Skip to contentSkip to conversation

nfsd: use vfs setgid helper

PATCH In review Christian Brauner 5 participants 11 messages May 2, 2023 – May 15, 2023

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 nfsd: use vfs setgid helper

We've aligned setgid behavior over multiple kernel releases. The details can be found in commit cf619f891971 ("Merge tag 'fs.ovl.setgid.v6.2' of
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping") and commit 426b4ca2d6a5 ("Merge tag 'fs.setgid.v6.0' of
git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux").
Consistent setgid stripping behavior is now encapsulated in the setattr_should_drop_sgid() helper which is used by all filesystems that strip setgid bits outside of vfs proper. Usually ATTR_KILL_SGID is raised in e.g., chown_common() and is subject to the setattr_should_drop_sgid() check to determine whether the setgid bit can be retained. Since nfsd is raising ATTR_KILL_SGID unconditionally it will cause notify_change() to strip it even if the caller had the necessary privileges to retain it. Ensure that nfsd only raises ATR_KILL_SGID if the caller lacks the necessary privileges to retain the setgid bit.

Without this patch the setgid stripping tests in LTP will fail:

Quoted 2 lines

As you can see, the problem is S_ISGID (0002000) was dropped on a non-group-executable file while chown was invoked by super-user, while

[...]

Quoted 1 line

fchown02.c:66: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

[...]

Quoted 1 line

chown02.c:57: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

With this patch all tests pass.

---

ubuntu@imp1-vm:~/ltp-install$ sudo ./runltp -d /mnt -s chown02
INFO: ltp-pan reported all tests PASS
LTP Version: 20230127-112-gf41e8a2fa
ubuntu@imp1-vm:~/ltp-install$ sudo ./runltp -d /mnt -s fchown02
INFO: ltp-pan reported all tests PASS
LTP Version: 20230127-112-gf41e8a2fa
ubuntu@imp1-vm:~/src/git/xfstests$ sudo ./check -g perms
FSTYP         -- nfs
PLATFORM      -- Linux/x86_64 imp1-vm 6.3.0-nfs-setgid-3a3cfe624076 #20 SMP PREEMPT_DYNAMIC Tue May  2 12:35:51 UTC 2023
MKFS_OPTIONS  -- 127.0.0.1:/nfsscratch
MOUNT_OPTIONS -- -o vers=3,noacl 127.0.0.1:/nfsscratch /mnt/scratch
Passed all 41 tests
---
 fs/nfsd/vfs.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
fs/nfsd/vfs.c +3 -1
index bb9d47172162..c4ef24c5ffd0 100644
@@ -388,7 +388,9 @@ nfsd_sanitize_attrs(struct inode *inode, struct iattr *iap)
 				iap->ia_mode &= ~S_ISGID;
 		} else {
 			/* set ATTR_KILL_* bits and let VFS handle it */
-			iap->ia_valid |= (ATTR_KILL_SUID | ATTR_KILL_SGID);
+			iap->ia_valid |= ATTR_KILL_SUID;
+			iap->ia_valid |=
+				setattr_should_drop_sgid(&nop_mnt_idmap, inode);
 		}
 	}
 }
Signature · 1 line
2.34.1
#1 lore raw
Quoted 28 lines · Christian Brauner

We've aligned setgid behavior over multiple kernel releases. The details can be found in commit cf619f891971 ("Merge tag 'fs.ovl.setgid.v6.2' of
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping") and commit 426b4ca2d6a5 ("Merge tag 'fs.setgid.v6.0' of
git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux").
Consistent setgid stripping behavior is now encapsulated in the setattr_should_drop_sgid() helper which is used by all filesystems that strip setgid bits outside of vfs proper. Usually ATTR_KILL_SGID is raised in e.g., chown_common() and is subject to the setattr_should_drop_sgid() check to determine whether the setgid bit can be retained. Since nfsd is raising ATTR_KILL_SGID unconditionally it will cause notify_change() to strip it even if the caller had the necessary privileges to retain it. Ensure that nfsd only raises ATR_KILL_SGID if the caller lacks the necessary privileges to retain the setgid bit.

Without this patch the setgid stripping tests in LTP will fail:

Quoted 2 lines

As you can see, the problem is S_ISGID (0002000) was dropped on a non-group-executable file while chown was invoked by super-user, while

[...]

