Skip to contentSkip to conversation

fs: call security_d_instantiate in d_obtain_alias V2

PATCH In review Josef Bacik 4 participants 7 messages Nov 19, 2010 – Feb 14, 2011

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

PATCH 1/1 fs: call security_d_instantiate in d_obtain_alias V2

While trying to track down some NFS problems with BTRFS, I kept noticing I was getting -EACCESS for no apparent reason. Eric Paris and printk() helped me figure out that it was SELinux that was giving me grief, with the following denial

type=AVC msg=audit(1290013638.413:95): avc: denied { 0x800000 } for pid=1772 comm="nfsd" name="" dev=sda1 ino=256 scontext=system_u:system_r:kernel_t:s0 tcontext=system_u:object_r:unlabeled_t:s0 tclass=file

Turns out this is because in d_obtain_alias if we can't find an alias we create one and do all the normal instantiation stuff, but we don't do the security_d_instantiate.

Usually we are protected from getting a hashed dentry that hasn't yet run security_d_instantiate() by the parent's i_mutex, but obviously this isn't an option there, so in order to deal with the case that a second thread comes in and finds our new dentry before we get to run security_d_instantiate(), we go ahead and call it if we find a dentry already. Eric assures me that this is ok as the code checks to see if the dentry has been initialized already so calling security_d_instantiate() against the same dentry multiple times is ok. With this patch I'm no longer getting errant -EACCESS values.

---
V1->V2:
-added second security_d_instantiate() call

 fs/dcache.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)
fs/dcache.c +3 -0
index 23702a9..119d489 100644
@@ -1201,9 +1201,12 @@ struct dentry *d_obtain_alias(struct inode *inode)
 	spin_unlock(&tmp->d_lock);
 
 	spin_unlock(&dcache_lock);
+	security_d_instantiate(tmp, inode);
 	return tmp;
 
  out_iput:
+	if (res && !IS_ERR(res))
+		security_d_instantiate(res, inode);
 	iput(inode);
 	return res;
 }
Signature · 1 line
1.6.6.1
#1 lore raw
Quoted 13 lines · Josef Bacik

While trying to track down some NFS problems with BTRFS, I kept noticing I was getting -EACCESS for no apparent reason. Eric Paris and printk() helped me figure out that it was SELinux that was giving me grief, with the following denial

type=AVC msg=audit(1290013638.413:95): avc: denied { 0x800000 } for pid=1772 comm="nfsd" name="" dev=sda1 ino=256 scontext=system_u:system_r:kernel_t:s0 tcontext=system_u:object_r:unlabeled_t:s0 tclass=file

Turns out this is because in d_obtain_alias if we can't find an alias we create one and do all the normal instantiation stuff, but we don't do the security_d_instantiate.

Quoted 8 lines

Usually we are protected from getting a hashed dentry that hasn't yet run security_d_instantiate() by the parent's i_mutex, but obviously this isn't an option there, so in order to deal with the case that a second thread comes in and finds our new dentry before we get to run security_d_instantiate(), we go ahead and call it if we find a dentry already. Eric assures me that this is ok as the code checks to see if the dentry has been initialized already so calling security_d_instantiate() against the same dentry multiple times is ok. With this patch I'm no longer getting errant -EACCESS values.

Thanks, I can't see any reason that wouldn't work.

--b.

Quoted 29 lines

---
V1->V2:

(quoted diff) +0 -1
-added second security_d_instantiate() call
 
 fs/dcache.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)
fs/dcache.c +3 -0
index 23702a9..119d489 100644
@@ -1201,9 +1201,12 @@ struct dentry *d_obtain_alias(struct inode *inode)
 	spin_unlock(&tmp->d_lock);
 
 	spin_unlock(&dcache_lock);
+	security_d_instantiate(tmp, inode);
 	return tmp;
 
  out_iput:
+	if (res && !IS_ERR(res))
+		security_d_instantiate(res, inode);
 	iput(inode);
 	return res;
 }
Signature · 1 line
1.6.6.1
#2 lore raw
Quoted 23 lines · J. Bruce Fields
Quoted 13 lines · Josef Bacik

While trying to track down some NFS problems with BTRFS, I kept noticing I was getting -EACCESS for no apparent reason. Eric Paris and printk() helped me figure out that it was SELinux that was giving me grief, with the following denial

type=AVC msg=audit(1290013638.413:95): avc: denied { 0x800000 } for pid=1772 comm="nfsd" name="" dev=sda1 ino=256 scontext=system_u:system_r:kernel_t:s0 tcontext=system_u:object_r:unlabeled_t:s0 tclass=file

Turns out this is because in d_obtain_alias if we can't find an alias we create one and do all the normal instantiation stuff, but we don't do the security_d_instantiate.

