Re: [PATCH] char: misc: Use __register_chrdev() to register the actual minor counts

["Hou Wenlong" <houwenlong.hwl@antgroup.com>]

On Fri, Mar 21, 2025 at 12:44:54PM +0800, Greg Kroah-Hartman wrote:
> On Fri, Mar 21, 2025 at 11:54:42AM +0800, Hou Wenlong wrote:
> > In commit ab760791c0cf ("char: misc: Increase the maximum number of
> > dynamic misc devices to 1048448"), the minor number of misc char device
> > is increased to MINORMASK + 1. However, it still uses register_chrdev(),
> > which defaults to registering the misc char device with only 256 minor
> > numbers.  Therefore, when a misc device is given a minor number greater
> > than 255, the device file fails to open in kobj_lookup(). As a solution,
> > use __register_chrdev() to register the actual minor counts to
> > accommodate this increase.
> > 
> > Signed-off-by: Hou Wenlong <houwenlong.hwl@antgroup.com>
> > ---
> > In our environment, the per-cpu coresight_tmc driver occupies too many
> > minor numbers, which ultimately allocates minor number 256 for our misc
> > device, resulting in a failure to open the device file in /dev. I'm not
> > sure if this is a bug or if userspace can only use fewer than 256 minor
> > numbers. Therefore, I didn't add a fix tag here.
> > ---
> >  drivers/char/misc.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/drivers/char/misc.c b/drivers/char/misc.c
> > index f7dd455dd0dd..dda466f9181a 100644
> > --- a/drivers/char/misc.c
> > +++ b/drivers/char/misc.c
> > @@ -315,7 +315,7 @@ static int __init misc_init(void)
> >  		goto fail_remove;
> >  
> >  	err = -EIO;
> > -	if (register_chrdev(MISC_MAJOR, "misc", &misc_fops))
> > +	if (__register_chrdev(MISC_MAJOR, 0, MINORMASK + 1, "misc", &misc_fops))
> >  		goto fail_printk;
> >  	return 0;
> >  
> > -- 
> > 2.31.1
> > 
> 
> I think this is the same change as this one:
> 	https://lore.kernel.org/r/20250317-misc-chrdev-v1-1-6cd05da11aef@igalia.com
> 
> right?
>
Yes, it's the same problem. I have replied to that one. Thank you!
 
> If so, can you provide a reviewed-by or tested-by on that one?  I
> haven't picked it up yet, sorry.
> 
> greg k-h