Skip to contentSkip to conversation

MMC: Fix multiblock SDIO transfers in AT91 MCI

PATCH New Yauhen Kharuzhy 4 participants 9 messages Nov 25, 2010 – Dec 12, 2010

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 MMC: Fix multiblock SDIO transfers in AT91 MCI

The AT91 MCI has special SDIO transfer types: SDIO block and SDIO byte transfers, but at91_mci driver doesn't use them and handles all SDIO transfers as ordinary MMC block transfers. This causes problems for multiple-block SDIO transfers (in particular for 256-bytes blocks).

Fix this situation by check opcode for SDIO CMD53 and set transfer type in AT91_MCI_CMDR register properly.

This patch was tested with libertas SDIO driver: problem with TX timeouts on big packets was eliminated.

---
 arch/arm/mach-at91/include/mach/at91_mci.h |    2 ++
 drivers/mmc/host/at91_mci.c                |   13 +++++++++----
 2 files changed, 11 insertions(+), 4 deletions(-)
arch/arm/mach-at91/include/mach/at91_mci.h +2 -0
index 57f8ee1..27ac6f5 100644
@@ -74,6 +74,8 @@
 #define			AT91_MCI_TRTYP_BLOCK	(0 << 19)
 #define			AT91_MCI_TRTYP_MULTIPLE	(1 << 19)
 #define			AT91_MCI_TRTYP_STREAM	(2 << 19)
+#define			AT91_MCI_TRTYP_SDIO_BYTE	(4 << 19)
+#define			AT91_MCI_TRTYP_SDIO_BLOCK	(5 << 19)
 
 #define AT91_MCI_BLKR		0x18		/* Block Register */
 #define		AT91_MCI_BLKR_BCNT(n)	((0xffff & (n)) << 0)	/* Block count */
drivers/mmc/host/at91_mci.c +9 -4
index 591ab54..d3e6a96 100644
@@ -69,6 +69,7 @@
 #include <linux/highmem.h>
 
 #include <linux/mmc/host.h>
+#include <linux/mmc/sdio.h>
 
 #include <asm/io.h>
 #include <asm/irq.h>
@@ -493,10 +494,14 @@ static void at91_mci_send_command(struct at91mci_host *host, struct mmc_command
 		else if (data->flags & MMC_DATA_WRITE)
 			cmdr |= AT91_MCI_TRCMD_START;
 
-		if (data->flags & MMC_DATA_STREAM)
-			cmdr |= AT91_MCI_TRTYP_STREAM;
-		if (data->blocks > 1)
-			cmdr |= AT91_MCI_TRTYP_MULTIPLE;
+		if (cmd->opcode == SD_IO_RW_EXTENDED) {
+			cmdr |= AT91_MCI_TRTYP_SDIO_BLOCK;
+		} else {
+			if (data->flags & MMC_DATA_STREAM)
+				cmdr |= AT91_MCI_TRTYP_STREAM;
+			if (data->blocks > 1)
+				cmdr |= AT91_MCI_TRTYP_MULTIPLE;
+		}
 	}
 	else {
 		block_length = 0;
Signature · 1 line
1.7.2.3
#1 lore raw

PATCH 1/1 MMC: Fix multiblock SDIO transfers in AT91 MCI

Le 25/11/2010 11:11, Yauhen Kharuzhy :

Quoted 5 lines

The AT91 MCI has special SDIO transfer types: SDIO block and SDIO byte transfers, but at91_mci driver doesn't use them and handles all SDIO transfers as ordinary MMC block transfers. This causes problems for multiple-block SDIO transfers (in particular for 256-bytes blocks).

Quoted 7 lines

Fix this situation by check opcode for SDIO CMD53 and set transfer type in AT91_MCI_CMDR register properly.

This patch was tested with libertas SDIO driver: problem with TX timeouts on big packets was eliminated.

