让Android系统支持ubifs文件系统

原文地址：http://www.cnblogs.com/linucos/p/3279381.html

1. ubifs号称性能比yaffs2 好，同时压缩可读写，文件系统image体较小同时可写，相当不错

2. ubifs制作

(1) mkfs.ubifs

mkfs.ubifs -r system -m 2048 -e 126976 -c 1057 -x zlib -o system.ubifs

-r说的目录

-m说的是nand flash的页大小，一般都是2048或者4096，当然也有些是512，硬件特性决定好了，或者说驱动规定好了，有些平台4096页当做2048用也是用的

-e说的是逻辑擦除块大小，大家知道nand flash页读页写块擦，一个设备多个块，一个块多个页，一般也都是一个块是64个页，这样算一下无论擦除块大小就是2048*64=131072，-e的算法是物理擦除块大小-2*页大小，这里就是131072-2*2048=126976

-c说的是最大逻辑块编号，这个很重要，不能大也不能小，太小也要大于image大小，太大mount有问题，计算起点是分区的物理块数量，比如128MiB的mtd分区，物理块数量是128MiB/2048/64 = 1024个，需要减去2个坏块保留块，减去1个wear-leveling块，还要减去1个eba的块，等等，比如最终的值是1022，注意，如果物理上这个分区有坏块的话，kernel会扫描到的，这时候，我们计算的这个值就要减去坏块数了，否则会有逻辑块大于物理块数的内核问题mount失败，确切知道坏块数是比较困难的，一般做法是做一个坏块容忍数，比如20个，这样我们再减去20个坏块，不要担心这个会浪费空间，ubinize的autoresize选项就是解决这个问题的。具体的这个值需要计算。！！！！

-o说的image名字

-x说的是压缩方法，默认是lzo，还支持zlib，zlib压缩率高些，但是lzo压缩解压速度快

(2)  ubinize

ubinize -o system.ubi -m 2048 -p 131072 ubinize.cfg

-o说的是输出image

-m还是页大小

-p是物理擦除块大小

ubinize.cfg是volume配置文件，例子如下：

1. [ubifs]
2. mode=ubi
3. image=system.ubifs #说的是mkfs.ubifs的结果
4. vol_id=0
5. vol_size=100MiB #说的是volume大小，用-e和-c的值做乘法计算，一般不用写，autoresize会自动根据mtd分区大小适应，默认值是image大小，写了这个作用是帮助检查image是否超过了分区限制，制作时候就提示，否则mount会出错。-c的值是经过计算的最大值了，不过autoresize参数会自适应大小，不会浪费空间的。
6. vol_type=dynamic
7. vol_alignment=1
8. vol_name=system #说的是分区名字
9. vol_flags=autoresize

(4) uboot支持

-0-  需要打开的配置 - 需要烧写mkfs.ubifs结果时候需要打开，ubinize处理过的不需要

CONFIG_CMD_UBI

CONFIG_CMD_UBIFS

CONFIG_LZO

CONFIG_RBTREE

CONFIG_ZLIB

CONFIG_GZIP

-1- ubifs烧写

mkfs.ubifs工具制作的结果，就是ubifs镜像，不包含volume信息，需要用

nand erase.part system

ubi part system - 激活分区

ubi create system - 创建分区

ubi write 84000000 system xxxxx - xxxxx表示镜像实际大小

-2- ubi volume bin烧写

mkfs.ubifs后，使用ubinize处理了ubifs的镜像后，镜像含有了volume信息，直接

nand write 84000000 system xxxxx - xxxxx表示镜像实际大小

(5) kernel支持

• Device Drivers --->
• Memory Technology Device(MTD) support --->
• UBI - Unsorted block images --->
• <*> Enable UBI - Unsorted block images
• File systems --->
• Miscellaneous filesystems --->
• <*> UBIFS file system support
• [*] Advanced compression options
• [*] LZO compression support
• [*] ZLIB compression support

