Skip to contentSkip to conversation

9pfs: fix invalid union V9fsFidOpenState access

PATCH In review Christian Schoenebeck 2 participants 11 messages Jun 16 – Jun 26

lore
Share & tools
lore
#0 lore raw

cover 9pfs: fix invalid union V9fsFidOpenState access

The individual FID types (P9_FID_NONE, P9_FID_FILE, P9_FID_DIR, P9_FID_XATTR) share union V9fsFidOpenState with FID-type specific fields. Accessing any of the union fields must comply with the FID-type to avoid undefined behaviour or information disclosure.

This series fixes invalid access of this union type at several locations.

* Patch 1 and Patch 2 are the core fixes checking the FID type on protocol level (9p.c) before allowing access to a FID-type specific union field.

* Patch 3 adds another safety layer by returning -1 from local_fid_fd() if the FID type would not have a valid file descriptor.

Christian Schoenebeck (3):
hw/9pfs: fix invalid union access by v9fs_co_fsync() hw/9pfs: fix invalid union access by v9fs_co_fstat() hw/9pfs/local: harden local_fid_fd() on FID types

 hw/9pfs/9p-local.c |  5 ++++-
 hw/9pfs/9p.c       | 17 +++++++++++++++++
 2 files changed, 21 insertions(+), 1 deletion(-)
Signature · 1 line
2.47.3
#3 lore raw

PATCH 3/3 hw/9pfs/local: harden local_fid_fd() on FID types

local_fid_fd() returns fs->fd for any FID type that is not P9_FID_DIR.

Since P9_FID_XATTR and P9_FID_NONE share union V9fsFidOpenState, calling local_fid_fd() on these types misinterprets xattr state as a file descriptor, potentially leading to undefined behaviour or information disclosure.

Even though we are catching these FID type mismatches on protocol level in 9p.c already, previous patches proofed this to be error prone.

So let's add another safety layer in local_fid_fd() that would return -1 if the FID type would not possess a valid file descriptor, to prevent wrong file descriptors from reaching fs backend calls.

---
 hw/9pfs/9p-local.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
hw/9pfs/9p-local.c +4 -1
index aa48306b0e..724f57dc3d 100644
@@ -775,8 +775,11 @@ static int local_fid_fd(int fid_type, V9fsFidOpenState *fs)
 {
     if (fid_type == P9_FID_DIR) {
         return dirfd(fs->dir.stream);
-    } else {
+    } else if (fid_type == P9_FID_FILE) {
         return fs->fd;
+    } else {
+        errno = EBADF;
+        return -1;
     }
 }
 
Signature · 1 line
2.47.3
#2 lore raw

PATCH 2/3 hw/9pfs: fix invalid union access by v9fs_co_fstat()

In reply to Christian Schoenebeck

The individual FID types (P9_FID_NONE, P9_FID_FILE, P9_FID_DIR, P9_FID_XATTR) share union V9fsFidOpenState with FID-type specific fields. Accessing any of the union fields must comply with the FID-type to avoid undefined behaviour or information disclosure.

Fix this in v9fs_lock() and v9fs_getlock() by checking if FID has a valid file descriptor before calling v9fs_co_fstat().

Fixes 10b468bdc533 ("virtio-9p: Implement TXATTRCREATE")
---
 hw/9pfs/9p.c | 8 ++++++++
 1 file changed, 8 insertions(+)
hw/9pfs/9p.c +8 -0
index 6dd6a57e7a..c83df57698 100644
@@ -3852,6 +3852,10 @@ static void coroutine_fn v9fs_lock(void *opaque)
         err = -ENOENT;
         goto out_nofid;
     }
+    if (!fid_has_valid_file_handle(pdu->s, fidp)) {
+        err = -EBADF;
+        goto out;
+    }
     err = v9fs_co_fstat(pdu, fidp, &stbuf);
     if (err < 0) {
         goto out;
@@ -3897,6 +3901,10 @@ static void coroutine_fn v9fs_getlock(void *opaque)
         err = -ENOENT;
         goto out_nofid;
     }
+    if (!fid_has_valid_file_handle(pdu->s, fidp)) {
+        err = -EBADF;
+        goto out;
+    }
     err = v9fs_co_fstat(pdu, fidp, &stbuf);
     if (err < 0) {
         goto out;
Signature · 1 line
2.47.3
#1 lore raw

PATCH 1/3 hw/9pfs: fix invalid union access by v9fs_co_fsync()

In reply to Christian Schoenebeck

The individual FID types (P9_FID_NONE, P9_FID_FILE, P9_FID_DIR, P9_FID_XATTR) share union V9fsFidOpenState with FID-type specific fields. Accessing any of the union fields must comply with the FID-type to avoid undefined behaviour or information disclosure.

