Re: [PATCH v4 4/5] mmc: sdhci: move SDHCI_QUIRK2_CLOCK_DIV_ZERO_BROKEN frequency limit

From: Adrian Hunter
Date: Thu Aug 12 2021 - 06:01:23 EST


On 7/08/21 5:12 pm, Michał Mirosław wrote:
> On Wed, Aug 04, 2021 at 03:33:56PM +0300, Adrian Hunter wrote:
>> On 25/07/21 12:20 pm, Michał Mirosław wrote:
>>> Push handling of clock frequency dependence for
>>> SDHCI_QUIRK2_CLOCK_DIV_ZERO_BROKEN quirk to the drivers that use it.
>> What is the max_clk dependency for, and why push it down?
>
> I guess this is a workaround for a hardware issue. When I wrote this,
> there was only a single user. Now I don't know if the second user got
> the limit by accident or just uses the flag not knowing it doesn't work
> as the quirk name suggests. IOW this makes it easier to fix in drivers
> if the limit is wrong or irrelevant. The dependency doesn't feel like
> it belongs to the generic driver anyway.

Would you mind reaching out to the authors of the relevant patches
and drivers to try to find out the purpose of the max_clk dependency,
before we make any changes?


>
> [...]
>>> @@ -318,6 +317,9 @@ static int dwcmshc_rk3568_init(struct sdhci_host *host, struct dwcmshc_priv *dwc
>>> sdhci_writel(host, 0, DWCMSHC_EMMC_DLL_TXCLK);
>>> sdhci_writel(host, 0, DWCMSHC_EMMC_DLL_STRBIN);
>>>
>>> + if (sdhci_pltfm_clk_get_max_clock(host) <= 25000000)
>>> + host->quirks2 |= SDHCI_QUIRK2_CLOCK_DIV_ZERO_BROKEN;
>>> +
>>> return 0;
>>> }
>>>
>>> diff --git a/drivers/mmc/host/sdhci.c b/drivers/mmc/host/sdhci.c
>>> index 0993f7d0ce8e..cfa314e659bc 100644
>>> --- a/drivers/mmc/host/sdhci.c
>>> +++ b/drivers/mmc/host/sdhci.c
>>> @@ -1905,8 +1905,7 @@ u16 sdhci_calc_clk(struct sdhci_host *host, unsigned int clock,
>>> /* Version 3.00 divisors must be a multiple of 2. */
>>> if (host->max_clk <= clock) {
>>> div = 1;
>>> - if ((host->quirks2 & SDHCI_QUIRK2_CLOCK_DIV_ZERO_BROKEN)
>>> - && host->max_clk <= 25000000)
>>> + if (host->quirks2 & SDHCI_QUIRK2_CLOCK_DIV_ZERO_BROKEN)
>>> div = 2;
>>> } else {
>>> for (div = 2; div < SDHCI_MAX_DIV_SPEC_300;