Skip to contentSkip to conversation

mmc: introduce mmc_power_init

PATCH Superseded Peng Fan 3 participants 4 messages Oct 8, 2016 – Oct 11, 2016 MMC, SD and SDIO

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Peng Fan

author
#0 lore raw

PATCH 1/1 mmc: introduce mmc_power_init

In device tree, there is vmmc-supply property for SD/MMC. Introduce mmc_power_init function to handle vmmc-supply.

mmc_power_init will first invoke board_mmc_power_init to avoid break boards which already implement board_mmc_power_init.

If DM_MMC and DM_REGULATOR is defined, the regulator will be enabled to power up the device.

Cc Jaehoon Chung <[email protected]>

---

The RFC patset thread: http://lists.denx.de/pipermail/u-boot/2016-April/251019.html
V1: Use a generic way to handle vmmc supply, but not let vendor driver
to handle it.

 drivers/mmc/mmc.c | 30 +++++++++++++++++++++++++++++-
 1 file changed, 29 insertions(+), 1 deletion(-)
drivers/mmc/mmc.c +29 -1
index 0312da9..c361098 100644
@@ -15,6 +15,7 @@
 #include <errno.h>
 #include <mmc.h>
 #include <part.h>
+#include <power/regulator.h>
 #include <malloc.h>
 #include <memalign.h>
 #include <linux/list.h>
@@ -1582,6 +1583,31 @@ __weak void board_mmc_power_init(void)
 {
 }
 
+int mmc_power_init(struct mmc *mmc)
+{
+	board_mmc_power_init();
+
+#if defined(CONFIG_DM_MMC) && defined(CONFIG_DM_REGULATOR) && \
+	!defined(CONFIG_SPL_BUILD)
+	struct udevice *vmmc_supply;
+	int ret;
+
+	ret = device_get_supply_regulator(mmc->dev, "vmmc-supply",
+					  &vmmc_supply);
+	if (ret) {
+		debug("No vmmc supply\n");
+		return 0;
+	}
+
+	ret = regulator_set_enable(vmmc_supply, true);
+	if (ret) {
+		puts("Error enabling VMMC supply\n");
+		return ret;
+	}
+#endif
+	return 0;
+}
+
 int mmc_start_init(struct mmc *mmc)
 {
 	bool no_card;
@@ -1606,7 +1632,9 @@ int mmc_start_init(struct mmc *mmc)
 #ifdef CONFIG_FSL_ESDHC_ADAPTER_IDENT
 	mmc_adapter_card_type_ident();
 #endif
-	board_mmc_power_init();
+	err = mmc_power_init(mmc);
+	if (err)
+		return err;
 
 #ifdef CONFIG_DM_MMC_OPS
 	/* The device has already been probed ready for use */
Signature · 1 line
2.6.2
#1 lore raw

PATCH 1/1 mmc: introduce mmc_power_init

Hi Peng,

Quoted 2 lines · Peng Fan

In device tree, there is vmmc-supply property for SD/MMC. Introduce mmc_power_init function to handle vmmc-supply.

As i know, vqmmc-supply should be optional. Do you have a plan to add this?

Quoted 31 lines

mmc_power_init will first invoke board_mmc_power_init to avoid break boards which already implement board_mmc_power_init.

If DM_MMC and DM_REGULATOR is defined, the regulator will be enabled to power up the device.

Cc Jaehoon Chung <[email protected]>

---

The RFC patset thread: http://lists.denx.de/pipermail/u-boot/2016-April/251019.html
V1: Use a generic way to handle vmmc supply, but not let vendor driver
to handle it.

 drivers/mmc/mmc.c | 30 +++++++++++++++++++++++++++++-
 1 file changed, 29 insertions(+), 1 deletion(-)
drivers/mmc/mmc.c +1 -0
index 0312da9..c361098 100644
@@ -15,6 +15,7 @@
 #include <errno.h>
 #include <mmc.h>
 #include <part.h>
+#include <power/regulator.h>
 #include <malloc.h>
 #include <memalign.h>
 #include <linux/list.h>

drivers/mmc/mmc.c __weak void board_mmc_power_init(void) · line 1586

Quoted 4 lines
drivers/mmc/mmc.c +1 -0
@@ -1582,6 +1583,31 @@ __weak void board_mmc_power_init(void)
 {
 }
 
