Skip to contentSkip to conversation

DFU: fix for raw data write

PATCH New Mateusz Zalega 3 participants 6 messages Aug 6, 2013 – Sep 5, 2013

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 DFU: fix for raw data write

This commit fixes DFU raw (performed w/o partition interface, via mmc_block_op()) write.

Mateusz Zalega (1):
DFU: fix for raw data write

 drivers/dfu/dfu.c     | 22 ++++++++++++++++++++++
 drivers/dfu/dfu_mmc.c |  3 +++
 include/dfu.h         |  7 ++-----
 3 files changed, 27 insertions(+), 5 deletions(-)
Signature · 1 line
1.8.2.1
#1 lore raw

PATCH 1/1 DFU: fix for raw data write

When user attempted to perform a raw write using DFU (vide dfu_fill_entity_mmc) with MMC interface not initialized before, get_mmc_blk_size() reported invalid (zero) block size - it wasn't possible to write ie. a new u-boot image.

This commit fixes that by initializing device in get_mmc_blk_size() when needed.

Tested on Samsung Goni.

---
 drivers/dfu/dfu.c     | 22 ++++++++++++++++++++++
 drivers/dfu/dfu_mmc.c |  3 +++
 include/dfu.h         |  7 ++-----
 3 files changed, 27 insertions(+), 5 deletions(-)
drivers/dfu/dfu.c +22 -0
index d73d510..6979112 100644
@@ -414,3 +414,25 @@ struct dfu_entity *dfu_get_entity(int alt)
 
 	return NULL;
 }
+
+int get_mmc_blk_size(int dev)
+{
+	struct mmc *mmc = find_mmc_device(dev);
+
+	if (mmc == NULL) {
+		error("Couldn't find MMC device no. %d.\n", dev);
+		return -ENODEV;
+	}
+	if (!mmc->has_init) {
+		if (!mmc_init(mmc)) {
+			if (!mmc->read_bl_len) {
+				error("invalid block length\n");
+				return -ENODEV;
+			}
+		} else {
+			error("Couldn't init MMC device.\n");
+			return -ENODEV;
+		}
+	}
+	return mmc->read_bl_len;
+}
drivers/dfu/dfu_mmc.c +3 -0
index 0871a77..c776e26 100644
@@ -235,5 +235,8 @@ int dfu_fill_entity_mmc(struct dfu_entity *dfu, char *s)
 	/* initial state */
 	dfu->inited = 0;
 
+	/* better safe than sorry */
+	assert(dfu->data.mmc.lba_blk_size > 0);
+
 	return 0;
 }
include/dfu.h +2 -5
index 1d4006d..67ca711 100644
@@ -40,6 +40,8 @@ struct mmc_internal_data {
 	unsigned int part;
 };
 
+int get_mmc_blk_size(int dev);
+
 struct nand_internal_data {
 	/* RAW programming */
 	u64 start;
@@ -49,11 +51,6 @@ struct nand_internal_data {
 	unsigned int part;
 };
 
-static inline unsigned int get_mmc_blk_size(int dev)
-{
-	return find_mmc_device(dev)->read_bl_len;
-}
-
 #define DFU_NAME_SIZE			32
 #define DFU_CMD_BUF_SIZE		128
 #ifndef CONFIG_SYS_DFU_DATA_BUF_SIZE
Signature · 1 line
1.8.2.1
#2 lore raw

PATCH 1/1 DFU: fix for raw data write

Hi Mateusz,

drivers/dfu/dfu.c

Quoted 24 lines

When user attempted to perform a raw write using DFU (vide dfu_fill_entity_mmc) with MMC interface not initialized before, get_mmc_blk_size() reported invalid (zero) block size - it wasn't possible to write ie. a new u-boot image.

This commit fixes that by initializing device in get_mmc_blk_size() when needed.

Tested on Samsung Goni.

---
 drivers/dfu/dfu.c     | 22 ++++++++++++++++++++++
 drivers/dfu/dfu_mmc.c |  3 +++
 include/dfu.h         |  7 ++-----
 3 files changed, 27 insertions(+), 5 deletions(-)
drivers/dfu/dfu.c +0 -0
index d73d510..6979112 100644

The mmc (or any other medium) related code shall NOT be added to generic dfu.c file.

