Skip to contentSkip to conversation

mmc: omap_hsmmc: Reduce the footprint of the driver and fix am335x clock

PATCH New Jean-Jacques Hiblot 4 participants 17 messages Feb 23, 2018 – Feb 26, 2018

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

v2 cover mmc: omap_hsmmc: Reduce the footprint of the driver and fix am335x clock

This series aims at reducing the footprint of the omap_hsmmc driver in the SPL (1.5 kB gain in the case of the SPL for the am335x evm). It also fixes an issue with the am335x_evm by setting a default maximum frequency if none is defined in the dts.

tested on am335x_evm, bbb, bbb (vboot), and dra7 evm

Changes in v2:
- Use a Kconfig option to compile out the ADMA support instead of relying on the platform architecture

Jean-Jacques Hiblot (4):
mmc: omap_hsmmc: do not embed struct mmc in struct omap_hsmmc_plat
mmc: omap_hsmmc: compile out write support if not needed
mmc: omap_hsmmc: make it possible to compile out ADMA support
mmc: omap_hsmmc: use a default 52MHz max clock rate if none is
specified

 arch/arm/include/asm/omap_mmc.h |  2 +-
 drivers/mmc/Kconfig             |  9 +++++++++
 drivers/mmc/omap_hsmmc.c        | 31 ++++++++++++++++++++-----------
 3 files changed, 30 insertions(+), 12 deletions(-)
Signature · 1 line
1.9.1
#1 lore raw

PATCH v2 1/4 mmc: omap_hsmmc: do not embed struct mmc in struct omap_hsmmc_plat

The area for struct mmc can be allocated dynamically. It greatly reduces the size of struct omap_hsmmc_plat. This is useful in cases where the board level code declares one or two struct omap_hsmmc_plat because it doesn't use the Driver Model.

This saves around 740 bytes for the am335x_evm SPL.

---

Changes in v2: None

 arch/arm/include/asm/omap_mmc.h | 2 +-
 drivers/mmc/omap_hsmmc.c        | 6 +++---
 2 files changed, 4 insertions(+), 4 deletions(-)
arch/arm/include/asm/omap_mmc.h +1 -1
index 3d70148..42ce8dc 100644
@@ -67,7 +67,7 @@ struct hsmmc {
 struct omap_hsmmc_plat {
 	struct mmc_config cfg;
 	struct hsmmc *base_addr;
-	struct mmc mmc;
+	struct mmc *mmc;
 	bool cd_inverted;
 	u32 controller_flags;
 	const char *hw_rev;
drivers/mmc/omap_hsmmc.c +3 -3
index 02970f2..e0b679a 100644
@@ -1858,8 +1858,8 @@ static int omap_hsmmc_ofdata_to_platdata(struct udevice *dev)
 static int omap_hsmmc_bind(struct udevice *dev)
 {
 	struct omap_hsmmc_plat *plat = dev_get_platdata(dev);
-
-	return mmc_bind(dev, &plat->mmc, &plat->cfg);
+	plat->mmc = calloc(1, sizeof(struct mmc));
+	return mmc_bind(dev, plat->mmc, &plat->cfg);
 }
 #endif
 static int omap_hsmmc_probe(struct udevice *dev)
@@ -1882,7 +1882,7 @@ static int omap_hsmmc_probe(struct udevice *dev)
 #endif
 
 #ifdef CONFIG_BLK
-	mmc = &plat->mmc;
+	mmc = plat->mmc;
 #else
 	mmc = mmc_create(cfg, priv);
 	if (mmc == NULL)
Signature · 1 line
1.9.1
#2 lore raw

PATCH v2 1/4 mmc: omap_hsmmc: do not embed struct mmc in struct omap_hsmmc_plat

Quoted 8 lines · Jean-Jacques Hiblot

The area for struct mmc can be allocated dynamically. It greatly reduces the size of struct omap_hsmmc_plat. This is useful in cases where the board level code declares one or two struct omap_hsmmc_plat because it doesn't use the Driver Model.

This saves around 740 bytes for the am335x_evm SPL.

Signature · 8 lines
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20180223/a24ff617/attachment.sig>
#3 lore raw

PATCH v2 1/4 mmc: omap_hsmmc: do not embed struct mmc in struct omap_hsmmc_plat

Quoted 7 lines · Tom Rini
Quoted 5 lines · Jean-Jacques Hiblot

The area for struct mmc can be allocated dynamically. It greatly reduces the size of struct omap_hsmmc_plat. This is useful in cases where the board level code declares one or two struct omap_hsmmc_plat because it doesn't use the Driver Model.

