Skip to contentSkip to conversation

mmc: sandbox: Some small fixes

PATCH In review Sean Anderson 3 participants 12 messages Mar 30, 2022 – Sep 8, 2022 MMC, SD and SDIO

lore
Share & tools
lore
#0 lore raw

cover mmc: sandbox: Some small fixes

Various small fixes which were found with valgrind.

Sean Anderson (4):
mmc: sandbox: Initialize the status register
mmc: sandbox: Initialize backing buffer
mmc: Import some defines for R1 responses
mmc: sandbox: Set the response

 drivers/mmc/sandbox_mmc.c | 13 +++++++---
 include/mmc.h             | 52 +++++++++++++++++++++++++++++++++++++--
 2 files changed, 60 insertions(+), 5 deletions(-)
Signature · 1 line
2.35.1
#1 lore raw

PATCH 1/4 mmc: sandbox: Initialize the status register

The send status command expects the status register to be returned as a response. Without writing data back, whatever is on the stack will be interpreted as the status register. There are a lot of fields in this register, but fortunately all zeros is interpreted as "we don't support anything."

---

 drivers/mmc/sandbox_mmc.c | 1 +
 1 file changed, 1 insertion(+)
drivers/mmc/sandbox_mmc.c +1 -0
index 451fe4a4e5..35159451e1 100644
@@ -57,6 +57,7 @@ static int sandbox_mmc_send_cmd(struct udevice *dev, struct mmc_cmd *cmd,
 		break;
 	case MMC_CMD_SEND_STATUS:
 		cmd->response[0] = MMC_STATUS_RDY_FOR_DATA;
+		memset(data->dest, '\0', data->blocks * data->blocksize);
 		break;
 	case MMC_CMD_SELECT_CARD:
 		break;
Signature · 1 line
2.35.1
Reviewed-by on v1 1/4
#2 lore raw

Re: [PATCH 1/4] mmc: sandbox: Initialize the status register

Hi

Quoted 7 lines · Sean Anderson

The send status command expects the status register to be returned as a response. Without writing data back, whatever is on the stack will be interpreted as the status register. There are a lot of fields in this register, but fortunately all zeros is interpreted as "we don't support anything."

Best Regards,
Jaehoon Chung

Quoted 17 lines

---

 drivers/mmc/sandbox_mmc.c | 1 +
 1 file changed, 1 insertion(+)
drivers/mmc/sandbox_mmc.c +1 -0
index 451fe4a4e5..35159451e1 100644
@@ -57,6 +57,7 @@ static int sandbox_mmc_send_cmd(struct udevice *dev, struct mmc_cmd *cmd,
 		break;
 	case MMC_CMD_SEND_STATUS:
 		cmd->response[0] = MMC_STATUS_RDY_FOR_DATA;
+		memset(data->dest, '\0', data->blocks * data->blocksize);
 		break;
 	case MMC_CMD_SELECT_CARD:
 		break;
#3 lore raw

Re: [PATCH 1/4] mmc: sandbox: Initialize the status register

Dear Sean,

Quoted 9 lines · Jaehoon Chung

Hi

Quoted 6 lines · Sean Anderson

The send status command expects the status register to be returned as a response. Without writing data back, whatever is on the stack will be interpreted as the status register. There are a lot of fields in this register, but fortunately all zeros is interpreted as "we don't support anything."

Quoted 3 lines
Quoted 1 line

After applied your patch, CI testing was failed.
I have been checking what's problem. So this patch set will be applied after fixing it.

Best Regards,
Jaehoon Chung

Quoted 23 lines

Best Regards,
Jaehoon Chung

Quoted 17 lines

---

 drivers/mmc/sandbox_mmc.c | 1 +
 1 file changed, 1 insertion(+)
drivers/mmc/sandbox_mmc.c +1 -0
index 451fe4a4e5..35159451e1 100644
@@ -57,6 +57,7 @@ static int sandbox_mmc_send_cmd(struct udevice *dev, struct mmc_cmd *cmd,
 		break;
 	case MMC_CMD_SEND_STATUS:
 		cmd->response[0] = MMC_STATUS_RDY_FOR_DATA;
+		memset(data->dest, '\0', data->blocks * data->blocksize);
 		break;
 	case MMC_CMD_SELECT_CARD:
 		break;
#4 lore raw

Re: [PATCH 1/4] mmc: sandbox: Initialize the status register

Hi Jaehoon,

Quoted 12 lines · Jaehoon Chung

Dear Sean,

Quoted 9 lines · Jaehoon Chung

Hi

Quoted 6 lines · Sean Anderson

The send status command expects the status register to be returned as a response. Without writing data back, whatever is on the stack will be interpreted as the status register. There are a lot of fields in this register, but fortunately all zeros is interpreted as "we don't support anything."

