Re: [PATCH dhowells/mount-context] fs: don't call fs_context->free() from fsmount()

[Andrei Vagin <avagin@virtuozzo.com>]

On Tue, Jul 31, 2018 at 09:52:36AM +0100, David Howells wrote:
> Andrei Vagin <avagin@openvz.org> wrote:
> 
> > @@ -3435,9 +3435,6 @@ SYSCALL_DEFINE3(fsmount, int, fs_fd, unsigned int, flags, unsigned int, ms_flags
> >  	 * do any memory allocation or anything like that at this point as we
> >  	 * don't want to have to handle any errors incurred.
> >  	 */
> > -	if (fc->ops && fc->ops->free)
> > -		fc->ops->free(fc);
> > -	fc->fs_private = NULL;
> >  	fc->s_fs_info = NULL;
> >  	fc->sb_flags = 0;
> >  	fc->sloppy = false;
> 
> This isn't the right fix.  The context needs to be reset at this point so that
> it's prepared to be reinitialised into in the same state as one generated by
> fspick().

I understand this. init_fs_context() is called from fspick() and
fs_context->free() is called for contexts which have been created in
fspick().

> 
> I can do this two ways: (1) stick a flag in the context that says if ->free()
> needs calling, (2) make all the ->free() routines aware that they may see the
> reset state.  I think (1) is less error prone.

Does it mean that fc->fs_type->init_fs_context() should not be called
contexts which are created from fspick()?

> 
> David