Skip to contentSkip to conversation

media: em28xx: fix corrupted list

PATCH In review Pavel Skripkin 2 participants 9 messages Jul 6, 2021 – Jul 29, 2021

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 media: em28xx: fix corrupted list

Syzbot reported corrupted list in em28xx driver. The problem was in non-reinitialized lists on disconnect. Since all 2 lists are global variables and driver can be connected and disconnected many times we should call INIT_LIST_HEAD() in .disconnect method to prevent corrupted list entries.

Fixes 1a23f81b7dc3 ("V4L/DVB (9979): em28xx: move usb probe code to a proper place")
Reported-by [email protected]
---
 drivers/media/usb/em28xx/em28xx-cards.c | 2 ++
 drivers/media/usb/em28xx/em28xx-core.c  | 6 ++++++
 drivers/media/usb/em28xx/em28xx.h       | 1 +
 3 files changed, 9 insertions(+)
drivers/media/usb/em28xx/em28xx-cards.c +2 -0
index ba9292e2a587..8b1ff79c37a0 100644
@@ -4148,6 +4148,8 @@ static void em28xx_usb_disconnect(struct usb_interface *intf)
 		dev->dev_next = NULL;
 	}
 	kref_put(&dev->ref, em28xx_free_device);
+
+	em28xx_reset_lists();
 }
 
 static int em28xx_usb_suspend(struct usb_interface *intf,
drivers/media/usb/em28xx/em28xx-core.c +6 -0
index 584fa400cd7d..03970ed00dba 100644
@@ -1131,6 +1131,12 @@ void em28xx_init_extension(struct em28xx *dev)
 	mutex_unlock(&em28xx_devlist_mutex);
 }
 
+void em28xx_reset_lists(void)
+{
+	INIT_LIST_HEAD(&em28xx_devlist);
+	INIT_LIST_HEAD(&em28xx_extension_devlist);
+}
+
 void em28xx_close_extension(struct em28xx *dev)
 {
 	const struct em28xx_ops *ops = NULL;
drivers/media/usb/em28xx/em28xx.h +1 -0
index ab167cd1f400..73caaaa398d3 100644
@@ -835,6 +835,7 @@ void em28xx_stop_urbs(struct em28xx *dev);
 int em28xx_set_mode(struct em28xx *dev, enum em28xx_mode set_mode);
 int em28xx_gpio_set(struct em28xx *dev, const struct em28xx_reg_seq *gpio);
 int em28xx_register_extension(struct em28xx_ops *dev);
+void em28xx_reset_lists(void);
 void em28xx_unregister_extension(struct em28xx_ops *dev);
 void em28xx_init_extension(struct em28xx *dev);
 void em28xx_close_extension(struct em28xx *dev);
Signature · 1 line
2.32.0
#1 lore raw
Quoted 33 lines · Pavel Skripkin

Syzbot reported corrupted list in em28xx driver. The problem was in non-reinitialized lists on disconnect. Since all 2 lists are global variables and driver can be connected and disconnected many times we should call INIT_LIST_HEAD() in .disconnect method to prevent corrupted list entries.

Fixes 1a23f81b7dc3 ("V4L/DVB (9979): em28xx: move usb probe code to a proper place")
Reported-by [email protected]
---
 drivers/media/usb/em28xx/em28xx-cards.c | 2 ++
 drivers/media/usb/em28xx/em28xx-core.c  | 6 ++++++
 drivers/media/usb/em28xx/em28xx.h       | 1 +
 3 files changed, 9 insertions(+)
drivers/media/usb/em28xx/em28xx-cards.c +2 -0
index ba9292e2a587..8b1ff79c37a0 100644
@@ -4148,6 +4148,8 @@ static void em28xx_usb_disconnect(struct usb_interface *intf)
 		dev->dev_next = NULL;
 	}
 	kref_put(&dev->ref, em28xx_free_device);
+
+	em28xx_reset_lists();
 }
 
 static int em28xx_usb_suspend(struct usb_interface *intf,
drivers/media/usb/em28xx/em28xx-core.c +0 -0
index 584fa400cd7d..03970ed00dba 100644

drivers/media/usb/em28xx/em28xx-core.c void em28xx_init_extension(struct em28xx *dev) · line 1137

Quoted 7 lines
drivers/media/usb/em28xx/em28xx-core.c +4 -0
@@ -1131,6 +1131,12 @@ void em28xx_init_extension(struct em28xx *dev)
 	mutex_unlock(&em28xx_devlist_mutex);
 }
 
+void em28xx_reset_lists(void)
+{
+	INIT_LIST_HEAD(&em28xx_devlist);
+	INIT_LIST_HEAD(&em28xx_extension_devlist);

This needs a mutex_lock(&em28xx_devlist_mutex);

But actually, I don't think this is right: if there are multiple em28xx devices, then I think if you disconnect one, then the other is - with this code - also removed from the list.