Quoted 6 lines
Quoted 3 lines
Quoted 1 line

After applied your patch, CI testing was failed.
I have been checking what's problem. So this patch set will be applied after fixing it.

Did you ever figure out what was causing CI failures? If you don't have time to look into it, could you send me the log?

--Sean

#5 lore raw

Re: [PATCH 1/4] mmc: sandbox: Initialize the status register

Hi,

Quoted 17 lines · Sean Anderson

Hi Jaehoon,

Quoted 14 lines · Jaehoon Chung

Dear Sean,

Quoted 11 lines · Jaehoon Chung

Hi

Quoted 7 lines · Sean Anderson

The send status command expects the status register to be returned as a response. Without writing data back, whatever is on the stack will be interpreted as the status register. There are a lot of fields in this register, but fortunately all zeros is interpreted as "we don't support anything."

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

After applied your patch, CI testing was failed.
I have been checking what's problem. So this patch set will be applied after fixing it.

Did you ever figure out what was causing CI failures? If you don't have time to look into it, could you send me the log?

I will check it with latest u-boot base. And If there is a failed, I will share a log. Sorry for too late.

Best Regards,
Jaehoon Chung

Quoted 2 lines

--Sean

#6 lore raw

PATCH 2/4 mmc: sandbox: Initialize backing buffer

In reply to Sean Anderson

Private data is initialized to all zeros by DM. Malloc does not do this. Initialize it. This fixes partition detection logic from trying to detect partitions in uninitialized memory.

Fixes 0bf61aced2 ("sandbox: mmc: Support a backing file")

---

 drivers/mmc/sandbox_mmc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mmc/sandbox_mmc.c +1 -1
index 35159451e1..60a6be0add 100644
@@ -178,7 +178,7 @@ static int sandbox_mmc_probe(struct udevice *dev)
 		priv->csize = 0;
 		priv->size = (priv->csize + 1) * SIZE_MULTIPLE; /* 1 MiB */
 
-		priv->buf = malloc(priv->size);
+		priv->buf = calloc(1, priv->size);
 		if (!priv->buf) {
 			log_err("%s: Not enough memory (%x bytes)\n",
 				dev->name, priv->size);
Signature · 1 line
2.35.1
Reviewed-by on v1 2/4
#7 lore raw

Re: [PATCH 2/4] mmc: sandbox: Initialize backing buffer

Quoted 6 lines · Sean Anderson

Private data is initialized to all zeros by DM. Malloc does not do this. Initialize it. This fixes partition detection logic from trying to detect partitions in uninitialized memory.

Fixes 0bf61aced2 ("sandbox: mmc: Support a backing file")

Best Regards,
Jaehoon Chung

Quoted 18 lines

---

 drivers/mmc/sandbox_mmc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mmc/sandbox_mmc.c +1 -1
index 35159451e1..60a6be0add 100644
@@ -178,7 +178,7 @@ static int sandbox_mmc_probe(struct udevice *dev)
 		priv->csize = 0;
 		priv->size = (priv->csize + 1) * SIZE_MULTIPLE; /* 1 MiB */
 
-		priv->buf = malloc(priv->size);
+		priv->buf = calloc(1, priv->size);
 		if (!priv->buf) {
 			log_err("%s: Not enough memory (%x bytes)\n",
 				dev->name, priv->size);
#8 lore raw

PATCH 3/4 mmc: Import some defines for R1 responses

In reply to Sean Anderson

This imports defines for R1 responses from include/linux/mmc/mmc.h from Linux 5.10.

