Skip to contentSkip to conversation

media: usb: fix memory leak in stk_camera_probe

PATCH In review Dongliang Mu 6 participants 15 messages Jul 14, 2021 – Sep 2, 2021 Media (V4L/DVB)

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 media: usb: fix memory leak in stk_camera_probe

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Fix this by removing the execution of usb_get_intf.

Fixes 0aa77f6c2954 ("[media] move the remaining USB drivers to drivers/media/usb")
---
 drivers/media/usb/stkwebcam/stk-webcam.c | 1 -
 1 file changed, 1 deletion(-)
drivers/media/usb/stkwebcam/stk-webcam.c +0 -1
index a45d464427c4..5bd8e85b9452 100644
@@ -1311,7 +1311,6 @@ static int stk_camera_probe(struct usb_interface *interface,
 
 	dev->udev = udev;
 	dev->interface = interface;
-	usb_get_intf(interface);
 
 	if (hflip != -1)
 		dev->vsettings.hflip = hflip;
Signature · 1 line
2.25.1
#1 lore raw
Quoted 5 lines · Dongliang Mu

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Fix this by removing the execution of usb_get_intf.

Any idea about this patch?

Quoted 23 lines
Fixes 0aa77f6c2954 ("[media] move the remaining USB drivers to drivers/media/usb")
---
 drivers/media/usb/stkwebcam/stk-webcam.c | 1 -
 1 file changed, 1 deletion(-)
drivers/media/usb/stkwebcam/stk-webcam.c +0 -2
index a45d464427c4..5bd8e85b9452 100644
@@ -1311,7 +1311,6 @@ static int stk_camera_probe(struct usb_interface *interface,
 
        dev->udev = udev;
        dev->interface = interface;
-       usb_get_intf(interface);
 
        if (hflip != -1)
                dev->vsettings.hflip = hflip;
--

2.25.1

#2 lore raw
Quoted 9 lines · Dongliang Mu
Quoted 5 lines · Dongliang Mu

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Fix this by removing the execution of usb_get_intf.

Any idea about this patch?

+cc Dan Carpenter, gregkh

There is no reply in this thread in one month. Can someone give some feedback on this patch?

Quoted 24 lines
Quoted 24 lines
Fixes 0aa77f6c2954 ("[media] move the remaining USB drivers to drivers/media/usb")
---
 drivers/media/usb/stkwebcam/stk-webcam.c | 1 -
 1 file changed, 1 deletion(-)
drivers/media/usb/stkwebcam/stk-webcam.c +0 -2
index a45d464427c4..5bd8e85b9452 100644
@@ -1311,7 +1311,6 @@ static int stk_camera_probe(struct usb_interface *interface,
 
        dev->udev = udev;
        dev->interface = interface;
-       usb_get_intf(interface);
 
        if (hflip != -1)
                dev->vsettings.hflip = hflip;
--

2.25.1

#3 lore raw
Quoted 7 lines · Dongliang Mu
Quoted 6 lines · Dongliang Mu
Quoted 4 lines · Dongliang Mu

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

Fix this by removing the execution of usb_get_intf.

Any idea about this patch?

+cc Dan Carpenter, gregkh

There is no reply in this thread in one month. Can someone give some feedback on this patch?

This is the media developers domain, not much I can do here.

thanks,

greg k-h

#4 lore raw

Em Thu, 2 Sep 2021 12:39:37 +0200
Greg KH <[email protected]> escreveu:

Quoted 10 lines · Dongliang Mu
Quoted 9 lines · Dongliang Mu
Quoted 8 lines · Dongliang Mu
Quoted 5 lines

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Fix this by removing the execution of usb_get_intf.

Quoted 8 lines
Quoted 6 lines
Quoted 1 line

Any idea about this patch?

+cc Dan Carpenter, gregkh

There is no reply in this thread in one month. Can someone give some feedback on this patch?

This is the media developers domain, not much I can do here.