Quoted 5 lines
Quoted 3 lines

This saves around 740 bytes for the am335x_evm SPL.

Applied to u-boot-mmc. Thanks!

Best Regards,
Jaehoon Chung

Quoted 1 line
#4 lore raw

PATCH v2 2/4 mmc: omap_hsmmc: compile out write support if not needed

In reply to Jean-Jacques Hiblot

This reduces the size of the binary by about 196 bytes.

---

Changes in v2: None

 drivers/mmc/omap_hsmmc.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)
drivers/mmc/omap_hsmmc.c +9 -2
index e0b679a..8b57edc 100644
@@ -1181,8 +1181,9 @@ static int mmc_read_data(struct hsmmc *mmc_base, char *buf, unsigned int size)
 	return 0;
 }
 
+#if CONFIG_IS_ENABLED(MMC_WRITE)
 static int mmc_write_data(struct hsmmc *mmc_base, const char *buf,
-				unsigned int size)
+			  unsigned int size)
 {
 	unsigned int *input_buf = (unsigned int *)buf;
 	unsigned int mmc_stat;
@@ -1235,7 +1236,13 @@ static int mmc_write_data(struct hsmmc *mmc_base, const char *buf,
 	}
 	return 0;
 }
-
+#else
+static int mmc_write_data(struct hsmmc *mmc_base, const char *buf,
+			  unsigned int size)
+{
+	return -ENOTSUPP;
+}
+#endif
 static void omap_hsmmc_stop_clock(struct hsmmc *mmc_base)
 {
 	writel(readl(&mmc_base->sysctl) & ~CEN_ENABLE, &mmc_base->sysctl);
Signature · 1 line
1.9.1
#5 lore raw

PATCH v2 2/4 mmc: omap_hsmmc: compile out write support if not needed

Quoted 3 lines · Jean-Jacques Hiblot

This reduces the size of the binary by about 196 bytes.

Signature · 8 lines
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20180223/f0daad84/attachment.sig>
#7 lore raw

PATCH v2 3/4 mmc: omap_hsmmc: make it possible to compile out ADMA support

In reply to Jean-Jacques Hiblot

Some platforms don't have ADMA controllers. For those platforms, compiling it out reduces the size of the binary by about 600 bytes. Leaving the support in doesn't break things as the driver checks at runtime if the ADMA2 controller is present.

---

Changes in v2:
- Use a Kconfig option to compile out the ADMA support instead of relying on the platform architecture

 drivers/mmc/Kconfig      |  9 +++++++++
 drivers/mmc/omap_hsmmc.c | 12 ++++++------
 2 files changed, 15 insertions(+), 6 deletions(-)
drivers/mmc/Kconfig +9 -0
index f2d8256..88a1359 100644
@@ -239,6 +239,15 @@ config MMC_OMAP_HS
 
 	  If unsure, say N.
 
+config MMC_OMAP_HS_ADMA
+	bool "ADMA support for OMAP HS MMC"
+	depends on MMC_OMAP_HS && !OMAP34XX
+	default y if !AM33XX
+	help
+	  This enables support for the ADMA2 controller (SDA3.00 Part A2 DMA
+	  controller). If supported by the hardware, selecting this option will
+	  increase performances.
+
 config MMC_OMAP36XX_PINS
 	bool "Enable MMC1 on OMAP36xx/37xx"
 	depends on OMAP34XX && MMC_OMAP_HS
drivers/mmc/omap_hsmmc.c +6 -6
index 8b57edc..3d2836d 100644
@@ -93,7 +93,7 @@ struct omap_hsmmc_data {
 	enum bus_mode mode;
 #endif
 	u8 controller_flags;
-#ifndef CONFIG_OMAP34XX
+#ifdef CONFIG_MMC_OMAP_HS_ADMA
 	struct omap_hsmmc_adma_desc *adma_desc_table;
 	uint desc_slot;
 #endif
@@ -117,7 +117,7 @@ struct omap_mmc_of_data {
 	u8 controller_flags;
 };
 
-#ifndef CONFIG_OMAP34XX
+#ifdef CONFIG_MMC_OMAP_HS_ADMA
 struct omap_hsmmc_adma_desc {
 	u8 attr;
 	u8 reserved;
@@ -741,7 +741,7 @@ static int omap_hsmmc_init_setup(struct mmc *mmc)
 			return -ETIMEDOUT;
 		}
 	}
-#ifndef CONFIG_OMAP34XX
+#ifdef CONFIG_MMC_OMAP_HS_ADMA
 	reg_val = readl(&mmc_base->hl_hwinfo);
 	if (reg_val & MADMA_EN)
 		priv->controller_flags |= OMAP_HSMMC_USE_ADMA;
@@ -834,7 +834,7 @@ static void mmc_reset_controller_fsm(struct hsmmc *mmc_base, u32 bit)
 	}
 }
 