Quoted 49 lines
---
 arch/arm/mach-at91/include/mach/at91_mci.h |    2 ++
 drivers/mmc/host/at91_mci.c                |   13 +++++++++----
 2 files changed, 11 insertions(+), 4 deletions(-)
arch/arm/mach-at91/include/mach/at91_mci.h +2 -0
index 57f8ee1..27ac6f5 100644
@@ -74,6 +74,8 @@
 #define			AT91_MCI_TRTYP_BLOCK	(0 << 19)
 #define			AT91_MCI_TRTYP_MULTIPLE	(1 << 19)
 #define			AT91_MCI_TRTYP_STREAM	(2 << 19)
+#define			AT91_MCI_TRTYP_SDIO_BYTE	(4 << 19)
+#define			AT91_MCI_TRTYP_SDIO_BLOCK	(5 << 19)
 
 #define AT91_MCI_BLKR		0x18		/* Block Register */
 #define		AT91_MCI_BLKR_BCNT(n)	((0xffff & (n)) << 0)	/* Block count */
drivers/mmc/host/at91_mci.c +9 -4
index 591ab54..d3e6a96 100644
@@ -69,6 +69,7 @@
 #include <linux/highmem.h>
 
 #include <linux/mmc/host.h>
+#include <linux/mmc/sdio.h>
 
 #include <asm/io.h>
 #include <asm/irq.h>
@@ -493,10 +494,14 @@ static void at91_mci_send_command(struct at91mci_host *host, struct mmc_command
 		else if (data->flags & MMC_DATA_WRITE)
 			cmdr |= AT91_MCI_TRCMD_START;
 
-		if (data->flags & MMC_DATA_STREAM)
-			cmdr |= AT91_MCI_TRTYP_STREAM;
-		if (data->blocks > 1)
-			cmdr |= AT91_MCI_TRTYP_MULTIPLE;
+		if (cmd->opcode == SD_IO_RW_EXTENDED) {
+			cmdr |= AT91_MCI_TRTYP_SDIO_BLOCK;
+		} else {
+			if (data->flags & MMC_DATA_STREAM)
+				cmdr |= AT91_MCI_TRTYP_STREAM;
+			if (data->blocks > 1)
+				cmdr |= AT91_MCI_TRTYP_MULTIPLE;
+		}
 	}
 	else {
 		block_length = 0;
Signature · 1 line
Nicolas Ferre
#2 lore raw

PATCH 1/1 MMC: Fix multiblock SDIO transfers in AT91 MCI

Le 26/11/2010 12:17, Nicolas Ferre :

Quoted 9 lines

Le 25/11/2010 11:11, Yauhen Kharuzhy :

Quoted 8 lines

The AT91 MCI has special SDIO transfer types: SDIO block and SDIO byte transfers, but at91_mci driver doesn't use them and handles all SDIO transfers as ordinary MMC block transfers. This causes problems for multiple-block SDIO transfers (in particular for 256-bytes blocks).

Fix this situation by check opcode for SDIO CMD53 and set transfer type in AT91_MCI_CMDR register properly.

Quoted 6 lines
Quoted 4 lines

This patch was tested with libertas SDIO driver: problem with TX timeouts on big packets was eliminated.

Hi Chris,

A little "ping"!

Do you want that I re-send you the patch with relevant S-O-B lines or you can include them yourself?

Best regards,

Quoted 51 lines
Quoted 49 lines
---
 arch/arm/mach-at91/include/mach/at91_mci.h |    2 ++
 drivers/mmc/host/at91_mci.c                |   13 +++++++++----
 2 files changed, 11 insertions(+), 4 deletions(-)
arch/arm/mach-at91/include/mach/at91_mci.h +2 -0
index 57f8ee1..27ac6f5 100644
@@ -74,6 +74,8 @@
 #define			AT91_MCI_TRTYP_BLOCK	(0 << 19)
 #define			AT91_MCI_TRTYP_MULTIPLE	(1 << 19)
 #define			AT91_MCI_TRTYP_STREAM	(2 << 19)
+#define			AT91_MCI_TRTYP_SDIO_BYTE	(4 << 19)
+#define			AT91_MCI_TRTYP_SDIO_BLOCK	(5 << 19)
 
