kernfs: Fix UAF in polling when open file is released

commit 3c9ba27 upstream.

JIRA: https://issues.redhat.com/browse/RHEL-122088
Conflicts: Dropped llseek bits, as commit 0fedefd (kernfs: sysfs:
	support custom llseek method for sysfs entries) is not part of RHEL-9
CVE: CVE-2025-39881

A use-after-free (UAF) vulnerability was identified in the PSI (Pressure
Stall Information) monitoring mechanism:

BUG: KASAN: slab-use-after-free in psi_trigger_poll+0x3c/0x140
Read of size 8 at addr ffff3de3d50bd308 by task systemd/1

psi_trigger_poll+0x3c/0x140
cgroup_pressure_poll+0x70/0xa0
cgroup_file_poll+0x8c/0x100
kernfs_fop_poll+0x11c/0x1c0
ep_item_poll.isra.0+0x188/0x2c0

Allocated by task 1:
cgroup_file_open+0x88/0x388
kernfs_fop_open+0x73c/0xaf0
do_dentry_open+0x5fc/0x1200
vfs_open+0xa0/0x3f0
do_open+0x7e8/0xd08
path_openat+0x2fc/0x6b0
do_filp_open+0x174/0x368

Freed by task 8462:
cgroup_file_release+0x130/0x1f8
kernfs_drain_open_files+0x17c/0x440
kernfs_drain+0x2dc/0x360
kernfs_show+0x1b8/0x288
cgroup_file_show+0x150/0x268
cgroup_pressure_write+0x1dc/0x340
cgroup_file_write+0x274/0x548

Reproduction Steps:
1. Open test/cpu.pressure and establish epoll monitoring
2. Disable monitoring: echo 0 > test/cgroup.pressure
3. Re-enable monitoring: echo 1 > test/cgroup.pressure

The race condition occurs because:
1. When cgroup.pressure is disabled (echo 0 > cgroup.pressure), it:
   - Releases PSI triggers via cgroup_file_release()
   - Frees of->priv through kernfs_drain_open_files()
2. While epoll still holds reference to the file and continues polling
3. Re-enabling (echo 1 > cgroup.pressure) accesses freed of->priv

epolling			disable/enable cgroup.pressure
fd=open(cpu.pressure)
while(1)
...
epoll_wait
kernfs_fop_poll
kernfs_get_active = true	echo 0 > cgroup.pressure
...				cgroup_file_show
				kernfs_show
				// inactive kn
				kernfs_drain_open_files
				cft->release(of);
				kfree(ctx);
				...
kernfs_get_active = false
				echo 1 > cgroup.pressure
				kernfs_show
				kernfs_activate_one(kn);
kernfs_fop_poll
kernfs_get_active = true
cgroup_file_poll
psi_trigger_poll
// UAF
...
end: close(fd)

To address this issue, introduce kernfs_get_active_of() for kernfs open
files to obtain active references. This function will fail if the open file
has been released. Replace kernfs_get_active() with kernfs_get_active_of()
to prevent further operations on released file descriptors.

Fixes: 34f26a1 ("sched/psi: Per-cgroup PSI accounting disable/re-enable interface")
Cc: stable <stable@kernel.org>
Reported-by: Zhang Zhaotian <zhangzhaotian@huawei.com>
Signed-off-by: Chen Ridong <chenridong@huawei.com>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20250822070715.1565236-2-chenridong@huaweicloud.com
[ Drop llseek bits ]
Signed-off-by: Sasha Levin <sashal@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
(cherry picked from commit 854baafc00c433cccbe0ab4231b77aeb9b637b77)
Signed-off-by: Pavel Reichl <preichl@redhat.com>

Author: preichl

fs/kernfs/file.c

@@ -70,6 +70,24 @@ static struct kernfs_open_node *of_on(struct kernfs_open_file *of)
 					 !list_empty(&of->list));
 }
 
+/* Get active reference to kernfs node for an open file */
+static struct kernfs_open_file *kernfs_get_active_of(struct kernfs_open_file *of)
+{
+	/* Skip if file was already released */
+	if (unlikely(of->released))
+		return NULL;
+
+	if (!kernfs_get_active(of->kn))
+		return NULL;
+
+	return of;
+}
+
+static void kernfs_put_active_of(struct kernfs_open_file *of)
+{
+	return kernfs_put_active(of->kn);
+}
+
 /**
  * kernfs_deref_open_node_locked - Get kernfs_open_node corresponding to @kn
  *
@@ -139,7 +157,7 @@ static void kernfs_seq_stop_active(struct seq_file *sf, void *v)
 
 	if (ops->seq_stop)
 		ops->seq_stop(sf, v);
-	kernfs_put_active(of->kn);
+	kernfs_put_active_of(of);
 }
 
 static void *kernfs_seq_start(struct seq_file *sf, loff_t *ppos)
@@ -152,7 +170,7 @@ static void *kernfs_seq_start(struct seq_file *sf, loff_t *ppos)
 	 * the ops aren't called concurrently for the same open file.
 	 */
 	mutex_lock(&of->mutex);
-	if (!kernfs_get_active(of->kn))
+	if (!kernfs_get_active_of(of))
 		return ERR_PTR(-ENODEV);
 
 	ops = kernfs_ops(of->kn);
@@ -243,7 +261,7 @@ static ssize_t kernfs_file_read_iter(struct kiocb *iocb, struct iov_iter *iter)
 	 * the ops aren't called concurrently for the same open file.
 	 */
 	mutex_lock(&of->mutex);
-	if (!kernfs_get_active(of->kn)) {
+	if (!kernfs_get_active_of(of)) {
 		len = -ENODEV;
 		mutex_unlock(&of->mutex);
 		goto out_free;
@@ -257,7 +275,7 @@ static ssize_t kernfs_file_read_iter(struct kiocb *iocb, struct iov_iter *iter)
 	else
 		len = -EINVAL;
 