---
Yes, this is an old version, but it's what I had checked out, and I don't think there are any new fields we need to handle :)

 include/mmc.h | 52 +++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 50 insertions(+), 2 deletions(-)
include/mmc.h +50 -2
index 6bdcce881d..7304eee0d4 100644
@@ -311,8 +311,56 @@ static inline bool mmc_is_tuning_cmd(uint cmdidx)
 #define EXT_CSD_WR_DATA_REL_USR		(1 << 0)	/* user data area WR_REL */
 #define EXT_CSD_WR_DATA_REL_GP(x)	(1 << ((x)+1))	/* GP part (x+1) WR_REL */
 
-#define R1_ILLEGAL_COMMAND		(1 << 22)
-#define R1_APP_CMD			(1 << 5)
+/*
+  MMC status in R1, for native mode (SPI bits are different)
+  Type
+	e : error bit
+	s : status bit
+	r : detected and set for the actual command response
+	x : detected and set during command execution. the host must poll
+            the card by sending status command in order to read these bits.
+  Clear condition
+	a : according to the card state
+	b : always related to the previous command. Reception of
+            a valid command will clear it (with a delay of one command)
+	c : clear by read
+ */
+
+#define R1_OUT_OF_RANGE		(1 << 31)	/* er, c */
+#define R1_ADDRESS_ERROR	(1 << 30)	/* erx, c */
+#define R1_BLOCK_LEN_ERROR	(1 << 29)	/* er, c */
+#define R1_ERASE_SEQ_ERROR      (1 << 28)	/* er, c */
+#define R1_ERASE_PARAM		(1 << 27)	/* ex, c */
+#define R1_WP_VIOLATION		(1 << 26)	/* erx, c */
+#define R1_CARD_IS_LOCKED	(1 << 25)	/* sx, a */
+#define R1_LOCK_UNLOCK_FAILED	(1 << 24)	/* erx, c */
+#define R1_COM_CRC_ERROR	(1 << 23)	/* er, b */
+#define R1_ILLEGAL_COMMAND	(1 << 22)	/* er, b */
+#define R1_CARD_ECC_FAILED	(1 << 21)	/* ex, c */
+#define R1_CC_ERROR		(1 << 20)	/* erx, c */
+#define R1_ERROR		(1 << 19)	/* erx, c */
+#define R1_UNDERRUN		(1 << 18)	/* ex, c */
+#define R1_OVERRUN		(1 << 17)	/* ex, c */
+#define R1_CID_CSD_OVERWRITE	(1 << 16)	/* erx, c, CID/CSD overwrite */
+#define R1_WP_ERASE_SKIP	(1 << 15)	/* sx, c */
+#define R1_CARD_ECC_DISABLED	(1 << 14)	/* sx, a */
+#define R1_ERASE_RESET		(1 << 13)	/* sr, c */
+#define R1_STATUS(x)            (x & 0xFFF9A000)
+#define R1_CURRENT_STATE(x)	((x & 0x00001E00) >> 9)	/* sx, b (4 bits) */
+#define R1_READY_FOR_DATA	(1 << 8)	/* sx, a */
+#define R1_SWITCH_ERROR		(1 << 7)	/* sx, c */
+#define R1_EXCEPTION_EVENT	(1 << 6)	/* sr, a */
+#define R1_APP_CMD		(1 << 5)	/* sr, c */
+
+#define R1_STATE_IDLE	0
+#define R1_STATE_READY	1
+#define R1_STATE_IDENT	2
+#define R1_STATE_STBY	3
+#define R1_STATE_TRAN	4
+#define R1_STATE_DATA	5
+#define R1_STATE_RCV	6
+#define R1_STATE_PRG	7
+#define R1_STATE_DIS	8
 
 #define MMC_RSP_PRESENT (1 << 0)
 #define MMC_RSP_136	(1 << 1)		/* 136 bit response */
Signature · 1 line
2.35.1
Reviewed-by on v1 3/4
#9 lore raw

Re: [PATCH 3/4] mmc: Import some defines for R1 responses

Quoted 4 lines · Sean Anderson

This imports defines for R1 responses from include/linux/mmc/mmc.h from Linux 5.10.

Best Regards,
Jaehoon Chung

Quoted 70 lines

---
Yes, this is an old version, but it's what I had checked out, and I don't think there are any new fields we need to handle :)

 include/mmc.h | 52 +++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 50 insertions(+), 2 deletions(-)
include/mmc.h +50 -2
index 6bdcce881d..7304eee0d4 100644
@@ -311,8 +311,56 @@ static inline bool mmc_is_tuning_cmd(uint cmdidx)
 #define EXT_CSD_WR_DATA_REL_USR		(1 << 0)	/* user data area WR_REL */
 #define EXT_CSD_WR_DATA_REL_GP(x)	(1 << ((x)+1))	/* GP part (x+1) WR_REL */
 