Quoted 1 line

fchown02.c:66: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

Quoted 8 lines

[...]

Quoted 1 line

chown02.c:57: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

With this patch all tests pass.

There are some similar fstests failures this fix might address.

I've applied this patch to the nfsd-fixes tree for broader testing. Thanks, Christian and Sherry!

Quoted 37 lines

---

ubuntu@imp1-vm:~/ltp-install$ sudo ./runltp -d /mnt -s chown02
INFO: ltp-pan reported all tests PASS
LTP Version: 20230127-112-gf41e8a2fa
ubuntu@imp1-vm:~/ltp-install$ sudo ./runltp -d /mnt -s fchown02
INFO: ltp-pan reported all tests PASS
LTP Version: 20230127-112-gf41e8a2fa
ubuntu@imp1-vm:~/src/git/xfstests$ sudo ./check -g perms
FSTYP         -- nfs
PLATFORM      -- Linux/x86_64 imp1-vm 6.3.0-nfs-setgid-3a3cfe624076 #20 SMP PREEMPT_DYNAMIC Tue May  2 12:35:51 UTC 2023
MKFS_OPTIONS  -- 127.0.0.1:/nfsscratch
MOUNT_OPTIONS -- -o vers=3,noacl 127.0.0.1:/nfsscratch /mnt/scratch
Passed all 41 tests
---
fs/nfsd/vfs.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
fs/nfsd/vfs.c +3 -1
index bb9d47172162..c4ef24c5ffd0 100644
@@ -388,7 +388,9 @@ nfsd_sanitize_attrs(struct inode *inode, struct iattr *iap)
 iap->ia_mode &= ~S_ISGID;
 } else {
 /* set ATTR_KILL_* bits and let VFS handle it */
- iap->ia_valid |= (ATTR_KILL_SUID | ATTR_KILL_SGID);
+ iap->ia_valid |= ATTR_KILL_SUID;
+ iap->ia_valid |=
+ setattr_should_drop_sgid(&nop_mnt_idmap, inode);
 }
 }
 }
Signature · 1 line
2.34.1

--
Chuck Lever

#2 lore raw
Quoted 38 lines · Chuck Lever III
Quoted 35 lines · Christian Brauner

We've aligned setgid behavior over multiple kernel releases. The details can be found in commit cf619f891971 ("Merge tag 'fs.ovl.setgid.v6.2' of
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping") and commit 426b4ca2d6a5 ("Merge tag 'fs.setgid.v6.0' of
git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux").
Consistent setgid stripping behavior is now encapsulated in the setattr_should_drop_sgid() helper which is used by all filesystems that strip setgid bits outside of vfs proper. Usually ATTR_KILL_SGID is raised in e.g., chown_common() and is subject to the setattr_should_drop_sgid() check to determine whether the setgid bit can be retained. Since nfsd is raising ATTR_KILL_SGID unconditionally it will cause notify_change() to strip it even if the caller had the necessary privileges to retain it. Ensure that nfsd only raises ATR_KILL_SGID if the caller lacks the necessary privileges to retain the setgid bit.

Without this patch the setgid stripping tests in LTP will fail:

Quoted 2 lines

As you can see, the problem is S_ISGID (0002000) was dropped on a non-group-executable file while chown was invoked by super-user, while

[...]

Quoted 1 line

fchown02.c:66: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

[...]

Quoted 1 line

chown02.c:57: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

With this patch all tests pass.

Quoted 7 lines
Quoted 2 lines

There are some similar fstests failures this fix might address.

I've applied this patch to the nfsd-fixes tree for broader testing.

ERROR: modpost: "setattr_should_drop_sgid" [fs/nfsd/nfsd.ko] undefined!

Did I apply this patch to the wrong kernel?

Quoted 35 lines

Thanks, Christian and Sherry!