Fix this in v9fs_fsync() and v9fs_wstat() by checking if FID has a valid file descriptor before calling v9fs_co_fsync().

Fixes 10b468bdc533 ("virtio-9p: Implement TXATTRCREATE")
---
 hw/9pfs/9p.c | 9 +++++++++
 1 file changed, 9 insertions(+)
hw/9pfs/9p.c +9 -0
index b4314d2549..6dd6a57e7a 100644
@@ -2247,10 +2247,15 @@ static void coroutine_fn v9fs_fsync(void *opaque)
         err = -ENOENT;
         goto out_nofid;
     }
+    if (!fid_has_valid_file_handle(pdu->s, fidp)) {
+        err = -EBADF;
+        goto out;
+    }
     err = v9fs_co_fsync(pdu, fidp, datasync);
     if (!err) {
         err = offset;
     }
+out:
     put_fid(pdu, fidp);
 out_nofid:
     pdu_complete(pdu, err);
@@ -3584,6 +3589,10 @@ static void coroutine_fn v9fs_wstat(void *opaque)
     }
     /* do we need to sync the file? */
     if (donttouch_stat(&v9stat)) {
+        if (!fid_has_valid_file_handle(s, fidp)) {
+            err = -EBADF;
+            goto out;
+        }
         err = v9fs_co_fsync(pdu, fidp, 0);
         goto out;
     }
Signature · 1 line
2.47.3
#4 lore raw
In reply to Christian Schoenebeck
Quoted 4 lines · Christian Schoenebeck

The individual FID types (P9_FID_NONE, P9_FID_FILE, P9_FID_DIR, P9_FID_XATTR) share union V9fsFidOpenState with FID-type specific fields. Accessing any of the union fields must comply with the FID-type to avoid undefined behaviour or information disclosure.

..

Hi!

Next batch of qemu stable series is scheduled for Jun-25 (patch freeze at Jun-23). It would be great if this and other 9pfs fixes landed in the master branch before the freeze. As far as I can see, we've several issues in there.

Thanks,

/mjt

#5 lore raw
Quoted 6 lines · On Wednesday, 17 June 2026 09:27:31 CEST Michael Tokarev wrote:
Quoted 4 lines · Christian Schoenebeck

The individual FID types (P9_FID_NONE, P9_FID_FILE, P9_FID_DIR, P9_FID_XATTR) share union V9fsFidOpenState with FID-type specific fields. Accessing any of the union fields must comply with the FID-type to avoid undefined behaviour or information disclosure.

Quoted 8 lines

..

Hi!

Next batch of qemu stable series is scheduled for Jun-25 (patch freeze at Jun-23). It would be great if this and other 9pfs fixes landed in the master branch before the freeze. As far as I can see, we've several issues in there.

Oh, that's quite close! What would be the next window?

This fix is already queued on my end:
https://lore.kernel.org/qemu-devel/[email protected]/
https://github.com/cschoenebeck/qemu/commits/9p.next

And these fixes are yet to be queued:
https://lore.kernel.org/qemu-devel/[email protected]/
https://lore.kernel.org/qemu-devel/[email protected]/

And of course this series here:
https://lore.kernel.org/qemu-devel/[email protected]/

At least I currently don't have further reports to work on. So all bugfix patches are posted.

/Christian

#6 lore raw

On 17.06.2026 11:10, Christian Schoenebeck wrote:
...

Quoted 6 lines
Quoted 4 lines

Next batch of qemu stable series is scheduled for Jun-25 (patch freeze at Jun-23). It would be great if this and other 9pfs fixes landed in the master branch before the freeze. As far as I can see, we've several issues in there.

Oh, that's quite close! What would be the next window?