-#define R1_ILLEGAL_COMMAND		(1 << 22)
-#define R1_APP_CMD			(1 << 5)
+/*
+  MMC status in R1, for native mode (SPI bits are different)
+  Type
+	e : error bit
+	s : status bit
+	r : detected and set for the actual command response
+	x : detected and set during command execution. the host must poll
+            the card by sending status command in order to read these bits.
+  Clear condition
+	a : according to the card state
+	b : always related to the previous command. Reception of
+            a valid command will clear it (with a delay of one command)
+	c : clear by read
+ */
+
+#define R1_OUT_OF_RANGE		(1 << 31)	/* er, c */
+#define R1_ADDRESS_ERROR	(1 << 30)	/* erx, c */
+#define R1_BLOCK_LEN_ERROR	(1 << 29)	/* er, c */
+#define R1_ERASE_SEQ_ERROR      (1 << 28)	/* er, c */
+#define R1_ERASE_PARAM		(1 << 27)	/* ex, c */
+#define R1_WP_VIOLATION		(1 << 26)	/* erx, c */
+#define R1_CARD_IS_LOCKED	(1 << 25)	/* sx, a */
+#define R1_LOCK_UNLOCK_FAILED	(1 << 24)	/* erx, c */
+#define R1_COM_CRC_ERROR	(1 << 23)	/* er, b */
+#define R1_ILLEGAL_COMMAND	(1 << 22)	/* er, b */
+#define R1_CARD_ECC_FAILED	(1 << 21)	/* ex, c */
+#define R1_CC_ERROR		(1 << 20)	/* erx, c */
+#define R1_ERROR		(1 << 19)	/* erx, c */
+#define R1_UNDERRUN		(1 << 18)	/* ex, c */
+#define R1_OVERRUN		(1 << 17)	/* ex, c */
+#define R1_CID_CSD_OVERWRITE	(1 << 16)	/* erx, c, CID/CSD overwrite */
+#define R1_WP_ERASE_SKIP	(1 << 15)	/* sx, c */
+#define R1_CARD_ECC_DISABLED	(1 << 14)	/* sx, a */
+#define R1_ERASE_RESET		(1 << 13)	/* sr, c */
+#define R1_STATUS(x)            (x & 0xFFF9A000)
+#define R1_CURRENT_STATE(x)	((x & 0x00001E00) >> 9)	/* sx, b (4 bits) */
+#define R1_READY_FOR_DATA	(1 << 8)	/* sx, a */
+#define R1_SWITCH_ERROR		(1 << 7)	/* sx, c */
+#define R1_EXCEPTION_EVENT	(1 << 6)	/* sr, a */
+#define R1_APP_CMD		(1 << 5)	/* sr, c */
+
+#define R1_STATE_IDLE	0
+#define R1_STATE_READY	1
+#define R1_STATE_IDENT	2
+#define R1_STATE_STBY	3
+#define R1_STATE_TRAN	4
+#define R1_STATE_DATA	5
+#define R1_STATE_RCV	6
+#define R1_STATE_PRG	7
+#define R1_STATE_DIS	8
 
 #define MMC_RSP_PRESENT (1 << 0)
 #define MMC_RSP_136	(1 << 1)		/* 136 bit response */
#10 lore raw

PATCH 4/4 mmc: sandbox: Set the response

In reply to Sean Anderson

The mmc subsystem checks the response, but we (almost) never set it. Add a bare-bones implementation. Technically, we are supposed to return our current state in our responses, but I don't think the subsystem checks it.

---

 drivers/mmc/sandbox_mmc.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)
drivers/mmc/sandbox_mmc.c +8 -2
index 60a6be0add..8922eeba2a 100644
@@ -44,6 +44,9 @@ static int sandbox_mmc_send_cmd(struct udevice *dev, struct mmc_cmd *cmd,
 	struct mmc *mmc = mmc_get_mmc_dev(dev);
 	static ulong erase_start, erase_end;
 
