nfsd: nfsd_do_file_acquire should hold rcu_read_lock while getting refs
Subsystems (2)
Jeff Layton
maintainerPATCH v2 1/1 nfsd: nfsd_do_file_acquire should hold rcu_read_lock while getting refs
nfsd_file is RCU-freed, so it's possible that one could be found that's in the process of being freed and the memory recycled. Ensure we hold the rcu_read_lock while attempting to get a reference on the object.
--- fs/nfsd/filecache.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-)
fs/nfsd/filecache.c +8 -6
@@ -1077,10 +1077,12 @@ nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp, retry:/* Avoid allocation if the item is already in cache */ +rcu_read_lock (); nf =rhashtable_lookup_fast (&nfsd_file_rhash_tbl, &key, nfsd_file_rhash_params);if (nf) nf =nfsd_file_get (nf); +rcu_read_unlock ();if (nf) goto wait_for_construction;
@@ -1090,21 +1092,21 @@ nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp, goto out_status; } +rcu_read_lock (); nf =rhashtable_lookup_get_insert_key (&nfsd_file_rhash_tbl, &key, &new->nf_rhash, nfsd_file_rhash_params); +if (!IS_ERR_OR_NULL (nf)) { + nf =nfsd_file_get (nf); +nfsd_file_slab_free (&new->nf_rcu); + } +rcu_read_unlock ();if (!nf) { nf = new; goto open_file; }if (IS_ERR (nf)) goto insert_err; - nf =nfsd_file_get (nf); -if (nf ==NULL ) { - nf = new; - goto open_file; - } -nfsd_file_slab_free (&new->nf_rcu); wait_for_construction:wait_on_bit (&nf->nf_flags,NFSD_FILE_PENDING ,TASK_UNINTERRUPTIBLE );
Signature · 1 line
2.37.3
Hi Jeff-
Quoted 5 lines · Jeff Layton
nfsd_fileis RCU-freed, so it's possible that one could be found that's in the process of being freed and the memory recycled. Ensure we hold thercu_read_lockwhile attempting to get a reference on the object.
I'm OK with entertaining clean-up patches in this code, but I am skeptical that this patch addresses the concern enumerated in bug #394. As you've pointed out to me before, the "UAF on DELEGRETURN crashes" appeared well before v5.19, which is the kernel release where this bit of code was introduced.
Quoted 19 lines
Cc NeilBrown <[email protected]>Signed-off-by Jeff Layton <[email protected]>---
fs/nfsd/filecache.c| 14 ++++++++------1 file changed, 8 insertions(+), 6 deletions(-)fs/nfsd/filecache.c +0
-0@@ -1077,10 +1077,12 @@ nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp,retry:
/* Avoid allocation if the item is already in cache */fs/nfsd/filecache.c +1
-0+rcu_read_lock ();nf = rhashtable_lookup_fast(&nfsd_file_rhash_tbl, &key, nfsd_file_rhash_params); if (nf) nf = nfsd_file_get(nf);
fs/nfsd/filecache.c
Quoted 1 line
fs/nfsd/filecache.c +1
-0+rcu_read_unlock ();
Again:
657 static inline void *rhashtable_lookup_fast(
658 struct rhashtable *ht, const void *key,
659 const struct rhashtable_params params)
660 {
661 void *obj;
662
663 rcu_read_lock(); 664 obj = rhashtable_lookup(ht, key, params); 665 rcu_read_unlock();
666
667 return obj;
668 }
Since rhashtable_lookup() itself is a public API, please just call that in nfsd_file_do_acquire() after explicitly taking the RCU read lock.
fs/nfsd/filecache.c nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp, · line 1100
Quoted 13 lines
if (nf) goto wait_for_construction;fs/nfsd/filecache.c +3
-0@@ -1090,21 +1092,21 @@ nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp, goto out_status; } +rcu_read_lock (); nf =rhashtable_lookup_get_insert_key (&nfsd_file_rhash_tbl, &key, &new->nf_rhash, nfsd_file_rhash_params); +if (!IS_ERR_OR_NULL (nf)) { + nf =nfsd_file_get (nf);
Note that nfsd_file_get() can still return NULL.
fs/nfsd/filecache.c nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp, · line 1101
Quoted 1 line
fs/nfsd/filecache.c +1
-0+nfsd_file_slab_free (&new->nf_rcu);
Why is the slab_free call now inside the RCU critical section? Granted this should be a rare case, but this adds unnecessary latency while the read lock is held.
fs/nfsd/filecache.c nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp, · line 1103
Quoted 2 lines
fs/nfsd/filecache.c +2
-0+ } +rcu_read_unlock ();
Is there a problem replacing rhashtable_lookup_get_insert_key() with rhashtable_lookup_insert_key() and just retrying the normal lookup if insertion returns EEXIST? That way, an nfsd_file_get() is necessary only at the rhashtable_lookup() call site above.
fs/nfsd/filecache.c nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp, · line 1105
Quoted 2 lines
fs/nfsd/filecache.c +0
-0if (!nf) { nf = new;
@new was just released above, so won't this set @nf to point to freed memory in some cases?
Quoted 12 lines
fs/nfsd/filecache.c +0
-6goto open_file; }if (IS_ERR (nf)) goto insert_err; - nf =nfsd_file_get (nf); -if (nf ==NULL ) { - nf = new; - goto open_file; - } -nfsd_file_slab_free (&new->nf_rcu);
wait_for_construction:
Quoted 3 lines
wait_on_bit(&nf->nf_flags, NFSD_FILE_PENDING, TASK_UNINTERRUPTIBLE);Signature · 1 line
2.37.3
--
Chuck Lever
Jeff Layton
maintainerQuoted 8 lines · Chuck Lever III
Hi Jeff-
Quoted 5 lines · Jeff Layton
nfsd_fileis RCU-freed, so it's possible that one could be found that's in the process of being freed and the memory recycled. Ensure we hold thercu_read_lockwhile attempting to get a reference on the object.
