Re: linux-next: manual merge of the pidfd tree with the vfs tree

[Stephen Rothwell]

Hi all,

On Mon, 25 Jan 2021 17:00:54 +1100 Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx> wrote:
>
> Today's linux-next merge of the pidfd tree got a conflict in:
> 
>   fs/namei.c
> 
> between commit:
> 
>   e36cffed20a3 ("fs: make unlazy_walk() error handling consistent")
>   1e8f44f159b3 ("do_tmpfile(): don't mess with finish_open()")
> 
> from the vfs tree and commit:
> 
>   47291baa8ddf ("namei: make permission helpers idmapped mount aware")
>   ba73d98745be ("namei: handle idmapped mounts in may_*() helpers")
>   549c7297717c ("fs: make helpers idmap mount aware")
> 
> from the pidfd tree.
> 
> I fixed it up (see below) and can carry the fix as necessary. This
> is now fixed as far as linux-next is concerned, but any non trivial
> conflicts should be mentioned to your upstream maintainer when your tree
> is submitted for merging.  You may also want to consider cooperating
> with the maintainer of the conflicting tree to minimise any particularly
> complex conflicts.
> 
> diff --cc fs/namei.c
> index 4cae88733a5c,dbf53b325ac9..000000000000
> --- a/fs/namei.c
> +++ b/fs/namei.c
> @@@ -1568,14 -1639,18 +1644,16 @@@ static struct dentry *lookup_slow(cons
>   	return res;
>   }
>   
> - static inline int may_lookup(struct nameidata *nd)
> + static inline int may_lookup(struct user_namespace *mnt_userns,
> + 			     struct nameidata *nd)
>   {
>   	if (nd->flags & LOOKUP_RCU) {
> - 		int err = inode_permission(nd->inode, MAY_EXEC|MAY_NOT_BLOCK);
> + 		int err = inode_permission(mnt_userns, nd->inode,
> + 					   MAY_EXEC | MAY_NOT_BLOCK);
>  -		if (err != -ECHILD)
>  +		if (err != -ECHILD || !try_to_unlazy(nd))
>   			return err;
>  -		if (unlazy_walk(nd))
>  -			return -ECHILD;
>   	}
> - 	return inode_permission(nd->inode, MAY_EXEC);
> + 	return inode_permission(mnt_userns, nd->inode, MAY_EXEC);
>   }
>   
>   static int reserve_stack(struct nameidata *nd, struct path *link, unsigned seq)
> @@@ -3324,9 -3453,11 +3453,9 @@@ static int do_tmpfile(struct nameidata 
>   	path.dentry = child;
>   	audit_inode(nd->name, child, 0);
>   	/* Don't check for other permissions, the inode was just created */
> - 	error = may_open(&path, 0, op->open_flag);
> + 	error = may_open(mnt_userns, &path, 0, op->open_flag);
>  -	if (error)
>  -		goto out2;
>  -	file->f_path.mnt = path.mnt;
>  -	error = finish_open(file, child, NULL);
>  +	if (!error)
>  +		error = vfs_open(&path, file);
>   out2:
>   	mnt_drop_write(path.mnt);
>   out:

With the merge window about to open, this is a reminder that this
conflict still exists.

Those vfs tree commits have also been merged into the block tree.

-- 
Cheers,
Stephen Rothwell

Attachment: pgpUQL41ugSVI.pgp
Description: OpenPGP digital signature