+	/* Default R1 response */
+	cmd->response[0] = R1_READY_FOR_DATA;
+
 	switch (cmd->cmdidx) {
 	case MMC_CMD_ALL_SEND_CID:
 		memset(cmd->response, '\0', sizeof(cmd->response));
@@ -55,8 +58,8 @@ static int sandbox_mmc_send_cmd(struct udevice *dev, struct mmc_cmd *cmd,
 	case SD_CMD_SEND_IF_COND:
 		cmd->response[0] = 0xaa;
 		break;
-	case MMC_CMD_SEND_STATUS:
-		cmd->response[0] = MMC_STATUS_RDY_FOR_DATA;
+	case SD_CMD_APP_SD_STATUS:
+		cmd->response[0] |= R1_APP_CMD;
 		memset(data->dest, '\0', data->blocks * data->blocksize);
 		break;
 	case MMC_CMD_SELECT_CARD:
@@ -106,6 +109,7 @@ static int sandbox_mmc_send_cmd(struct udevice *dev, struct mmc_cmd *cmd,
 		cmd->response[2] = 0;
 		break;
 	case MMC_CMD_APP_CMD:
+		cmd->response[0] |= R1_APP_CMD;
 		break;
 	case MMC_CMD_SET_BLOCKLEN:
 		debug("block len %d\n", cmd->cmdarg);
@@ -113,11 +117,13 @@ static int sandbox_mmc_send_cmd(struct udevice *dev, struct mmc_cmd *cmd,
 	case SD_CMD_APP_SEND_SCR: {
 		u32 *scr = (u32 *)data->dest;
 
+		cmd->response[0] |= R1_APP_CMD;
 		scr[0] = cpu_to_be32(2 << 24 | 1 << 15);  /* SD version 3 */
 		break;
 	}
 	default:
 		debug("%s: Unknown command %d\n", __func__, cmd->cmdidx);
+		cmd->response[0] |= R1_ILLEGAL_COMMAND;
 		break;
 	}
 
Signature · 1 line
2.35.1
Reviewed-by on v1 4/4
#11 lore raw

Re: [PATCH 4/4] mmc: sandbox: Set the response

Quoted 5 lines · Sean Anderson

The mmc subsystem checks the response, but we (almost) never set it. Add a bare-bones implementation. Technically, we are supposed to return our current state in our responses, but I don't think the subsystem checks it.

Best Regards,
Jaehoon Chung

Quoted 52 lines

---

 drivers/mmc/sandbox_mmc.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)
drivers/mmc/sandbox_mmc.c +8 -2
index 60a6be0add..8922eeba2a 100644
@@ -44,6 +44,9 @@ static int sandbox_mmc_send_cmd(struct udevice *dev, struct mmc_cmd *cmd,
 	struct mmc *mmc = mmc_get_mmc_dev(dev);
 	static ulong erase_start, erase_end;
 
+	/* Default R1 response */
+	cmd->response[0] = R1_READY_FOR_DATA;
+
 	switch (cmd->cmdidx) {
 	case MMC_CMD_ALL_SEND_CID:
 		memset(cmd->response, '\0', sizeof(cmd->response));
@@ -55,8 +58,8 @@ static int sandbox_mmc_send_cmd(struct udevice *dev, struct mmc_cmd *cmd,
 	case SD_CMD_SEND_IF_COND:
 		cmd->response[0] = 0xaa;
 		break;
-	case MMC_CMD_SEND_STATUS:
-		cmd->response[0] = MMC_STATUS_RDY_FOR_DATA;
+	case SD_CMD_APP_SD_STATUS:
+		cmd->response[0] |= R1_APP_CMD;
 		memset(data->dest, '\0', data->blocks * data->blocksize);
 		break;
 	case MMC_CMD_SELECT_CARD:
@@ -106,6 +109,7 @@ static int sandbox_mmc_send_cmd(struct udevice *dev, struct mmc_cmd *cmd,
 		cmd->response[2] = 0;
 		break;
 	case MMC_CMD_APP_CMD:
+		cmd->response[0] |= R1_APP_CMD;
 		break;
 	case MMC_CMD_SET_BLOCKLEN:
 		debug("block len %d\n", cmd->cmdarg);
@@ -113,11 +117,13 @@ static int sandbox_mmc_send_cmd(struct udevice *dev, struct mmc_cmd *cmd,
 	case SD_CMD_APP_SEND_SCR: {
 		u32 *scr = (u32 *)data->dest;
 
+		cmd->response[0] |= R1_APP_CMD;
 		scr[0] = cpu_to_be32(2 << 24 | 1 << 15);  /* SD version 3 */
 		break;
 	}
 	default:
 		debug("%s: Unknown command %d\n", __func__, cmd->cmdidx);
+		cmd->response[0] |= R1_ILLEGAL_COMMAND;
 		break;
 	}