Skip to contentSkip to conversation

mmc: fix response timeout after switch command

PATCH RESEND New Stefan Bosch 2 participants 5 messages Jan 23, 2021 – Jan 28, 2021 MMC, SD and SDIO

lore
Share & tools
lore
#0 lore raw

cover mmc: fix response timeout after switch command

Currently I am implementing SPL for frienlyARM's NanoPC-T2 board (SoC S5P4418). Loading of U-Boot from SD-card fails if CONFIG_SPL_MMC_TINY=y. I.e. mmc_set_blocklen() inside mmc_bread() fails (Response Timeout), caused by the previous call of __mmc_switch() in drivers/mmc/mmc.c:
Here the 'current state' of the card status keeps 'data' for several 100ms after issuing the switch command. Because of this state the next command issued fails (in this case CMD16 = SET_BLOCKLEN). Patch: Wait for 'current state' = 'tran' in __mmc_switch().

Stefan Bosch (1):
mmc: fix response timeout after switch command

 drivers/mmc/mmc.c | 3 ++-
 include/mmc.h     | 1 +
 2 files changed, 3 insertions(+), 1 deletion(-)
Signature · 1 line
2.17.1
#1 lore raw

PATCH RESEND 1/1 mmc: fix response timeout after switch command

After issuing the switch command: Wait until 'current state' of the card status becomes 'tran'. This prevents from response timeout at the next command because of 'current state' = 'data'.

---

 drivers/mmc/mmc.c | 3 ++-
 include/mmc.h     | 1 +
 2 files changed, 3 insertions(+), 1 deletion(-)
drivers/mmc/mmc.c +2 -1
index a47700e313..8ccd2058a9 100644
@@ -823,7 +823,8 @@ static int __mmc_switch(struct mmc *mmc, u8 set, u8 index, u8 value,
 				 value);
 			return -EIO;
 		}
-		if (!ret && (status & MMC_STATUS_RDY_FOR_DATA))
+		if (!ret && (status & MMC_STATUS_RDY_FOR_DATA) &&
+		    (status & MMC_STATUS_CURR_STATE) == MMC_STATE_TRANS)
 			return 0;
 		udelay(100);
 	} while (get_timer(start) < timeout_ms);
include/mmc.h +1 -0
index 1d377e0281..18402494c6 100644
@@ -178,6 +178,7 @@ static inline bool mmc_is_tuning_cmd(uint cmdidx)
 #define MMC_STATUS_ERROR	(1 << 19)
 
 #define MMC_STATE_PRG		(7 << 9)
+#define MMC_STATE_TRANS		(4 << 9)
 
 #define MMC_VDD_165_195		0x00000080	/* VDD voltage 1.65 - 1.95 */
 #define MMC_VDD_20_21		0x00000100	/* VDD voltage 2.0 ~ 2.1 */
Signature · 1 line
2.17.1
#2 lore raw

PATCH RESEND 1/1 mmc: fix response timeout after switch command

Hi,

Quoted 15 lines · Stefan Bosch

After issuing the switch command: Wait until 'current state' of the card status becomes 'tran'. This prevents from response timeout at the next command because of 'current state' = 'data'.>

---

 drivers/mmc/mmc.c | 3 ++-
 include/mmc.h     | 1 +
 2 files changed, 3 insertions(+), 1 deletion(-)
drivers/mmc/mmc.c +0 -0
index a47700e313..8ccd2058a9 100644

drivers/mmc/mmc.c static int __mmc_switch(struct mmc *mmc, u8 set, u8 index, u8 value, · line 827

Quoted 6 lines
drivers/mmc/mmc.c +2 -1
@@ -823,7 +823,8 @@ static int __mmc_switch(struct mmc *mmc, u8 set, u8 index, u8 value,
 				 value);
 			return -EIO;
 		}
-		if (!ret && (status & MMC_STATUS_RDY_FOR_DATA))
+		if (!ret && (status & MMC_STATUS_RDY_FOR_DATA) &&
+		    (status & MMC_STATUS_CURR_STATE) == MMC_STATE_TRANS)

AFAIK, MMC_STATUS_RDY_FOR_DATA bit means the signaling about "buffer is empty". Even though its bit was set, does it need to check TRANS state?

Best Regards,
Jaehoon Chung

Quoted 16 lines
drivers/mmc/mmc.c +0 -0
 			return 0;
 		udelay(100);
 	} while (get_timer(start) < timeout_ms);
include/mmc.h +1 -0
index 1d377e0281..18402494c6 100644
@@ -178,6 +178,7 @@ static inline bool mmc_is_tuning_cmd(uint cmdidx)
 #define MMC_STATUS_ERROR	(1 << 19)
 
 #define MMC_STATE_PRG		(7 << 9)
+#define MMC_STATE_TRANS		(4 << 9)
 
 #define MMC_VDD_165_195		0x00000080	/* VDD voltage 1.65 - 1.95 */
 #define MMC_VDD_20_21		0x00000100	/* VDD voltage 2.0 ~ 2.1 */
#3 lore raw

PATCH RESEND 1/1 mmc: fix response timeout after switch command

Hi,

Quoted 28 lines · Jaehoon Chung

Hi,

Quoted 15 lines · Stefan Bosch

After issuing the switch command: Wait until 'current state' of the card status becomes 'tran'. This prevents from response timeout at the next command because of 'current state' = 'data'.>

