Disable the output of debugging information about unused file systems in Linux kernels #170

Activity

1. p0i5k

   added

   enhancementNew feature or requestNew feature or request

   kernel

   on Jun 15, 2022
2. widgetii

   added this to firmwareon Apr 13, 2026
3. widgetii

   added

   platform:UnknownUnknown/unidentified platformUnknown/unidentified platform

   on Apr 13, 2026
4. widgetii

   added 2 commits that reference this issue on Jun 19, 2026

   hi3516cv200_neo + hi3516av100_neo: add V2/V2A modern-kernel firmware …

   ...

   27aeedd

   hi3516cv200_neo + hi3516av100_neo: add V2/V2A modern-kernel firmware …

   ...

   e11d139
5. openipc-ai commented on Oct 26, 2026

   

   openipc-ai

   on Oct 26, 2026

   Collaborator

   More actions

   This is fixed on master by #2413 (mdev: mount a card with no partition table, and say so when nothing mounts). It was a userspace problem, not a kernel one, so nothing needs silencing in the kernel.

   Cause. lib/mdev/automount.sh used to mount cards with mount -t auto. busybox then tries every block filesystem in /proc/filesystems, in order. On these kernels yaffs/yaffs2 are built in and come before vfat (a module on most boards), so every card insert first tried a raw-NAND filesystem on the SD card, and F2FS too on kernels that have it. The lines in this issue are those failed attempts. #2413 replaced auto with a named list (disk_fstypes: vfat first, then the other block filesystems the board configs enable, with yaffs, jffs2, ubifs and squashfs left out).

   The noise was the harmless part. When yaffs2 doesn't decline, mount -t auto can crash the kernel, and #2413 recorded two such Oopses on a hi3516av300.

   Before/after, reproduced in qemu-bench (hi3516ev300, booted from NOR, stock nightly-20261006-8a81fed). after is that image as shipped. before is the same image with only automount.sh swapped for the pre-#2413 version. The kernel lines come from dmesg, because /etc/sysctl.conf keeps them off the console.

   card	before (mount -t auto)	after (master)
   FAT32, 1 partition	the yaffs lines from this issue, then mounts vfat	mounts vfat, no lines
   FAT32, dirty flag set	yaffs lines + FAT-fs: Volume was not properly unmounted	FAT-fs: Volume was not properly unmounted only
   FAT32, no partition table	yaffs lines on mmcblk0, then mounts vfat	mounts vfat, no lines
   blank or ext4 p1, FAT32 p2	yaffs lines on both; on p2 yaffs2 gets past probing → kernel BUG at fs/yaffs2/yaffs_getblockinfo.h:31, Internal error: Oops - BUG in mount (2 of 3 runs)	p1: one automount: cannot mount /dev/mmcblk0p1 as vfat exfat ... line; p2 mounts vfat; no kernel lines

   The output in this issue, as it appears in the before run:

   yaffs: dev is 187695105 name is "mmcblk0p1" rw
   yaffs: passed flags ""
   yaffs: yaffs: Attempting MTD mount of 179.1,"mmcblk0p1"
   yaffs: MTD device does not support have the right page sizes
   yaffs: yaffs: Attempting MTD mount of 179.1,"mmcblk0p1"
   yaffs: NAND geometry problems: chunk size 1, type is yaffs2, inband_tags 1
   

   What stays, on purpose. FAT-fs (...): Volume was not properly unmounted is the kernel reporting a real dirty flag on the card. It means the card should be fsck'd and should stay visible. The automount: cannot mount ... line is #2413's diagnosis for a card with nothing mountable on it.

   Gaps. This was verified in emulation only, on the ev300 kernel (yaffs built in, no F2FS), not on a camera with a card in it. F2FS lines on boards whose kernel has F2FS weren't separately reproduced. On master f2fs is tried only after vfat has declined, so a FAT card never reaches it.

   Closing as fixed by #2413. Please reopen with dmesg from a current build if you still see foreign-filesystem lines on a FAT card.
6. openipc-ai

   closed this as completedon Oct 26, 2026