Quoted 32 lines

---

ubuntu@imp1-vm:~/ltp-install$ sudo ./runltp -d /mnt -s chown02
INFO: ltp-pan reported all tests PASS
LTP Version: 20230127-112-gf41e8a2fa
ubuntu@imp1-vm:~/ltp-install$ sudo ./runltp -d /mnt -s fchown02
INFO: ltp-pan reported all tests PASS
LTP Version: 20230127-112-gf41e8a2fa
ubuntu@imp1-vm:~/src/git/xfstests$ sudo ./check -g perms
FSTYP         -- nfs
PLATFORM      -- Linux/x86_64 imp1-vm 6.3.0-nfs-setgid-3a3cfe624076 #20 SMP PREEMPT_DYNAMIC Tue May  2 12:35:51 UTC 2023
MKFS_OPTIONS  -- 127.0.0.1:/nfsscratch
MOUNT_OPTIONS -- -o vers=3,noacl 127.0.0.1:/nfsscratch /mnt/scratch
Passed all 41 tests
---
fs/nfsd/vfs.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
fs/nfsd/vfs.c +3 -1
index bb9d47172162..c4ef24c5ffd0 100644
@@ -388,7 +388,9 @@ nfsd_sanitize_attrs(struct inode *inode, struct iattr *iap)
 iap->ia_mode &= ~S_ISGID;
 } else {
 /* set ATTR_KILL_* bits and let VFS handle it */
- iap->ia_valid |= (ATTR_KILL_SUID | ATTR_KILL_SGID);
+ iap->ia_valid |= ATTR_KILL_SUID;
+ iap->ia_valid |=
+ setattr_should_drop_sgid(&nop_mnt_idmap, inode);
 }
Quoted 8 lines
Quoted 5 lines
fs/nfsd/vfs.c +0 -0
 }
 }
Signature · 1 line
2.34.1

--
Chuck Lever

--
Chuck Lever

#3 lore raw
Quoted 45 lines · Chuck Lever III
Quoted 42 lines · Chuck Lever III
Quoted 37 lines · Christian Brauner

We've aligned setgid behavior over multiple kernel releases. The details can be found in commit cf619f891971 ("Merge tag 'fs.ovl.setgid.v6.2' of
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping") and commit 426b4ca2d6a5 ("Merge tag 'fs.setgid.v6.0' of
git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux").
Consistent setgid stripping behavior is now encapsulated in the setattr_should_drop_sgid() helper which is used by all filesystems that strip setgid bits outside of vfs proper. Usually ATTR_KILL_SGID is raised in e.g., chown_common() and is subject to the setattr_should_drop_sgid() check to determine whether the setgid bit can be retained. Since nfsd is raising ATTR_KILL_SGID unconditionally it will cause notify_change() to strip it even if the caller had the necessary privileges to retain it. Ensure that nfsd only raises ATR_KILL_SGID if the caller lacks the necessary privileges to retain the setgid bit.

Without this patch the setgid stripping tests in LTP will fail:

Quoted 2 lines

As you can see, the problem is S_ISGID (0002000) was dropped on a non-group-executable file while chown was invoked by super-user, while

[...]

Quoted 1 line

fchown02.c:66: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

[...]

Quoted 1 line

chown02.c:57: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

With this patch all tests pass.

Quoted 8 lines
Quoted 4 lines

There are some similar fstests failures this fix might address.

I've applied this patch to the nfsd-fixes tree for broader testing.

ERROR: modpost: "setattr_should_drop_sgid" [fs/nfsd/nfsd.ko] undefined!

Did I apply this patch to the wrong kernel?

setattr_should_drop_sgid() is not available to callers built as modules. It needs an EXPORT_SYMBOL or _GPL.

Quoted 45 lines
Quoted 35 lines

Thanks, Christian and Sherry!

Quoted 32 lines

---

ubuntu@imp1-vm:~/ltp-install$ sudo ./runltp -d /mnt -s chown02
INFO: ltp-pan reported all tests PASS
LTP Version: 20230127-112-gf41e8a2fa
ubuntu@imp1-vm:~/ltp-install$ sudo ./runltp -d /mnt -s fchown02
INFO: ltp-pan reported all tests PASS
LTP Version: 20230127-112-gf41e8a2fa
ubuntu@imp1-vm:~/src/git/xfstests$ sudo ./check -g perms
FSTYP         -- nfs
PLATFORM      -- Linux/x86_64 imp1-vm 6.3.0-nfs-setgid-3a3cfe624076 #20 SMP PREEMPT_DYNAMIC Tue May  2 12:35:51 UTC 2023
MKFS_OPTIONS  -- 127.0.0.1:/nfsscratch
MOUNT_OPTIONS -- -o vers=3,noacl 127.0.0.1:/nfsscratch /mnt/scratch
Passed all 41 tests
---
fs/nfsd/vfs.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
fs/nfsd/vfs.c +3 -1
index bb9d47172162..c4ef24c5ffd0 100644
@@ -388,7 +388,9 @@ nfsd_sanitize_attrs(struct inode *inode, struct iattr *iap)
 iap->ia_mode &= ~S_ISGID;
 } else {
 /* set ATTR_KILL_* bits and let VFS handle it */
- iap->ia_valid |= (ATTR_KILL_SUID | ATTR_KILL_SGID);
+ iap->ia_valid |= ATTR_KILL_SUID;
+ iap->ia_valid |=
+ setattr_should_drop_sgid(&nop_mnt_idmap, inode);
 }
Quoted 8 lines
Quoted 5 lines
fs/nfsd/vfs.c +0 -0
 }
 }
Signature · 1 line
2.34.1