+int mmc_power_init(struct mmc *mmc)

Can be static?

drivers/mmc/mmc.c __weak void board_mmc_power_init(void) · line 1599

Quoted 13 lines
drivers/mmc/mmc.c +13 -0
+{
+	board_mmc_power_init();
+
+#if defined(CONFIG_DM_MMC) && defined(CONFIG_DM_REGULATOR) && \
+	!defined(CONFIG_SPL_BUILD)
+	struct udevice *vmmc_supply;
+	int ret;
+
+	ret = device_get_supply_regulator(mmc->dev, "vmmc-supply",
+					  &vmmc_supply);
+	if (ret) {
+		debug("No vmmc supply\n");
+		return 0;

"return 0" is Right? Doesn't need to return error?

Quoted 26 lines
drivers/mmc/mmc.c +14 -1
+	}
+
+	ret = regulator_set_enable(vmmc_supply, true);
+	if (ret) {
+		puts("Error enabling VMMC supply\n");
+		return ret;
+	}
+#endif
+	return 0;
+}
+
 int mmc_start_init(struct mmc *mmc)
 {
 	bool no_card;
@@ -1606,7 +1632,9 @@ int mmc_start_init(struct mmc *mmc)
 #ifdef CONFIG_FSL_ESDHC_ADAPTER_IDENT
 	mmc_adapter_card_type_ident();
 #endif
-	board_mmc_power_init();
+	err = mmc_power_init(mmc);
+	if (err)
+		return err;
 
 #ifdef CONFIG_DM_MMC_OPS
 	/* The device has already been probed ready for use */
#2 lore raw

PATCH 1/1 mmc: introduce mmc_power_init

Hi Jaehoon,

Quoted 7 lines · Jaehoon Chung

Hi Peng,

Quoted 2 lines · Peng Fan

In device tree, there is vmmc-supply property for SD/MMC. Introduce mmc_power_init function to handle vmmc-supply.

As i know, vqmmc-supply should be optional. Do you have a plan to add this?

In the dts for my board, there is no vqmmc-supply. So I did not add this. Let me add it in V2 as the following.

"
ret = device_get_supply_regulator(mmc->dev, "vqmmc-supply",

				  &vqmmc_supply);
if (ret) {
	debug("No vqmmc supply\n");
	return 0;
}

ret = regulator_set_enable(vqmmc_supply, true);

if (ret) {
	puts("Error enabling VQMMC supply\n");
	return ret;
}

"

Quoted 35 lines
Quoted 35 lines

mmc_power_init will first invoke board_mmc_power_init to avoid break boards which already implement board_mmc_power_init.

If DM_MMC and DM_REGULATOR is defined, the regulator will be enabled to power up the device.

Cc Jaehoon Chung <[email protected]>

---

The RFC patset thread: http://lists.denx.de/pipermail/u-boot/2016-April/251019.html
V1: Use a generic way to handle vmmc supply, but not let vendor driver
to handle it.

 drivers/mmc/mmc.c | 30 +++++++++++++++++++++++++++++-
 1 file changed, 29 insertions(+), 1 deletion(-)
drivers/mmc/mmc.c +1 -0
index 0312da9..c361098 100644
@@ -15,6 +15,7 @@
 #include <errno.h>
 #include <mmc.h>
 #include <part.h>
+#include <power/regulator.h>
 #include <malloc.h>
 #include <memalign.h>
 #include <linux/list.h>
@@ -1582,6 +1583,31 @@ __weak void board_mmc_power_init(void)
 {
 }
 
Quoted 3 lines

drivers/mmc/mmc.c __weak void board_mmc_power_init(void) · line 1586

Quoted 1 line
drivers/mmc/mmc.c +1 -0
+int mmc_power_init(struct mmc *mmc)

Can be static?

Sure. Fix in V2.

