On Tue, May 22, 2018 at 04:32:09PM +0200, Ulf Hansson wrote:
> On 18 April 2018 at 11:56, Simon Horman <horms+renesas@xxxxxxxxxxxx> wrote:
> > This adds two new HS400 tuning operations:
> > * prepare_hs400_tuning_downgrade
> > * complete_hs400_tuning
> >
> > These supplement the existing HS400 operation:
> > * prepare_hs400_tuning
> >
> > This is motivated by a requirement of Renesas SDHI for the following:
> > 1. Disabling SCC before selecting to HS if selection of HS400 has occurred.
> > This can be done in an implementation of prepare_hs400_tuning_downgrade
> > 2. Updating registers after switching to HS400
> > This can be done in an implementation of complete_hs400_tuning
> >
> > After this patch the call sequence is as follows:
> >
> > * Initial tuning
> > i. prepare_hs400_tuning
> > 2. Tuning procedure
> > 3. Select HS400
> > 4. complete_hs400_tuning
> >
> > * Retune
> > 1. prepare_hs400_tuning_downgrade
> > 2. Select HS200
> > 3. prepare_hs400_tuning
> > 4. Tuning procedure
> > 5. Select HS400
> > 6. complete_hs400_tuning
> >
> > If prepare_hs400_tuning or complete_hs400_tuning are not implemented they
> > are not called. And if neither are called the procedure is the same as
> > before this patch.
> >
> > Design consideration: In the case of Renesas SDHI it is likely that
> > prepare_hs400_tuning_downgrade and prepare_hs400_tuning could be combined
> > if the latter was called before selecting HS200 during tuning. When I say
> > likely, I believe it matches my understanding of the hardware. However, I
> > did not test this as I am entirely unsure if moving the
> > prepare_hs400_tuning call would work for other hardware that uses this
> > operation and I am in no position to test such hardware.
> >
> > Signed-off-by: Simon Horman <horms+renesas@xxxxxxxxxxxx>
> > ---
> > v4
> > * New patch
> > ---
> > drivers/mmc/core/host.c | 13 ++++++++++++-
> > drivers/mmc/core/mmc.c | 19 ++++++++++++++-----
> > include/linux/mmc/host.h | 26 +++++++++++++++++++++++++-
> > 3 files changed, 51 insertions(+), 7 deletions(-)
> >
> > diff --git a/drivers/mmc/core/host.c b/drivers/mmc/core/host.c
> > index 64b03d6eaf18..5e60df7ca11f 100644
> > --- a/drivers/mmc/core/host.c
> > +++ b/drivers/mmc/core/host.c
> > @@ -138,6 +138,10 @@ int mmc_retune(struct mmc_host *host)
> > host->doing_retune = 1;
> >
> > if (host->ios.timing == MMC_TIMING_MMC_HS400) {
> > + if (host->ops->prepare_hs400_tuning_downgrade)
> > + host->ops->prepare_hs400_tuning_downgrade(host,
> > + &host->ios);
> > +
>
> Quite a long lame for the callback, perhaps "hs400_downgrade" is sufficient?
I struggled with the names (more than any other aspect of the patch). So I
decided to go for something very descriptive in the hope you would suggest
better names. Thanks for doing so!
> Moreover, I suggest you move this new code snippet into
> mmc_hs400_to_hs200() instead.
Thanks, I also wondered about that. I think it makes a lot of sense.
And although I haven't prototyped that I believe it should work quite
nicely.
>
> > err = mmc_hs400_to_hs200(host->card);
> > if (err)
> > goto out;
> > @@ -152,8 +156,15 @@ int mmc_retune(struct mmc_host *host)
> > if (err)
> > goto out;
> >
> > - if (return_to_hs400)
> > + if (return_to_hs400) {
> > err = mmc_hs200_to_hs400(host->card);
> > + if (err)
> > + goto out;
> > +
> > + if (host->ops->complete_hs400_tuning)
> > + host->ops->complete_hs400_tuning(host, &host->ios);
>
> Perhaps rename callback to "hs400_complete"?
Will do.
> And, please move this new code into mmc_select_hs400() (which is
> called from mmc_hs200_to_hs400()), as I think it better belongs there.
Thanks, I will see about doing so.
