Re: [PATCH v2 1/3] acl: move idmapped mount fixup into vfs_{g,s}etxattr()

From: Seth Forshee <sforshee@digitalocean.com>
To: Christian Brauner <brauner@kernel.org>
Cc: Amir Goldstein <amir73il@gmail.com>,
	Miklos Szeredi <mszeredi@redhat.com>,
	Vivek Goyal <vgoyal@redhat.com>, Christoph Hellwig <hch@lst.de>,
	Aleksa Sarai <cyphar@cyphar.com>,
	linux-unionfs@vger.kernel.org, linux-fsdevel@vger.kernel.org
Subject: Re: [PATCH v2 1/3] acl: move idmapped mount fixup into vfs_{g,s}etxattr()
Date: Thu, 14 Jul 2022 16:36:21 -0500	[thread overview]
Message-ID: <YtCMVQIm0cU1wnYU@do-x1extreme> (raw)
In-Reply-To: <20220708090134.385160-2-brauner@kernel.org>

On Fri, Jul 08, 2022 at 11:01:32AM +0200, Christian Brauner wrote:
> This cycle we added support for mounting overlayfs on top of idmapped mounts.
> Recently I've started looking into potential corner cases when trying to add
> additional tests and I noticed that reporting for POSIX ACLs is currently wrong
> when using idmapped layers with overlayfs mounted on top of it.

<snip detailed explanation>

Beyond the issues described here, it also looks like the vfs_*() calls
are been inconsistent wrt idmapped mounts. With acls it takes/returns
unmapped ids, but other interfaces like vfs_getattr() return mapped ids.
So it makes sense to make vfs_{get,set}xattr() behave likewise.

I have one small suggestion below, but I think this looks good.

Reviewed-by: Seth Forshee <sforshee@digitalocean.com>

> +static inline bool is_posix_acl_xattr(const char *name)
> +{
> +	return (strcmp(name, XATTR_NAME_POSIX_ACL_ACCESS) == 0) ||
> +	       (strcmp(name, XATTR_NAME_POSIX_ACL_DEFAULT) == 0);
> +}
> +

There are locations still open-coding this check -- setxattr_convert()
and do_getxattr(). Maybe consider adding a follow-on patch to convert
those too.

Seth