On Mon, 12 Apr 2021, at 16:20, Steven Lee wrote:

The 04/09/2021 12:14, Andrew Jeffery wrote:

quoted

Hi Steven,

On Thu, 8 Apr 2021, at 11:22, Steven Lee wrote:

quoted

AST2600-A2 EVB provides reference design to support toggling signal
voltage between 3.3v and 1.8v by power-switch-gpio pin that defined in
the device tree.

Is this something you think we need support for beyond the EVB? It 
sounds a lot like a knob to enable testing of different SD/MMC power 
configurations and not something that you'd otherwise see in a system 
design.

It can be used for testing different SD power configuration.
The main purpose of the patch is letting the driver has the ability to 
change the bus voltage for switching SD bus speed between UHS-I mode and
normal/high speed mode in AST2600-A2 EVB according to the value of
power-switch-gpio that defined in the device tree. 

Ah okay. I'm not strong on all the power requirements for each of the 
bus modes :)

But I am not sure whether it needs to be support in mainline kernel.
Since it requires a particular hardware circuit design outside of ASPEED
AST2600 SoC, and AST2600-A2 EVB does have that design.

This shouldn't be a problem with Uffe's suggestion on the binding patch.

Better to have support in mainline than maintain out-of-tree patches!

quoted

quoted

It also supporting enabling/disabling SD bus power by
power-gpio.

This sounds like it could be useful but I'll defer to Ulf with regards 
to the binding.

quoted

In the reference design, GPIOV0 of AST2600-A2 EVB is connected to power
load switch that providing 3.3v to SD1 bus vdd. GPIOV1 is connected to
a 1.8v and a 3.3v power load switch that providing signal voltage to
SD1 bus.
If GPIOV0 is active high, SD1 bus is enabled. Otherwise, SD1 bus is
disabled.
If GPIOV1 is active high, 3.3v power load switch is enabled, SD1 signal
voltage is 3.3v, otherwise, 1.8v power load switch will be enabled, SD1
signal voltage becomes 1.8v.

AST2600-A2 EVB also support toggling signal voltage for SD2 bus.
The design is the same as SD1 bus. It uses GPIOV2 as power-gpio and
GPIOV3 as power-switch-gpio.

Signed-off-by: Steven Lee <redacted>
---
 drivers/mmc/host/sdhci-of-aspeed.c | 155 +++++++++++++++++++++++++----
 1 file changed, 137 insertions(+), 18 deletions(-)

diff --git a/drivers/mmc/host/sdhci-of-aspeed.c 

b/drivers/mmc/host/sdhci-of-aspeed.c
index 7d8692e90996..a74a03d37915 100644

--- a/drivers/mmc/host/sdhci-of-aspeed.c
+++ b/drivers/mmc/host/sdhci-of-aspeed.c

@@ -5,6 +5,7 @@
 #include <linux/clk.h>
 #include <linux/delay.h>
 #include <linux/device.h>
+#include <linux/gpio/consumer.h>
 #include <linux/io.h>
 #include <linux/math64.h>
 #include <linux/mmc/host.h>

@@ -30,6 +31,7 @@
 #define   ASPEED_SDC_S0_PHASE_IN_EN	BIT(2)
 #define   ASPEED_SDC_S0_PHASE_OUT_EN	GENMASK(1, 0)
 #define   ASPEED_SDC_PHASE_MAX		31
+#define ASPEED_CLOCK_PHASE 0xf4

This isn't related to the power GPIOs, and we already have phase 
support as suggested by the macros immediately above the one you've 
added here.

Please remove it and make use of the existing mmc phase devicetree 
binding and driver support.

Will remove it.

quoted

quoted

 
 struct aspeed_sdc {
 	struct clk *clk;

@@ -58,18 +60,21 @@ struct aspeed_sdhci_phase_desc {
 	struct aspeed_sdhci_tap_desc out;
 };
 
-struct aspeed_sdhci_pdata {
+struct aspeed_sdhci_data {

Why are we renaming this? It looks like it creates a lot of noise in 
the patch. The data it captured was platform data, hence 'pdata' in the 
name. In my opinon it's fine if we also have a member called pdata.

