Re: [PATCH 1/2] quota: Add mountpath based quota support

From: Jan Kara <jack@suse.cz>
To: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Christoph Hellwig <hch@infradead.org>,
	linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
	kernel@pengutronix.de, Jan Kara <jack@suse.com>,
	Richard Weinberger <richard@nod.at>
Subject: Re: [PATCH 1/2] quota: Add mountpath based quota support
Date: Fri, 12 Feb 2021 11:05:05 +0100	[thread overview]
Message-ID: <20210212100505.GT19070@quack2.suse.cz> (raw)
In-Reply-To: <20210212083835.GF19583@pengutronix.de>

On Fri 12-02-21 09:38:35, Sascha Hauer wrote:
> On Thu, Feb 11, 2021 at 03:38:13PM +0000, Christoph Hellwig wrote:
> > > +	if (!mountpoint)
> > > +		return -ENODEV;
> > > +
> > > +	ret = user_path_at(AT_FDCWD, mountpoint,
> > > +			     LOOKUP_FOLLOW | LOOKUP_AUTOMOUNT, &mountpath);
> > 
> > user_path_at handles an empty path, although you'll get EFAULT instead.
> > Do we care about the -ENODEV here?
> 
> The quotactl manpage documents EFAULT as error code for invalid addr or
> special argument, so we really should return -EFAULT here.
> 
> Existing quotactl gets this wrong as well:
> 
> 	if (!special) {
> 		if (cmds == Q_SYNC)
> 			return quota_sync_all(type);
> 		return -ENODEV;
> 	}
> 
> Should we fix this or is there userspace code that is confused by a changed
> return value?

I'd leave the original quotactl(2) as is. There's no strong reason to risk
breaking some userspace. For the new syscall, I agree we can just
standardize the return value, there ENODEV makes even less sense since
there's no device in that call.

								Honza

-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR