On Wed, Jan 27, 2021 at 11:24:41AM +1100, Stephen Rothwell wrote:

Hi all,

On Mon, 25 Jan 2021 17:14:14 +1100 Stephen Rothwell [off-list ref] wrote:

quoted

Today's linux-next merge of the pidfd tree got a conflict in:

  fs/xfs/xfs_inode.c

between commit:

  01ea173e103e ("xfs: fix up non-directory creation in SGID directories")

from the xfs tree and commit:

  f736d93d76d3 ("xfs: support idmapped mounts")

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/xfs/xfs_inode.c
index e2a1db4cee43,95b7f2ba4e06..000000000000

--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c

@@@ -809,13 -810,13 +810,13 @@@ xfs_init_new_inode
  	inode->i_rdev = rdev;
  	ip->i_d.di_projid = prid;
  
 -	if (pip && XFS_INHERIT_GID(pip)) {
 -		inode->i_gid = VFS_I(pip)->i_gid;
 -		if ((VFS_I(pip)->i_mode & S_ISGID) && S_ISDIR(mode))
 -			inode->i_mode |= S_ISGID;
 +	if (dir && !(dir->i_mode & S_ISGID) &&
 +	    (mp->m_flags & XFS_MOUNT_GRPID)) {
 +		inode->i_uid = current_fsuid();

Looking a bit harder, I replaced the above line with
		inode->i_uid = fsuid_into_mnt(mnt_userns);

I think that looks good, though Mr. Brauner is probably better equipped
to tell if that change is correct.

(He says watching kernel.org mail take nearly a day to come through...)

--D

quoted

 +		inode->i_gid = dir->i_gid;
 +		inode->i_mode = mode;
  	} else {
- 		inode_init_owner(inode, dir, mode);
 -		inode->i_gid = fsgid_into_mnt(mnt_userns);
++		inode_init_owner(mnt_userns, inode, dir, mode);
  	}
  
  	/*

-- 
Cheers,
Stephen Rothwell