[PATCH] ALSA: pcm: Move rwsem lock inside snd_ctl_elem_read to prevent UAF

[PATCH] ALSA: pcm: Move rwsem lock inside snd_ctl_elem_read to prevent UAF

[Takashi Iwai]

From: Clement Lecigne <clecigne@google.com>

Takes rwsem lock inside snd_ctl_elem_read instead of snd_ctl_elem_read_user
like it was done for write in commit 1fa4445f9adf1 ("ALSA: control - introduce
snd_ctl_notify_one() helper"). Doing this way we are also fixing the following
locking issue happening in the compat path which can be easily triggered and
turned into an use-after-free.

64-bits:
snd_ctl_ioctl
  snd_ctl_elem_read_user
    [takes controls_rwsem]
    snd_ctl_elem_read [lock properly held, all good]
    [drops controls_rwsem]

32-bits:
snd_ctl_ioctl_compat
  snd_ctl_elem_write_read_compat
    ctl_elem_write_read
      snd_ctl_elem_read [missing lock, not good]

CVE-2023-0266 was assigned for this issue.

Cc: stable@kernel.org # 5.13+
Signed-off-by: Clement Lecigne <clecigne@google.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
---

Note: the fixes for older stable trees will be submitted in another patch

 sound/core/control.c | 24 +++++++++++++++---------
 1 file changed, 15 insertions(+), 9 deletions(-)

diff --git a/sound/core/control.c b/sound/core/control.c
index 50e7ba66f187..82aa1af1d1d8 100644
--- a/sound/core/control.c
+++ b/sound/core/control.c
@@ -1203,14 +1203,19 @@ static int snd_ctl_elem_read(struct snd_card *card,
 	const u32 pattern = 0xdeadbeef;
 	int ret;
 
+	down_read(&card->controls_rwsem);
 	kctl = snd_ctl_find_id(card, &control->id);
-	if (kctl == NULL)
-		return -ENOENT;
+	if (kctl == NULL) {
+		ret = -ENOENT;
+		goto unlock;
+	}
 
 	index_offset = snd_ctl_get_ioff(kctl, &control->id);
 	vd = &kctl->vd[index_offset];
-	if (!(vd->access & SNDRV_CTL_ELEM_ACCESS_READ) || kctl->get == NULL)
-		return -EPERM;
+	if (!(vd->access & SNDRV_CTL_ELEM_ACCESS_READ) || kctl->get == NULL) {
+		ret = -EPERM;
+		goto unlock;
+	}
 
 	snd_ctl_build_ioff(&control->id, kctl, index_offset);
 
@@ -1220,7 +1225,7 @@ static int snd_ctl_elem_read(struct snd_card *card,
 	info.id = control->id;
 	ret = __snd_ctl_elem_info(card, kctl, &info, NULL);
 	if (ret < 0)
-		return ret;
+		goto unlock;
 #endif
 
 	if (!snd_ctl_skip_validation(&info))
@@ -1230,7 +1235,7 @@ static int snd_ctl_elem_read(struct snd_card *card,
 		ret = kctl->get(kctl, control);
 	snd_power_unref(card);
 	if (ret < 0)
-		return ret;
+		goto unlock;
 	if (!snd_ctl_skip_validation(&info) &&
 	    sanity_check_elem_value(card, control, &info, pattern) < 0) {
 		dev_err(card->dev,
@@ -1238,8 +1243,11 @@ static int snd_ctl_elem_read(struct snd_card *card,
 			control->id.iface, control->id.device,
 			control->id.subdevice, control->id.name,
 			control->id.index);
-		return -EINVAL;
+		ret = -EINVAL;
+		goto unlock;
 	}
+unlock:
+	up_read(&card->controls_rwsem);
 	return ret;
 }
 
@@ -1253,9 +1261,7 @@ static int snd_ctl_elem_read_user(struct snd_card *card,
 	if (IS_ERR(control))
 		return PTR_ERR(control);
 
-	down_read(&card->controls_rwsem);
 	result = snd_ctl_elem_read(card, control);
-	up_read(&card->controls_rwsem);
 	if (result < 0)
 		goto error;
 
-- 
2.35.3

Re: [PATCH] ALSA: pcm: Move rwsem lock inside snd_ctl_elem_read to prevent UAF

[Jaroslav Kysela]

On 13. 01. 23 13:07, Takashi Iwai wrote:
> From: Clement Lecigne <clecigne@google.com>
> 
> Takes rwsem lock inside snd_ctl_elem_read instead of snd_ctl_elem_read_user
> like it was done for write in commit 1fa4445f9adf1 ("ALSA: control - introduce
> snd_ctl_notify_one() helper"). Doing this way we are also fixing the following
> locking issue happening in the compat path which can be easily triggered and
> turned into an use-after-free.
> 
> 64-bits:
> snd_ctl_ioctl
>    snd_ctl_elem_read_user
>      [takes controls_rwsem]
>      snd_ctl_elem_read [lock properly held, all good]
>      [drops controls_rwsem]
> 
> 32-bits:
> snd_ctl_ioctl_compat
>    snd_ctl_elem_write_read_compat
>      ctl_elem_write_read
>        snd_ctl_elem_read [missing lock, not good]
> 
> CVE-2023-0266 was assigned for this issue.
> 
> Cc: stable@kernel.org # 5.13+
> Signed-off-by: Clement Lecigne <clecigne@google.com>
> Signed-off-by: Takashi Iwai <tiwai@suse.de>

Reviewed-by: Jaroslav Kysela <perex@perex.cz>

-- 
Jaroslav Kysela <perex@perex.cz>
Linux Sound Maintainer; ALSA Project; Red Hat, Inc.