-	kernfs_put_active(of->kn);
+	kernfs_put_active_of(of);
 	mutex_unlock(&of->mutex);
 
 	if (len < 0)
@@ -328,7 +346,7 @@ static ssize_t kernfs_fop_write_iter(struct kiocb *iocb, struct iov_iter *iter)
 	 * the ops aren't called concurrently for the same open file.
 	 */
 	mutex_lock(&of->mutex);
-	if (!kernfs_get_active(of->kn)) {
+	if (!kernfs_get_active_of(of)) {
 		mutex_unlock(&of->mutex);
 		len = -ENODEV;
 		goto out_free;
@@ -340,7 +358,7 @@ static ssize_t kernfs_fop_write_iter(struct kiocb *iocb, struct iov_iter *iter)
 	else
 		len = -EINVAL;
 
-	kernfs_put_active(of->kn);
+	kernfs_put_active_of(of);
 	mutex_unlock(&of->mutex);
 
 	if (len > 0)
@@ -362,13 +380,13 @@ static void kernfs_vma_open(struct vm_area_struct *vma)
 	if (!of->vm_ops)
 		return;
 
-	if (!kernfs_get_active(of->kn))
+	if (!kernfs_get_active_of(of))
 		return;
 
 	if (of->vm_ops->open)
 		of->vm_ops->open(vma);
 
-	kernfs_put_active(of->kn);
+	kernfs_put_active_of(of);
 }
 
 static vm_fault_t kernfs_vma_fault(struct vm_fault *vmf)
@@ -380,14 +398,14 @@ static vm_fault_t kernfs_vma_fault(struct vm_fault *vmf)
 	if (!of->vm_ops)
 		return VM_FAULT_SIGBUS;
 
-	if (!kernfs_get_active(of->kn))
+	if (!kernfs_get_active_of(of))
 		return VM_FAULT_SIGBUS;
 
 	ret = VM_FAULT_SIGBUS;
 	if (of->vm_ops->fault)
 		ret = of->vm_ops->fault(vmf);
 
-	kernfs_put_active(of->kn);
+	kernfs_put_active_of(of);
 	return ret;
 }
 
@@ -400,7 +418,7 @@ static vm_fault_t kernfs_vma_page_mkwrite(struct vm_fault *vmf)
 	if (!of->vm_ops)
 		return VM_FAULT_SIGBUS;
 
-	if (!kernfs_get_active(of->kn))
+	if (!kernfs_get_active_of(of))
 		return VM_FAULT_SIGBUS;
 
 	ret = 0;
@@ -409,7 +427,7 @@ static vm_fault_t kernfs_vma_page_mkwrite(struct vm_fault *vmf)
 	else
 		file_update_time(file);
 
-	kernfs_put_active(of->kn);
+	kernfs_put_active_of(of);
 	return ret;
 }
 
@@ -423,14 +441,14 @@ static int kernfs_vma_access(struct vm_area_struct *vma, unsigned long addr,
 	if (!of->vm_ops)
 		return -EINVAL;
 
-	if (!kernfs_get_active(of->kn))
+	if (!kernfs_get_active_of(of))
 		return -EINVAL;
 
 	ret = -EINVAL;
 	if (of->vm_ops->access)
 		ret = of->vm_ops->access(vma, addr, buf, len, write);
 
-	kernfs_put_active(of->kn);
+	kernfs_put_active_of(of);
 	return ret;
 }
 
@@ -460,7 +478,7 @@ static int kernfs_fop_mmap(struct file *file, struct vm_area_struct *vma)
 	mutex_lock(&of->mutex);
 
 	rc = -ENODEV;
-	if (!kernfs_get_active(of->kn))
+	if (!kernfs_get_active_of(of))
 		goto out_unlock;
 
 	ops = kernfs_ops(of->kn);
@@ -493,7 +511,7 @@ static int kernfs_fop_mmap(struct file *file, struct vm_area_struct *vma)
 	of->vm_ops = vma->vm_ops;
 	vma->vm_ops = &kernfs_vm_ops;
 out_put:
-	kernfs_put_active(of->kn);
+	kernfs_put_active_of(of);
 out_unlock:
 	mutex_unlock(&of->mutex);
 
@@ -847,15 +865,15 @@ static __poll_t kernfs_fop_poll(struct file *filp, poll_table *wait)
 	struct kernfs_node *kn = kernfs_dentry_node(filp->f_path.dentry);
 	__poll_t ret;
 
-	if (!kernfs_get_active(kn))
+	if (!kernfs_get_active_of(of))
 		return DEFAULT_POLLMASK|EPOLLERR|EPOLLPRI;
 
 	if (kn->attr.ops->poll)
 		ret = kn->attr.ops->poll(of, wait);
 	else
 		ret = kernfs_generic_poll(of, wait);
 
-	kernfs_put_active(kn);
+	kernfs_put_active_of(of);
 	return ret;
 }