Can you give a link to the actual syzbot bug? I'm not at all sure you are fixing the right thing here.

Regards,

Hans

Quoted 18 lines
drivers/media/usb/em28xx/em28xx-core.c +2 -0
+}
+
 void em28xx_close_extension(struct em28xx *dev)
 {
 	const struct em28xx_ops *ops = NULL;
drivers/media/usb/em28xx/em28xx.h +1 -0
index ab167cd1f400..73caaaa398d3 100644
@@ -835,6 +835,7 @@ void em28xx_stop_urbs(struct em28xx *dev);
 int em28xx_set_mode(struct em28xx *dev, enum em28xx_mode set_mode);
 int em28xx_gpio_set(struct em28xx *dev, const struct em28xx_reg_seq *gpio);
 int em28xx_register_extension(struct em28xx_ops *dev);
+void em28xx_reset_lists(void);
 void em28xx_unregister_extension(struct em28xx_ops *dev);
 void em28xx_init_extension(struct em28xx *dev);
 void em28xx_close_extension(struct em28xx *dev);
#2 lore raw
Quoted 44 lines · On Tue, 20 Jul 2021 14:13:13 +0200 Hans Verkuil <[email protected]> wrote:
Quoted 36 lines · Pavel Skripkin

Syzbot reported corrupted list in em28xx driver. The problem was in non-reinitialized lists on disconnect. Since all 2 lists are global variables and driver can be connected and disconnected many times we should call INIT_LIST_HEAD() in .disconnect method to prevent corrupted list entries.

Fixes 1a23f81b7dc3 ("V4L/DVB (9979): em28xx: move usb probe code

to a proper place") Reported-by:
[email protected]

 drivers/media/usb/em28xx/em28xx-cards.c | 2 ++
 drivers/media/usb/em28xx/em28xx-core.c  | 6 ++++++
 drivers/media/usb/em28xx/em28xx.h       | 1 +
 3 files changed, 9 insertions(+)
(quoted diff) +0 -0

b/drivers/media/usb/em28xx/em28xx-cards.c index ba9292e2a587..8b1ff79c37a0 100644 --- a/drivers/media/usb/em28xx/em28xx-cards.c +++ b/drivers/media/usb/em28xx/em28xx-cards.c @@ -4148,6 +4148,8 @@ static void em28xx_usb_disconnect(struct usb_interface *intf) dev->dev_next = NULL; }

(quoted diff) +2 -0
 	kref_put(&dev->ref, em28xx_free_device);
+
+	em28xx_reset_lists();
 }
 
 static int em28xx_usb_suspend(struct usb_interface *intf,
(quoted diff) +0 -0

b/drivers/media/usb/em28xx/em28xx-core.c index 584fa400cd7d..03970ed00dba 100644 --- a/drivers/media/usb/em28xx/em28xx-core.c +++ b/drivers/media/usb/em28xx/em28xx-core.c @@ -1131,6 +1131,12 @@ void em28xx_init_extension(struct em28xx *dev) mutex_unlock(&em28xx_devlist_mutex); }

Review

Quoted 4 lines
(quoted diff) +4 -0
+void em28xx_reset_lists(void)
+{
+	INIT_LIST_HEAD(&em28xx_devlist);
+	INIT_LIST_HEAD(&em28xx_extension_devlist);

This needs a mutex_lock(&em28xx_devlist_mutex);

Quoted 6 lines

But actually, I don't think this is right: if there are multiple em28xx devices, then I think if you disconnect one, then the other is
- with this code - also removed from the list.

Can you give a link to the actual syzbot bug? I'm not at all sure you are fixing the right thing here.

Hi, Hans!

I guess, I missed my coffee that morning, sorry :) This patch looks complety wrong, of course.

I've took a close look at this bug one more time, and I found the true root case of this bug.

