在linux-4.19.78中使用UBIFS

最近的一块板上使用了一颗大容量的NAND Flash，MLC架构，内核启动时，显示相关信息如下：

nand: device found, Manufacturer ID: 0x2c, Chip ID: 0x88
nand: Micron MT29F64G08CBAAAWP
nand: 8192 MiB, MLC, erase size: 2048 KiB, page size: 8192, OOB size: 448
4 fixed-partitions partitions found on MTD device atmel_nand
Creating 4 MTD partitions on "atmel_nand":
0x000000400000-0x000001400000 : "device tree"
0x000001400000-0x000002000000 : "kernel"
0x000002000000-0x000040000000 : "rootfs"
0x000040000000-0x000200000000 : "database"

ubinize需要指定一个配置文件，内容如下：
    [ubifs]
    mode=ubi
    image=ubifs.img
    vol_id=0
    vol_size=768MiB
    vol_type=dynamic
    vol_alignment=1
    vol_name=rootfs
    vol_flags=autoresize

• 1、ubifs镜像的制作：

# mkfs.ubifs -x lzo -m 8KiB -e 2032KiB -c 496 -o ubifs.img -d rootfs
# ubinize -o rootfs.ubi -m 8KiB -p 2048KiB -s 8KiB -O 8192 ubinize.cfg -v
ubinize: LEB size:                  2080768
ubinize: PEB size:                  2097152
ubinize: min. I/O size:             8192
ubinize: sub-page size:             8192
ubinize: VID offset:                8192
ubinize: data offset:               16384
ubinize: UBI image sequence number: 662004650
ubinize: loaded the ini-file "ubinize.cfg"
ubinize: count of sections: 1

ubinize: parsing section "ubifs"
ubinize: mode=ubi, keep parsing
ubinize: volume type: dynamic
ubinize: volume ID: 0
ubinize: volume size: 805306368 bytes
ubinize: volume name: rootfs
ubinize: volume alignment: 1
ubinize: autoresize flags found
ubinize: adding volume 0
ubinize: writing volume 0
ubinize: image file: ubifs.img

ubinize: writing layout volume
ubinize: done

• 2、在linux系统下NFS烧写ubifs：

  ubiformat /dev/mtd7 -s 8192 -O 8192 -f rootfs.ubi

因NAND中有坏块，未完成退出。查看代码发现NAND Erase不能擦除坏块，遇到坏块时退出。
找到源码文件"./drivers/mtd/nand/raw/nand-base.c"中函数 "nand_erase_nand"体部分，
摘录代码如下：

while (len) {
		/* Check if we have a bad block, we do not erase bad blocks! */
		if (nand_block_checkbad(mtd, ((loff_t) page) <<
					chip->page_shift, allowbbt)) {
			pr_warn("%s: attempt to erase a bad block at page 0x%08x\n",
				    __func__, page);
			ret = -EIO;
			goto erase_exit;
		}
		……

修改为跳过坏块继续擦除，代码如下：

while (len) {
		/* Check if we have a bad block, we do not erase bad blocks! */
		if (nand_block_checkbad(mtd, ((loff_t) page) <<
					chip->page_shift, allowbbt)) {
			pr_warn("%s: attempt to erase a bad block at page 0x%08x\n",
				    __func__, page);
		//	ret = -EIO;
		//	goto erase_exit;

			/* Increment page address and decrement length */
			len -= (1ULL << chip->phys_erase_shift);
			page += pages_per_block;
			continue;
		}
        ……

