Am Dienstag, dem 11.05.2021 um 03:00 +0000 schrieb Bough Chen:

quoted

-----Original Message-----
From: Lucas Stach [mailto:l.stach@pengutronix.de]
Sent: 2021年5月11日 3:04
To: Ulf Hansson <redacted>; Adrian Hunter
[off-list ref]; Bough Chen [off-list ref]
Cc: Rob Herring <robh+dt@kernel.org>; dl-linux-imx <redacted>;
kernel@pengutronix.de; linux-mmc@vger.kernel.org;
devicetree@vger.kernel.org; linux-arm-kernel@lists.infradead.org
Subject: [PATCH v2 2/3] mmc: sdhci-esdhc-imx: advertise HS400 mode through
MMC caps

Instead of having an indirection through the SDHCI layer and emulating a
capability bit, that isn't there in hardware, do the same same thing as

with

quoted

HS400_ES and advertise the support for HS400 directly through the MMC

caps.

quoted

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
---
 drivers/mmc/host/sdhci-esdhc-imx.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/drivers/mmc/host/sdhci-esdhc-imx.c

b/drivers/mmc/host/sdhci-esdhc-imx.c
index a20459744d21..65a52586db36 100644

--- a/drivers/mmc/host/sdhci-esdhc-imx.c
+++ b/drivers/mmc/host/sdhci-esdhc-imx.c

@@ -427,9 +427,6 @@ static u32 esdhc_readl_le(struct sdhci_host *host, int

reg)
 					|

FIELD_PREP(SDHCI_RETUNING_MODE_MASK,

quoted

 						     SDHCI_TUNING_MODE_3);

-			if (imx_data->socdata->flags & ESDHC_FLAG_HS400)
-				val |= SDHCI_SUPPORT_HS400;
-
 			/*
 			 * Do not advertise faster UHS modes if there are no
 			 * pinctrl states for 100MHz/200MHz.

@@ -1603,7 +1600,7 @@ static int sdhci_esdhc_imx_probe(struct

platform_device *pdev)
 		host->quirks |= SDHCI_QUIRK_BROKEN_ADMA;

 	if (imx_data->socdata->flags & ESDHC_FLAG_HS400)
-		host->quirks2 |= SDHCI_QUIRK2_CAPS_BIT63_FOR_HS400;
+		host->mmc->caps2 |= MMC_CAP2_HS400;

Hi Lucas,

I think patch1 and patch 2 are enough to cover your requirement.
For this patch, I think it's unnecessary, sdhci-esdhc-imx.c need to reuse
sdhci.c as much as possible.
In sdhci.c, already contain the following logic. 

         if (host->quirks2 & SDHCI_QUIRK2_CAPS_BIT63_FOR_HS400 &&
             (host->caps1 & SDHCI_SUPPORT_HS400))
                 mmc->caps2 |= MMC_CAP2_HS400;

The reason why we directly use host->mmc->caps2 for HS400ES mode is that
sdhci.c do not contain the similar logic.

No, it's not enough. We call mmc_of_parse(), which clears the HS400
flags, before sdhci_setup_host() is called, which will then add the
HS400 flags again. So either I still need to evaluate the DT property
in the esdhc driver to make it return the right emulated SDHCI caps bit
for the HS400 case, or do it like in this patch.

While the way it is done here is a bit of a layering violation between
SDHCI and MMC, it still feels like the cleaner and more straight
forward solution.

Regards,
Lucas