There is a high volume of patches for the media subsystem. Anyway, as your patch is at our patchwork instance:

	https://patchwork.linuxtv.org/project/linux-media/patch/[email protected]/

It should be properly tracked, and likely handled after the end of the merge window.

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Fixes 0aa77f6c2954 ("[media] move the remaining USB drivers to drivers/media/usb")

If you're the author of the patch, it doesn't make much sense to add a "Reported-by:" tag there. We only use it in order to give someone's else credit to report an issue.

Thanks,
Mauro

#5 lore raw
Quoted 31 lines · Mauro Carvalho Chehab

Em Thu, 2 Sep 2021 12:39:37 +0200
Greg KH <[email protected]> escreveu:

Quoted 10 lines · Dongliang Mu
Quoted 9 lines · Dongliang Mu
Quoted 8 lines · Dongliang Mu
Quoted 5 lines

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Fix this by removing the execution of usb_get_intf.

Quoted 8 lines
Quoted 6 lines
Quoted 1 line

Any idea about this patch?

+cc Dan Carpenter, gregkh

There is no reply in this thread in one month. Can someone give some feedback on this patch?

This is the media developers domain, not much I can do here.

There is a high volume of patches for the media subsystem. Anyway, as your patch is at our patchwork instance:

        https://patchwork.linuxtv.org/project/linux-media/patch/[email protected]/

It should be properly tracked, and likely handled after the end of the merge window.

Quoted 7 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Fixes 0aa77f6c2954 ("[media] move the remaining USB drivers to drivers/media/usb")

If you're the author of the patch, it doesn't make much sense to add a "Reported-by:" tag there. We only use it in order to give someone's else credit to report an issue.

I see. Someone told me this rule in another thread. I will update this in the next version.

Quoted 3 lines

Thanks,
Mauro

#6 lore raw
Quoted 25 lines · Dongliang Mu
Quoted 23 lines · Mauro Carvalho Chehab

Em Thu, 2 Sep 2021 12:39:37 +0200
Greg KH <[email protected]> escreveu:

Quoted 18 lines · Dongliang Mu
Quoted 7 lines · Dongliang Mu
Quoted 6 lines · Dongliang Mu
Quoted 4 lines

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

Fix this by removing the execution of usb_get_intf.

Any idea about this patch?

+cc Dan Carpenter, gregkh

There is no reply in this thread in one month. Can someone give some feedback on this patch?

This is the media developers domain, not much I can do here.

Quoted 7 lines
Quoted 7 lines

There is a high volume of patches for the media subsystem. Anyway, as your patch is at our patchwork instance:

        https://patchwork.linuxtv.org/project/linux-media/patch/[email protected]/

It should be properly tracked, and likely handled after the end of the merge window.

Hi Mauro,

I found there is another fix [1] for the same memory leak from Pavel Skripkin (already cc-ed in this thread).

[1] https://www.spinics.net/lists/stable/msg479628.html

Quoted 15 lines
Quoted 8 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Fixes 0aa77f6c2954 ("[media] move the remaining USB drivers to drivers/media/usb")

If you're the author of the patch, it doesn't make much sense to add a "Reported-by:" tag there. We only use it in order to give someone's else credit to report an issue.

I see. Someone told me this rule in another thread. I will update this in the next version.

Quoted 3 lines

Thanks,
Mauro

#7 lore raw
Quoted 34 lines · Dongliang Mu
Quoted 25 lines · Dongliang Mu
Quoted 23 lines · Mauro Carvalho Chehab

Em Thu, 2 Sep 2021 12:39:37 +0200
Greg KH <[email protected]> escreveu:

Quoted 18 lines · Dongliang Mu
Quoted 7 lines · Dongliang Mu
Quoted 6 lines · Dongliang Mu
Quoted 4 lines

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

Fix this by removing the execution of usb_get_intf.

Any idea about this patch?

+cc Dan Carpenter, gregkh

There is no reply in this thread in one month. Can someone give some feedback on this patch?

This is the media developers domain, not much I can do here.