--
Chuck Lever

Quoted 2 lines

--
Chuck Lever

--
Chuck Lever

#4 lore raw
Quoted 50 lines · Chuck Lever III
Quoted 50 lines · Chuck Lever III
Quoted 46 lines · Chuck Lever III
Quoted 29 lines · Christian Brauner

We've aligned setgid behavior over multiple kernel releases. The details can be found in commit cf619f891971 ("Merge tag 'fs.ovl.setgid.v6.2' of
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping") and commit 426b4ca2d6a5 ("Merge tag 'fs.setgid.v6.0' of
git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux").
Consistent setgid stripping behavior is now encapsulated in the setattr_should_drop_sgid() helper which is used by all filesystems that strip setgid bits outside of vfs proper. Usually ATTR_KILL_SGID is raised in e.g., chown_common() and is subject to the setattr_should_drop_sgid() check to determine whether the setgid bit can be retained. Since nfsd is raising ATTR_KILL_SGID unconditionally it will cause notify_change() to strip it even if the caller had the necessary privileges to retain it. Ensure that nfsd only raises ATR_KILL_SGID if the caller lacks the necessary privileges to retain the setgid bit.

Without this patch the setgid stripping tests in LTP will fail:

Quoted 2 lines

As you can see, the problem is S_ISGID (0002000) was dropped on a non-group-executable file while chown was invoked by super-user, while

[...]

Quoted 1 line

fchown02.c:66: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

Quoted 8 lines

[...]

Quoted 1 line

chown02.c:57: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

With this patch all tests pass.

There are some similar fstests failures this fix might address.

I've applied this patch to the nfsd-fixes tree for broader testing.

Quoted 6 lines
Quoted 3 lines

ERROR: modpost: "setattr_should_drop_sgid" [fs/nfsd/nfsd.ko] undefined!

Did I apply this patch to the wrong kernel?

setattr_should_drop_sgid() is not available to callers built as modules. It needs an EXPORT_SYMBOL or _GPL.

Hey Chuck,

Thanks for taking a look!
The required export is part of
commit 4f704d9a835 ("nfs: use vfs setgid helper")
which is in current mainline as of Monday this week which was part of my v6.4/vfs.misc PR that was merged.

So this should all work fine on mainline. Seems I didn't use --base which is why that info was missing.

Thanks!
Christian

#5 lore raw

Hi Christian,

Quoted 39 lines · Christian Brauner
Quoted 38 lines · Chuck Lever III
Quoted 38 lines · Chuck Lever III
Quoted 35 lines · Chuck Lever III
Quoted 31 lines · Christian Brauner

We've aligned setgid behavior over multiple kernel releases. The details can be found in commit cf619f891971 ("Merge tag 'fs.ovl.setgid.v6.2' of
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping") and commit 426b4ca2d6a5 ("Merge tag 'fs.setgid.v6.0' of
git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux").
Consistent setgid stripping behavior is now encapsulated in the setattr_should_drop_sgid() helper which is used by all filesystems that strip setgid bits outside of vfs proper. Usually ATTR_KILL_SGID is raised in e.g., chown_common() and is subject to the setattr_should_drop_sgid() check to determine whether the setgid bit can be retained. Since nfsd is raising ATTR_KILL_SGID unconditionally it will cause notify_change() to strip it even if the caller had the necessary privileges to retain it. Ensure that nfsd only raises ATR_KILL_SGID if the caller lacks the necessary privileges to retain the setgid bit.