I'm trying to make stable releases every month, but it rarely happens this way :) Previous delay between releases was more than 2 months.

Quoted 8 lines

This fix is already queued on my end:
https://lore.kernel.org/qemu-devel/[email protected]/
https://github.com/cschoenebeck/qemu/commits/9p.next

And these fixes are yet to be queued:
https://lore.kernel.org/qemu-devel/[email protected]/
https://lore.kernel.org/qemu-devel/[email protected]/

Quoted 5 lines

And of course this series here:
https://lore.kernel.org/qemu-devel/[email protected]/

At least I currently don't have further reports to work on. So all bugfix patches are posted.

Yeah, that's excellent. Thank you for this stuff!
The next step is for all this to land in the master branch in the main qemu repository - where I can pick it up for the stable branches :)

Thanks,

/mjt

#7 lore raw
Quoted 1 line · On Wednesday, 17 June 2026 11:35:04 CEST Michael Tokarev wrote:

On 17.06.2026 11:10, Christian Schoenebeck wrote:

[...]

Quoted 6 lines
Quoted 2 lines

At least I currently don't have further reports to work on. So all bugfix patches are posted.

Yeah, that's excellent. Thank you for this stuff!
The next step is for all this to land in the master branch in the main qemu repository - where I can pick it up for the stable branches :)

I know, I just wanted to point out that I am usually not inclined to send out a PR with patches that only have been posted few days before, especially as they are going to pushed through several stables branches. You know, regressions, etc. ;-)

What about delaying the stable release for a week? Does that sound like a viable compromise?

/Christian

#8 lore raw

On 17.06.2026 12:59, Christian Schoenebeck wrote:
...

Quoted 4 lines

I know, I just wanted to point out that I am usually not inclined to send out a PR with patches that only have been posted few days before, especially as they are going to pushed through several stables branches. You know, regressions, etc. ;-)

Ah, it's the opposite concern, - to introduce a regression, not to miss the fix in time. It's definitely a valid concern.

Quoted 2 lines

What about delaying the stable release for a week? Does that sound like a viable compromise?

Let's make it into the next stable releases instead. You can include the set which is already published in your tree, and for the rest, let's wait for the next round. I don't plan to delay next stable releases.

Thanks,

/mjt

#9 lore raw
Quoted 1 line · On Wednesday, 17 June 2026 12:43:09 CEST Michael Tokarev wrote:

On 17.06.2026 12:59, Christian Schoenebeck wrote:

[...]

Quoted 6 lines
Quoted 2 lines

What about delaying the stable release for a week? Does that sound like a viable compromise?

Let's make it into the next stable releases instead. You can include the set which is already published in your tree, and for the rest, let's wait for the next round. I don't plan to delay next stable releases.

Agreed.

The already queued 3 patches will certainly make it to master in-time for the stable release next week, but all other patches will miss that release window.

Thanks!

/Christian

#10 lore raw
In reply to Christian Schoenebeck
Quoted 13 lines · On Tuesday, 16 June 2026 17:00:11 CEST Christian Schoenebeck wrote:

The individual FID types (P9_FID_NONE, P9_FID_FILE, P9_FID_DIR, P9_FID_XATTR) share union V9fsFidOpenState with FID-type specific fields. Accessing any of the union fields must comply with the FID-type to avoid undefined behaviour or information disclosure.

This series fixes invalid access of this union type at several locations.

* Patch 1 and Patch 2 are the core fixes checking the FID type on protocol level (9p.c) before allowing access to a FID-type specific union field.

* Patch 3 adds another safety layer by returning -1 from local_fid_fd() if the FID type would not have a valid file descriptor.

Quoted 8 lines

Christian Schoenebeck (3):
hw/9pfs: fix invalid union access by v9fs_co_fsync() hw/9pfs: fix invalid union access by v9fs_co_fstat() hw/9pfs/local: harden local_fid_fd() on FID types

 hw/9pfs/9p-local.c |  5 ++++-
 hw/9pfs/9p.c       | 17 +++++++++++++++++
 2 files changed, 21 insertions(+), 1 deletion(-)

Queued on 9p.next:
https://github.com/cschoenebeck/qemu/commits/9p.next

Thanks!

/Christian