Quoted 7 lines
Quoted 7 lines

There is a high volume of patches for the media subsystem. Anyway, as your patch is at our patchwork instance:

        https://patchwork.linuxtv.org/project/linux-media/patch/[email protected]/

It should be properly tracked, and likely handled after the end of the merge window.

Quoted 6 lines

Hi Mauro,

I found there is another fix [1] for the same memory leak from Pavel Skripkin (already cc-ed in this thread).

[1] https://www.spinics.net/lists/stable/msg479628.html

Ah, that's why I marked it as Obsoleted :-)

Regards,

Hans

Quoted 16 lines
Quoted 16 lines
Quoted 9 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Fixes 0aa77f6c2954 ("[media] move the remaining USB drivers to drivers/media/usb")

If you're the author of the patch, it doesn't make much sense to add a "Reported-by:" tag there. We only use it in order to give someone's else credit to report an issue.

I see. Someone told me this rule in another thread. I will update this in the next version.

Quoted 3 lines

Thanks,
Mauro

#8 lore raw
Quoted 36 lines · Hans Verkuil
Quoted 34 lines · Dongliang Mu
Quoted 32 lines · Dongliang Mu
Quoted 30 lines · Mauro Carvalho Chehab

Em Thu, 2 Sep 2021 12:39:37 +0200
Greg KH <[email protected]> escreveu:

Quoted 10 lines · Dongliang Mu
Quoted 9 lines · Dongliang Mu
Quoted 8 lines · Dongliang Mu
Quoted 5 lines

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Fix this by removing the execution of usb_get_intf.

Quoted 8 lines
Quoted 6 lines
Quoted 1 line

Any idea about this patch?

+cc Dan Carpenter, gregkh

There is no reply in this thread in one month. Can someone give some feedback on this patch?

This is the media developers domain, not much I can do here.

There is a high volume of patches for the media subsystem. Anyway, as your patch is at our patchwork instance:

        https://patchwork.linuxtv.org/project/linux-media/patch/[email protected]/

It should be properly tracked, and likely handled after the end of the merge window.

Quoted 8 lines
Quoted 6 lines

Hi Mauro,

I found there is another fix [1] for the same memory leak from Pavel Skripkin (already cc-ed in this thread).

[1] https://www.spinics.net/lists/stable/msg479628.html

Ah, that's why I marked it as Obsoleted :-)

Oh, I see. If that patch is already merged, please remark my patch as Obsoleted.

Curiously, I did not get an email notification to mark my patch as Obsoleted before. Why?

Quoted 22 lines

Regards,

Hans

Quoted 16 lines
Quoted 16 lines
Quoted 9 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Fixes 0aa77f6c2954 ("[media] move the remaining USB drivers to drivers/media/usb")

If you're the author of the patch, it doesn't make much sense to add a "Reported-by:" tag there. We only use it in order to give someone's else credit to report an issue.

I see. Someone told me this rule in another thread. I will update this in the next version.

Quoted 3 lines

Thanks,
Mauro

#9 lore raw
Quoted 42 lines · Dongliang Mu
Quoted 41 lines · Hans Verkuil
Quoted 39 lines · Dongliang Mu
Quoted 25 lines · Dongliang Mu
Quoted 23 lines · Mauro Carvalho Chehab

Em Thu, 2 Sep 2021 12:39:37 +0200
Greg KH <[email protected]> escreveu:

Quoted 18 lines · Dongliang Mu
Quoted 7 lines · Dongliang Mu
Quoted 6 lines · Dongliang Mu
Quoted 4 lines

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

Fix this by removing the execution of usb_get_intf.

Any idea about this patch?

+cc Dan Carpenter, gregkh

There is no reply in this thread in one month. Can someone give some feedback on this patch?

This is the media developers domain, not much I can do here.

Quoted 7 lines
Quoted 7 lines

There is a high volume of patches for the media subsystem. Anyway, as your patch is at our patchwork instance:

        https://patchwork.linuxtv.org/project/linux-media/patch/[email protected]/

It should be properly tracked, and likely handled after the end of the merge window.