Quoted 16 lines
Quoted 4 lines
drivers/mmc/mmc.c +3 -0
+{
+	board_mmc_power_init();
+

drivers/mmc/mmc.c __weak void board_mmc_power_init(void) · line 1599

Quoted 10 lines
drivers/mmc/mmc.c +10 -0
+#if defined(CONFIG_DM_MMC) && defined(CONFIG_DM_REGULATOR) && \
+	!defined(CONFIG_SPL_BUILD)
+	struct udevice *vmmc_supply;
+	int ret;
+
+	ret = device_get_supply_regulator(mmc->dev, "vmmc-supply",
+					  &vmmc_supply);
+	if (ret) {
+		debug("No vmmc supply\n");
+		return 0;

"return 0" is Right? Doesn't need to return error?

In U-Boot, not every board supports regulator now. Some boards may supports DM MMC, but DM REGULATOR not supported now. So I did not use error code here. Or use puts, but not debug?

Any comments?

Regards,
Peng.

#3 lore raw

PATCH 1/1 mmc: introduce mmc_power_init

Hi Peng,

Quoted 4 lines · Peng Fan

Hi Jaehoon,

Quoted 2 lines · Jaehoon Chung

Hi Peng,

Quoted 8 lines
Quoted 5 lines · Peng Fan
Quoted 2 lines

In device tree, there is vmmc-supply property for SD/MMC. Introduce mmc_power_init function to handle vmmc-supply.

As i know, vqmmc-supply should be optional. Do you have a plan to add this?

In the dts for my board, there is no vqmmc-supply. So I did not add this. Let me add it in V2 as the following.

Then just remain this as future work. :)
I think there is no use-case in uboot yet..

Quoted 75 lines

"
ret = device_get_supply_regulator(mmc->dev, "vqmmc-supply",

				  &vqmmc_supply);
if (ret) {
	debug("No vqmmc supply\n");
	return 0;
}

ret = regulator_set_enable(vqmmc_supply, true);

if (ret) {
	puts("Error enabling VQMMC supply\n");
	return ret;
}

"

Quoted 35 lines
Quoted 35 lines

mmc_power_init will first invoke board_mmc_power_init to avoid break boards which already implement board_mmc_power_init.

If DM_MMC and DM_REGULATOR is defined, the regulator will be enabled to power up the device.

Cc Jaehoon Chung <[email protected]>

---

The RFC patset thread: http://lists.denx.de/pipermail/u-boot/2016-April/251019.html
V1: Use a generic way to handle vmmc supply, but not let vendor driver
to handle it.

 drivers/mmc/mmc.c | 30 +++++++++++++++++++++++++++++-
 1 file changed, 29 insertions(+), 1 deletion(-)
drivers/mmc/mmc.c +1 -0
index 0312da9..c361098 100644
@@ -15,6 +15,7 @@
 #include <errno.h>
 #include <mmc.h>
 #include <part.h>
+#include <power/regulator.h>
 #include <malloc.h>
 #include <memalign.h>
 #include <linux/list.h>
@@ -1582,6 +1583,31 @@ __weak void board_mmc_power_init(void)
 {
 }
 
Quoted 3 lines

drivers/mmc/mmc.c __weak void board_mmc_power_init(void) · line 1586

Quoted 1 line
drivers/mmc/mmc.c +1 -0
+int mmc_power_init(struct mmc *mmc)

Can be static?

Sure. Fix in V2.

Quoted 16 lines
Quoted 4 lines
drivers/mmc/mmc.c +3 -0
+{
+	board_mmc_power_init();
+

drivers/mmc/mmc.c __weak void board_mmc_power_init(void) · line 1599

Quoted 10 lines
drivers/mmc/mmc.c +10 -0
+#if defined(CONFIG_DM_MMC) && defined(CONFIG_DM_REGULATOR) && \
+	!defined(CONFIG_SPL_BUILD)
+	struct udevice *vmmc_supply;
+	int ret;
+
+	ret = device_get_supply_regulator(mmc->dev, "vmmc-supply",
+					  &vmmc_supply);
+	if (ret) {
+		debug("No vmmc supply\n");
+		return 0;

"return 0" is Right? Doesn't need to return error?

Quoted 3 lines

In U-Boot, not every board supports regulator now. Some boards may supports DM MMC, but DM REGULATOR not supported now. So I did not use error code here. Or use puts, but not debug?

Ok. Nothing. When you send patch v2, I will apply on u-boot-mmc, after checking. Thanks!

Best Reagrds,
Jaehoon Chung

Quoted 8 lines

Any comments?

Regards,
Peng.