UBIFS replay journal failure on power cuts

[Rock Lee]

On Mon, Apr 17, 2017 at 7:29 AM, Raphael Pereira
<raphaelpereira at gmail.com> wrote:
> Hello,
>
> I have been using UBIFS on a device and have been getting some
> unrecoverable errors on journal replay after some power cuts.
>
> I have not stablished the specific way to reproduce the problem, but
> it is happening with some repeatability.
>
> The error seems quite like both threads:
>
> http://lists.infradead.org/pipermail/linux-mtd/2014-July/054620.html
> http://lists.infradead.org/pipermail/linux-mtd/2016-June/068339.html
>
> And before you ask, I do indeed use fastmap.
>
> I understand that it seems the journal is trying to reproduce an inode
> removal that has been interrupted by a power cut, so the block on NAND
> seems to be in an intermediary state.
>
> Wouldn't be enough to just erase the block instead of reading it
> (which is where it seems to fail)?
>
> The error log is below:
>
> kernel: UBI: default fastmap pool size: 190
> kernel: UBI: default fastmap WL pool size: 25
> kernel: UBI: attaching mtd2 to ubi0
> kernel: UBI: attached by fastmap
> kernel: UBI: fastmap pool size: 190
> kernel: UBI: fastmap WL pool size: 25
> kernel: UBI: attached mtd2 (name "besav2rx_rootfs", size 476 MiB) to ubi0
> kernel: UBI: PEB size: 131072 bytes (128 KiB), LEB size: 126976 bytes
> kernel: UBI: min./max. I/O unit sizes: 2048/2048, sub-page size 2048
> kernel: UBI: VID header offset: 2048 (aligned 2048), data offset: 4096
> kernel: UBI: good PEBs: 3804, bad PEBs: 4, corrupted PEBs: 0
> kernel: UBI: user volume: 1, internal volumes: 1, max. volumes count: 128
> kernel: UBI: max/mean erase counter: 63062/239, WL threshold: 4096,
> image sequence number: 377102491
> kernel: UBI: available PEBs: 2, total reserved PEBs: 3802, PEBs
> reserved for bad PEB handling: 76
> kernel: UBI: background thread "ubi_bgt0d" started, PID 459
> kernel: UBIFS: recovery needed
> kernel: UBIFS error (pid 460): ubifs_read_node: bad node type (0 but expected 3)
> kernel: UBIFS error (pid 460): ubifs_read_node: bad node at LEB
> 3246:96960, LEB mapping status 1
> kernel: Not a node, first 24 bytes:
> kernel: 00000000: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> 00 00 00 00 00 00 00                          ........................
> kernel: CPU: 0 PID: 460 Comm: mount Not tainted 3.18.48 #15
> kernel: [<c001491c>] (unwind_backtrace) from [<c0011fe4>] (show_stack+0x10/0x14)
> kernel: [<c0011fe4>] (show_stack) from [<bf158c50>]
> (ubifs_read_node+0x260/0x2f0 [ubifs])
> kernel: [<bf158c50>] (ubifs_read_node [ubifs]) from [<bf176300>]
> (ubifs_tnc_read_node+0x4c/0x140 [ubifs])
> kernel: [<bf176300>] (ubifs_tnc_read_node [ubifs]) from [<bf159e70>]
> (tnc_read_node_nm+0xbc/0x1dc [ubifs])
> kernel: [<bf159e70>] (tnc_read_node_nm [ubifs]) from [<bf15d610>]
> (ubifs_tnc_next_ent+0x138/0x1a0 [ubifs])
> kernel: [<bf15d610>] (ubifs_tnc_next_ent [ubifs]) from [<bf15d748>]
> (ubifs_tnc_remove_ino+0xd0/0x150 [ubifs])
> kernel: [<bf15d748>] (ubifs_tnc_remove_ino [ubifs]) from [<bf1607d4>]
> (ubifs_replay_journal+0xf88/0x1510 [ubifs])
> kernel: [<bf1607d4>] (ubifs_replay_journal [ubifs]) from [<bf154aa0>]
> (ubifs_mount+0x1194/0x1750 [ubifs])
> kernel: [<bf154aa0>] (ubifs_mount [ubifs]) from [<c00e6b84>]
> (mount_fs+0x14/0xd0)
> kernel: [<c00e6b84>] (mount_fs) from [<c0101e94>] (vfs_kern_mount+0x54/0x12c)
> kernel: [<c0101e94>] (vfs_kern_mount) from [<c01054b4>] (do_mount+0x18c/0xbe8)
> kernel: [<c01054b4>] (do_mount) from [<c0106230>] (SyS_mount+0x74/0xa0)
> kernel: [<c0106230>] (SyS_mount) from [<c000ed60>] (ret_fast_syscall+0x0/0x4c)
>
> --
> Raphael Derosso Pereira
>
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/

Hi:

It looks like the problem I've been experienced. Maybe this patch is helpfull.
https://patchwork.kernel.org/patch/9534489/

-- 
Cheers,
Rock