Quoted 16 lines
drivers/dfu/dfu.c +12 -0
@@ -414,3 +414,25 @@ struct dfu_entity *dfu_get_entity(int alt)
 
 	return NULL;
 }
+
+int get_mmc_blk_size(int dev)
+{
+	struct mmc *mmc = find_mmc_device(dev);
+
+	if (mmc == NULL) {
+		error("Couldn't find MMC device no. %d.\n", dev);
+		return -ENODEV;
+	}
+	if (!mmc->has_init) {
+		if (!mmc_init(mmc)) {
+			if (!mmc->read_bl_len) {

drivers/dfu/dfu.c struct dfu_entity *dfu_get_entity(int alt) · line 438

Quoted 10 lines
drivers/dfu/dfu.c +10 -0
+				error("invalid block length\n");
+				return -ENODEV;
+			}
+		} else {
+			error("Couldn't init MMC device.\n");
+			return -ENODEV;
+		}
+	}
+	return mmc->read_bl_len;
+}

The above code fits better at dfu_mmc.c file. My proposition would be to clean up the dfu_fill_entity_mmc() function and add the above code there.

Quoted 38 lines
drivers/dfu/dfu_mmc.c +0 -0
index 0871a77..c776e26 100644
@@ -235,5 +235,8 @@ int dfu_fill_entity_mmc(struct dfu_entity *dfu,

char *s) /* initial state */

drivers/dfu/dfu_mmc.c +3 -0
 	dfu->inited = 0;
 
+	/* better safe than sorry */
+	assert(dfu->data.mmc.lba_blk_size > 0);
+
 	return 0;
 }
include/dfu.h +2 -5
index 1d4006d..67ca711 100644
@@ -40,6 +40,8 @@ struct mmc_internal_data {
 	unsigned int part;
 };
 
+int get_mmc_blk_size(int dev);
+
 struct nand_internal_data {
 	/* RAW programming */
 	u64 start;
@@ -49,11 +51,6 @@ struct nand_internal_data {
 	unsigned int part;
 };
 
-static inline unsigned int get_mmc_blk_size(int dev)
-{
-	return find_mmc_device(dev)->read_bl_len;
-}
-
 #define DFU_NAME_SIZE			32
 #define DFU_CMD_BUF_SIZE		128
 #ifndef CONFIG_SYS_DFU_DATA_BUF_SIZE
Signature · 5 lines
Best regards,

Lukasz Majewski

Samsung R&D Institute Poland (SRPOL) | Linux Platform Group
#3 lore raw

PATCH v2 1/1 dfu:mmc: raw data write fix

In reply to Mateusz Zalega

When user attempted to perform a raw write using DFU (vide dfu_fill_entity_mmc) with MMC interface not initialized before, get_mmc_blk_size() reported invalid (zero) block size - it wasn't possible to write ie. a new u-boot image.

This commit fixes that by initializing device in get_mmc_blk_size() when needed.

Tested on Samsung Goni.

---
Changes for v2:
- code cleanup
- minor dfu_alt_info format change

---
 drivers/dfu/dfu_mmc.c        | 110 +++++++++++++++++++++++++++----------------
 include/configs/am335x_evm.h |   4 +-
 include/configs/s5p_goni.h   |   2 +-
 include/configs/trats.h      |   2 +-
 include/dfu.h                |   5 --
 5 files changed, 73 insertions(+), 50 deletions(-)
drivers/dfu/dfu_mmc.c +69 -41
index 0871a77..a91793b 100644
@@ -173,66 +173,94 @@ int dfu_read_medium_mmc(struct dfu_entity *dfu, u64 offset, void *buf,
 	return ret;
 }
 
