Re: [PATCH] erofs: fix lockdep false positives on initializing erofs_pseudo_mnt

[Baokun Li]

On 2024/3/7 16:46, Al Viro wrote:
> On Thu, Mar 07, 2024 at 07:21:12AM +0000, Al Viro wrote:
> > On Thu, Mar 07, 2024 at 03:06:49PM +0800, Baokun Li wrote:
> > > > > +int erofs_anon_register_fs(void)
> > > > > +{
> > > > > +	return register_filesystem(&erofs_anon_fs_type);
> > > > > +}
> > > > What for?  The only thing it gives you is an ability to look it up by
> > > > name.  Which is completely pointless, IMO,
> > > The helper function here is to avoid extern erofs_anon_fs_type(), because
> > > we define it in fscache.c, but also use it in super.c. Moreover, we don't
> > > need
> > > to register it when CONFIG_EROFS_FS_ONDEMAND is not enabled, so we
> > You don't need to register it at all.
> >
> > The one and only effect of register_filesystem() is making file_system_type
> > instance visible to get_fs_type() (and making it show up in /proc/filesystems).
> >
> > That's it.  If you want to have it looked up by name (e.g. for userland
> > mounts), you need to register.  If not, you do not need to do that.
> >
> > Note that kern_mount() take a pointer to struct file_system_type,
> > not its (string) name.  So all you get from registration is an extra line
> > in /proc/filesystems.  What's the point?
> PS: at one point I considered renaming it to something that would sound
> less vague, but the best variant I'd been able to come up with was
> "publish_filesystem()", which is not much better and has an extra problem -
> how do you describe the reverse of that?  "withdraw_filesystem()"?
> Decided that it wasn't worth the amount of noise and headache...
I feel the emphasis on "fs_name" rather than "filesystem" is less
likely to be misunderstood. What do you think about renaming
to add_fs_name/remove_fs_name?

--
With Best Regards,
Baokun Li
.