If em28xx dev has dev_next pointer we need to close dev->next extension on disconnect to avoid UAF and corrupted list bug. So, something like this should work:

drivers/media/usb/em28xx/em28xx-cards.c +3 -1
index c1e0dccb7408..d56b040e1bd7 100644
@@ -4139,8 +4139,10 @@ static void em28xx_usb_disconnect(struct usb_interface *intf)
 
 	em28xx_close_extension(dev);
 
-	if (dev->dev_next)
+	if (dev->dev_next) {
 		em28xx_release_resources(dev->dev_next);
+		em28xx_close_extension(dev->dev_next);
+	}
 	em28xx_release_resources(dev);
 
 	if (dev->dev_next) {

How to You feel about it? I am going to send this patch for syzbot testing

Syzbot link:
https://syzkaller.appspot.com/bug?id=3609bbf45bf63a8f6032f330eb3d34f51cc81493

Again, sorry for v1 patch, I don't know how I came up with this fix :(

With regards,
Pavel Skripkin

#3 lore raw
Quoted 83 lines · Pavel Skripkin
Quoted 44 lines · On Tue, 20 Jul 2021 14:13:13 +0200 Hans Verkuil <[email protected]> wrote:
Quoted 36 lines · Pavel Skripkin

Syzbot reported corrupted list in em28xx driver. The problem was in non-reinitialized lists on disconnect. Since all 2 lists are global variables and driver can be connected and disconnected many times we should call INIT_LIST_HEAD() in .disconnect method to prevent corrupted list entries.

Fixes 1a23f81b7dc3 ("V4L/DVB (9979): em28xx: move usb probe code

to a proper place") Reported-by:
[email protected]

 drivers/media/usb/em28xx/em28xx-cards.c | 2 ++
 drivers/media/usb/em28xx/em28xx-core.c  | 6 ++++++
 drivers/media/usb/em28xx/em28xx.h       | 1 +
 3 files changed, 9 insertions(+)
(quoted diff) +0 -0

b/drivers/media/usb/em28xx/em28xx-cards.c index ba9292e2a587..8b1ff79c37a0 100644 --- a/drivers/media/usb/em28xx/em28xx-cards.c +++ b/drivers/media/usb/em28xx/em28xx-cards.c @@ -4148,6 +4148,8 @@ static void em28xx_usb_disconnect(struct usb_interface *intf) dev->dev_next = NULL; }

(quoted diff) +2 -0
 	kref_put(&dev->ref, em28xx_free_device);
+
+	em28xx_reset_lists();
 }
 
 static int em28xx_usb_suspend(struct usb_interface *intf,
(quoted diff) +0 -0

b/drivers/media/usb/em28xx/em28xx-core.c index 584fa400cd7d..03970ed00dba 100644 --- a/drivers/media/usb/em28xx/em28xx-core.c +++ b/drivers/media/usb/em28xx/em28xx-core.c @@ -1131,6 +1131,12 @@ void em28xx_init_extension(struct em28xx *dev) mutex_unlock(&em28xx_devlist_mutex); }

Review

Quoted 4 lines
(quoted diff) +4 -0
+void em28xx_reset_lists(void)
+{
+	INIT_LIST_HEAD(&em28xx_devlist);
+	INIT_LIST_HEAD(&em28xx_extension_devlist);

This needs a mutex_lock(&em28xx_devlist_mutex);

Quoted 6 lines

But actually, I don't think this is right: if there are multiple em28xx devices, then I think if you disconnect one, then the other is
- with this code - also removed from the list.

Can you give a link to the actual syzbot bug? I'm not at all sure you are fixing the right thing here.

Hi, Hans!

I guess, I missed my coffee that morning, sorry :) This patch looks complety wrong, of course.

I've took a close look at this bug one more time, and I found the true root case of this bug.

If em28xx dev has dev_next pointer we need to close dev->next extension on disconnect to avoid UAF and corrupted list bug. So, something like this should work:

drivers/media/usb/em28xx/em28xx-cards.c +3 -1
index c1e0dccb7408..d56b040e1bd7 100644
@@ -4139,8 +4139,10 @@ static void em28xx_usb_disconnect(struct usb_interface *intf)
 
 	em28xx_close_extension(dev);
 
-	if (dev->dev_next)
+	if (dev->dev_next) {
 		em28xx_release_resources(dev->dev_next);
+		em28xx_close_extension(dev->dev_next);
+	}
 	em28xx_release_resources(dev);
 
 	if (dev->dev_next) {
Quoted 2 lines

How to You feel about it? I am going to send this patch for syzbot testing

That looks a lot saner :-)

Quoted 7 lines

Syzbot link:
https://syzkaller.appspot.com/bug?id=3609bbf45bf63a8f6032f330eb3d34f51cc81493

Again, sorry for v1 patch, I don't know how I came up with this fix :(

No problem, just post a v2 if syzbot passes.