重新编译后，执行ubiformat：

# ubiformat /dev/mtd7 -s 8192 -O 8192 -f rootfs.ubi
ubiformat: mtd7 (mlc-nand), size 8556380160 bytes (8.0 GiB), 4080 eraseblocks of 2097152 bytes (2.0 MiB), min. I/O size 8192 bytes
libscan: scanning eraseblock 4079 -- 100 % complete
ubiformat: 74 eraseblocks have valid erase counter, mean value is 0
ubiformat: 4002 eraseblocks are supposedly empty
ubiformat: warning!: 4 of 4080 eraseblocks contain non-ubifs data
ubiformat: continue? (yes/no)  y
ubiformat: warning!: only 74 of 4080 eraseblocks have valid erase counter
ubiformat: erase counter 0 will be used for all eraseblocks
ubiformat: note, arbitrary erase counter value may be specified using -e option
ubiformat: continue? (yes/no)  y
ubiformat: use erase counter 0 for all eraseblocks
ubiformat: flashing eraseblock 28 -- 100 % complete
ubiformat: formatting eraseblock 70 --  1 % complenand: biformat: formatting eraseblock 71 --  1 % cubiformat: formatting eraseblock 73 --  1 % complete  nand_erase_nand: attempt to erase a bad block at page 0x00005a00
ubiformat: fnamatting eraseblock 74 --  1 % complete  nd: nand_erase_nand: attempt to erase a bad block at page 0x00005b00
ubiformat: formatting eraseblock 4048 -- 99 % complete  and: nand_erase_nand: attempt to erase a bad block at page 0x000ffc00
ubiformat: formatting eraseblock 406d: nand_erase_nand: attempt to erase a bad block at page 0x000ffd00
ubiformat: formatting eraseblock 4077 -- 99 % complete  and: nand_erase_nand: attempt to erase a bad block at page 0x000ffe00
ubiformat: formanand: nraseblock 4078 -- 99 % complete  and_erase_nand: attempt to erase a bad block at page 0x000fff00
ubiformat: formatting eraseblock 4079 -- 100 % complete

• 3、挂载、查看ubifs

ubiattach /dev/ubi_ctrl -m 7 -O 8192
    ubi: refuse attaching mtd7 - MLC NAND is not supported

经查，发现高版本内核已屏蔽对MLC NAND的支持。对比linux-4.9.36，没有屏蔽MLC，跳过下面这段代码：源码文件"./drivers/mtd/ubi/build.c"中函数 “ubi_attach_mtd_dev”

if (mtd->type == MTD_MLCNANDFLASH) {
	pr_err("ubi: refuse attaching mtd%d - MLC NAND is not supported\n",
		mtd->index);
	return -EINVAL;
}

重新编译后，执行：

# ubiattach /dev/ubi_ctrl -m 7 -O 8192
ubi0: default fastmap pool size: 20
ubi0: default fastmap WL pool size: 10
ubi0: attaching mtd7
ubi0: scanning is finished
gluebi (pid 153): gluebi_resized: got update notification for unknown UBI device 0 volume 0
ubi0: volume 0 ("rootfs") re-sized from 388 to 410 LEBs
ubi0: attached mtd7 (name "rootfs", size 992 MiB)
ubi0: PEB size: 2097152 bytes (2048 KiB), LEB size: 2080768 bytes
ubi0: min./max. I/O unit sizes: 8192/8192, sub-page size 8192
ubi0: VID header offset: 8192 (aligned 8192), data offset: 16384
ubi0: good PEBs: 494, bad PEBs: 2, corrupted PEBs: 0
ubi0: user volume: 1, internal volumes: 1, max. volumes count: 128
ubi0: max/mean erase counter: 0/0, WL threshold: 4096, image sequence number: 773013103
ubi0: available PEBs: 0, total reserved PEBs: 494, PEBs reserved for bad PEB handling: 78
ubi0: background thread "ubi_bgt0d" started, PID 154
UBI device number 0, total 494 LEBs (1027899392 bytes, 980.3 MiB), available 0 LEBs (0 bytes), LEB size 2080768 bytes (2.0 MiB)

重启目标板，在u-boot中设置rootfs为此分区，linux正常启动。

总结：为什么在高版本内核中暂时放弃MLC支持，肯定是有原因的，目前的几块测试板暂时先这样吧，下版用eMMC 换掉 MLC NAND。