(6) android支持

-1- andriod/system/core/rootdir/init.rc

mount yaffs2 mtd@userdata /data nosuid nodev

改为：

mount ubifs ubi@userdata /data nosuid nodev

-2- 增加对ubifs的mount支持

• andriod/system/core/init/builtins.c
• 在 } else if (!strncmp(source, "loop@", 5)) {
• 之前加上
• + }else if (!strncmp(source, "ubi@", 4)) {
• + n = ubi_attach_mtd(source + 4);
• + if (n < 0) {
• + return -1;
• + }
• + sprintf(tmp, "/dev/ubi%d_0", n);
• + if (wait)
• + wait_for_file(tmp, COMMAND_RETRY_TIMEOUT);
• + if (mount(tmp, target, system, flags, options) < 0) {
• + ubi_detach_dev(n);
• + return -1;
• + }
• + return 0;
• + }else if (!strncmp(source, "loop@", 5)) {
• 3. andriod/system/core/init/init.c
• 在static int property_triggers_enabled = 0;之后加上
• +static unsigned ubifs_enabled = 1;
• static int set_init_properties_action(int nargs, char **args)
• {
• property_set("ro.revision", tmp);
• + property_set("ro.ubifs",ubifs_enabled ? "1" : "0");
• return 0;
• }
• int main(int argc, char **argv)
• {
• action_for_each_trigger("post-fs", action_add_queue_tail);
• + action_for_each_trigger("ubi-fs", action_add_queue_tail);
• }
• 4. andriod/system/core/init/util.h
• void get_hardware_name(char *hardware, unsigned int *revision);
• +int ubi_attach_mtd(const char *name);
• +int ubi_detach_dev(int dev);
• 5. andriod/system/core/init/util.c
• #include <sys/un.h>
• +#include <sys/ioctl.h>
• #include "util.h"
• +#include "ubi-user.h"
• +#define UBI_CTRL_DEV "/dev/ubi_ctrl"
• +#define UBI_SYS_PATH "/sys/class/ubi"
• 在最后添加下面三个函数
• static int ubi_dev_read_int(int dev, const char *file, int def)
• {
• int fd, val = def;
• char path[128], buf[64];
• sprintf(path, UBI_SYS_PATH "/ubi%d/%s", dev, file);
• wait_for_file(path, 5);
• fd = open(path, O_RDONLY);
• if (fd == -1) {
• return val;
• }
• if (read(fd, buf, 64) > 0) {
• val = atoi(buf);
• }
• close(fd);
• return val;
• }
• int ubi_attach_mtd(const char *name)
• {
• int ret;
• int mtd_num, ubi_num;
• int ubi_ctrl, ubi_dev;
• int vols, avail_lebs, leb_size;
• char path[128];
• struct ubi_attach_req attach_req;
• struct ubi_mkvol_req mkvol_req;
• mtd_num = mtd_name_to_number(name);
• if (mtd_num == -1) {
• return -1;
• }
• ubi_ctrl = open(UBI_CTRL_DEV, O_RDONLY);
• if (ubi_ctrl == -1) {
• return -1;
• }
• memset(&attach_req, 0, sizeof(struct ubi_attach_req));
• attach_req.ubi_num = UBI_DEV_NUM_AUTO;
• attach_req.mtd_num = mtd_num;
• ret = ioctl(ubi_ctrl, UBI_IOCATT, &attach_req);
• if (ret == -1) {
• close(ubi_ctrl);
• return -1;
• }
• ubi_num = attach_req.ubi_num;
• vols = ubi_dev_read_int(ubi_num, "volumes_count", -1);
• if (vols == 0) {
• sprintf(path, "/dev/ubi%d", ubi_num);
• ubi_dev = open(path, O_RDONLY);
• if (ubi_dev == -1) {
• close(ubi_ctrl);
• return ubi_num;
• }
• avail_lebs = ubi_dev_read_int(ubi_num, "avail_eraseblocks", 0);
• leb_size = ubi_dev_read_int(ubi_num, "eraseblock_size", 0);
• memset(&mkvol_req, 0, sizeof(struct ubi_mkvol_req));
• mkvol_req.vol_id = UBI_VOL_NUM_AUTO;
• mkvol_req.alignment = 1;
• mkvol_req.bytes = (long long)avail_lebs * leb_size;
• mkvol_req.vol_type = UBI_DYNAMIC_VOLUME;
• ret = snprintf(mkvol_req.name, UBI_MAX_VOLUME_NAME + 1, "%s", name);
• mkvol_req.name_len = ret;
• ioctl(ubi_dev, UBI_IOCMKVOL, &mkvol_req);
• close(ubi_dev);
• }
• close(ubi_ctrl);
• return ubi_num;
• }
• int ubi_detach_dev(int dev)
• {
• int ret, ubi_ctrl;
• ubi_ctrl = open(UBI_CTRL_DEV, O_RDONLY);
• if (ubi_ctrl == -1) {
• return -1;
• }
• ret = ioctl(ubi_ctrl, UBI_IOCDET, &dev);
• close(ubi_ctrl);
• return ret;
• }
• 6.增加文件
• andriod/system/core/init/ubi-user.h
• /*
• * Copyright (c) International Business Machines Corp., 2006
• *
• * This program is free software; you can redistribute it and/or modify
• * it under the terms of the GNU General Public License as published by
• * the Free Software Foundation; either version 2 of the License, or
• * (at your option) any later version.
• *
• * This program is distributed in the hope that it will be useful,
• * but WITHOUT ANY WARRANTY; without even the implied warranty of
• * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See
• * the GNU General Public License for more details.
• *
• * You should have received a copy of the GNU General Public License
• * along with this program; if not, write to the Free Software
• * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
• *
• * Author: Artem Bityutskiy (Битюцкий Артём)
• */
• #ifndef __UBI_USER_H__
• #define __UBI_USER_H__
• /*
• * UBI device creation (the same as MTD device attachment)
• * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
