user namespace compile fix for 3.7

Linus,

Please pull the for-linus git tree from:

git://git.kernel.org:/pub/scm/linux/kernel/git/ebiederm/user-namespace.git for-linus

HEAD: 702e490211b2b7e448ebe1b3a07d97ad2fc07e03 userns: Fix build of drivers/staging/dgrp

This tree contains a trivial build fix for one of the staging drivers when user namespace support is enabled.

---
Subject: [PATCH] userns: Fix build of drivers/staging/dgrp

Explicitly test for GLOBAL_ROOT_UID and GLOBAL_ROOT_GID instead of using 0, allowing dgrp to compile and work properly when user namespace support is enabled.

---
 drivers/staging/dgrp/dgrp_common.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

index fce74e7..3553998 100644
@@ -179,9 +179,9 @@ void dgrp_carrier(struct ch_struct *ch)
  */
 int dgrp_chk_perm(int mode, int op)
 {
-	if (!current_euid())
+	if (!uid_eq(GLOBAL_ROOT_UID, current_euid()))
 		mode >>= 6;
-	else if (in_egroup_p(0))
+	else if (in_egroup_p(GLOBAL_ROOT_GID))
 		mode >>= 3;
 
 	if ((mode & op & 0007) == op)

1.7.5.4

Is this something new? Why didn't it show up in the linux-next builds?

I'll be glad to take this single patch myself and queue it up with other staging tree fixes before 3.7-final is out, what is the big rush?

thanks,

greg k-h

Yes this appears to have been a silent merge conflict between my user namespace tree and your staging tree.

I care because it breaks my builds.

This build failure only shows up when CONFIG_USER_NS is enabled which would have required a randconfig build in linux-next for it to show up.

When I get the last of the filesystems converted for 3.8 this kind of failure should stop hiding in the more customary build checks.

I figured I found it, I fixed it, and I am going to be carrying the fix anyway in my tree, so why not make it easy on everyone and make it a patch that can be pulled.

Eric