+/*
+ * @param s Parameter string containing space-separated arguments:
+ *	1st:
+ *		raw	(raw read/write)
+ *		fat	(files)
+ *		ext4	(^)
+ *		part	(partition image)
+ *	2nd and 3rd:
+ *		lba_start and lba_size, for raw write
+ *		mmc_dev and mmc_part, for filesystems and part
+ */
 int dfu_fill_entity_mmc(struct dfu_entity *dfu, char *s)
 {
-	int dev, part;
-	struct mmc *mmc;
-	block_dev_desc_t *blk_dev;
-	disk_partition_t partinfo;
-	char *st;
-
-	dfu->dev_type = DFU_DEV_MMC;
-	st = strsep(&s, " ");
-	if (!strcmp(st, "mmc")) {
-		dfu->layout = DFU_RAW_ADDR;
-		dfu->data.mmc.lba_start = simple_strtoul(s, &s, 16);
-		dfu->data.mmc.lba_size = simple_strtoul(++s, &s, 16);
-		dfu->data.mmc.lba_blk_size = get_mmc_blk_size(dfu->dev_num);
-	} else if (!strcmp(st, "fat")) {
-		dfu->layout = DFU_FS_FAT;
-	} else if (!strcmp(st, "ext4")) {
-		dfu->layout = DFU_FS_EXT4;
-	} else if (!strcmp(st, "part")) {
-
-		dfu->layout = DFU_RAW_ADDR;
+	const char *argv[3];
+	const char **parg = argv;
+	for (; parg < argv + sizeof(argv) / sizeof(*argv); ++parg) {
+		*parg = strsep(&s, " ");
+		if (*parg == NULL) {
+			error("Invalid number of arguments.\n");
+			return -ENODEV;
+		}
+	}
 
-		dev = simple_strtoul(s, &s, 10);
-		s++;
-		part = simple_strtoul(s, &s, 10);
+	const char *entity_type = argv[0];
+	/*
+	 * Base 0 means we'll accept (prefixed with 0x or 0) base 16, 8,
+	 * with default 10.
+	 */
+	size_t second_arg = simple_strtoul(argv[1], NULL, 0);
+	size_t third_arg = simple_strtoul(argv[2], NULL, 0);
 
-		mmc = find_mmc_device(dev);
-		if (mmc == NULL || mmc_init(mmc)) {
-			printf("%s: could not find mmc device #%d!\n",
-			       __func__, dev);
+	struct mmc *mmc = find_mmc_device(dfu->dev_num);
+	if (mmc == NULL) {
+		error("Couldn't find MMC device no. %d.\n", dfu->dev_num);
+		return -ENODEV;
+	}
+	if (!mmc->has_init) {
+		if (!mmc_init(mmc)) {
+			if (!mmc->read_bl_len) {
+				error("invalid block length\n");
+				return -ENODEV;
+			}
+		} else {
+			error("Couldn't init MMC device.\n");
 			return -ENODEV;
 		}
+	}
 
-		blk_dev = &mmc->block_dev;
-		if (get_partition_info(blk_dev, part, &partinfo) != 0) {
-			printf("%s: could not find partition #%d on mmc device #%d!\n",
-			       __func__, part, dev);
+	if (!strcmp(entity_type, "raw")) {
+		dfu->layout			= DFU_RAW_ADDR;
+		dfu->data.mmc.lba_start		= second_arg;
+		dfu->data.mmc.lba_size		= third_arg;
+		dfu->data.mmc.lba_blk_size	= mmc->read_bl_len;
+	} else if (!strcmp(entity_type, "part")) {
+		disk_partition_t partinfo;
+		block_dev_desc_t *blk_dev = &mmc->block_dev;
+		int mmcdev = second_arg;
+		int mmcpart = third_arg;
+
+		if (get_partition_info(blk_dev, mmcpart, &partinfo) != 0) {
+			error("Couldn't find part #%d on mmc device #%d\n",
+			      mmcpart, mmcdev);
 			return -ENODEV;
 		}
 
-		dfu->data.mmc.lba_start = partinfo.start;
-		dfu->data.mmc.lba_size = partinfo.size;
-		dfu->data.mmc.lba_blk_size = partinfo.blksz;
-
+		dfu->layout			= DFU_RAW_ADDR;
+		dfu->data.mmc.lba_start		= partinfo.start;
+		dfu->data.mmc.lba_size		= partinfo.size;
+		dfu->data.mmc.lba_blk_size	= partinfo.blksz;
+	} else if (!strcmp(entity_type, "fat")) {
+		dfu->layout = DFU_FS_FAT;
+	} else if (!strcmp(entity_type, "ext4")) {
+		dfu->layout = DFU_FS_EXT4;
 	} else {
-		printf("%s: Memory layout (%s) not supported!\n", __func__, st);
+		error("Memory layout (%s) not supported!\n", entity_type);
 		return -ENODEV;
 	}
 
