[PATCH v4 2/7] mmc: mediatek: Add Mediatek MMC driver

[Ulf Hansson]

[...]

>> You are invoking msdc_gate_clock() and msdc_ungate_clock() in a
>> balanced manner, thus hclk_enabled is redundant. Please remove it.
>
> on drv->probe(), already invoke the msdc_ungate_clock(), so, when the
> runtime pm resume invoke the msdc_ungate_clock(), the hclk already
> enabled.


That's why you invoke pm_runtime_set_active() during ->probe() when
deploying PM support in patch3. It's not an issue then.

[...]

>> I assume it's possible to gate the clock by updating a MSDC register
>> instead!? That would be prefereable since then you can leave clock
>> gating/ungating via the clk API, to be dealt from runtime PM. That
>> would also make "sclk_enabled" in the struct msdc_host redundant.
>>
>> Adopting to above, obviously requires MSDC to be able to ungate the
>> clock by also updating a MSDC register. I assume that's possible as
>> well!?
>>
> We can set the bit1 of MSDC_CFG, when this bit is 0, the bus clock was
> gated to 0 if no command or data is transmitted.
> And, from our designer, when we operate the MSDC register, we need make
> sure both HCLK and source are enabled, if source clock was disabled,
> write some MSDC registers will have no effect(eg. send CMD, without
> source clock, not only cannot send CMD, but also cannot get CMD timeout
> interrupt.)

Thanks, that answered my question. As I understand it you should be
able to adopt to my propsual.

[...]

>> > +{
>> > +       unsigned long tmo = jiffies + msecs_to_jiffies(20);
>> > +
>> > +       while ((readl(host->base + SDC_STS) & SDC_STS_CMDBUSY)
>> > +                       && time_before(jiffies, tmo))
>> > +               continue;
>> > +
>> > +       if (readl(host->base + SDC_STS) & SDC_STS_CMDBUSY) {
>> > +               dev_err(host->dev, "CMD bus busy detected\n");
>> > +               host->error |= REQ_CMD_BUSY;
>> > +               msdc_cmd_done(host, MSDC_INT_CMDTMO, mrq, cmd);
>> > +               return false;
>> > +       }
>> > +
>> > +       if (mmc_resp_type(cmd) == MMC_RSP_R1B || cmd->data) {
>> > +               /* R1B or with data, should check SDCBUSY */
>> > +               while (readl(host->base + SDC_STS) & SDC_STS_SDCBUSY)
>> > +                       cpu_relax();
>> > +       }
>>
>> MSDC seems to be handling card busy detection in HW, right?
>>
> Do not have this ability, HW only know if CMD/DAT is low, but do not
> have any interrupt for it,

I see, but doesn't the above polling mean that msdc will not propagate
the response until the card have stopped signal busy? That's what
MMC_CAP_WAIT_WHILE_BUSY shall be used for.

Perhaps you should remove the above polling, and rely on the MMC core
to poll with CMD13 instead?

>> If so, you should enable MMC_CAP_WAIT_WHILE_BUSY and set
>> "max_busy_timeout" to DAT_TIMEOUT to inform the mmc core about it.
>>

[...]

Kind regards
Uffe