Regards,

Hans

Quoted 6 lines

With regards,
Pavel Skripkin

#4 lore raw

PATCH v2 1/1 media: em28xx: add missing em28xx_close_extension

If em28xx dev has ->dev_next pointer, we need to delete dev_next list node from em28xx_extension_devlist on disconnect to avoid UAF bugs and corrupted list bugs, since driver frees this pointer on disconnect.

Fixes 1a23f81b7dc3 ("V4L/DVB (9979): em28xx: move usb probe code to a proper place")
Reported-and-tested-by [email protected]

---

Changes in v2:
Previous patch was completely broken. I've done some debugging again and found true root case of the reported bug.

---
 drivers/media/usb/em28xx/em28xx-cards.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/media/usb/em28xx/em28xx-cards.c +3 -1
index c1e0dccb7408..d56b040e1bd7 100644
@@ -4139,8 +4139,10 @@ static void em28xx_usb_disconnect(struct usb_interface *intf)
 
 	em28xx_close_extension(dev);
 
-	if (dev->dev_next)
+	if (dev->dev_next) {
 		em28xx_release_resources(dev->dev_next);
+		em28xx_close_extension(dev->dev_next);
+	}
 	em28xx_release_resources(dev);
 
 	if (dev->dev_next) {
Signature · 1 line
2.32.0
#5 lore raw

Re: [PATCH v2] media: em28xx: add missing em28xx_close_extension

Quoted 22 lines · Pavel Skripkin

If em28xx dev has ->dev_next pointer, we need to delete dev_next list node from em28xx_extension_devlist on disconnect to avoid UAF bugs and corrupted list bugs, since driver frees this pointer on disconnect.

Fixes 1a23f81b7dc3 ("V4L/DVB (9979): em28xx: move usb probe code to a proper place")
Reported-and-tested-by [email protected]

---

Changes in v2:
Previous patch was completely broken. I've done some debugging again and found true root case of the reported bug.

---
 drivers/media/usb/em28xx/em28xx-cards.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/media/usb/em28xx/em28xx-cards.c +0 -0
index c1e0dccb7408..d56b040e1bd7 100644

drivers/media/usb/em28xx/em28xx-cards.c static void em28xx_usb_disconnect(struct usb_interface *intf) · line 4144

Quoted 7 lines
drivers/media/usb/em28xx/em28xx-cards.c +2 -1
@@ -4139,8 +4139,10 @@ static void em28xx_usb_disconnect(struct usb_interface *intf)
 
 	em28xx_close_extension(dev);
 
-	if (dev->dev_next)
+	if (dev->dev_next) {
 		em28xx_release_resources(dev->dev_next);
+		em28xx_close_extension(dev->dev_next);

Wouldn't it be better to swap these two?

That order is also used for em28xx_close_extension(dev) and em28xx_release_resources(dev).

You do need to store dev->dev_next in a temp variable, though.

Regards,

Hans

Quoted 5 lines
drivers/media/usb/em28xx/em28xx-cards.c +1 -0
+	}
 	em28xx_release_resources(dev);
 
 	if (dev->dev_next) {
#6 lore raw

Re: [PATCH v2] media: em28xx: add missing em28xx_close_extension

Quoted 34 lines · On Thu, 29 Jul 2021 11:45:19 +0200 Hans Verkuil <[email protected]> wrote:
Quoted 24 lines · Pavel Skripkin

If em28xx dev has ->dev_next pointer, we need to delete dev_next list node from em28xx_extension_devlist on disconnect to avoid UAF bugs and corrupted list bugs, since driver frees this pointer on disconnect.

Fixes 1a23f81b7dc3 ("V4L/DVB (9979): em28xx: move usb probe code

to a proper place") Reported-and-tested-by:
[email protected]

Changes in v2:
Previous patch was completely broken. I've done some
debugging again and found true root case of the reported bug.

---
 drivers/media/usb/em28xx/em28xx-cards.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
(quoted diff) +0 -0

b/drivers/media/usb/em28xx/em28xx-cards.c index c1e0dccb7408..d56b040e1bd7 100644 --- a/drivers/media/usb/em28xx/em28xx-cards.c +++ b/drivers/media/usb/em28xx/em28xx-cards.c @@ -4139,8 +4139,10 @@ static void em28xx_usb_disconnect(struct usb_interface *intf)

drivers/media/usb/em28xx/em28xx-cards.c static void em28xx_usb_disconnect(struct usb_interface *intf) · line 4144

Quoted 6 lines
drivers/media/usb/em28xx/em28xx-cards.c +2 -1
 	em28xx_close_extension(dev);
 
-	if (dev->dev_next)
+	if (dev->dev_next) {
 		em28xx_release_resources(dev->dev_next);
+		em28xx_close_extension(dev->dev_next);

Wouldn't it be better to swap these two?

