• State Resolved
  • +1 person also asked this people also asked this
  • Locked Locked
  • Replies 2 replies
  • Subscribers 99 subscribers
  • Views 4491 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Linux/AM5726: EXT4-fs errors running from eMMC

Mark Lombardi
Prodigy 120 points
Part Number: AM5726

Tool/software: Linux

Hi There,

I'm just finishing up a migration from an AM5726 board from TI-SDK 3.02.00.05 to TI-SDK 5.01.00.11.

After running swupdate to flash the eMMC from the SDCard, I am seeing a series of ext4-fs errors. Also after booting from the eMMC there appears to be considerable file system corruption.

This seems to be a known Linux Kernel issue. It appears a fix was made for Kernel versions 4.19x and 4.20.x. I was wondering if a back ported fix was available for Linux Kernel 4.14.x which is included in TI SDK 5.x? I did not find this particular issue listed in the Release Notes.


Incorrect disk IO caused by blk-mq direct issue can lead to file system corruption

EXT4-fs error (device sda7): ext4_iget:4831: inode

 

Partitioning Script

======================

prepare_emmc() {
    
    fw_setenv factorymode 0

    echo 0 > /sys/block/mmcblk1boot0/force_ro
    dd if=/dev/zero of=/dev/mmcblk1boot0 bs=512 count=4096
    dd if=/run/media/mmcblk0p1/MLO of=/dev/mmcblk1boot0 bs=512
    dd if=/run/media/mmcblk0p1/u-boot.img of=/dev/mmcblk1boot0 bs=512 seek=768

    unmount /dev/mmcblk1p1
    unmount /dev/mmcblk1p2
    unmount /dev/mmcblk1p3
    sleep 1
    echo yes | parted -s /dev/mmcblk1 rm 1
    echo yes | parted -s /dev/mmcblk1 rm 2
    echo yes | parted -s /dev/mmcblk1 rm 3

    parted -s /dev/mmcblk1 mkpart primary ext4 0% 40%
    parted -s /dev/mmcblk1 mkpart primary ext4 40% 80%
    parted -s /dev/mmcblk1 mkpart primary ext4 80% 100%
    sleep 1
    unmount /dev/mmcblk1p1
    unmount /dev/mmcblk1p2
    unmount /dev/mmcblk1p3
    sleep 1
    echo yes | mkfs.ext4 /dev/mmcblk1p1
    echo yes | mkfs.ext4 /dev/mmcblk1p2
    echo yes | mkfs.ext4 /dev/mmcblk1p3
    mount_if mmcblk1p1
    mount_if mmcblk1p2
    mount_if mmcblk1p3
    cd /run/media/mmcblk0p3

    swupdate -v -H board:3.50 -i board_3.5.swu -e board,eMMC-install >> recovery.$timestamp.log 2>&1
    fw_setenv mmcdev 1
    reboot
}

 

 

dmesg logs listed below
=======================

root@am57xx-evm:~# stat /
  File: /
  Size: 4096            Blocks: 8          IO Block: 4096   directory
Device: b311h/45841d    Inode: 2           Links: 21
Access: (0755/drwxr-xr-x)  Uid: (    0/    root)   Gid: (    0/    root)
Access: 2019-02-26 06:27:20.000000000
Modify: 2019-02-26 06:34:41.000000000
Change: 2019-02-26 06:34:41.000000000


