openwrt/target/linux/generic-2.6/patches-2.6.31/065-rootfs_split.patch

--- a/drivers/mtd/Kconfig
+++ b/drivers/mtd/Kconfig
@@ -53,6 +53,16 @@ config MTD_TESTS
 	  should normally be compiled as kernel modules. The modules perform
 	  various checks and verifications when loaded.
 
+config MTD_ROOTFS_ROOT_DEV
+	bool "Automatically set 'rootfs' partition to be root filesystem"
+	depends on MTD_PARTITIONS
+	default y
+
+config MTD_ROOTFS_SPLIT
+	bool "Automatically split 'rootfs' partition for squashfs"
+	depends on MTD_PARTITIONS
+	default y
+
 config MTD_REDBOOT_PARTS
 	tristate "RedBoot partition table parsing"
 	depends on MTD_PARTITIONS
--- a/drivers/mtd/mtdpart.c
+++ b/drivers/mtd/mtdpart.c
@@ -18,6 +18,8 @@
 #include <linux/mtd/mtd.h>
 #include <linux/mtd/partitions.h>
 #include <linux/mtd/compatmac.h>
+#include <linux/root_dev.h>
+#include <linux/magic.h>
 
 /* Our partition linked list */
 static LIST_HEAD(mtd_partitions);
@@ -35,7 +37,7 @@ struct mtd_part {
  * the pointer to that structure with this macro.
  */
 #define PART(x)  ((struct mtd_part *)(x))
-
+#define IS_PART(mtd) (mtd->read == part_read)
 
 /*
  * MTD methods which simply translate the effective address and pass through
@@ -502,6 +504,150 @@ out_register:
 	return slave;
 }
 