Without this patch the setgid stripping tests in LTP will fail:

Quoted 2 lines

As you can see, the problem is S_ISGID (0002000) was dropped on a non-group-executable file while chown was invoked by super-user, while

[...]

Quoted 1 line

fchown02.c:66: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

[...]

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 1 line

chown02.c:57: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

With this patch all tests pass.

We had a very similar report from kernel-test-robot.

https://lore.kernel.org/all/[email protected]/

Which points to commit: ed5a7047d201 ("attr: use consistent sgid stripping checks")

And the above commit is backported to LTS kernels -- 5.10.y,5.15.y and 6.1.y

So would it be better to tag it to stable by adding "Cc: <[email protected]>" ?

Thanks,
Harshit

Quoted 27 lines
Quoted 11 lines
Quoted 8 lines
Quoted 4 lines

There are some similar fstests failures this fix might address.

I've applied this patch to the nfsd-fixes tree for broader testing.

ERROR: modpost: "setattr_should_drop_sgid" [fs/nfsd/nfsd.ko] undefined!

Did I apply this patch to the wrong kernel?

setattr_should_drop_sgid() is not available to callers built as modules. It needs an EXPORT_SYMBOL or _GPL.

Hey Chuck,

Thanks for taking a look!
The required export is part of
commit 4f704d9a835 ("nfs: use vfs setgid helper")
which is in current mainline as of Monday this week which was part of my v6.4/vfs.misc PR that was merged.

So this should all work fine on mainline. Seems I didn't use --base which is why that info was missing.

Thanks!
Christian

#6 lore raw
Quoted 58 lines · Harshit Mogalapalli

Hi Christian,

Quoted 39 lines · Christian Brauner
Quoted 38 lines · Chuck Lever III
Quoted 38 lines · Chuck Lever III
Quoted 35 lines · Chuck Lever III
Quoted 31 lines · Christian Brauner

We've aligned setgid behavior over multiple kernel releases. The details can be found in commit cf619f891971 ("Merge tag 'fs.ovl.setgid.v6.2' of
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping") and commit 426b4ca2d6a5 ("Merge tag 'fs.setgid.v6.0' of
git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux").
Consistent setgid stripping behavior is now encapsulated in the setattr_should_drop_sgid() helper which is used by all filesystems that strip setgid bits outside of vfs proper. Usually ATTR_KILL_SGID is raised in e.g., chown_common() and is subject to the setattr_should_drop_sgid() check to determine whether the setgid bit can be retained. Since nfsd is raising ATTR_KILL_SGID unconditionally it will cause notify_change() to strip it even if the caller had the necessary privileges to retain it. Ensure that nfsd only raises ATR_KILL_SGID if the caller lacks the necessary privileges to retain the setgid bit.

Without this patch the setgid stripping tests in LTP will fail:

Quoted 2 lines

As you can see, the problem is S_ISGID (0002000) was dropped on a non-group-executable file while chown was invoked by super-user, while

[...]

Quoted 1 line

fchown02.c:66: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

[...]

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 1 line

chown02.c:57: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

With this patch all tests pass.

We had a very similar report from kernel-test-robot.

https://lore.kernel.org/all/[email protected]/

Quoted 30 lines

Which points to commit: ed5a7047d201 ("attr: use consistent sgid stripping checks")

And the above commit is backported to LTS kernels -- 5.10.y,5.15.y and 6.1.y

So would it be better to tag it to stable by adding "Cc: <[email protected]>" ?

Thanks,
Harshit

Quoted 21 lines
Quoted 11 lines
Quoted 8 lines
Quoted 4 lines

There are some similar fstests failures this fix might address.

I've applied this patch to the nfsd-fixes tree for broader testing.

ERROR: modpost: "setattr_should_drop_sgid" [fs/nfsd/nfsd.ko] undefined!

Did I apply this patch to the wrong kernel?

setattr_should_drop_sgid() is not available to callers built as modules. It needs an EXPORT_SYMBOL or _GPL.

Hey Chuck,
Thanks for taking a look!
The required export is part of
commit 4f704d9a835 ("nfs: use vfs setgid helper")
which is in current mainline as of Monday this week which was part of my v6.4/vfs.misc PR that was merged.
So this should all work fine on mainline. Seems I didn't use --base which is why that info was missing.
Thanks!
Christian

I'll apply this to nfsd-next (6.5) and add Jeff's Reviewed-by and a Cc: stable.

