740,
>>>          JZ_MMC_JZ4725B,
>>>          JZ_MMC_JZ4760,
>>>  +       JZ_MMC_JZ4770,
>>>          JZ_MMC_JZ4780,
>>>          JZ_MMC_X1000,
>>>   };
>>>  @@ -887,7 +888,13 @@ static int jz4740_mmc_set_clock_rate(struct
>>> jz4740_mmc_host *host, int rate)
>>>          int real_rate;
>>>          jz4740_mmc_clock_disable(host);
>>>  -       clk_set_rate(host->clk, host->mmc->f_max);
>>>  +
>>>  +       /*
>>>  +        * Changing rate on these SoCs affects other MMC units too.
>>>  +        * Make sure the rate is configured properly by the CGU driver.
>>>  +        */
>>>  +       if (host->version != JZ_MMC_JZ4725B && host->version !=
>>> JZ_MMC_JZ4760)
>>>  +               clk_set_rate(host->clk, host->mmc->f_max);
>>>          real_rate = clk_get_rate(host->clk);
>>>  @@ -992,6 +999,7 @@ static const struct of_device_id jz4740_mmc_of_match[]
>>> = {
>>>          { .compatible = "ingenic,jz4740-mmc", .data = (void *)
>>> JZ_MMC_JZ4740 },
>>>          { .compatible = "ingenic,jz4725b-mmc", .data = (void
>>> *)JZ_MMC_JZ4725B },
>>>          { .compatible = "ingenic,jz4760-mmc", .data = (void *)
>>> JZ_MMC_JZ4760 },
>>>  +       { .compatible = "ingenic,jz4770-mmc", .data = (void *)
>>> JZ_MMC_JZ4770 },
>>>          { .compatible = "ingenic,jz4775-mmc", .data = (void *)
>>> JZ_MMC_JZ4780 },
>>>          { .compatible = "ingenic,jz4780-mmc", .data = (void *)
>>> JZ_MMC_JZ4780 },
>>>          { .compatible = "ingenic,x1000-mmc", .data = (void *) JZ_MMC_X1000
>>> },
>>>  --
>>>  2.36.1
>>>
  

Patch

diff --git a/drivers/mmc/host/jz4740_mmc.c b/drivers/mmc/host/jz4740_mmc.c
index dc2db9c18..d390ff31d 100644
--- a/drivers/mmc/host/jz4740_mmc.c
+++ b/drivers/mmc/host/jz4740_mmc.c
@@ -114,6 +114,7 @@  enum jz4740_mmc_version {
 	JZ_MMC_JZ4740,
 	JZ_MMC_JZ4725B,
 	JZ_MMC_JZ4760,
+	JZ_MMC_JZ4770,
 	JZ_MMC_JZ4780,
 	JZ_MMC_X1000,
 };
@@ -887,7 +888,13 @@  static int jz4740_mmc_set_clock_rate(struct jz4740_mmc_host *host, int rate)
 	int real_rate;
 
 	jz4740_mmc_clock_disable(host);
-	clk_set_rate(host->clk, host->mmc->f_max);
+
+	/*
+	 * Changing rate on these SoCs affects other MMC units too.
+	 * Make sure the rate is configured properly by the CGU driver.
+	 */
+	if (host->version != JZ_MMC_JZ4725B && host->version != JZ_MMC_JZ4760)
+		clk_set_rate(host->clk, host->mmc->f_max);
 
 	real_rate = clk_get_rate(host->clk);
 
@@ -992,6 +999,7 @@  static const struct of_device_id jz4740_mmc_of_match[] = {
 	{ .compatible = "ingenic,jz4740-mmc", .data = (void *) JZ_MMC_JZ4740 },
 	{ .compatible = "ingenic,jz4725b-mmc", .data = (void *)JZ_MMC_JZ4725B },
 	{ .compatible = "ingenic,jz4760-mmc", .data = (void *) JZ_MMC_JZ4760 },
+	{ .compatible = "ingenic,jz4770-mmc", .data = (void *) JZ_MMC_JZ4770 },
 	{ .compatible = "ingenic,jz4775-mmc", .data = (void *) JZ_MMC_JZ4780 },
 	{ .compatible = "ingenic,jz4780-mmc", .data = (void *) JZ_MMC_JZ4780 },
 	{ .compatible = "ingenic,x1000-mmc", .data = (void *) JZ_MMC_X1000 },