Re: [PATCH] debugfs: fix mount options not being applied

[Eric Sandeen <sandeen@redhat.com>]

On 8/4/25 12:22 PM, Eric Sandeen wrote:
> On 8/4/25 9:30 AM, Charalampos Mitrodimas wrote:
>> Mount options (uid, gid, mode) are silently ignored when debugfs is
>> mounted. This is a regression introduced during the conversion to the
>> new mount API.
>>
>> When the mount API conversion was done, the line that sets
>> sb->s_fs_info to the parsed options was removed. This causes
>> debugfs_apply_options() to operate on a NULL pointer.
>>
>> As an example, with the bug the "mode" mount option is ignored:
>>
>>   $ mount -o mode=0666 -t debugfs debugfs /tmp/debugfs_test
>>   $ mount | grep debugfs_test
>>   debugfs on /tmp/debugfs_test type debugfs (rw,relatime)
>>   $ ls -ld /tmp/debugfs_test
>>   drwx------ 25 root root 0 Aug  4 14:16 /tmp/debugfs_test
> 
> Argh. So, this looks a lot like the issue that got fixed for tracefs in:
> 
> e4d32142d1de tracing: Fix tracefs mount options
> 
> Let me look at this; tracefs & debugfs are quite similar, so perhaps
> keeping the fix consistent would make sense as well but I'll dig
> into it a bit more.

So, yes - a fix following the pattern of e4d32142d1de does seem to resolve
this issue.

However, I think we might be playing whack-a-mole here (fixing one fs at a time,
when the problem is systemic) among filesystems that use get_tree_single()
and have configurable options. For example, pstore:

# umount /sys/fs/pstore 

# mount -t pstore -o kmsg_bytes=65536 none /sys/fs/pstore
# mount | grep pstore
none on /sys/fs/pstore type pstore (rw,relatime,seclabel)

# mount -o remount,kmsg_bytes=65536 /sys/fs/pstore
# mount | grep pstore
none on /sys/fs/pstore type pstore (rw,relatime,seclabel,kmsg_bytes=65536)
#

I think gadgetfs most likely has the same problem but I'm not yet sure
how to test that.

I have no real objection to merging your patch, though I like the
consistency of following e4d32142d1de a bit more. But I think we should
find a graceful solution so that any filesystem using get_tree_single
can avoid this pitfall, if possible.

-Eric