root@am57xx-evm:~# dmesg | grep -i ext4
[    0.000000] Kernel command line: console=ttyS2,115200n8 net.ifnames=0 biosdevname=0 root=PARTUUID=d093c9e2-01 rw rootfstype=ext4 rootwait
[    3.795105] EXT4-fs (mmcblk1p1): warning: mounting fs with errors, running e2fsck is recommended
[    3.805298] EXT4-fs (mmcblk1p1): mounted filesystem with ordered data mode. Opts: (null)
[    3.813458] VFS: Mounted root (ext4 filesystem) on device 179:17.
[    4.999205] EXT4-fs (mmcblk1p1): re-mounted. Opts: (null)
[    8.168201] EXT4-fs (mmcblk1p2): mounted filesystem with ordered data mode. Opts: (null)
[    8.213794] EXT4-fs (mmcblk1p3): mounted filesystem with ordered data mode. Opts: (null)
[    8.228680] EXT4-fs (mmcblk0p3): mounting ext3 file system using the ext4 subsystem
[    8.528818] EXT4-fs error (device mmcblk1p1): ext4_iget:4723: inode #4331: comm grep: checksum invalid
[    8.537106] EXT4-fs error (device mmcblk1p1): ext4_iget:4723: inode #4033: comm grep: checksum invalid
[    8.550659] EXT4-fs error (device mmcblk1p1): ext4_iget:4723: inode #4019: comm grep: checksum invalid
[    8.552701] EXT4-fs error (device mmcblk1p1): ext4_iget:4723: inode #4509: comm grep: checksum invalid
[    8.555045] EXT4-fs error (device mmcblk1p1): ext4_iget:4723: inode #4697: comm grep: checksum invalid
[    8.556233] EXT4-fs error (device mmcblk1p1): ext4_iget:4723: inode #3908: comm grep: checksum invalid
[    8.557062] EXT4-fs error (device mmcblk1p1): ext4_iget:4723: inode #3872: comm grep: checksum invalid
[    8.558947] EXT4-fs error (device mmcblk1p1): ext4_iget:4723: inode #4474: comm grep: checksum invalid
[    8.575134] EXT4-fs error (device mmcblk1p1): ext4_iget:4723: inode #4216: comm grep: checksum invalid
[    8.576337] EXT4-fs error (device mmcblk1p1): ext4_iget:4723: inode #4078: comm grep: checksum invalid
[    9.029010] EXT4-fs (mmcblk0p3): mounted filesystem with ordered data mode. Opts: (null)
[    9.042961] EXT4-fs (mmcblk0p2): mounting ext3 file system using the ext4 subsystem
[    9.139316] EXT4-fs (mmcblk0p2): mounted filesystem with ordered data mode. Opts: (null)
...
[    8.852095] EXT4-fs error: 356 callbacks suppressed
[    8.852102] EXT4-fs error (device mmcblk1p1): ext4_iget:4703: inode #3742: comm modprobe: bad extra_isize 11640 (inode size 256)
[    8.871310] EXT4-fs error (device mmcblk1p1): ext4_iget:4703: inode #3742: comm mount.sh: bad extra_isize 11640 (inode size 256)
[    8.885920] EXT4-fs error (device mmcblk1p1): ext4_iget:4703: inode #3742: comm mkdir: bad extra_isize 11640 (inode size 256)
[    8.901446] EXT4-fs error (device mmcblk1p1): ext4_iget:4703: inode #3742: comm basename: bad extra_isize 11640 (inode size 256)
[    8.915776] EXT4-fs error (device mmcblk1p1): ext4_iget:4703: inode #3742: comm grep: bad extra_isize 11640 (inode size 256)
[    8.930319] EXT4-fs error (device mmcblk1p1): ext4_iget:4703: inode #3742: comm readlink: bad extra_isize 11640 (inode size 256)
[    8.945352] EXT4-fs error (device mmcblk1p1): ext4_iget:4703: inode #3742: comm mkdir: bad extra_isize 11640 (inode size 256)
[    8.959794] EXT4-fs error (device mmcblk1p1): ext4_iget:4703: inode #3742: comm expr: bad extra_isize 11640 (inode size 256)
[    8.974192] EXT4-fs error (device mmcblk1p1): ext4_iget:4703: inode #3742: comm mount: bad extra_isize 11640 (inode size 256)
[    8.988109] EXT4-fs error (device mmcblk1p1): ext4_iget:4703: inode #3742: comm readlink: bad extra_isize 11640 (inode size 256)

...

[ 320.470812] EXT4-fs (mmcblk1p1): error count since last fsck: 1354
[ 320.477043] EXT4-fs (mmcblk1p1): initial error at time 1551162840: ext4_iget:4703: inode 3729
[ 320.485638] EXT4-fs (mmcblk1p1): last error at time 1551162334: ext4_iget:4723: inode 4427

 

  • 0
    TI__Mastermind 30536 points

    Thanks for your post.

    I looked through your referenced threads and it seems that that issue was a regression from 4.19.0 to 4.19.3 or so and was then fixed. I don't see any evidence that it was believed to exist in earlier kernels. Did I miss something that indicates this?

    If you start with a clean filesystem do you see this error? Ideally, using the FS provided with the SDK simply untarred to a partition, much the same way an SD Card is created.

    It would be good to establish a baseline using the files provided with the SDK. We haven't had any other reports of this to my knowledge.

    Thanks.

  • 0 in reply to RonB
    Prodigy 120 points

    Hi Ron,

    Thanks for replying.

    Sorry for the delay, I wanted to ensure the problem was fully resolved and it took me a while to go through the process. Your suggestion of starting with a clean filesystem showed there was no problem with the pre-built binaries.

    The problem ended up being a build issue. A mix of TI SDK and Yocto u-boot and Linux kernel dependencies were being blended together.

    Thanks for the tip to try loading the pre-built binaries.