+#ifdef CONFIG_MTD_ROOTFS_SPLIT
+#define ROOTFS_SPLIT_NAME "rootfs_data"
+#define ROOTFS_REMOVED_NAME "<removed>"
+
+struct squashfs_super_block {
+	__le32 s_magic;
+	__le32 pad0[9];
+	__le64 bytes_used;
+};
+
+
+static int split_squashfs(struct mtd_info *master, int offset, int *split_offset)
+{
+	struct squashfs_super_block sb;
+	int len, ret;
+
+	ret = master->read(master, offset, sizeof(sb), &len, (void *) &sb);
+	if (ret || (len != sizeof(sb))) {
+		printk(KERN_ALERT "split_squashfs: error occured while reading "
+			"from \"%s\"\n", master->name);
+		return -EINVAL;
+	}
+
+	if (SQUASHFS_MAGIC != le32_to_cpu(sb.s_magic) ) {
+		printk(KERN_ALERT "split_squashfs: no squashfs found in \"%s\"\n",
+			master->name);
+		*split_offset = 0;
+		return 0;
+	}
+
+	if (le64_to_cpu((sb.bytes_used)) <= 0) {
+		printk(KERN_ALERT "split_squashfs: squashfs is empty in \"%s\"\n",
+			master->name);
+		*split_offset = 0;
+		return 0;
+	}
+
+	len = (u32) le64_to_cpu(sb.bytes_used);
+	len += (offset & 0x000fffff);
+	len +=  (master->erasesize - 1);
+	len &= ~(master->erasesize - 1);
+	len -= (offset & 0x000fffff);
+	*split_offset = offset + len;
+
+	return 0;
+}
+
+static int split_rootfs_data(struct mtd_info *master, struct mtd_info *rpart, const struct mtd_partition *part)
+{
+	struct mtd_partition *dpart;
+	struct mtd_part *slave = NULL;
+	int split_offset = 0;
+	int ret;
+
+	ret = split_squashfs(master, part->offset, &split_offset);
+	if (ret)
+		return ret;
+
+	if (split_offset <= 0)
+		return 0;
+
+	dpart = kmalloc(sizeof(*part)+sizeof(ROOTFS_SPLIT_NAME)+1, GFP_KERNEL);
+	if (dpart == NULL) {
+		printk(KERN_INFO "split_squashfs: no memory for partition \"%s\"\n",
+			ROOTFS_SPLIT_NAME);
+		return -ENOMEM;
+	}
+
+	memcpy(dpart, part, sizeof(*part));
+	dpart->name = (unsigned char *)&dpart[1];
+	strcpy(dpart->name, ROOTFS_SPLIT_NAME);
+
+	dpart->size -= split_offset - dpart->offset;
+	dpart->offset = split_offset;
+
+	if (dpart == NULL)
+		return 1;
+
+	printk(KERN_INFO "mtd: partition \"%s\" created automatically, ofs=%llX, len=%llX \n",
+		ROOTFS_SPLIT_NAME, dpart->offset, dpart->size);
+
+	slave = add_one_partition(master, dpart, 0, split_offset);
+	if (!slave) {
+		kfree(dpart);
+		return -ENOMEM;
+	}
+	rpart->split = &slave->mtd;
+
+	return 0;
+}
+
+static int refresh_rootfs_split(struct mtd_info *mtd)
+{
+	struct mtd_partition tpart;
+	struct mtd_part *part;
+	char *name;
+	//int index = 0;
+	int offset, size;
+	int ret;
+
+	part = PART(mtd);
+
+	/* check for the new squashfs offset first */
+	ret = split_squashfs(part->master, part->offset, &offset);
+	if (ret)
+		return ret;
+
+	if ((offset > 0) && !mtd->split) {
+		printk(KERN_INFO "%s: creating new split partition for \"%s\"\n", __func__, mtd->name);
+		/* if we don't have a rootfs split partition, create a new one */
+		tpart.name = (char *) mtd->name;
+		tpart.size = mtd->size;
+		tpart.offset = part->offset;
+
+		return split_rootfs_data(part->master, &part->mtd, &tpart);
+	} else if ((offset > 0) && mtd->split) {
+		/* update the offsets of the existing partition */
+		size = mtd->size + part->offset - offset;
+
+		part = PART(mtd->split);
+		part->offset = offset;
+		part->mtd.size = size;
+		printk(KERN_INFO "%s: %s partition \"" ROOTFS_SPLIT_NAME "\", offset: 0x%06x (0x%06x)\n",
+			__func__, (!strcmp(part->mtd.name, ROOTFS_SPLIT_NAME) ? "updating" : "creating"),
+			(u32) part->offset, (u32) part->mtd.size);
+		name = kmalloc(sizeof(ROOTFS_SPLIT_NAME) + 1, GFP_KERNEL);
+		strcpy(name, ROOTFS_SPLIT_NAME);
+		part->mtd.name = name;
+	} else if ((offset <= 0) && mtd->split) {
+		printk(KERN_INFO "%s: removing partition \"%s\"\n", __func__, mtd->split->name);
+
+		/* mark existing partition as removed */
+		part = PART(mtd->split);
+		name = kmalloc(sizeof(ROOTFS_SPLIT_NAME) + 1, GFP_KERNEL);
+		strcpy(name, ROOTFS_REMOVED_NAME);
+		part->mtd.name = name;
+		part->offset = 0;
+		part->mtd.size = 0;
+	}
+
+	return 0;
+}
+#endif /* CONFIG_MTD_ROOTFS_SPLIT */
+
 /*
  * This function, given a master MTD object and a partition table, creates
  * and registers slave MTD objects which are bound to the master according to
@@ -517,7 +663,7 @@ int add_mtd_partitions(struct mtd_info *
 {
 	struct mtd_part *slave;
 	uint64_t cur_offset = 0;
-	int i;
+	int i, ret;
 
 	printk(KERN_NOTICE "Creating %d MTD partitions on \"%s\":\n", nbparts, master->name);
 
@@ -525,6 +671,21 @@ int add_mtd_partitions(struct mtd_info *
 		slave = add_one_partition(master, parts + i, i, cur_offset);
 		if (!slave)
 			return -ENOMEM;
+
+		if (!strcmp(parts[i].name, "rootfs")) {
+#ifdef CONFIG_MTD_ROOTFS_ROOT_DEV
+			if (ROOT_DEV == 0) {
+				printk(KERN_NOTICE "mtd: partition \"rootfs\" "
+					"set to be root filesystem\n");
+				ROOT_DEV = MKDEV(MTD_BLOCK_MAJOR, i);
+			}
+#endif
+#ifdef CONFIG_MTD_ROOTFS_SPLIT
+			ret = split_rootfs_data(master, &slave->mtd, &parts[i]);
+			/* if (ret == 0)
+				j++; */
+#endif
+		}
 		cur_offset = slave->offset + slave->mtd.size;
 	}
 
@@ -532,6 +693,32 @@ int add_mtd_partitions(struct mtd_info *
 }
 EXPORT_SYMBOL(add_mtd_partitions);
 
