[PATCH -next] ALSA: timer: Fix use-after-free problem
When the timer instance was add into ack_list but was not currently in process, the user could stop it via snd_timer_stop1() without delete it from the ack_list. Then the user could free the timer instance and when it was actually processed UAF occurred.
This issue could be reproduced via testcase snd_timer01 in ltp - running several instances of that testcase at the same time.
What I actually met was that the ack_list of the timer broken and the kernel went into deadloop with irqoff. That could be detected by hardlockup detector on board or when we run it on qemu, we could use gdb to dump the ack_list when the console has no response.
To fix this issue, we introduce a new flag SNDRV_TIMER_IFLG_ACKING to indicate the state where the timer instance is in ack_list but not currently processed and check against the new flag in snd_timer_stop1() and delete it from ack_list if the flag is set.
Signed-off-by: Wang Wensheng wangwensheng4@huawei.com --- include/sound/timer.h | 1 + sound/core/timer.c | 12 +++++++++--- 2 files changed, 10 insertions(+), 3 deletions(-)
diff --git a/include/sound/timer.h b/include/sound/timer.h index 760e132cc0cd..549288e94a39 100644 --- a/include/sound/timer.h +++ b/include/sound/timer.h @@ -31,6 +31,7 @@ #define SNDRV_TIMER_IFLG_CALLBACK 0x00000020 /* timer callback is active */ #define SNDRV_TIMER_IFLG_EXCLUSIVE 0x00000040 /* exclusive owner - no more instances */ #define SNDRV_TIMER_IFLG_EARLY_EVENT 0x00000080 /* write early event to the poll queue */ +#define SNDRV_TIMER_IFLG_ACKING 0x00000100 /* the timeri was added to ack_list */
#define SNDRV_TIMER_FLG_CHANGE 0x00000001 #define SNDRV_TIMER_FLG_RESCHED 0x00000002 /* need reschedule */ diff --git a/sound/core/timer.c b/sound/core/timer.c index 92b7008fcdb8..1d1e4274919c 100644 --- a/sound/core/timer.c +++ b/sound/core/timer.c @@ -625,10 +625,12 @@ static int snd_timer_stop1(struct snd_timer_instance *timeri, bool stop) return -EINVAL; spin_lock_irqsave(&timer->lock, flags); if (!(timeri->flags & (SNDRV_TIMER_IFLG_RUNNING | - SNDRV_TIMER_IFLG_START))) { + SNDRV_TIMER_IFLG_START | + SNDRV_TIMER_IFLG_ACKING))) { result = -EBUSY; goto unlock; } + list_del_init(&timeri->ack_list); list_del_init(&timeri->active_list); if (timer->card && timer->card->shutdown) @@ -649,7 +651,8 @@ static int snd_timer_stop1(struct snd_timer_instance *timeri, bool stop) } } } - timeri->flags &= ~(SNDRV_TIMER_IFLG_RUNNING | SNDRV_TIMER_IFLG_START); + timeri->flags &= ~(SNDRV_TIMER_IFLG_RUNNING | SNDRV_TIMER_IFLG_START | + SNDRV_TIMER_IFLG_ACKING); if (stop) timeri->flags &= ~SNDRV_TIMER_IFLG_PAUSED; else @@ -786,6 +789,7 @@ static void snd_timer_process_callbacks(struct snd_timer *timer,
/* remove from ack_list and make empty */ list_del_init(&ti->ack_list); + ti->flags &= ~SNDRV_TIMER_IFLG_ACKING;
if (!(ti->flags & SNDRV_TIMER_IFLG_DEAD)) { ticks = ti->pticks; @@ -890,8 +894,10 @@ void snd_timer_interrupt(struct snd_timer * timer, unsigned long ticks_left) ack_list_head = &timer->ack_list_head; else ack_list_head = &timer->sack_list_head; - if (list_empty(&ti->ack_list)) + if (list_empty(&ti->ack_list)) { list_add_tail(&ti->ack_list, ack_list_head); + ti->flags |= SNDRV_TIMER_IFLG_ACKING; + } list_for_each_entry(ts, &ti->slave_active_head, active_list) { ts->pticks = ti->pticks; ts->resolution = resolution;
On Tue, 02 Nov 2021 14:41:07 +0100, Wang Wensheng wrote:
When the timer instance was add into ack_list but was not currently in process, the user could stop it via snd_timer_stop1() without delete it from the ack_list. Then the user could free the timer instance and when it was actually processed UAF occurred.
This issue could be reproduced via testcase snd_timer01 in ltp - running several instances of that testcase at the same time.
What I actually met was that the ack_list of the timer broken and the kernel went into deadloop with irqoff. That could be detected by hardlockup detector on board or when we run it on qemu, we could use gdb to dump the ack_list when the console has no response.
To fix this issue, we introduce a new flag SNDRV_TIMER_IFLG_ACKING to indicate the state where the timer instance is in ack_list but not currently processed and check against the new flag in snd_timer_stop1() and delete it from ack_list if the flag is set.