• *
• * MTD devices may be attached using %UBI_IOCATT ioctl command of the UBI
• * control device. The caller has to properly fill and pass
• * &struct ubi_attach_req object - UBI will attach the MTD device specified in
• * the request and return the newly created UBI device number as the ioctl
• * return value.
• *
• * UBI device deletion (the same as MTD device detachment)
• * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
• *
• * An UBI device maybe deleted with %UBI_IOCDET ioctl command of the UBI
• * control device.
• *
• * UBI volume creation
• * ~~~~~~~~~~~~~~~~~~~
• *
• * UBI volumes are created via the %UBI_IOCMKVOL IOCTL command of UBI character
• * device. A &struct ubi_mkvol_req object has to be properly filled and a
• * pointer to it has to be passed to the IOCTL.
• *
• * UBI volume deletion
• * ~~~~~~~~~~~~~~~~~~~
• *
• * To delete a volume, the %UBI_IOCRMVOL IOCTL command of the UBI character
• * device should be used. A pointer to the 32-bit volume ID hast to be passed
• * to the IOCTL.
• *
• * UBI volume re-size
• * ~~~~~~~~~~~~~~~~~~
• *
• * To re-size a volume, the %UBI_IOCRSVOL IOCTL command of the UBI character
• * device should be used. A &struct ubi_rsvol_req object has to be properly
• * filled and a pointer to it has to be passed to the IOCTL.
• *
• * UBI volume update
• * ~~~~~~~~~~~~~~~~~
• *
• * Volume update should be done via the %UBI_IOCVOLUP IOCTL command of the
• * corresponding UBI volume character device. A pointer to a 64-bit update
• * size should be passed to the IOCTL. After this, UBI expects user to write
• * this number of bytes to the volume character device. The update is finished
• * when the claimed number of bytes is passed. So, the volume update sequence
• * is something like:
• *
• * fd = open("/dev/my_volume");
• * ioctl(fd, UBI_IOCVOLUP, &image_size);
• * write(fd, buf, image_size);
• * close(fd);
• *
• * Atomic eraseblock change
• * ~~~~~~~~~~~~~~~~~~~~~~~~
• *
• * Atomic eraseblock change operation is done via the %UBI_IOCEBCH IOCTL
• * command of the corresponding UBI volume character device. A pointer to
• * &struct ubi_leb_change_req has to be passed to the IOCTL. Then the user is
• * expected to write the requested amount of bytes. This is similar to the
• * "volume update" IOCTL.
• */
• /*
• * When a new UBI volume or UBI device is created, users may either specify the
• * volume/device number they want to create or to let UBI automatically assign
• * the number using these constants.
• */
• #define UBI_VOL_NUM_AUTO (-1)
• #define UBI_DEV_NUM_AUTO (-1)
• /* Maximum volume name length */
• #define UBI_MAX_VOLUME_NAME 127
• /* IOCTL commands of UBI character devices */
• #define UBI_IOC_MAGIC 'o'
• /* Create an UBI volume */
• #define UBI_IOCMKVOL _IOW(UBI_IOC_MAGIC, 0, struct ubi_mkvol_req)
• /* Remove an UBI volume */
• #define UBI_IOCRMVOL _IOW(UBI_IOC_MAGIC, 1, int32_t)
• /* Re-size an UBI volume */
• #define UBI_IOCRSVOL _IOW(UBI_IOC_MAGIC, 2, struct ubi_rsvol_req)
• /* IOCTL commands of the UBI control character device */
• #define UBI_CTRL_IOC_MAGIC 'o'
• /* Attach an MTD device */
• #define UBI_IOCATT _IOW(UBI_CTRL_IOC_MAGIC, 64, struct ubi_attach_req)
• /* Detach an MTD device */
• #define UBI_IOCDET _IOW(UBI_CTRL_IOC_MAGIC, 65, int32_t)
• /* IOCTL commands of UBI volume character devices */
• #define UBI_VOL_IOC_MAGIC 'O'
• /* Start UBI volume update */
• #define UBI_IOCVOLUP _IOW(UBI_VOL_IOC_MAGIC, 0, int64_t)
• /* An eraseblock erasure command, used for debugging, disabled by default */
• #define UBI_IOCEBER _IOW(UBI_VOL_IOC_MAGIC, 1, int32_t)
• /* An atomic eraseblock change command */
• #define UBI_IOCEBCH _IOW(UBI_VOL_IOC_MAGIC, 2, int32_t)
• /* Maximum MTD device name length supported by UBI */
• #define MAX_UBI_MTD_NAME_LEN 127
• /*
• * UBI data type hint constants.
• *
• * UBI_LONGTERM: long-term data
• * UBI_SHORTTERM: short-term data
• * UBI_UNKNOWN: data persistence is unknown
• *
• * These constants are used when data is written to UBI volumes in order to
• * help the UBI wear-leveling unit to find more appropriate physical
• * eraseblocks.
• */
• enum {
• UBI_LONGTERM = 1,
• UBI_SHORTTERM = 2,
• UBI_UNKNOWN = 3,
• };
• /*
• * UBI volume type constants.
• *
• * @UBI_DYNAMIC_VOLUME: dynamic volume
• * @UBI_STATIC_VOLUME: static volume
• */
• enum {
• UBI_DYNAMIC_VOLUME = 3,
• UBI_STATIC_VOLUME = 4,
• };
• /**
• * struct ubi_attach_req - attach MTD device request.
• * @ubi_num: UBI device number to create
• * @mtd_num: MTD device number to attach