+int refresh_mtd_partitions(struct mtd_info *mtd)
+{
+	int ret = 0;
+
+	if (IS_PART(mtd)) {
+		struct mtd_part *part;
+		struct mtd_info *master;
+
+		part = PART(mtd);
+		master = part->master;
+		if (master->refresh_device)
+			ret = master->refresh_device(master);
+	}
+
+	if (!ret && mtd->refresh_device)
+		ret = mtd->refresh_device(mtd);
+
+#ifdef CONFIG_MTD_ROOTFS_SPLIT
+	if (!ret && IS_PART(mtd) && !strcmp(mtd->name, "rootfs"))
+		refresh_rootfs_split(mtd);
+#endif
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(refresh_mtd_partitions);
+
 static DEFINE_SPINLOCK(part_parser_lock);
 static LIST_HEAD(part_parsers);
 
--- a/drivers/mtd/devices/block2mtd.c
+++ b/drivers/mtd/devices/block2mtd.c
@@ -29,6 +29,8 @@ struct block2mtd_dev {
 	struct block_device *blkdev;
 	struct mtd_info mtd;
 	struct mutex write_mutex;
+	rwlock_t bdev_mutex;
+	char devname[0];
 };
 
 
@@ -81,6 +83,12 @@ static int block2mtd_erase(struct mtd_in
 	size_t len = instr->len;
 	int err;
 
+	read_lock(&dev->bdev_mutex);
+	if (!dev->blkdev) {
+		err = -EINVAL;
+		goto done;
+	}
+
 	instr->state = MTD_ERASING;
 	mutex_lock(&dev->write_mutex);
 	err = _block2mtd_erase(dev, from, len);
@@ -93,6 +101,10 @@ static int block2mtd_erase(struct mtd_in
 
 	instr->state = MTD_ERASE_DONE;
 	mtd_erase_callback(instr);
+
+done:
+	read_unlock(&dev->bdev_mutex);
+
 	return err;
 }
 
@@ -104,10 +116,14 @@ static int block2mtd_read(struct mtd_inf
 	struct page *page;
 	int index = from >> PAGE_SHIFT;
 	int offset = from & (PAGE_SIZE-1);
-	int cpylen;
+	int cpylen, err = 0;
+
+	read_lock(&dev->bdev_mutex);
+	if (!dev->blkdev || (from > mtd->size)) {
+		err = -EINVAL;
+		goto done;
+	}
 
-	if (from > mtd->size)
-		return -EINVAL;
 	if (from + len > mtd->size)
 		len = mtd->size - from;
 
@@ -122,10 +138,14 @@ static int block2mtd_read(struct mtd_inf
 		len = len - cpylen;
 
 		page = page_read(dev->blkdev->bd_inode->i_mapping, index);
-		if (!page)
-			return -ENOMEM;
-		if (IS_ERR(page))
-			return PTR_ERR(page);
+		if (!page) {
+			err = -ENOMEM;
+			goto done;
+		}
+		if (IS_ERR(page)) {
+			err = PTR_ERR(page);
+			goto done;
+		}
 
 		memcpy(buf, page_address(page) + offset, cpylen);
 		page_cache_release(page);
@@ -136,7 +156,10 @@ static int block2mtd_read(struct mtd_inf
 		offset = 0;
 		index++;
 	}
-	return 0;
+
+done:
+	read_unlock(&dev->bdev_mutex);
+	return err;
 }
 
 
@@ -188,12 +211,22 @@ static int block2mtd_write(struct mtd_in
 		size_t *retlen, const u_char *buf)
 {
 	struct block2mtd_dev *dev = mtd->priv;
-	int err;
+	int err = 0;
+
+	read_lock(&dev->bdev_mutex);
+	if (!dev->blkdev) {
+		err = -EINVAL;
+		goto done;
+	}
 
 	if (!len)
-		return 0;
-	if (to >= mtd->size)
-		return -ENOSPC;
+		goto done;
+
+	if (to >= mtd->size) {
+		err = -ENOSPC;
+		goto done;
+	}
+
 	if (to + len > mtd->size)
 		len = mtd->size - to;
 
@@ -202,6 +235,9 @@ static int block2mtd_write(struct mtd_in
 	mutex_unlock(&dev->write_mutex);
 	if (err > 0)
 		err = 0;
+
+done:
+	read_unlock(&dev->bdev_mutex);
 	return err;
 }
 
