Should the "loop" driver reject using pseudo files as backing file?

Should the "loop" driver reject using pseudo files as backing file?

[Tetsuo Handa]

I noticed that /dev/loopX accepts pseudo files, for currently
loop_validate_file() does

	if (!S_ISREG(inode->i_mode) && !S_ISBLK(inode->i_mode))
		return -EINVAL;

and pseudo files are treated as S_ISREG().

Reading pseudo files via /dev/loopX causes bogus results (tries to
repeatedly read the entire content up to the size visible to "ls"
command). I think that allowing such usage will confuse userspace
programs.

[root@localhost ~]# ls -l /sys/power/pm_test
-rw-r--r-- 1 root root 4096 May 26 22:14 /sys/power/pm_test
[root@localhost ~]# cat /sys/power/pm_test | wc
      1       6      48
[root@localhost ~]# cat $(losetup -f --show /sys/power/pm_test) | wc
     85     513    4096

Writing to pseudo files via /dev/loopX seems to work (at least works for
/sys/power/pm_test ), but can/should we forbid binding to pseudo files?

[root@localhost ~]# echo none > /sys/power/pm_test
[root@localhost ~]# echo none > $(losetup -f --show /sys/power/pm_test)

F.Y.I. An analysis by Google AI mode (expires in 7 days) is at https://share.google/aimode/prhETTZMQEzsw5HV9 .

Re: Should the "loop" driver reject using pseudo files as backing file?

[Christoph Hellwig]

On Wed, May 27, 2026 at 10:52:02AM +0900, Tetsuo Handa wrote:
> I noticed that /dev/loopX accepts pseudo files, for currently
> loop_validate_file() does
> 
> 	if (!S_ISREG(inode->i_mode) && !S_ISBLK(inode->i_mode))
> 		return -EINVAL;
> 
> and pseudo files are treated as S_ISREG().
> 
> Reading pseudo files via /dev/loopX causes bogus results (tries to
> repeatedly read the entire content up to the size visible to "ls"
> command). I think that allowing such usage will confuse userspace
> programs.

You get what you pay for.  Many other in-kernel and userspace users
will be just as confused by them.