On Sat, 16 Oct 2021, Tetsuo Handa wrote:
> Commit bf9c0538e485b591 ("ataflop: use a separate gendisk for each media
> format") introduced ataflop_probe_lock mutex, but forgot to unlock the
> mutex when atari_floppy_init() (i.e. module loading) succeeded. If
> ataflop_probe() is called, it will deadlock on ataflop_probe_lock mutex.
>
> Signed-off-by: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx>
> Fixes: bf9c0538e485b591 ("ataflop: use a separate gendisk for each media format")
> ---
> To m68k users
>
> This patch suggests that nobody is testing this module using a real hardware.
> Can somebody test this module?
> Is current m68k hardware still supporting Atari floppy?
> If Atari floppy is no longer supported, do we still need this module?
>
It is only to be expected that no-one would have reported this bug yet.
2 months ago, Debian 11 shipped with a 5.10 kernel, but the bug you found
first appeared in Linux 5.11.
The existence of buggy drivers in mainline is undesirable but the real
problem here is the rate at which new bugs get added.
So I wonder if it would have been possible to use Aranym to find the
regression, or avoid it in the first place?