Quoted 8 lines

Usually we are protected from getting a hashed dentry that hasn't yet run security_d_instantiate() by the parent's i_mutex, but obviously this isn't an option there, so in order to deal with the case that a second thread comes in and finds our new dentry before we get to run security_d_instantiate(), we go ahead and call it if we find a dentry already. Eric assures me that this is ok as the code checks to see if the dentry has been initialized already so calling security_d_instantiate() against the same dentry multiple times is ok. With this patch I'm no longer getting errant -EACCESS values.

Quoted 1 line

Thanks, I can't see any reason that wouldn't work.

(FWIW, I also ran my usual nfs regression tests with this applied. They don't exercise the problem you were seeing, but maybe it's at least some sort of sanity check.)

--b.

Quoted 36 lines

--b.

Quoted 29 lines

---
V1->V2:

(quoted diff) +0 -1
-added second security_d_instantiate() call
 
 fs/dcache.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)
fs/dcache.c +3 -0
index 23702a9..119d489 100644
@@ -1201,9 +1201,12 @@ struct dentry *d_obtain_alias(struct inode *inode)
 	spin_unlock(&tmp->d_lock);
 
 	spin_unlock(&dcache_lock);
+	security_d_instantiate(tmp, inode);
 	return tmp;
 
  out_iput:
+	if (res && !IS_ERR(res))
+		security_d_instantiate(res, inode);
 	iput(inode);
 	return res;
 }
Signature · 1 line
1.6.6.1

--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#3 lore raw
Quoted 24 lines · J. Bruce Fields
Quoted 23 lines · J. Bruce Fields
Quoted 21 lines · Josef Bacik

While trying to track down some NFS problems with BTRFS, I kept noticing I was getting -EACCESS for no apparent reason. Eric Paris and printk() helped me figure out that it was SELinux that was giving me grief, with the following denial

type=AVC msg=audit(1290013638.413:95): avc: denied { 0x800000 } for pid=1772 comm="nfsd" name="" dev=sda1 ino=256 scontext=system_u:system_r:kernel_t:s0 tcontext=system_u:object_r:unlabeled_t:s0 tclass=file

Turns out this is because in d_obtain_alias if we can't find an alias we create one and do all the normal instantiation stuff, but we don't do the security_d_instantiate.

Usually we are protected from getting a hashed dentry that hasn't yet run security_d_instantiate() by the parent's i_mutex, but obviously this isn't an option there, so in order to deal with the case that a second thread comes in and finds our new dentry before we get to run security_d_instantiate(), we go ahead and call it if we find a dentry already. Eric assures me that this is ok as the code checks to see if the dentry has been initialized already so calling security_d_instantiate() against the same dentry multiple times is ok. With this patch I'm no longer getting errant -EACCESS values.

Quoted 6 lines
Quoted 1 line

Thanks, I can't see any reason that wouldn't work.

(FWIW, I also ran my usual nfs regression tests with this applied. They don't exercise the problem you were seeing, but maybe it's at least some sort of sanity check.)

Oh I should have mentioned the testing I gave it. With BTRFS I could regularly reproduce this by doing the following

SERVER
mount /dev/sda1 /mnt/btrfs
btrfs subvol create /mnt/btrfs/foo
cp mutt.tar.gz /mnt/btrfs/foo

CLIENT
mount server:/mnt/btrfs /mnt/test
cd /mnt/test/foo
tar xzvf mutt.tar.gz
cd mutt
ls

SERVER
echo 3 > /proc/sys/vm/drop_caches

CLIENT
ls

and bam I'd get -EACCESS every time. I did this a bunch of times to make sure it didn't happen anymore plus variations of the above. Thanks,

Josef

#4 lore raw
In reply to Josef Bacik
Quoted 50 lines · Josef Bacik

While trying to track down some NFS problems with BTRFS, I kept noticing I was getting -EACCESS for no apparent reason. Eric Paris and printk() helped me figure out that it was SELinux that was giving me grief, with the following denial

type=AVC msg=audit(1290013638.413:95): avc: denied { 0x800000 } for pid=1772 comm="nfsd" name="" dev=sda1 ino=256 scontext=system_u:system_r:kernel_t:s0 tcontext=system_u:object_r:unlabeled_t:s0 tclass=file

Turns out this is because in d_obtain_alias if we can't find an alias we create one and do all the normal instantiation stuff, but we don't do the security_d_instantiate.

Usually we are protected from getting a hashed dentry that hasn't yet run security_d_instantiate() by the parent's i_mutex, but obviously this isn't an option there, so in order to deal with the case that a second thread comes in and finds our new dentry before we get to run security_d_instantiate(), we go ahead and call it if we find a dentry already. Eric assures me that this is ok as the code checks to see if the dentry has been initialized already so calling security_d_instantiate() against the same dentry multiple times is ok. With this patch I'm no longer getting errant -EACCESS values.