Signed-off-by: Wang Wensheng wangwensheng4@huawei.com
Thanks for the patch. Just through a quick glance, I wonder whether it'd be easier to do list_del_init(&timeri->ack_list) unconditionally before the check of timeri->flags in snd_timer1_stop(). Ditto for active_list. So something like:
--- a/sound/core/timer.c +++ b/sound/core/timer.c @@ -624,13 +624,13 @@ static int snd_timer_stop1(struct snd_timer_instance *timeri, bool stop) if (!timer) return -EINVAL; spin_lock_irqsave(&timer->lock, flags); + list_del_init(&timeri->ack_list); + list_del_init(&timeri->active_list); if (!(timeri->flags & (SNDRV_TIMER_IFLG_RUNNING | SNDRV_TIMER_IFLG_START))) { result = -EBUSY; goto unlock; } - list_del_init(&timeri->ack_list); - list_del_init(&timeri->active_list); if (timer->card && timer->card->shutdown) goto unlock; if (stop) {
Takashi
include/sound/timer.h | 1 + sound/core/timer.c | 12 +++++++++--- 2 files changed, 10 insertions(+), 3 deletions(-)
diff --git a/include/sound/timer.h b/include/sound/timer.h index 760e132cc0cd..549288e94a39 100644 --- a/include/sound/timer.h +++ b/include/sound/timer.h @@ -31,6 +31,7 @@ #define SNDRV_TIMER_IFLG_CALLBACK 0x00000020 /* timer callback is active */ #define SNDRV_TIMER_IFLG_EXCLUSIVE 0x00000040 /* exclusive owner - no more instances */ #define SNDRV_TIMER_IFLG_EARLY_EVENT 0x00000080 /* write early event to the poll queue */ +#define SNDRV_TIMER_IFLG_ACKING 0x00000100 /* the timeri was added to ack_list */
#define SNDRV_TIMER_FLG_CHANGE 0x00000001 #define SNDRV_TIMER_FLG_RESCHED 0x00000002 /* need reschedule */ diff --git a/sound/core/timer.c b/sound/core/timer.c index 92b7008fcdb8..1d1e4274919c 100644 --- a/sound/core/timer.c +++ b/sound/core/timer.c @@ -625,10 +625,12 @@ static int snd_timer_stop1(struct snd_timer_instance *timeri, bool stop) return -EINVAL; spin_lock_irqsave(&timer->lock, flags); if (!(timeri->flags & (SNDRV_TIMER_IFLG_RUNNING |
SNDRV_TIMER_IFLG_START))) {
SNDRV_TIMER_IFLG_START |
result = -EBUSY; goto unlock; }SNDRV_TIMER_IFLG_ACKING))) {
- list_del_init(&timeri->ack_list); list_del_init(&timeri->active_list); if (timer->card && timer->card->shutdown)
@@ -649,7 +651,8 @@ static int snd_timer_stop1(struct snd_timer_instance *timeri, bool stop) } } }
- timeri->flags &= ~(SNDRV_TIMER_IFLG_RUNNING | SNDRV_TIMER_IFLG_START);
- timeri->flags &= ~(SNDRV_TIMER_IFLG_RUNNING | SNDRV_TIMER_IFLG_START |
if (stop) timeri->flags &= ~SNDRV_TIMER_IFLG_PAUSED; elseSNDRV_TIMER_IFLG_ACKING);
@@ -786,6 +789,7 @@ static void snd_timer_process_callbacks(struct snd_timer *timer,
/* remove from ack_list and make empty */ list_del_init(&ti->ack_list);
ti->flags &= ~SNDRV_TIMER_IFLG_ACKING;
if (!(ti->flags & SNDRV_TIMER_IFLG_DEAD)) { ticks = ti->pticks;
@@ -890,8 +894,10 @@ void snd_timer_interrupt(struct snd_timer * timer, unsigned long ticks_left) ack_list_head = &timer->ack_list_head; else ack_list_head = &timer->sack_list_head;
if (list_empty(&ti->ack_list))
if (list_empty(&ti->ack_list)) { list_add_tail(&ti->ack_list, ack_list_head);
ti->flags |= SNDRV_TIMER_IFLG_ACKING;
list_for_each_entry(ts, &ti->slave_active_head, active_list) { ts->pticks = ti->pticks; ts->resolution = resolution;}
-- 2.17.1
在 2021/11/2 22:10, Takashi Iwai 写道:
On Tue, 02 Nov 2021 14:41:07 +0100, Wang Wensheng wrote:
When the timer instance was add into ack_list but was not currently in process, the user could stop it via snd_timer_stop1() without delete it from the ack_list. Then the user could free the timer instance and when it was actually processed UAF occurred.
This issue could be reproduced via testcase snd_timer01 in ltp - running several instances of that testcase at the same time.
What I actually met was that the ack_list of the timer broken and the kernel went into deadloop with irqoff. That could be detected by hardlockup detector on board or when we run it on qemu, we could use gdb to dump the ack_list when the console has no response.
To fix this issue, we introduce a new flag SNDRV_TIMER_IFLG_ACKING to indicate the state where the timer instance is in ack_list but not currently processed and check against the new flag in snd_timer_stop1() and delete it from ack_list if the flag is set.
Signed-off-by: Wang Wensheng wangwensheng4@huawei.com
Thanks for the patch. Just through a quick glance, I wonder whether it'd be easier to do list_del_init(&timeri->ack_list) unconditionally before the check of timeri->flags in snd_timer1_stop(). Ditto for active_list. So something like:
Thanks for your suggestions. It looks much more efficient and easier and is also effective. I will take this as a second version.
--- a/sound/core/timer.c +++ b/sound/core/timer.c @@ -624,13 +624,13 @@ static int snd_timer_stop1(struct snd_timer_instance *timeri, bool stop) if (!timer) return -EINVAL; spin_lock_irqsave(&timer->lock, flags);
- list_del_init(&timeri->ack_list);
- list_del_init(&timeri->active_list); if (!(timeri->flags & (SNDRV_TIMER_IFLG_RUNNING | SNDRV_TIMER_IFLG_START))) { result = -EBUSY; goto unlock; }
- list_del_init(&timeri->ack_list);
- list_del_init(&timeri->active_list); if (timer->card && timer->card->shutdown) goto unlock; if (stop) {
Takashi
participants (3)
-
Takashi Iwai
-
Wang Wensheng
-
wangwensheng (C)