Since I add a new member in aspeed_sdhci_pdata(In the patch, it is named
aspeed_sdhci_data) for separating the platform data of aspeed-g5 and
aspped-g6, the type of the member is sdhci_pltfm_data.
Therefore, I use the pdata for my new created member and renamed the
original pdata as data as follows.

static const struct of_device_id aspeed_sdhci_of_match[] = {
  { .compatible = "aspeed,ast2400-sdhci", .data = &ast2400_sdhci_data, },
};

static const struct aspeed_sdhci_data ast2400_sdhci_data = {
  .clk_div_start = 2,
  .pdata = &ast2400_sdhci_pdata,
};


Regardless, I will revert it to the original name and rename
my new created variable.

quoted

quoted

 	unsigned int clk_div_start;
 	const struct aspeed_sdhci_phase_desc *phase_desc;
 	size_t nr_phase_descs;
+	const struct sdhci_pltfm_data *pdata;
 };
 
 struct aspeed_sdhci {
-	const struct aspeed_sdhci_pdata *pdata;
+	const struct aspeed_sdhci_data *data;
 	struct aspeed_sdc *parent;
 	u32 width_mask;
 	struct mmc_clk_phase_map phase_map;
 	const struct aspeed_sdhci_phase_desc *phase_desc;
+	struct gpio_desc *pwr_pin;
+	struct gpio_desc *pwr_sw_pin;
 };
 
 static void aspeed_sdc_configure_8bit_mode(struct aspeed_sdc *sdc,

@@ -209,7 +214,6 @@ static void aspeed_sdhci_set_clock(struct 

sdhci_host *host, unsigned int clock)
 	sdhci = sdhci_pltfm_priv(pltfm_host);
 
 	parent = clk_get_rate(pltfm_host->clk);
-

Unrelated whitespace cleanup.

quoted

 	sdhci_writew(host, 0, SDHCI_CLOCK_CONTROL);
 
 	if (clock == 0)

@@ -234,14 +238,13 @@ static void aspeed_sdhci_set_clock(struct 

sdhci_host *host, unsigned int clock)
 	 * supporting the value 0 in (EMMC12C[7:6], EMMC12C[15:8]), and 
capture
 	 * the 0-value capability in clk_div_start.
 	 */
-	for (div = sdhci->pdata->clk_div_start; div < 256; div *= 2) {
+	for (div = sdhci->data->clk_div_start; div < 256; div *= 2) {
 		bus = parent / div;
 		if (bus <= clock)
 			break;
 	}
 
 	div >>= 1;
-

Unrelated whitespace cleanup.

quoted

 	clk = div << SDHCI_DIVIDER_SHIFT;
 
 	aspeed_sdhci_configure_phase(host, bus);

@@ -292,8 +295,78 @@ static u32 aspeed_sdhci_readl(struct sdhci_host 

*host, int reg)
 	return val;
 }
 
+static void sdhci_aspeed_set_power(struct sdhci_host *host, unsigned char mode,
+				   unsigned short vdd)
+{
+	struct sdhci_pltfm_host *pltfm_priv = sdhci_priv(host);
+	struct aspeed_sdhci *dev = sdhci_pltfm_priv(pltfm_priv);
+	u8 pwr = 0;
+
+	if (!dev->pwr_pin)
+		return sdhci_set_power(host, mode, vdd);
+
+	if (mode != MMC_POWER_OFF) {
+		switch (1 << vdd) {
+		case MMC_VDD_165_195:
+		/*
+		 * Without a regulator, SDHCI does not support 2.0v
+		 * so we only get here if the driver deliberately
+		 * added the 2.0v range to ocr_avail. Map it to 1.8v
+		 * for the purpose of turning on the power.
+		 */
+		case MMC_VDD_20_21:
+				pwr = SDHCI_POWER_180;
+				break;
+		case MMC_VDD_29_30:
+		case MMC_VDD_30_31:
+				pwr = SDHCI_POWER_300;
+				break;
+		case MMC_VDD_32_33:
+		case MMC_VDD_33_34:
+				pwr = SDHCI_POWER_330;
+				break;
+		default:
+				WARN(1, "%s: Invalid vdd %#x\n",
+				     mmc_hostname(host->mmc), vdd);
+				break;
+		}
+	}
+
+	if (host->pwr == pwr)
+		return;
+
+	host->pwr = pwr;
+
+	if (pwr == 0) {

Shouldn't we be testing against an SDHCI_POWER_* macro?

It is copied from sdhci_set_power_noreg() of sdhci.c
The difference is we need to change the bus voltage before writing
SDHCI_POWER_CONTROL.
Since There are only SDHCI_POWER_ON, SDHCI_POWER_180, SDHCI_POWER_300,
and SDHCI_POWER_330 defined in sdhci.h.
Do you mean pwr should be checked with SDHCI_POWER_ON, SDHCI_POWER_180,
SDHCI_POWER_300, and SDHCI_POWER_330?

No, I was a bit lazy and didn't look at what macros were defined.

Cheers,

Andrew