--
Chuck Lever

#7 lore raw
Quoted 89 lines · Chuck Lever III
Quoted 58 lines · Harshit Mogalapalli

Hi Christian,

Quoted 39 lines · Christian Brauner
Quoted 38 lines · Chuck Lever III
Quoted 38 lines · Chuck Lever III
Quoted 35 lines · Chuck Lever III
Quoted 31 lines · Christian Brauner

We've aligned setgid behavior over multiple kernel releases. The details can be found in commit cf619f891971 ("Merge tag 'fs.ovl.setgid.v6.2' of
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping") and commit 426b4ca2d6a5 ("Merge tag 'fs.setgid.v6.0' of
git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux").
Consistent setgid stripping behavior is now encapsulated in the setattr_should_drop_sgid() helper which is used by all filesystems that strip setgid bits outside of vfs proper. Usually ATTR_KILL_SGID is raised in e.g., chown_common() and is subject to the setattr_should_drop_sgid() check to determine whether the setgid bit can be retained. Since nfsd is raising ATTR_KILL_SGID unconditionally it will cause notify_change() to strip it even if the caller had the necessary privileges to retain it. Ensure that nfsd only raises ATR_KILL_SGID if the caller lacks the necessary privileges to retain the setgid bit.

Without this patch the setgid stripping tests in LTP will fail:

Quoted 2 lines

As you can see, the problem is S_ISGID (0002000) was dropped on a non-group-executable file while chown was invoked by super-user, while

[...]

Quoted 1 line

fchown02.c:66: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

[...]

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 1 line

chown02.c:57: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

With this patch all tests pass.

We had a very similar report from kernel-test-robot.

https://lore.kernel.org/all/[email protected]/

Quoted 30 lines

Which points to commit: ed5a7047d201 ("attr: use consistent sgid stripping checks")

And the above commit is backported to LTS kernels -- 5.10.y,5.15.y and 6.1.y

So would it be better to tag it to stable by adding "Cc: <[email protected]>" ?

Thanks,
Harshit

Quoted 21 lines
Quoted 11 lines
Quoted 8 lines
Quoted 4 lines

There are some similar fstests failures this fix might address.

I've applied this patch to the nfsd-fixes tree for broader testing.

ERROR: modpost: "setattr_should_drop_sgid" [fs/nfsd/nfsd.ko] undefined!

Did I apply this patch to the wrong kernel?

setattr_should_drop_sgid() is not available to callers built as modules. It needs an EXPORT_SYMBOL or _GPL.

Hey Chuck,
Thanks for taking a look!
The required export is part of
commit 4f704d9a835 ("nfs: use vfs setgid helper")
which is in current mainline as of Monday this week which was part of my v6.4/vfs.misc PR that was merged.
So this should all work fine on mainline. Seems I didn't use --base which is why that info was missing.
Thanks!
Christian

Quoted 2 lines

I'll apply this to nfsd-next (6.5) and add Jeff's Reviewed-by and a Cc: stable.

Thanks all! Appreciate the Cc: stable fixup.

#8 lore raw
In reply to Chuck Lever III
Quoted 93 lines · Chuck Lever III
Quoted 59 lines · Harshit Mogalapalli

Hi Christian,

Quoted 39 lines · Christian Brauner
Quoted 38 lines · Chuck Lever III
Quoted 38 lines · Chuck Lever III
Quoted 35 lines · Chuck Lever III
Quoted 31 lines · Christian Brauner

We've aligned setgid behavior over multiple kernel releases. The details can be found in commit cf619f891971 ("Merge tag 'fs.ovl.setgid.v6.2' of
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping") and commit 426b4ca2d6a5 ("Merge tag 'fs.setgid.v6.0' of
git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux").
Consistent setgid stripping behavior is now encapsulated in the setattr_should_drop_sgid() helper which is used by all filesystems that strip setgid bits outside of vfs proper. Usually ATTR_KILL_SGID is raised in e.g., chown_common() and is subject to the setattr_should_drop_sgid() check to determine whether the setgid bit can be retained. Since nfsd is raising ATTR_KILL_SGID unconditionally it will cause notify_change() to strip it even if the caller had the necessary privileges to retain it. Ensure that nfsd only raises ATR_KILL_SGID if the caller lacks the necessary privileges to retain the setgid bit.