• * @vid_hdr_offset: VID header offset (use defaults if %0)
• * @padding: reserved for future, not used, has to be zeroed
• *
• * This data structure is used to specify MTD device UBI has to attach and the
• * parameters it has to use. The number which should be assigned to the new UBI
• * device is passed in @ubi_num. UBI may automatically assign the number if
• * @UBI_DEV_NUM_AUTO is passed. In this case, the device number is returned in
• * @ubi_num.
• *
• * Most applications should pass %0 in @vid_hdr_offset to make UBI use default
• * offset of the VID header within physical eraseblocks. The default offset is
• * the next min. I/O unit after the EC header. For example, it will be offset
• * 512 in case of a 512 bytes page NAND flash with no sub-page support. Or
• * it will be 512 in case of a 2KiB page NAND flash with 4 512-byte sub-pages.
• *
• * But in rare cases, if this optimizes things, the VID header may be placed to
• * a different offset. For example, the boot-loader might do things faster if the
• * VID header sits at the end of the first 2KiB NAND page with 4 sub-pages. As
• * the boot-loader would not normally need to read EC headers (unless it needs
• * UBI in RW mode), it might be faster to calculate ECC. This is weird example,
• * but it real-life example. So, in this example, @vid_hdr_offer would be
• * 2KiB-64 bytes = 1984. Note, that this position is not even 512-bytes
• * aligned, which is OK, as UBI is clever enough to realize this is 4th sub-page
• * of the first page and add needed padding.
• */
• struct ubi_attach_req {
• int32_t ubi_num;
• int32_t mtd_num;
• int32_t vid_hdr_offset;
• uint8_t padding[12];
• };
• /**
• * struct ubi_mkvol_req - volume description data structure used in
• * volume creation requests.
• * @vol_id: volume number
• * @alignment: volume alignment
• * @bytes: volume size in bytes
• * @vol_type: volume type (%UBI_DYNAMIC_VOLUME or %UBI_STATIC_VOLUME)
• * @padding1: reserved for future, not used, has to be zeroed
• * @name_len: volume name length
• * @padding2: reserved for future, not used, has to be zeroed
• * @name: volume name
• *
• * This structure is used by user-space programs when creating new volumes. The
• * @used_bytes field is only necessary when creating static volumes.
• *
• * The @alignment field specifies the required alignment of the volume logical
• * eraseblock. This means, that the size of logical eraseblocks will be aligned
• * to this number, i.e.,
• *    (UBI device logical eraseblock size) mod (@alignment) = 0.
• *
• * To put it differently, the logical eraseblock of this volume may be slightly
• * shortened in order to make it properly aligned. The alignment has to be
• * multiple of the flash minimal input/output unit, or %1 to utilize the entire
• * available space of logical eraseblocks.
• *
• * The @alignment field may be useful, for example, when one wants to maintain
• * a block device on top of an UBI volume. In this case, it is desirable to fit
• * an integer number of blocks in logical eraseblocks of this UBI volume. With
• * alignment it is possible to update this volume using plane UBI volume image
• * BLOBs, without caring about how to properly align them.
• */
• struct ubi_mkvol_req {
• int32_t vol_id;
• int32_t alignment;
• int64_t bytes;
• int8_t vol_type;
• int8_t padding1;
• int16_t name_len;
• int8_t padding2[4];
• char name[UBI_MAX_VOLUME_NAME + 1];
• } __attribute__ ((packed));
• /**
• * struct ubi_rsvol_req - a data structure used in volume re-size requests.
• * @vol_id: ID of the volume to re-size
• * @bytes: new size of the volume in bytes
• *
• * Re-sizing is possible for both dynamic and static volumes. But while dynamic
• * volumes may be re-sized arbitrarily, static volumes cannot be made to be
• * smaller then the number of bytes they bear. To arbitrarily shrink a static
• * volume, it must be wiped out first (by means of volume update operation with
• * zero number of bytes).
• */
• struct ubi_rsvol_req {
• int64_t bytes;
• int32_t vol_id;
• } __attribute__ ((packed));
• /**
• * struct ubi_leb_change_req - a data structure used in atomic logical
• * eraseblock change requests.
• * @lnum: logical eraseblock number to change
• * @bytes: how many bytes will be written to the logical eraseblock
• * @dtype: data type (%UBI_LONGTERM, %UBI_SHORTTERM, %UBI_UNKNOWN)
• * @padding: reserved for future, not used, has to be zeroed
• */
• struct ubi_leb_change_req {
• int32_t lnum;
• int32_t bytes;
• uint8_t dtype;
• uint8_t padding[7];
• } __attribute__ ((packed));
• #endif /* __UBI_USER_H__ */