@@ -210,52 +246,29 @@ static int block2mtd_write(struct mtd_in
 static void block2mtd_sync(struct mtd_info *mtd)
 {
 	struct block2mtd_dev *dev = mtd->priv;
-	sync_blockdev(dev->blkdev);
-	return;
-}
-
-
-static void block2mtd_free_device(struct block2mtd_dev *dev)
-{
-	if (!dev)
-		return;
-
-	kfree(dev->mtd.name);
 
-	if (dev->blkdev) {
-		invalidate_mapping_pages(dev->blkdev->bd_inode->i_mapping,
-					0, -1);
-		close_bdev_exclusive(dev->blkdev, FMODE_READ|FMODE_WRITE);
-	}
+	read_lock(&dev->bdev_mutex);
+	if (dev->blkdev)
+		sync_blockdev(dev->blkdev);
+	read_unlock(&dev->bdev_mutex);
 
-	kfree(dev);
+	return;
 }
 
 
-/* FIXME: ensure that mtd->size % erase_size == 0 */
-static struct block2mtd_dev *add_device(char *devname, int erase_size, const char *mtdname)
+static int _open_bdev(struct block2mtd_dev *dev)
 {
 	struct block_device *bdev;
-	struct block2mtd_dev *dev;
-	struct mtd_partition *part;
-	char *name;
-
-	if (!devname)
-		return NULL;
-
-	dev = kzalloc(sizeof(struct block2mtd_dev), GFP_KERNEL);
-	if (!dev)
-		return NULL;
 
 	/* Get a handle on the device */
-	bdev = open_bdev_exclusive(devname, FMODE_READ|FMODE_WRITE, NULL);
+	bdev = open_bdev_exclusive(dev->devname, FMODE_READ|FMODE_WRITE, NULL);
 #ifndef MODULE
 	if (IS_ERR(bdev)) {
 
 		/* We might not have rootfs mounted at this point. Try
 		   to resolve the device name by other means. */
 
-		dev_t devt = name_to_dev_t(devname);
+		dev_t devt = name_to_dev_t(dev->devname);
 		if (devt) {
 			bdev = open_by_devnum(devt, FMODE_WRITE | FMODE_READ);
 		}
@@ -263,17 +276,97 @@ static struct block2mtd_dev *add_device(
 #endif
 
 	if (IS_ERR(bdev)) {
-		ERROR("error: cannot open device %s", devname);
-		goto devinit_err;
+		ERROR("error: cannot open device %s", dev->devname);
+		return 1;
 	}
 	dev->blkdev = bdev;
 
 	if (MAJOR(bdev->bd_dev) == MTD_BLOCK_MAJOR) {
 		ERROR("attempting to use an MTD device as a block device");
-		goto devinit_err;
+		return 1;
 	}
 
+	return 0;
+}
+
+static void _close_bdev(struct block2mtd_dev *dev)
+{
+	struct block_device *bdev;
+
+	if (!dev->blkdev)
+		return;
+
+	bdev = dev->blkdev;
+	invalidate_mapping_pages(dev->blkdev->bd_inode->i_mapping, 0, -1);
+	close_bdev_exclusive(dev->blkdev, FMODE_READ|FMODE_WRITE);
+	dev->blkdev = NULL;
+}
+
+static void block2mtd_free_device(struct block2mtd_dev *dev)
+{
+	if (!dev)
+		return;
+
+	kfree(dev->mtd.name);
+	_close_bdev(dev);
+	kfree(dev);
+}
+
+
+static int block2mtd_refresh(struct mtd_info *mtd)
+{
+	struct block2mtd_dev *dev = mtd->priv;
+	struct block_device *bdev;
+	dev_t devt;
+	int err = 0;
+
+	/* no other mtd function can run at this point */
+	write_lock(&dev->bdev_mutex);
+
+	/* get the device number for the whole disk */
+	devt = MKDEV(MAJOR(dev->blkdev->bd_dev), 0);
+
+	/* close the old block device */
+	_close_bdev(dev);
+
+	/* open the whole disk, issue a partition rescan, then */
+	bdev = open_by_devnum(devt, FMODE_WRITE | FMODE_READ);
+	if (!bdev || !bdev->bd_disk)
+		err = -EINVAL;
+	else {
+		err = rescan_partitions(bdev->bd_disk, bdev);
+	}
+	if (bdev)
+		close_bdev_exclusive(bdev, FMODE_READ|FMODE_WRITE);
+
+	/* try to open the partition block device again */
+	_open_bdev(dev);
+	write_unlock(&dev->bdev_mutex);
+
+	return err;
+}
+
+/* FIXME: ensure that mtd->size % erase_size == 0 */
+static struct block2mtd_dev *add_device(char *devname, int erase_size, char *mtdname)
+{
+	struct block2mtd_dev *dev;
+	struct mtd_partition *part;
+	char *name;
+
+	if (!devname)
+		return NULL;
+
+	dev = kzalloc(sizeof(struct block2mtd_dev) + strlen(devname) + 1, GFP_KERNEL);
+	if (!dev)
+		return NULL;
+
+	strcpy(dev->devname, devname);
+
+	if (_open_bdev(dev))
+		goto devinit_err;
+
 	mutex_init(&dev->write_mutex);
+	rwlock_init(&dev->bdev_mutex);
 