Quoted 6 lines
I'm OK with entertaining clean-up patches in this code, but I am skeptical that this patch addresses the concern enumerated in bug #394. As you've pointed out to me before, the "UAF on DELEGRETURN crashes" appeared well before v5.19, which is the kernel release where this bit of code was introduced.
Yeah, there may be more than one bug here. In any case, these patches should close potential races, so I think we ought to take them.
Quoted 33 lines
Quoted 20 lines
Cc NeilBrown <[email protected]>Signed-off-by Jeff Layton <[email protected]>---
fs/nfsd/filecache.c| 14 ++++++++------1 file changed, 8 insertions(+), 6 deletions(-)fs/nfsd/filecache.c +0
-0@@ -1077,10 +1077,12 @@ nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp,retry:
/* Avoid allocation if the item is already in cache */fs/nfsd/filecache.c +1
-0+rcu_read_lock ();nf = rhashtable_lookup_fast(&nfsd_file_rhash_tbl, &key, nfsd_file_rhash_params); if (nf) nf = nfsd_file_get(nf);fs/nfsd/filecache.c
Quoted 1 line
fs/nfsd/filecache.c +1
-0+rcu_read_unlock ();Again:
657 static inline void *rhashtable_lookup_fast(
658 struct rhashtable *ht, const void *key,
659 const structrhashtable_paramsparams)
660 {
661 void *obj;
662663 rcu_read_lock(); 664 obj = rhashtable_lookup(ht, key, params); 665 rcu_read_unlock();
Quoted 9 lines
666
667 return obj;
668 }Since
rhashtable_lookup()itself is a public API, please just call that innfsd_file_do_acquire()after explicitly taking the RCU read lock.
Understood. Sorry I missed your point. I'll fix that up.
Quoted 7 lines
Quoted 7 lines
if (nf) goto wait_for_construction;fs/nfsd/filecache.c +0
-0@@ -1090,21 +1092,21 @@ nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp, goto out_status; }
Quoted 9 lines
fs/nfsd/filecache.c nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp, · line 1100
Quoted 6 lines
fs/nfsd/filecache.c +3
-0+rcu_read_lock (); nf =rhashtable_lookup_get_insert_key (&nfsd_file_rhash_tbl, &key, &new->nf_rhash, nfsd_file_rhash_params); +if (!IS_ERR_OR_NULL (nf)) { + nf =nfsd_file_get (nf);Note that
nfsd_file_get()can still return NULL.
True, and that would leak. Good catch.
Quoted 7 lines
fs/nfsd/filecache.c nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp, · line 1101
Quoted 2 lines
fs/nfsd/filecache.c +1
-0+nfsd_file_slab_free (&new->nf_rcu);Why is the
slab_freecall now inside the RCU critical section? Granted this should be a rare case, but this adds unnecessary latency while the read lock is held.
Fair point.
Quoted 10 lines
fs/nfsd/filecache.c nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp, · line 1103
Quoted 3 lines
fs/nfsd/filecache.c +2
-0+ } +rcu_read_unlock ();Is there a problem replacing
rhashtable_lookup_get_insert_key()withrhashtable_lookup_insert_key()and just retrying the normal lookup if insertion returns EEXIST? That way, annfsd_file_get()is necessary only at therhashtable_lookup()call site above.
I like this idea, and it allows for a rather nice cleanup of the code. I'll send a v3 set after I've had a chance to do some testing.
Quoted 29 lines
fs/nfsd/filecache.c nfsd_file_do_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp, · line 1105
Quoted 2 lines
fs/nfsd/filecache.c +0
-0if (!nf) { nf = new;@new was just released above, so won't this set @nf to point to freed memory in some cases?
Quoted 12 lines
fs/nfsd/filecache.c +0
-6goto open_file; }if (IS_ERR (nf)) goto insert_err; - nf =nfsd_file_get (nf); -if (nf ==NULL ) { - nf = new; - goto open_file; - } -nfsd_file_slab_free (&new->nf_rcu);
wait_for_construction:Quoted 3 lines
wait_on_bit(&nf->nf_flags, NFSD_FILE_PENDING, TASK_UNINTERRUPTIBLE);Signature · 1 line
2.37.3--
Chuck Lever
Signature · 1 line
Jeff Layton <[email protected]>
Quoted 20 lines · Jeff Layton
Quoted 8 lines · Chuck Lever III
Hi Jeff-
Quoted 5 lines · Jeff Layton
nfsd_fileis RCU-freed, so it's possible that one could be found that's in the process of being freed and the memory recycled. Ensure we hold thercu_read_lockwhile attempting to get a reference on the object.Quoted 6 lines
I'm OK with entertaining clean-up patches in this code, but I am skeptical that this patch addresses the concern enumerated in bug #394. As you've pointed out to me before, the "UAF on DELEGRETURN crashes" appeared well before v5.19, which is the kernel release where this bit of code was introduced.
Quoted 2 lines
Yeah, there may be more than one bug here. In any case, these patches should close potential races, so I think we ought to take them.
Agreed, all of these are valid and desirable improvements.
I've applied the two from yesterday to my internal tree for more testing. I plan to apply this one as well once the wrinkles are ironed out. Since these are a bit late in the cycle, I plan to push the set to Linus after the initial nfsd-6.1 PR, either near the end of the merge window or in -rc1.
--
Chuck Lever