UBIFS Problems After Upgrading Kernel
I'm looking at upgrading the kernel on an embedded product we have from 2.6.36 to 3.4.x (currently testing with 3.4.39).
The device has UBIFS that is created on 2.6.30 running on the device itself with the following commands:flash_eraseall /dev/mtd8
ubiformat -y /dev/mtd8 -s 512
ubiattach /dev/ubi_ctrl -m 8
ubimkvol /dev/ubi0 -N rootfs -m
(I'm not sure why there is the "-s 512" there but I'm sure there was a good reason at the time!).
The various UBI utilitites would appear to be v1.0.
The UBIFS generated in this way works fine under 2.6.36:
[ 0.580000] NAND device: Manufacturer ID: 0xad, Chip ID: 0xda (Hynix NAND 256MiB 3,3V 8-bit) [ 0.590000] AT91 NAND: 8-bit, Software ECC [ 0.600000] Scanning device for bad blocks
[ 0.790000] 0x000000800000-0x000010000000 : "rootfs" [ 0.800000] UBI: attaching mtd8 to ubi0 [ 0.800000] UBI: physical eraseblock size: 131072 bytes (128 KiB) [ 0.810000] UBI: logical eraseblock size: 129024 bytes [ 0.810000] UBI: smallest flash I/O unit: 2048 [ 0.820000] UBI: sub-page size: 512 [ 0.820000] UBI: VID header offset: 512 (aligned 512) [ 0.830000] UBI: data offset: 2048 [ 1.100000] UBI: max. sequence number: 675 [ 1.120000] UBI: attached mtd8 to ubi0 [ 1.130000] UBI: MTD device name: "rootfs" [ 1.130000] UBI: MTD device size: 248 MiB [ 1.140000] UBI: number of good PEBs: 1984 [ 1.140000] UBI: number of bad PEBs: 0 [ 1.150000] UBI: max. allowed volumes: 128 [ 1.150000] UBI: wear-leveling threshold: 4096 [ 1.160000] UBI: number of internal volumes: 1 [ 1.160000] UBI: number of user volumes: 1 [ 1.160000] UBI: available PEBs: 0 [ 1.170000] UBI: total number of reserved PEBs: 1984 [ 1.170000] UBI: number of PEBs reserved for bad PEB handling: 19 [ 1.180000] UBI: max/mean erase counter: 1/0 [ 1.180000] UBI: image sequence number: 0 [ 1.190000] UBI: background thread "ubi_bgt0d" started, PID 30
[ 1.530000] UBIFS: mounted UBI device 0, volume 0, name "rootfs" [ 1.530000] UBIFS: file system size: 251467776 bytes (245574 KiB, 239 MiB, 1949 LEBs) [ 1.540000] UBIFS: journal size: 12644352 bytes (12348 KiB, 12 MiB, 98 LEBs) [ 1.550000] UBIFS: media format: w4/r0 (latest is w4/r0) [ 1.560000] UBIFS: default compressor: lzo [ 1.560000] UBIFS: reserved for root: 4952683 bytes (4836 KiB) [ 1.570000] VFS: Mounted root (ubifs filesystem) on device 0:11.
However, after upgrading to 3.4.39, the root FS doesn't mount:
[ 0.710000] atmel_nand: Use On Flash BBT [ 0.720000] atmel_nand atmel_nand: No DMA support for NAND access. [ 0.720000] NAND device: Manufacturer ID: 0xad, Chip ID: 0xda (Hynix NAND 256MiB 3,3V 8-bit) [ 0.730000] Bad block table not found for chip 0 [ 0.740000] Bad block table not found for chip 0 [ 0.740000] Scanning device for bad blocks [ 0.870000] Bad block table written to 0x00000ffe0000, version 0x01 [ 0.880000] Bad block table written to 0x00000ffc0000, version 0x01
[ 0.960000] 0x000000800000-0x000010000000 : "rootfs" [ 0.970000] UBI: attaching mtd8 to ubi0 [ 0.970000] UBI: physical eraseblock size: 131072 bytes (128 KiB) [ 0.980000] UBI: logical eraseblock size: 129024 bytes [ 0.980000] UBI: smallest flash I/O unit: 2048 [ 0.990000] UBI: sub-page size: 512 [ 0.990000] UBI: VID header offset: 512 (aligned 512) [ 1.000000] UBI: data offset: 2048 [ 1.280000] UBI: max. sequence number: 745 [ 1.290000] UBI warning: print_rsvd_warning: cannot reserve enough PEBs for bad PEB handling, reserved 15, need 19 [ 1.310000] UBI: attached mtd8 to ubi0 [ 1.310000] UBI: MTD device name: "rootfs" [ 1.310000] UBI: MTD device size: 248 MiB [ 1.320000] UBI: number