3. 常见问题

(1) ubifs_check_node: bad CRC: calculated 0xca82b3d7, read 0x9be0e26

ubifs_check_node: bad node at LEB 51:45312

注意这个crc错误，说的错误并不是LED 0的问题，也就是说第一个逻辑块没问题，CRC是正确的，这个问题的一个解决办法是要精确mkfs.ubifs时候的-c参数的值，不能太大了

(2)

[    5.433349] UBIFS error (pid 71): ubifs_read_node: bad node type (150 but expected 1)
[    5.434204] UBIFS error (pid 71): ubifs_read_node: bad node at LEB 524:4072, LEB mapping status 1
[    5.435241] Not a node, first 24 bytes:
[    5.435729] 00000000: 34 fb 21 ee 84 18 69 2d 60 b0 33 e6 74 f8 1c 15 da ca a1 c9 96 e3 ac 51                          4.!...i-`.3.t..........Q

这个问题是，ubifs给nand驱动的buffer不一定是按照硬件对齐要求的，所以要驱动来判断，dma等地址需要页对齐等特性

(3) crc错误，但是LED 0:0

可能镜像没烧对，或者mtd-utils的版本和内核版本相差较远

(4) 可以擦掉一个分区，并不需要非要烧写image到那个分区，也可以直接mount的，这样可以做实验验证ubi的性能，前提是kernel配置好了，mount正确了

(5) 如果是跟文件系统或者要手动mount，以下步骤供参考:

uboot里, mtd命令后，看到需要用ubifs的mtd分区的编号，比如

device nand0 <rda_nand>, # parts = 10
 #: name                size            offset          mask_flags
 0: bootloader          0x00200000      0x00000000      0
 1: boot                0x00800000      0x00800000      0
 2: system              0x08000000      0x02000000      0

system分区mtd编号是2，命令如下:

nand erase.part system

命令行参数增加

ubi.mtd=2，如果有多个，可以增加，如ubi.mtd=2,ubi.mtd=1，这样传递后，内核启动后会做attach的操作，类似android的init中的attach那样，attach之后，/dev/下就建立好设备了，进入系统后用:

内核控制台里，mount -t ubifs /dev/ubi0_0 /mnt 即可mount