 #define AT91_MCI_BLKR		0x18		/* Block Register */
 #define		AT91_MCI_BLKR_BCNT(n)	((0xffff & (n)) << 0)	/* Block count */
drivers/mmc/host/at91_mci.c +9 -4
index 591ab54..d3e6a96 100644
@@ -69,6 +69,7 @@
 #include <linux/highmem.h>
 
 #include <linux/mmc/host.h>
+#include <linux/mmc/sdio.h>
 
 #include <asm/io.h>
 #include <asm/irq.h>
@@ -493,10 +494,14 @@ static void at91_mci_send_command(struct at91mci_host *host, struct mmc_command
 		else if (data->flags & MMC_DATA_WRITE)
 			cmdr |= AT91_MCI_TRCMD_START;
 
-		if (data->flags & MMC_DATA_STREAM)
-			cmdr |= AT91_MCI_TRTYP_STREAM;
-		if (data->blocks > 1)
-			cmdr |= AT91_MCI_TRTYP_MULTIPLE;
+		if (cmd->opcode == SD_IO_RW_EXTENDED) {
+			cmdr |= AT91_MCI_TRTYP_SDIO_BLOCK;
+		} else {
+			if (data->flags & MMC_DATA_STREAM)
+				cmdr |= AT91_MCI_TRTYP_STREAM;
+			if (data->blocks > 1)
+				cmdr |= AT91_MCI_TRTYP_MULTIPLE;
+		}
 	}
 	else {
 		block_length = 0;
Signature · 1 line
Nicolas Ferre
#3 lore raw

PATCH 1/1 MMC: Fix multiblock SDIO transfers in AT91 MCI

Hi Nicolas,

Quoted 4 lines · Nicolas Ferre

A little "ping"!

Do you want that I re-send you the patch with relevant S-O-B lines or you can include them yourself?

Thanks for the ping -- I've pushed this to mmc-next and queued it for 2.6.37 now. This looks like a candidate for the stable kernel to me; do you agree?

Thanks,

Signature · 2 lines
Chris Ball   <[email protected]>   <http://printf.net/>
One Laptop Per Child
#4 lore raw

PATCH 1/1 MMC: Fix multiblock SDIO transfers in ATMEL MCI

Based on report made by Yauhen in:
"MMC: Fix multiblock SDIO transfers in AT91 MCI" patch, I report those changes to the brother driver: atmel-mci.

So, this patch sets SDIO transfer types: SDIO block and SDIO byte transfers instead of using ordinary MMC block transfers. It is checking opcode for SDIO CMD53 and setting transfer type in MCI_CMDR register properly.

---

Hi Chris,

In addition to the at91_mci multiblock SDIO fix, I add this little one too as it concerns the brother driver atmel-mci with same issue.

I hope that you will be able to queue those fixes to .37-final and stable...

Thanks a lot, best regards,

 drivers/mmc/host/atmel-mci.c |   18 ++++++++++++------
 1 files changed, 12 insertions(+), 6 deletions(-)
drivers/mmc/host/atmel-mci.c +12 -6
index 301351a..ad2a7a0 100644
@@ -26,6 +26,7 @@
 #include <linux/stat.h>
 