Quoted 5 lines

That order is also used for em28xx_close_extension(dev) and em28xx_release_resources(dev).

You do need to store dev->dev_next in a temp variable, though.

Hi, Hans!

I don't understand why I need to store dev->dev_next in a temp variable. I don't see code in em28xx_release_resources() or em28xx_close_extension() that zeroes this pointer.

With regards,
Pavel Skripkin

#7 lore raw

Re: [PATCH v2] media: em28xx: add missing em28xx_close_extension

Quoted 40 lines · Pavel Skripkin
Quoted 37 lines · On Thu, 29 Jul 2021 11:45:19 +0200 Hans Verkuil <[email protected]> wrote:
Quoted 24 lines · Pavel Skripkin

If em28xx dev has ->dev_next pointer, we need to delete dev_next list node from em28xx_extension_devlist on disconnect to avoid UAF bugs and corrupted list bugs, since driver frees this pointer on disconnect.

Fixes 1a23f81b7dc3 ("V4L/DVB (9979): em28xx: move usb probe code

to a proper place") Reported-and-tested-by:
[email protected]

Changes in v2:
Previous patch was completely broken. I've done some
debugging again and found true root case of the reported bug.

---
 drivers/media/usb/em28xx/em28xx-cards.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
(quoted diff) +0 -0

b/drivers/media/usb/em28xx/em28xx-cards.c index c1e0dccb7408..d56b040e1bd7 100644 --- a/drivers/media/usb/em28xx/em28xx-cards.c +++ b/drivers/media/usb/em28xx/em28xx-cards.c @@ -4139,8 +4139,10 @@ static void em28xx_usb_disconnect(struct usb_interface *intf)

drivers/media/usb/em28xx/em28xx-cards.c static void em28xx_usb_disconnect(struct usb_interface *intf) · line 4144

Quoted 6 lines
drivers/media/usb/em28xx/em28xx-cards.c +2 -1
 	em28xx_close_extension(dev);
 
-	if (dev->dev_next)
+	if (dev->dev_next) {
 		em28xx_release_resources(dev->dev_next);
+		em28xx_close_extension(dev->dev_next);

Wouldn't it be better to swap these two?

That order is also used for em28xx_close_extension(dev) and em28xx_release_resources(dev).

Quoted 8 lines
Quoted 2 lines

You do need to store dev->dev_next in a temp variable, though.

Hi, Hans!

I don't understand why I need to store dev->dev_next in a temp variable. I don't see code in em28xx_release_resources() or em28xx_close_extension() that zeroes this pointer.

Apologies for the confusion, just ignore that bit. I misunderstood what dev_next was for.

So check if swapping these two lines passes syzbot, and then that's the final version.

Regards,

Hans

Quoted 7 lines

With regards,
Pavel Skripkin

#8 lore raw

PATCH v3 1/1 media: em28xx: add missing em28xx_close_extension

If em28xx dev has ->dev_next pointer, we need to delete ->dev_next list node from em28xx_extension_devlist on disconnect to avoid UAF bugs and corrupted list bugs, since driver frees this pointer on disconnect.

Fixes 1a23f81b7dc3 ("V4L/DVB (9979): em28xx: move usb probe code to a proper place")
Reported-and-tested-by [email protected]

---

Changes in v3:
Changed order of em28xx_close_extension() and em28xx_release_resources() as Hans suggested

Changes in v2:
Previous patch was completely broken. I've done some debugging again and found true root case of the reported bug.

---
 drivers/media/usb/em28xx/em28xx-cards.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
drivers/media/usb/em28xx/em28xx-cards.c +4 -1
index c1e0dccb7408..948e22e29b42 100644
@@ -4139,8 +4139,11 @@ static void em28xx_usb_disconnect(struct usb_interface *intf)
 
 	em28xx_close_extension(dev);
 
-	if (dev->dev_next)
+	if (dev->dev_next) {
+		em28xx_close_extension(dev->dev_next);
 		em28xx_release_resources(dev->dev_next);
+	}
+
 	em28xx_release_resources(dev);
 
 	if (dev->dev_next) {
Signature · 1 line
2.32.0