---

  drivers/mmc/mmc.c | 3 ++-
  include/mmc.h     | 1 +
  2 files changed, 3 insertions(+), 1 deletion(-)
drivers/mmc/mmc.c +0 -0
index a47700e313..8ccd2058a9 100644

drivers/mmc/mmc.c static int __mmc_switch(struct mmc *mmc, u8 set, u8 index, u8 value, · line 827

Quoted 6 lines
drivers/mmc/mmc.c +2 -1
@@ -823,7 +823,8 @@ static int __mmc_switch(struct mmc *mmc, u8 set, u8 index, u8 value,
  				 value);
  			return -EIO;
  		}
-		if (!ret && (status & MMC_STATUS_RDY_FOR_DATA))
+		if (!ret && (status & MMC_STATUS_RDY_FOR_DATA) &&
+		    (status & MMC_STATUS_CURR_STATE) == MMC_STATE_TRANS)

AFAIK, MMC_STATUS_RDY_FOR_DATA bit means the signaling about "buffer is empty". Even though its bit was set, does it need to check TRANS state?

Quoted 3 lines

Best Regards,
Jaehoon Chung

After issuing command 'MMC_CMD_SWITCH' = CMD6 the state transits to 'DATA'. But in DATA state it is not possible to issue a (new) CMD except CMD12 (operation complete) and CMD7 (select/deselect card). See appropriate State Diagram, e.g. "Figure 4-3: SD Memory Card State Diagram (data transfer mode)" in "SD Specifications Part 1 - Physical Layer - Simplified Specification - Version 2.00". Status 'Ready For Data' is in my opinion only relevant for CMD12 and CMD7.

Regards
Stefan Bosch

Quoted 17 lines
Quoted 16 lines
drivers/mmc/mmc.c +0 -0
  			return 0;
  		udelay(100);
  	} while (get_timer(start) < timeout_ms);
include/mmc.h +1 -0
index 1d377e0281..18402494c6 100644
@@ -178,6 +178,7 @@ static inline bool mmc_is_tuning_cmd(uint cmdidx)
  #define MMC_STATUS_ERROR	(1 << 19)
 
  #define MMC_STATE_PRG		(7 << 9)
+#define MMC_STATE_TRANS		(4 << 9)
 
  #define MMC_VDD_165_195		0x00000080	/* VDD voltage 1.65 - 1.95 */
  #define MMC_VDD_20_21		0x00000100	/* VDD voltage 2.0 ~ 2.1 */
#4 lore raw

PATCH RESEND 1/1 mmc: fix response timeout after switch command

Hi Stefan,

Quoted 31 lines · Stefan Bosch

Hi,

Quoted 28 lines · Jaehoon Chung

Hi,

Quoted 15 lines · Stefan Bosch

After issuing the switch command: Wait until 'current state' of the card status becomes 'tran'. This prevents from response timeout at the next command because of 'current state' = 'data'.>

---

? drivers/mmc/mmc.c | 3 ++-
? include/mmc.h???? | 1 +
? 2 files changed, 3 insertions(+), 1 deletion(-)

drivers/mmc/mmc.c +0 -0
index a47700e313..8ccd2058a9 100644

drivers/mmc/mmc.c static int __mmc_switch(struct mmc *mmc, u8 set, u8 index, u8 value, · line 827

Quoted 6 lines
drivers/mmc/mmc.c +2 -1
@@ -823,7 +823,8 @@ static int __mmc_switch(struct mmc *mmc, u8 set, u8 index, u8 value,
 ?????????????????? value);
 ????????????? return -EIO;
 ????????? }
-??????? if (!ret && (status & MMC_STATUS_RDY_FOR_DATA))
+??????? if (!ret && (status & MMC_STATUS_RDY_FOR_DATA) &&
+??????????? (status & MMC_STATUS_CURR_STATE) == MMC_STATE_TRANS)

AFAIK, MMC_STATUS_RDY_FOR_DATA bit means the signaling about "buffer is empty". Even though its bit was set, does it need to check TRANS state?

Quoted 5 lines
Quoted 3 lines

Best Regards,
Jaehoon Chung

After issuing command 'MMC_CMD_SWITCH' = CMD6 the state transits to 'DATA'. But in DATA state it is not possible to issue a (new) CMD except CMD12 (operation complete) and CMD7 (select/deselect card). See appropriate State Diagram, e.g. "Figure 4-3: SD Memory Card State Diagram (data transfer mode)" in "SD Specifications Part 1 - Physical Layer - Simplified Specification - Version 2.00". Status 'Ready For Data' is in my opinion only relevant for CMD12 and CMD7.

Thanks for explanation. I have checked Spec about SD and eMMC. It makes sense.

Best Regards,
Jaehoon Chung

Quoted 20 lines

Regards
Stefan Bosch

Quoted 15 lines
Quoted 14 lines
drivers/mmc/mmc.c +0 -0
 ????????????? return 0;
 ????????? udelay(100);
 ????? } while (get_timer(start) < timeout_ms);
include/mmc.h +1 -0
index 1d377e0281..18402494c6 100644
@@ -178,6 +178,7 @@ static inline bool mmc_is_tuning_cmd(uint cmdidx)
 ? #define MMC_STATUS_ERROR??? (1 << 19)
 ? ? #define MMC_STATE_PRG??????? (7 << 9)
+#define MMC_STATE_TRANS??????? (4 << 9)

? ? #define MMC_VDD_165_195??????? 0x00000080??? /* VDD voltage 1.65 - 1.95 */ ? #define MMC_VDD_20_21??????? 0x00000100??? /* VDD voltage 2.0 ~ 2.1 */