---
V1->V2:

(quoted diff) +0 -1
-added second security_d_instantiate() call
 
 fs/dcache.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)
fs/dcache.c +3 -0
index 23702a9..119d489 100644
@@ -1201,9 +1201,12 @@ struct dentry *d_obtain_alias(struct inode *inode)
 	spin_unlock(&tmp->d_lock);
 
 	spin_unlock(&dcache_lock);
+	security_d_instantiate(tmp, inode);
 	return tmp;
 
  out_iput:
+	if (res && !IS_ERR(res))
+		security_d_instantiate(res, inode);
 	iput(inode);
 	return res;
 }
Signature · 1 line
1.6.6.1

Hey Al,

I forgot to cc you directly, is this ok with you, and if it is would you mind picking it up? Thanks,

Josef

#5 lore raw
In reply to Josef Bacik
Quoted 22 lines · Josef Bacik

While trying to track down some NFS problems with BTRFS, I kept noticing I was getting -EACCESS for no apparent reason. Eric Paris and printk() helped me figure out that it was SELinux that was giving me grief, with the following denial

type=AVC msg=audit(1290013638.413:95): avc: denied { 0x800000 } for pid=1772 comm="nfsd" name="" dev=sda1 ino=256 scontext=system_u:system_r:kernel_t:s0 tcontext=system_u:object_r:unlabeled_t:s0 tclass=file

Turns out this is because in d_obtain_alias if we can't find an alias we create one and do all the normal instantiation stuff, but we don't do the security_d_instantiate.

Usually we are protected from getting a hashed dentry that hasn't yet run security_d_instantiate() by the parent's i_mutex, but obviously this isn't an option there, so in order to deal with the case that a second thread comes in and finds our new dentry before we get to run security_d_instantiate(), we go ahead and call it if we find a dentry already. Eric assures me that this is ok as the code checks to see if the dentry has been initialized already so calling security_d_instantiate() against the same dentry multiple times is ok. With this patch I'm no longer getting errant -EACCESS values.

Quoted 1 line

Thanks, this fixes xfsdump too, see also:

https://bugzilla.redhat.com/show_bug.cgi?id=662344
Bug 662344 - broken SELinux AVCs on XFS partition when running xfsdump

(xfsdump uses the open by handle library from xfsprogs)

-Eric

Quoted 24 lines

---
V1->V2:

(quoted diff) +0 -1
-added second security_d_instantiate() call
 
 fs/dcache.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)
fs/dcache.c +3 -0
index 23702a9..119d489 100644
@@ -1201,9 +1201,12 @@ struct dentry *d_obtain_alias(struct inode *inode)
 	spin_unlock(&tmp->d_lock);
 
 	spin_unlock(&dcache_lock);
+	security_d_instantiate(tmp, inode);
 	return tmp;
 
  out_iput:
+	if (res && !IS_ERR(res))
+		security_d_instantiate(res, inode);
 	iput(inode);
 	return res;
 }
Acked-by on v1 1/1
#6 lore raw
In reply to Josef Bacik
Quoted 22 lines · Josef Bacik

While trying to track down some NFS problems with BTRFS, I kept noticing I was getting -EACCESS for no apparent reason. Eric Paris and printk() helped me figure out that it was SELinux that was giving me grief, with the following denial

type=AVC msg=audit(1290013638.413:95): avc: denied { 0x800000 } for pid=1772 comm="nfsd" name="" dev=sda1 ino=256 scontext=system_u:system_r:kernel_t:s0 tcontext=system_u:object_r:unlabeled_t:s0 tclass=file

Turns out this is because in d_obtain_alias if we can't find an alias we create one and do all the normal instantiation stuff, but we don't do the security_d_instantiate.

Usually we are protected from getting a hashed dentry that hasn't yet run security_d_instantiate() by the parent's i_mutex, but obviously this isn't an option there, so in order to deal with the case that a second thread comes in and finds our new dentry before we get to run security_d_instantiate(), we go ahead and call it if we find a dentry already. Eric assures me that this is ok as the code checks to see if the dentry has been initialized already so calling security_d_instantiate() against the same dentry multiple times is ok. With this patch I'm no longer getting errant -EACCESS values.

Quoted 1 line

Al, any chance we can push this along?

-Eric

Quoted 24 lines

---
V1->V2:

(quoted diff) +0 -1
-added second security_d_instantiate() call
 
 fs/dcache.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)
fs/dcache.c +3 -0
index 23702a9..119d489 100644
@@ -1201,9 +1201,12 @@ struct dentry *d_obtain_alias(struct inode *inode)
 	spin_unlock(&tmp->d_lock);
 
 	spin_unlock(&dcache_lock);
+	security_d_instantiate(tmp, inode);
 	return tmp;
 
  out_iput:
+	if (res && !IS_ERR(res))
+		security_d_instantiate(res, inode);
 	iput(inode);
 	return res;
 }