of good PEBs: 1980 [ 1.320000] UBI: number of bad PEBs: 4 [ 1.330000] UBI: number of corrupted PEBs: 0 [ 1.330000] UBI: max. allowed volumes: 128 [ 1.340000] UBI: wear-leveling threshold: 4096 [ 1.340000] UBI: number of internal volumes: 1 [ 1.350000] UBI: number of user volumes: 1 [ 1.350000] UBI: available PEBs: 0 [ 1.360000] UBI: total number of reserved PEBs: 1980 [ 1.360000] UBI: number of PEBs reserved for bad PEB handling: 15 [ 1.370000] UBI: max/mean erase counter: 1/0 [ 1.370000] UBI: image sequence number: 0 [ 1.380000] UBI: background thread "ubi_bgt0d" started, PID 799
[ 1.650000] UBIFS error (pid 1): ubifs_read_node: bad node type (255 but expected 6) [ 1.660000] UBIFS error (pid 1): ubifs_read_node: bad node at LEB 0:0, LEB mapping status 0 [ 1.660000] Not a node, first 24 bytes: [ 1.670000] 00000000: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ........................ [ 1.680000] [<c000d2d0>] (unwind_backtrace+0x0/0xf0) from [<c00ccdf4>] (ubifs_read_node+0x22c/0x2a4) [ 1.690000] [<c00ccdf4>] (ubifs_read_node+0x22c/0x2a4) from [<c00c94b8>] (ubifs_read_sb_node+0x54/0x78) [ 1.700000] [<c00c94b8>] (ubifs_read_sb_node+0x54/0x78) from [<c00ca1d0>] (ubifs_read_superblock+0xca4/0x1674) [ 1.710000] [<c00ca1d0>] (ubifs_read_superblock+0xca4/0x1674) from [<c00c7e38>] (ubifs_mount+0x6dc/0x1b44) [ 1.720000] [<c00c7e38>] (ubifs_mount+0x6dc/0x1b44) from [<c00749ac>] (mount_fs+0x14/0xcc) [ 1.730000] [<c00749ac>] (mount_fs+0x14/0xcc) from [<c00888b4>] (vfs_kern_mount+0x4c/0xac) [ 1.740000] [<c00888b4>] (vfs_kern_mount+0x4c/0xac) from [<c0088b74>] (do_kern_mount+0x34/0xd4) [ 1.740000] [<c0088b74>] (do_kern_mount+0x34/0xd4) from [<c008a324>] (do_mount+0x3f8/0x6b4) [ 1.750000] [<c008a324>] (do_mount+0x3f8/0x6b4) from [<c008a66c>] (sys_mount+0x84/0xc4) [ 1.760000] [<c008a66c>] (sys_mount+0x84/0xc4) from [<c0346ab4>] (mount_block_root+0xf0/0x298) [ 1.770000] [<c0346ab4>] (mount_block_root+0xf0/0x298) from [<c0346d4c>] (prepare_namespace+0x8c/0x174) [ 1.780000] [<c0346d4c>] (prepare_namespace+0x8c/0x174) from [<c0346884>] (kernel_init+0x16c/0x1a8) [ 1.790000] [<c0346884>] (kernel_init+0x16c/0x1a8) from [<c0009c24>]
(kernel_thread_exit+0x0/0x8)
(This is with all the UBI/UBIFS debugging options turned on that I could find).
Is there a way that I can get the FS mounted under the later kernel?
Cheers.
Paul
Quoted 7 lines · Paul Black
[ 1.650000] UBIFS error (pid 1): ubifs_read_node: bad node type (255 but expected 6) [ 1.660000] UBIFS error (pid 1): ubifs_read_node: bad node at LEB 0:0, LEB mapping status 0 [ 1.660000] Not a node, first 24 bytes: [ 1.670000] 00000000: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ffff ff ff ff ff ff ff ff
UBIFS says that the eraseblock where it expects to find the Master node is empty. Somehow the data is not there anymore. Can you compare the contents of the corresponding /dev/mtd and /dev/ubi* before and after upgrade?
Signature · 2 lines
Best Regards, Artem Bityutskiy
Quoted 10 lines · Artem Bityutskiy
Quoted 7 lines · Paul Black
[ 1.650000] UBIFS error (pid 1): ubifs_read_node: bad node type (255 but expected 6) [ 1.660000] UBIFS error (pid 1): ubifs_read_node: bad node at LEB 0:0, LEB mapping status 0 [ 1.660000] Not a node, first 24 bytes: [ 1.670000] 00000000: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ffff ff ff ff ff ff ff ff
Quoted 4 lines
UBIFS says that the eraseblock where it expects to find the Master node is empty. Somehow the data is not there anymore. Can you compare the contents of the corresponding
/dev/mtdand/dev/ubi*before and after upgrade?