Without this patch the setgid stripping tests in LTP will fail:

Quoted 2 lines

As you can see, the problem is S_ISGID (0002000) was dropped on a non-group-executable file while chown was invoked by super-user, while

[...]

Quoted 1 line

fchown02.c:66: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

[...]

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 1 line

chown02.c:57: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

With this patch all tests pass.

We had a very similar report from kernel-test-robot.

https://lore.kernel.org/all/[email protected]/

Quoted 30 lines

Which points to commit: ed5a7047d201 ("attr: use consistent sgid stripping checks")

And the above commit is backported to LTS kernels -- 5.10.y,5.15.y and 6.1.y

So would it be better to tag it to stable by adding "Cc: <[email protected]>" ?

Thanks,
Harshit

Quoted 21 lines
Quoted 11 lines
Quoted 8 lines
Quoted 4 lines

There are some similar fstests failures this fix might address.

I've applied this patch to the nfsd-fixes tree for broader testing.

ERROR: modpost: "setattr_should_drop_sgid" [fs/nfsd/nfsd.ko] undefined!

Did I apply this patch to the wrong kernel?

setattr_should_drop_sgid() is not available to callers built as modules. It needs an EXPORT_SYMBOL or _GPL.

Hey Chuck,
Thanks for taking a look!
The required export is part of
commit 4f704d9a835 ("nfs: use vfs setgid helper")
which is in current mainline as of Monday this week which was part of my v6.4/vfs.misc PR that was merged.
So this should all work fine on mainline. Seems I didn't use --base which is why that info was missing.
Thanks!
Christian

Quoted 2 lines

I'll apply this to nfsd-next (6.5) and add Jeff's Reviewed-by and a Cc: stable.

Hi Chuck,

Just following up to see if you know when this patch will go to linux mainline and linux-stable?

Thanks,
Sherry

Quoted 6 lines

--
Chuck Lever

#9 lore raw
Quoted 97 lines · Sherry Yang
Quoted 94 lines · Chuck Lever III
Quoted 89 lines · Harshit Mogalapalli

Hi Christian,

Quoted 39 lines · Christian Brauner
Quoted 38 lines · Chuck Lever III
Quoted 38 lines · Chuck Lever III
Quoted 35 lines · Chuck Lever III
Quoted 31 lines · Christian Brauner

We've aligned setgid behavior over multiple kernel releases. The details can be found in commit cf619f891971 ("Merge tag 'fs.ovl.setgid.v6.2' of
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping") and commit 426b4ca2d6a5 ("Merge tag 'fs.setgid.v6.0' of
git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux").
Consistent setgid stripping behavior is now encapsulated in the setattr_should_drop_sgid() helper which is used by all filesystems that strip setgid bits outside of vfs proper. Usually ATTR_KILL_SGID is raised in e.g., chown_common() and is subject to the setattr_should_drop_sgid() check to determine whether the setgid bit can be retained. Since nfsd is raising ATTR_KILL_SGID unconditionally it will cause notify_change() to strip it even if the caller had the necessary privileges to retain it. Ensure that nfsd only raises ATR_KILL_SGID if the caller lacks the necessary privileges to retain the setgid bit.

Without this patch the setgid stripping tests in LTP will fail:

Quoted 2 lines

As you can see, the problem is S_ISGID (0002000) was dropped on a non-group-executable file while chown was invoked by super-user, while

[...]

Quoted 1 line

fchown02.c:66: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

[...]

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 1 line

chown02.c:57: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

With this patch all tests pass.

We had a very similar report from kernel-test-robot.

https://lore.kernel.org/all/[email protected]/

Which points to commit: ed5a7047d201 ("attr: use consistent sgid stripping checks")

And the above commit is backported to LTS kernels -- 5.10.y,5.15.y and 6.1.y

So would it be better to tag it to stable by adding "Cc: <[email protected]>" ?

Thanks,
Harshit

Quoted 21 lines
Quoted 11 lines
Quoted 8 lines
Quoted 4 lines

There are some similar fstests failures this fix might address.

I've applied this patch to the nfsd-fixes tree for broader testing.

ERROR: modpost: "setattr_should_drop_sgid" [fs/nfsd/nfsd.ko] undefined!