-#ifndef CONFIG_OMAP34XX
+#ifdef CONFIG_MMC_OMAP_HS_ADMA
 static void omap_hsmmc_adma_desc(struct mmc *mmc, char *buf, u16 len, bool end)
 {
 	struct omap_hsmmc_data *priv = omap_hsmmc_get_data(mmc);
@@ -1037,7 +1037,7 @@ static int omap_hsmmc_send_cmd(struct udevice *dev, struct mmc_cmd *cmd,
 		else
 			flags |= (DP_DATA | DDIR_WRITE);
 
-#ifndef CONFIG_OMAP34XX
+#ifdef CONFIG_MMC_OMAP_HS_ADMA
 		if ((priv->controller_flags & OMAP_HSMMC_USE_ADMA) &&
 		    !mmc_is_tuning_cmd(cmd->cmdidx)) {
 			omap_hsmmc_prepare_data(mmc, data);
@@ -1082,7 +1082,7 @@ static int omap_hsmmc_send_cmd(struct udevice *dev, struct mmc_cmd *cmd,
 		}
 	}
 
-#ifndef CONFIG_OMAP34XX
+#ifdef CONFIG_MMC_OMAP_HS_ADMA
 	if ((priv->controller_flags & OMAP_HSMMC_USE_ADMA) && data &&
 	    !mmc_is_tuning_cmd(cmd->cmdidx)) {
 		u32 sz_mb, timeout;
Signature · 1 line
1.9.1
#8 lore raw

PATCH v2 3/4 mmc: omap_hsmmc: make it possible to compile out ADMA support

Quoted 22 lines · Jean-Jacques Hiblot

Some platforms don't have ADMA controllers. For those platforms, compiling it out reduces the size of the binary by about 600 bytes. Leaving the support in doesn't break things as the driver checks at runtime if the ADMA2 controller is present.

---

Changes in v2:
- Use a Kconfig option to compile out the ADMA support instead of relying on the platform architecture

 drivers/mmc/Kconfig      |  9 +++++++++
 drivers/mmc/omap_hsmmc.c | 12 ++++++------
 2 files changed, 15 insertions(+), 6 deletions(-)
drivers/mmc/Kconfig +0 -0
index f2d8256..88a1359 100644

drivers/mmc/Kconfig config MMC_OMAP_HS · line 245

Quoted 7 lines
drivers/mmc/Kconfig +4 -0
@@ -239,6 +239,15 @@ config MMC_OMAP_HS
 
 	  If unsure, say N.
 
+config MMC_OMAP_HS_ADMA
+	bool "ADMA support for OMAP HS MMC"
+	depends on MMC_OMAP_HS && !OMAP34XX
+	default y if !AM33XX

Is this logic really the right restrictions? Does it work on some AM33XX platforms (AM43xx?) and just never OMAP34XX? Thanks!

Signature · 8 lines
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20180223/29571878/attachment.sig>
#9 lore raw

PATCH v2 3/4 mmc: omap_hsmmc: make it possible to compile out ADMA support

Quoted 24 lines · Tom Rini
Quoted 23 lines · Jean-Jacques Hiblot

Some platforms don't have ADMA controllers. For those platforms, compiling it out reduces the size of the binary by about 600 bytes. Leaving the support in doesn't break things as the driver checks at runtime if the ADMA2 controller is present.

---

Changes in v2:
- Use a Kconfig option to compile out the ADMA support instead of relying on the platform architecture

  drivers/mmc/Kconfig      |  9 +++++++++
  drivers/mmc/omap_hsmmc.c | 12 ++++++------
  2 files changed, 15 insertions(+), 6 deletions(-)
drivers/mmc/Kconfig +0 -0
index f2d8256..88a1359 100644
@@ -239,6 +239,15 @@ config MMC_OMAP_HS
 
Quoted 8 lines

drivers/mmc/Kconfig config MMC_OMAP_HS · line 245

Quoted 6 lines
drivers/mmc/Kconfig +4 -0
  	  If unsure, say N.
 