Thanks for the reply.
Got the following difference in the file systems before and after booting 3.4.40:
< 1737000000 U B I # 001 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0
< 4255 2349 0001 0000 0000 0000 0000 0000
< 1737000020 \0 \0 002 \0 \0 \0 \b \0 \0 \0 \0 \0 \0 \0 \0 \0
< 0000 0002 0000 0008 0000 0000 0000 0000
< 1737000040 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0
< 0000 0000 0000 0000 0000 0000 0000 0000
< 1737000060 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 N 204 I 204
< 0000 0000 0000 0000 0000 0000 844e 8449 < 1737000100 377 377 377 377 377 377 377 377 377 377 377 377 377 377 377 377 < ffff ffff ffff ffff ffff ffff ffff ffff
< *
< 1737400000 U B I # 001 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0
< 4255 2349 0001 0000 0000 0000 0000 0000
< 1737400020 \0 \0 002 \0 \0 \0 \b \0 \0 \0 \0 \0 \0 \0 \0 \0
< 0000 0002 0000 0008 0000 0000 0000 0000
< 1737400040 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0
< 0000 0000 0000 0000 0000 0000 0000 0000
< 1737400060 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 N 204 I 204
< 0000 0000 0000 0000 0000 0000 844e 8449 < 1737400100 377 377 377 377 377 377 377 377 377 377 377 377 377 377 377 377 < ffff ffff ffff ffff ffff ffff ffff ffff
< *
< 1737401000 U B I ! 001 001 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0
< 4255 2149 0101 0000 0000 0000 0000 0000
< 1737401020 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0
< 0000 0000 0000 0000 0000 0000 0000 0000
< 1737401040 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 004
< 0000 0000 0000 0000 0000 0000 0000 0400
< 1737401060 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 263 \b 301 301
< 0000 0000 0000 0000 0000 0000 08b3 c1c1
< 1737401100 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0
< 0000 0000 0000 0000 0000 0000 0000 0000
< *
< 1737402000 377 377 377 377 377 377 377 377 377 377 377 377 377 377 377 377
< ffff ffff ffff ffff ffff ffff ffff ffff
< *
< 1737404000 1 030 020 006 320 I y 335 001 \0 \0 \0 \0 \0 \0 \0
< 1831 0610 49d0 dd79 0001 0000 0000 0000
< 1737404020 \0 020 \0 \0 006 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0
< 1000 0000 0006 0000 0000 0000 0000 0000
< 1737404040 \0 \b \0 \0 \0 370 001 \0 251 \a \0 \0 251 \a \0 \0
< 0800 0000 f800 0001 07a9 0000 07a9 0000 < 1737404060 \0 265 \0 \0 \0 \0 \0 006 \0 \0 \0 002 \0 \0 \0 < 2000 00b5 0000 0000 0006 0000 0002 0000 < 1737404100 001 \0 \0 \0 001 \0 \0 \0 \b \0 \0 \0 \0 001 \0 \0 < 0001 0000 0001 0000 0008 0000 0100 0000 < 1737404120 004 \0 \0 \0 001 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 < 0004 0000 0001 0000 0000 0000 0000 0000 < 1737404140 \0 \0 P \0 \0 \0 \0 \0 \0 312 232 ; 332 - 305 213 < 0000 0050 0000 0000 ca00 3b9a 2dda 8bc5 < 1737404160 251 373 F 332 213 3 $ h 376 ! 300 r \0 \0 \0 \0 < fba9 da46 338b 6824 21fe 72c0 0000 0000
< 1737404200 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0
< 0000 0000 0000 0000 0000 0000 0000 0000
< *
< 1737414000 377 377 377 377 377 377 377 377 377 377 377 377 377 377 377 377
< ffff ffff ffff ffff ffff ffff ffff ffff
< *
During booting, this is printed out. The addresses shown above, with offsets accounted for and converted to octal, overlap with the addresses in the last two lines below.
[ 0.720000] atmel_nand: Use On Flash BBT [ 0.720000] atmel_nand atmel_nand: No DMA support for NAND access. [ 0.730000] NAND device: Manufacturer ID: 0xad, Chip ID: 0xda (Hynix NAND 256MiB 3,3V 8-bit) [ 0.740000] Bad block table not found for chip 0 [ 0.740000] Bad block table not found for chip 0 [ 0.750000] Scanning device for bad blocks [ 0.880000] Bad block table written to 0x00000ffe0000, version 0x01 [ 0.890000] Bad block table written to 0x00000ffc0000, version 0x01
I've also noticed that I don't have the "Use On Flash BBT" on my working 2.6.36 kernels. Finding and disabling this in the files for my boards of interest produces a working 3.4.40 kernel.
Thanks for the pointers.
--
Paul