Did I apply this patch to the wrong kernel?

setattr_should_drop_sgid() is not available to callers built as modules. It needs an EXPORT_SYMBOL or _GPL.

Hey Chuck,
Thanks for taking a look!
The required export is part of
commit 4f704d9a835 ("nfs: use vfs setgid helper")
which is in current mainline as of Monday this week which was part of my v6.4/vfs.misc PR that was merged.
So this should all work fine on mainline. Seems I didn't use --base which is why that info was missing.
Thanks!
Christian

Quoted 6 lines
Quoted 2 lines

I'll apply this to nfsd-next (6.5) and add Jeff's Reviewed-by and a Cc: stable.

Hi Chuck,

Just following up to see if you know when this patch will go to linux mainline and linux-stable?

Howdy- mentioned above: applied to nfsd-next for the v6.5 merge window, which will open in several weeks.

I added a Cc: stable, so once it is in 6.5-rc it will be automatically backported into the stable kernels.

--
Chuck Lever

maintainer Reviewed-by on v1 1/1
#10 lore raw
In reply to Christian Brauner
Quoted 57 lines · Christian Brauner

We've aligned setgid behavior over multiple kernel releases. The details can be found in commit cf619f891971 ("Merge tag 'fs.ovl.setgid.v6.2' of
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping") and commit 426b4ca2d6a5 ("Merge tag 'fs.setgid.v6.0' of
git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux").
Consistent setgid stripping behavior is now encapsulated in the setattr_should_drop_sgid() helper which is used by all filesystems that strip setgid bits outside of vfs proper. Usually ATTR_KILL_SGID is raised in e.g., chown_common() and is subject to the setattr_should_drop_sgid() check to determine whether the setgid bit can be retained. Since nfsd is raising ATTR_KILL_SGID unconditionally it will cause notify_change() to strip it even if the caller had the necessary privileges to retain it. Ensure that nfsd only raises ATR_KILL_SGID if the caller lacks the necessary privileges to retain the setgid bit.

Without this patch the setgid stripping tests in LTP will fail:

Quoted 2 lines

As you can see, the problem is S_ISGID (0002000) was dropped on a non-group-executable file while chown was invoked by super-user, while

[...]

Quoted 1 line

fchown02.c:66: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

[...]

Quoted 1 line

chown02.c:57: TFAIL: testfile2: wrong mode permissions 0100700, expected 0102700

With this patch all tests pass.

---

ubuntu@imp1-vm:~/ltp-install$ sudo ./runltp -d /mnt -s chown02
INFO: ltp-pan reported all tests PASS
LTP Version: 20230127-112-gf41e8a2fa
ubuntu@imp1-vm:~/ltp-install$ sudo ./runltp -d /mnt -s fchown02
INFO: ltp-pan reported all tests PASS
LTP Version: 20230127-112-gf41e8a2fa
ubuntu@imp1-vm:~/src/git/xfstests$ sudo ./check -g perms
FSTYP         -- nfs
PLATFORM      -- Linux/x86_64 imp1-vm 6.3.0-nfs-setgid-3a3cfe624076 #20 SMP PREEMPT_DYNAMIC Tue May  2 12:35:51 UTC 2023
MKFS_OPTIONS  -- 127.0.0.1:/nfsscratch
MOUNT_OPTIONS -- -o vers=3,noacl 127.0.0.1:/nfsscratch /mnt/scratch
Passed all 41 tests
---
 fs/nfsd/vfs.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
fs/nfsd/vfs.c +0 -0
index bb9d47172162..c4ef24c5ffd0 100644

fs/nfsd/vfs.c nfsd_sanitize_attrs(struct inode *inode, struct iattr *iap) · line 396

Quoted 10 lines
fs/nfsd/vfs.c +3 -1
@@ -388,7 +388,9 @@ nfsd_sanitize_attrs(struct inode *inode, struct iattr *iap)
 				iap->ia_mode &= ~S_ISGID;
 		} else {
 			/* set ATTR_KILL_* bits and let VFS handle it */
-			iap->ia_valid |= (ATTR_KILL_SUID | ATTR_KILL_SGID);
+			iap->ia_valid |= ATTR_KILL_SUID;
+			iap->ia_valid |=
+				setattr_should_drop_sgid(&nop_mnt_idmap, inode);
 		}
 	}
 }

Looks good to me. The thread should have assumed the user's creds by this point.