+config MMC_OMAP_HS_ADMA
+	bool "ADMA support for OMAP HS MMC"
+	depends on MMC_OMAP_HS && !OMAP34XX
+	default y if !AM33XX

Is this logic really the right restrictions? Does it work on some AM33XX platforms (AM43xx?) and just never OMAP34XX? Thanks!

Starting with platform am33xx there is a way to check if the ADMA controller is present or not by reading a register. This register is not available in OMAP34XX.
That being said I don't think that ADMA is available on any variant of am33xx.

JJ

Quoted 1 line
#10 lore raw

PATCH v2 3/4 mmc: omap_hsmmc: make it possible to compile out ADMA support

Quoted 33 lines · Jean-Jacques Hiblot
Quoted 18 lines · Tom Rini
Quoted 17 lines · Jean-Jacques Hiblot

Some platforms don't have ADMA controllers. For those platforms, compiling it out reduces the size of the binary by about 600 bytes. Leaving the support in doesn't break things as the driver checks at runtime if the ADMA2 controller is present.

---

Changes in v2:
- Use a Kconfig option to compile out the ADMA support instead of relying on the platform architecture

 drivers/mmc/Kconfig      |  9 +++++++++
 drivers/mmc/omap_hsmmc.c | 12 ++++++------
 2 files changed, 15 insertions(+), 6 deletions(-)
Quoted 12 lines

drivers/mmc/Kconfig config MMC_OMAP_HS · line 243

Quoted 10 lines
drivers/mmc/Kconfig +4 -0
index f2d8256..88a1359 100644
@@ -239,6 +239,15 @@ config MMC_OMAP_HS
 	  If unsure, say N.
+config MMC_OMAP_HS_ADMA
+	bool "ADMA support for OMAP HS MMC"
+	depends on MMC_OMAP_HS && !OMAP34XX
+	default y if !AM33XX

Is this logic really the right restrictions? Does it work on some AM33XX platforms (AM43xx?) and just never OMAP34XX? Thanks!

Quoted 5 lines

Starting with platform am33xx there is a way to check if the ADMA controller is present or not by reading a register. This register is not available in OMAP34XX.
That being said I don't think that ADMA is available on any variant of am33xx.

OK, thanks!

Signature · 8 lines
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20180223/5ed192e7/attachment.sig>
#11 lore raw

PATCH v2 3/4 mmc: omap_hsmmc: make it possible to compile out ADMA support

Quoted 40 lines · Tom Rini
Quoted 33 lines · Jean-Jacques Hiblot
Quoted 18 lines · Tom Rini
Quoted 17 lines · Jean-Jacques Hiblot

Some platforms don't have ADMA controllers. For those platforms, compiling it out reduces the size of the binary by about 600 bytes. Leaving the support in doesn't break things as the driver checks at runtime if the ADMA2 controller is present.

---

Changes in v2:
- Use a Kconfig option to compile out the ADMA support instead of relying on the platform architecture

 drivers/mmc/Kconfig      |  9 +++++++++
 drivers/mmc/omap_hsmmc.c | 12 ++++++------
 2 files changed, 15 insertions(+), 6 deletions(-)
Quoted 12 lines

drivers/mmc/Kconfig config MMC_OMAP_HS · line 243

Quoted 10 lines
drivers/mmc/Kconfig +4 -0
index f2d8256..88a1359 100644
@@ -239,6 +239,15 @@ config MMC_OMAP_HS
 	  If unsure, say N.
+config MMC_OMAP_HS_ADMA
+	bool "ADMA support for OMAP HS MMC"
+	depends on MMC_OMAP_HS && !OMAP34XX
+	default y if !AM33XX

Is this logic really the right restrictions? Does it work on some AM33XX platforms (AM43xx?) and just never OMAP34XX? Thanks!

Quoted 5 lines

Starting with platform am33xx there is a way to check if the ADMA controller is present or not by reading a register. This register is not available in OMAP34XX.
That being said I don't think that ADMA is available on any variant of am33xx.

Quoted 3 lines

OK, thanks!

Applied to u-boot-mmc. Thanks!

Best Regards,
Jaehoon Chung

Quoted 1 line
#12 lore raw

PATCH v2 4/4 mmc: omap_hsmmc: use a default 52MHz max clock rate if none is specified

In reply to Jean-Jacques Hiblot

mmc_of_parse() doesn't set a default value if none is available in DT. In that case, use a default 52MHz clock rate.