 	if (!mtdname)
 		mtdname = devname;
@@ -297,6 +390,7 @@ static struct block2mtd_dev *add_device(
 	dev->mtd.read = block2mtd_read;
 	dev->mtd.priv = dev;
 	dev->mtd.owner = THIS_MODULE;
+	dev->mtd.refresh_device = block2mtd_refresh;
 
 	part = kzalloc(sizeof(struct mtd_partition), GFP_KERNEL);
 	part->name = dev->mtd.name;
--- a/drivers/mtd/mtdchar.c
+++ b/drivers/mtd/mtdchar.c
@@ -18,6 +18,7 @@
 
 #include <linux/mtd/mtd.h>
 #include <linux/mtd/compatmac.h>
+#include <linux/mtd/partitions.h>
 
 #include <asm/uaccess.h>
 
@@ -814,6 +815,13 @@ static int mtd_ioctl(struct inode *inode
 		file->f_pos = 0;
 		break;
 	}
+#ifdef CONFIG_MTD_PARTITIONS
+	case MTDREFRESH:
+	{
+		ret = refresh_mtd_partitions(mtd);
+		break;
+	}
+#endif
 
 	default:
 		ret = -ENOTTY;
--- a/include/linux/mtd/mtd.h
+++ b/include/linux/mtd/mtd.h
@@ -101,6 +101,7 @@ struct mtd_oob_ops {
 	uint8_t		*oobbuf;
 };
 
+struct mtd_info;
 struct mtd_info {
 	u_char type;
 	uint32_t flags;
@@ -241,6 +242,9 @@ struct mtd_info {
 	struct device dev;
 	int usecount;
 
+	int (*refresh_device)(struct mtd_info *mtd);
+	struct mtd_info *split;
+
 	/* If the driver is something smart, like UBI, it may need to maintain
 	 * its own reference counting. The below functions are only for driver.
 	 * The driver may register its callbacks. These callbacks are not
--- a/include/linux/mtd/partitions.h
+++ b/include/linux/mtd/partitions.h
@@ -34,12 +34,14 @@
  * erasesize aligned (e.g. use MTDPART_OFS_NEXTBLK).
  */
 
+struct mtd_partition;
 struct mtd_partition {
 	char *name;			/* identifier string */
 	uint64_t size;			/* partition size */
 	uint64_t offset;		/* offset within the master MTD space */
 	uint32_t mask_flags;		/* master MTD flags to mask out for this partition */
 	struct nand_ecclayout *ecclayout;	/* out of band layout for this partition (NAND only)*/
+	int (*refresh_partition)(struct mtd_info *);
 };
 
 #define MTDPART_OFS_NXTBLK	(-2)
@@ -51,6 +53,7 @@ struct mtd_info;
 
 int add_mtd_partitions(struct mtd_info *, const struct mtd_partition *, int);
 int del_mtd_partitions(struct mtd_info *);
+int refresh_mtd_partitions(struct mtd_info *);
 
 /*
  * Functions dealing with the various ways of partitioning the space
--- a/include/mtd/mtd-abi.h
+++ b/include/mtd/mtd-abi.h
@@ -110,6 +110,7 @@ struct otp_info {
 #define MEMERASE64		_IOW('M', 20, struct erase_info_user64)
 #define MEMWRITEOOB64		_IOWR('M', 21, struct mtd_oob_buf64)
 #define MEMREADOOB64		_IOWR('M', 22, struct mtd_oob_buf64)
+#define MTDREFRESH		_IO('M', 23)
 
 /*
  * Obsolete legacy interface. Keep it in order not to break userspace