> > + }
> > +
> > out:
> > host->doing_retune = 0;
> >
> > diff --git a/drivers/mmc/core/mmc.c b/drivers/mmc/core/mmc.c
> > index 099b327e10ca..a108a1a3e27f 100644
> > --- a/drivers/mmc/core/mmc.c
> > +++ b/drivers/mmc/core/mmc.c
> > @@ -1508,22 +1508,31 @@ static int mmc_select_timing(struct mmc_card *card)
> > static int mmc_hs200_tuning(struct mmc_card *card)
> > {
> > struct mmc_host *host = card->host;
> > + bool run_hs400_ops;
> > int err;
> >
> > + run_hs400_ops = card->mmc_avail_type & EXT_CSD_CARD_TYPE_HS400 &&
> > + host->ios.bus_width == MMC_BUS_WIDTH_8;
> > +
> > /*
> > * Timing should be adjusted to the HS400 target
> > * operation frequency for tuning process
> > */
> > - if (card->mmc_avail_type & EXT_CSD_CARD_TYPE_HS400 &&
> > - host->ios.bus_width == MMC_BUS_WIDTH_8)
> > - if (host->ops->prepare_hs400_tuning)
> > - host->ops->prepare_hs400_tuning(host, &host->ios);
> > + if (run_hs400_ops && host->ops->prepare_hs400_tuning)
> > + host->ops->prepare_hs400_tuning(host, &host->ios);
> >
> > err = mmc_execute_tuning(card);
> > if (err)
> > return err;
> >
> > - return mmc_select_hs400(card);
> > + err = mmc_select_hs400(card);
> > + if (err)
> > + return err;
> > +
> > + if (run_hs400_ops && host->ops->complete_hs400_tuning)
> > + host->ops->complete_hs400_tuning(host, &host->ios);
> > +
>
> I would suggest you to drop patch 1, then re-base $subject patch on
> top of the patch I just sent ("mmc: core: Move calls to
> ->prepare_hs400_tuning() closer to mmc code").
>
> In this way, we get less card specific code in mmc_retune(), which is
> desirable - and in the end I think the code becomes a bit more easy to
> understand.
Understood. Thanks for your guidance. I think that should work out quite
nicely.
> > + return 0;
> > }
> >
> > /*
> > diff --git a/include/linux/mmc/host.h b/include/linux/mmc/host.h
> > index 85146235231e..5d3ae1071d2f 100644
> > --- a/include/linux/mmc/host.h
> > +++ b/include/linux/mmc/host.h
> > @@ -143,8 +143,32 @@ struct mmc_host_ops {
> > /* The tuning command opcode value is different for SD and eMMC cards */
> > int (*execute_tuning)(struct mmc_host *host, u32 opcode);
> >
> > - /* Prepare HS400 target operating frequency depending host driver */
> > + /* Prepare for HS400 downgrade during tuning of target operating frequency depending on host driver
> > + * If provided and retuning is in progress, this is called before:
> > + * 1. Switching from HS400 to HS200; which preceeds
> > + * 2. Calling .prepare_hs400_tuning, if present; which preceeds
> > + * 3. The HS400 tuning procedure
> > + */
> > + void (*prepare_hs400_tuning_downgrade)(struct mmc_host *host, struct mmc_ios *ios);
> > +
> > + /* Prepare for tuning HS400 target operating frequency depending on host driver
> > + * If provided, this called:
> > + * - In the case that retuning is in progress, after:
> > + * 1. .prepare_hs400_tuning_downgrade(), if present
> > + * 2. Switching from HS400 to HS200
> > + * - And in any case before:
> > + * 3. The HS400 tuning procedure
> > + */
> > +
> > int (*prepare_hs400_tuning)(struct mmc_host *host, struct mmc_ios *ios);
> > +
> > + /* Complete tuning HS400 target operating frequency depending host driver
> > + * If provided, this is called after:
> > + * 1. The HS400 tuning procedure
> > + * 2. Switching from HS200 to HS400
> > + */
> > + void (*complete_hs400_tuning)(struct mmc_host *host, struct mmc_ios *ios);
> > +
> > /* Prepare enhanced strobe depending host driver */
> > void (*hs400_enhanced_strobe)(struct mmc_host *host,
> > struct mmc_ios *ios);
> > --
> > 2.11.0
> >
>
> Otherwise, as said, I like the approach.
Great!