---

Changes in v2: None

 drivers/mmc/omap_hsmmc.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/mmc/omap_hsmmc.c +2 -0
index 3d2836d..caaa914 100644
@@ -1832,6 +1832,8 @@ static int omap_hsmmc_ofdata_to_platdata(struct udevice *dev)
 	if (ret < 0)
 		return ret;
 
+	if (!cfg->f_max)
+		cfg->f_max = 52000000;
 	cfg->host_caps |= MMC_MODE_HS_52MHz | MMC_MODE_HS;
 	cfg->f_min = 400000;
 	cfg->voltages = MMC_VDD_32_33 | MMC_VDD_33_34 | MMC_VDD_165_195;
Signature · 1 line
1.9.1
#13 lore raw

PATCH v2 4/4 mmc: omap_hsmmc: use a default 52MHz max clock rate if none is specified

Quoted 6 lines · Jean-Jacques Hiblot

mmc_of_parse() doesn't set a default value if none is available in DT. In that case, use a default 52MHz clock rate.

Signature · 8 lines
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20180223/a1fd5687/attachment.sig>
#14 lore raw

PATCH v2 4/4 mmc: omap_hsmmc: use a default 52MHz max clock rate if none is specified

Quoted 10 lines · Tom Rini
Quoted 6 lines · Jean-Jacques Hiblot

mmc_of_parse() doesn't set a default value if none is available in DT. In that case, use a default 52MHz clock rate.

Applied to u-boot-mmc. Thanks!

Best Regards,
Jaehoon Chung

Quoted 1 line
#15 lore raw

v2 cover mmc: omap_hsmmc: Reduce the footprint of the driver and fix am335x clock

In reply to Jean-Jacques Hiblot
Quoted 9 lines · Jean-Jacques Hiblot

This series aims at reducing the footprint of the omap_hsmmc driver in the SPL (1.5 kB gain in the case of the SPL for the am335x evm). It also fixes an issue with the am335x_evm by setting a default maximum frequency if none is defined in the dts.

tested on am335x_evm, bbb, bbb (vboot), and dra7 evm

I tested this version as well.

This series fixes 2d7482cf793f ("mmc: omap_hsmmc: use mmc_of_parse to populate mmc_config") which made my DM3730 unable to probe the MMC.

Quoted 20 lines

Changes in v2:
- Use a Kconfig option to compile out the ADMA support instead of relying on the platform architecture

Jean-Jacques Hiblot (4):
mmc: omap_hsmmc: do not embed struct mmc in struct omap_hsmmc_plat
mmc: omap_hsmmc: compile out write support if not needed
mmc: omap_hsmmc: make it possible to compile out ADMA support
mmc: omap_hsmmc: use a default 52MHz max clock rate if none is
specified

 arch/arm/include/asm/omap_mmc.h |  2 +-
 drivers/mmc/Kconfig             |  9 +++++++++
 drivers/mmc/omap_hsmmc.c        | 31 ++++++++++++++++++++-----------
 3 files changed, 30 insertions(+), 12 deletions(-)

--
1.9.1

#16 lore raw

v2 cover mmc: omap_hsmmc: Reduce the footprint of the driver and fix am335x clock

Quoted 12 lines · Adam Ford
Quoted 9 lines · Jean-Jacques Hiblot

This series aims at reducing the footprint of the omap_hsmmc driver in the SPL (1.5 kB gain in the case of the SPL for the am335x evm). It also fixes an issue with the am335x_evm by setting a default maximum frequency if none is defined in the dts.

tested on am335x_evm, bbb, bbb (vboot), and dra7 evm

Quoted 6 lines

I tested this version as well.

This series fixes 2d7482cf793f ("mmc: omap_hsmmc: use mmc_of_parse to populate mmc_config") which made my DM3730 unable to probe the MMC.

After testing, I will apply these patches with your tag.

Best Regards,
Jaehoon Chung

Quoted 23 lines
Quoted 22 lines

Changes in v2:
- Use a Kconfig option to compile out the ADMA support instead of relying on the platform architecture

Jean-Jacques Hiblot (4):
mmc: omap_hsmmc: do not embed struct mmc in struct omap_hsmmc_plat
mmc: omap_hsmmc: compile out write support if not needed
mmc: omap_hsmmc: make it possible to compile out ADMA support
mmc: omap_hsmmc: use a default 52MHz max clock rate if none is
specified

 arch/arm/include/asm/omap_mmc.h |  2 +-
 drivers/mmc/Kconfig             |  9 +++++++++
 drivers/mmc/omap_hsmmc.c        | 31 ++++++++++++++++++++-----------
 3 files changed, 30 insertions(+), 12 deletions(-)

--
1.9.1