-	if (dfu->layout == DFU_FS_EXT4 || dfu->layout == DFU_FS_FAT) {
-		dfu->data.mmc.dev = simple_strtoul(s, &s, 10);
-		dfu->data.mmc.part = simple_strtoul(++s, &s, 10);
+	/* if it's NOT a raw write */
+	if (strcmp(entity_type, "raw")) {
+		dfu->data.mmc.dev = second_arg;
+		dfu->data.mmc.part = third_arg;
 	}
 
+	dfu->dev_type = DFU_DEV_MMC;
 	dfu->read_medium = dfu_read_medium_mmc;
 	dfu->write_medium = dfu_write_medium_mmc;
 	dfu->flush_medium = dfu_flush_medium_mmc;
-
-	/* initial state */
 	dfu->inited = 0;
 
 	return 0;
include/configs/am335x_evm.h +2 -2
index e0a87f8..aa036b5 100644
@@ -188,8 +188,8 @@
 	"boot part 0 1;" \
 	"rootfs part 0 2;" \
 	"MLO fat 0 1;" \
-	"MLO.raw mmc 100 100;" \
-	"u-boot.img.raw mmc 300 400;" \
+	"MLO.raw mmc 0x100 0x100;" \
+	"u-boot.img.raw mmc 0x300 0x400;" \
 	"spl-os-args.raw mmc 80 80;" \
 	"spl-os-image.raw mmc 900 2000;" \
 	"spl-os-args fat 0 1;" \
include/configs/s5p_goni.h +1 -1
index ef5c421..8c21ab5 100644
@@ -86,7 +86,7 @@
 #define CONFIG_ZERO_BOOTDELAY_CHECK
 
 #define CONFIG_DFU_ALT \
-	"u-boot mmc 80 400;" \
+	"u-boot raw 0x80 0x400;" \
 	"uImage fat 0 2\0" \
 
 /* partitions definitions */
include/configs/trats.h +1 -1
index 9b6aac9..d40e583 100644
@@ -129,7 +129,7 @@
 	"name="PARTS_UMS",size=-,uuid=${uuid_gpt_"PARTS_UMS"}\0" \
 
 #define CONFIG_DFU_ALT \
-	"u-boot mmc 80 400;" \
+	"u-boot raw 0x80 0x400;" \
 	"uImage ext4 0 2;" \
 	"exynos4210-trats.dtb ext4 0 2\0"
 
include/dfu.h +0 -5
index 1d4006d..ea8eb09 100644
@@ -49,11 +49,6 @@ struct nand_internal_data {
 	unsigned int part;
 };
 
-static inline unsigned int get_mmc_blk_size(int dev)
-{
-	return find_mmc_device(dev)->read_bl_len;
-}
-
 #define DFU_NAME_SIZE			32
 #define DFU_CMD_BUF_SIZE		128
 #ifndef CONFIG_SYS_DFU_DATA_BUF_SIZE
Signature · 1 line
1.8.2.1
#4 lore raw

PATCH v2 1/1 dfu:mmc: raw data write fix

Quoted 15 lines · Mateusz Zalega

When user attempted to perform a raw write using DFU (vide dfu_fill_entity_mmc) with MMC interface not initialized before, get_mmc_blk_size() reported invalid (zero) block size - it wasn't possible to write ie. a new u-boot image.

This commit fixes that by initializing device in get_mmc_blk_size() when needed.

Tested on Samsung Goni.

Quoted 4 lines

---
Changes for v2:
- code cleanup
- minor dfu_alt_info format change

Changes depend on "[PATCH RESEND v3] arm:goni: Update GONI configuration".

Signature · 2 lines
Mateusz Zalega
Samsung R&D Institute Poland
#5 lore raw

PATCH v2 1/1 dfu:mmc: raw data write fix

In reply to Mateusz Zalega
Quoted 8 lines · Mateusz Zalega

When user attempted to perform a raw write using DFU (vide dfu_fill_entity_mmc) with MMC interface not initialized before, get_mmc_blk_size() reported invalid (zero) block size - it wasn't possible to write ie. a new u-boot image.

This commit fixes that by initializing device in get_mmc_blk_size() when needed.

Quoted 6 lines

Tested on Samsung Goni.

I poked Marek on IRC yesterday, and he was fine with me taking this into master, but since it depends on some other samsung patches, I'm fine with this going in via u-boot-samsung.

Signature · 1 line
Tom