 #include <linux/mmc/host.h>
+#include <linux/mmc/sdio.h>
 
 #include <mach/atmel-mci.h>
 #include <linux/atmel-mci.h>
@@ -532,12 +533,17 @@ static u32 atmci_prepare_command(struct mmc_host *mmc,
 	data = cmd->data;
 	if (data) {
 		cmdr |= MCI_CMDR_START_XFER;
-		if (data->flags & MMC_DATA_STREAM)
-			cmdr |= MCI_CMDR_STREAM;
-		else if (data->blocks > 1)
-			cmdr |= MCI_CMDR_MULTI_BLOCK;
-		else
-			cmdr |= MCI_CMDR_BLOCK;
+
+		if (cmd->opcode == SD_IO_RW_EXTENDED) {
+			cmdr |= MCI_CMDR_SDIO_BLOCK;
+		} else {
+			if (data->flags & MMC_DATA_STREAM)
+				cmdr |= MCI_CMDR_STREAM;
+			else if (data->blocks > 1)
+				cmdr |= MCI_CMDR_MULTI_BLOCK;
+			else
+				cmdr |= MCI_CMDR_BLOCK;
+		}
 
 		if (data->flags & MMC_DATA_READ)
 			cmdr |= MCI_CMDR_TRDIR_READ;
Signature · 1 line
1.7.3
#5 lore raw

PATCH 1/1 MMC: Fix multiblock SDIO transfers in ATMEL MCI

Hi Nicolas,

Quoted 11 lines · Nicolas Ferre

Based on report made by Yauhen in:
"MMC: Fix multiblock SDIO transfers in AT91 MCI" patch, I report those changes to the brother driver: atmel-mci.

So, this patch sets SDIO transfer types: SDIO block and SDIO byte transfers instead of using ordinary MMC block transfers. It is checking opcode for SDIO CMD53 and setting transfer type in MCI_CMDR register properly.

Quoted 8 lines

---

Hi Chris,

In addition to the at91_mci multiblock SDIO fix, I add this little one too as it concerns the brother driver atmel-mci with same issue.

I hope that you will be able to queue those fixes to .37-final and stable...

That's fine, this is queued now too. Thanks!

Signature · 2 lines
Chris Ball   <[email protected]>   <http://printf.net/>
One Laptop Per Child
#6 lore raw

PATCH 1/1 MMC: Fix multiblock SDIO transfers in ATMEL MCI

In reply to Nicolas Ferre
Quoted 11 lines · Nicolas Ferre

Based on report made by Yauhen in:
"MMC: Fix multiblock SDIO transfers in AT91 MCI" patch, I report those changes to the brother driver: atmel-mci.

So, this patch sets SDIO transfer types: SDIO block and SDIO byte transfers instead of using ordinary MMC block transfers. It is checking opcode for SDIO CMD53 and setting transfer type in MCI_CMDR register properly.

Best Regards,
J.

#7 lore raw

PATCH 1/1 MMC: Fix multiblock SDIO transfers in ATMEL MCI

Quoted 4 lines · Jean-Christophe PLAGNIOL-VILLARD
Quoted 4 lines

Based on report made by Yauhen in:
"MMC: Fix multiblock SDIO transfers in AT91 MCI" patch, I report those changes to the brother driver: atmel-mci.

Quoted 8 lines
Quoted 7 lines

So, this patch sets SDIO transfer types: SDIO block and SDIO byte transfers instead of using ordinary MMC block transfers. It is checking opcode for SDIO CMD53 and setting transfer type in MCI_CMDR register properly.

Thanks, pushed to mmc-next and queued for .37/stable.

Signature · 2 lines
Chris Ball   <[email protected]>   <http://printf.net/>
One Laptop Per Child
#8 lore raw

PATCH 1/1 MMC: Fix multiblock SDIO transfers in AT91 MCI

In reply to Yauhen Kharuzhy
Quoted 5 lines · Yauhen Kharuzhy

The AT91 MCI has special SDIO transfer types: SDIO block and SDIO byte transfers, but at91_mci driver doesn't use them and handles all SDIO transfers as ordinary MMC block transfers. This causes problems for multiple-block SDIO transfers (in particular for 256-bytes blocks).

Quoted 7 lines

Fix this situation by check opcode for SDIO CMD53 and set transfer type in AT91_MCI_CMDR register properly.

This patch was tested with libertas SDIO driver: problem with TX timeouts on big packets was eliminated.

Best Regards,
J.