Hi Mauro,

I found there is another fix [1] for the same memory leak from Pavel Skripkin (already cc-ed in this thread).

Quoted 10 lines
Quoted 3 lines
Quoted 1 line

[1] https://www.spinics.net/lists/stable/msg479628.html

Ah, that's why I marked it as Obsoleted :-)

Oh, I see. If that patch is already merged, please remark my patch as Obsoleted.

Curiously, I did not get an email notification to mark my patch as Obsoleted before. Why?

Quoted 1 line

Hi, Dongliang!

Yep my patch has been merged already (1 month ago, I guess).

With regards,
Pavel Skripkin

#10 lore raw
Quoted 50 lines · Pavel Skripkin
Quoted 48 lines · Dongliang Mu
Quoted 36 lines · Hans Verkuil
Quoted 34 lines · Dongliang Mu
Quoted 32 lines · Dongliang Mu
Quoted 30 lines · Mauro Carvalho Chehab

Em Thu, 2 Sep 2021 12:39:37 +0200
Greg KH <[email protected]> escreveu:

Quoted 10 lines · Dongliang Mu
Quoted 9 lines · Dongliang Mu
Quoted 8 lines · Dongliang Mu
Quoted 5 lines

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Fix this by removing the execution of usb_get_intf.

Quoted 8 lines
Quoted 6 lines
Quoted 1 line

Any idea about this patch?

+cc Dan Carpenter, gregkh

There is no reply in this thread in one month. Can someone give some feedback on this patch?

This is the media developers domain, not much I can do here.

There is a high volume of patches for the media subsystem. Anyway, as your patch is at our patchwork instance:

        https://patchwork.linuxtv.org/project/linux-media/patch/[email protected]/

It should be properly tracked, and likely handled after the end of the merge window.

Quoted 8 lines
Quoted 6 lines

Hi Mauro,

I found there is another fix [1] for the same memory leak from Pavel Skripkin (already cc-ed in this thread).

[1] https://www.spinics.net/lists/stable/msg479628.html

Ah, that's why I marked it as Obsoleted :-)

Oh, I see. If that patch is already merged, please remark my patch as Obsoleted.

Quoted 8 lines
Quoted 4 lines

Curiously, I did not get an email notification to mark my patch as Obsoleted before. Why?

Quoted 1 line

Hi, Dongliang!

Yep my patch has been merged already (1 month ago, I guess).

Yes. When I submit this patch, your patch is still pending. I did not know someone is already sending a patch with good quality.

So I am curious if there are any methods to notify people there are already some similar patches in the patchwork or some other resources.

Quoted 6 lines

With regards,
Pavel Skripkin

#11 lore raw
In reply to Dongliang Mu

Hi Dongliang,

Quoted 7 lines · Dongliang Mu
Quoted 6 lines · Dongliang Mu
Quoted 4 lines · Dongliang Mu

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

Fix this by removing the execution of usb_get_intf.

Any idea about this patch?

+cc Dan Carpenter, gregkh

There is no reply in this thread in one month. Can someone give some feedback on this patch?

I saw that it was marked as Obsoleted in patchwork, but I might have confused this patch with similar, but not identical, patches for this driver.

I've moved the state back to New.

Comments follow below:

Quoted 16 lines
Quoted 16 lines
Quoted 16 lines
Fixes 0aa77f6c2954 ("[media] move the remaining USB drivers to drivers/media/usb")
---
 drivers/media/usb/stkwebcam/stk-webcam.c | 1 -
 1 file changed, 1 deletion(-)
drivers/media/usb/stkwebcam/stk-webcam.c +0 -0
index a45d464427c4..5bd8e85b9452 100644
@@ -1311,7 +1311,6 @@ static int stk_camera_probe(struct usb_interface *interface,
 
Quoted 3 lines
Quoted 3 lines

drivers/media/usb/stkwebcam/stk-webcam.c static int stk_camera_probe(struct usb_interface *interface, · line 1314

Quoted 3 lines
drivers/media/usb/stkwebcam/stk-webcam.c +0 -1
        dev->udev = udev;
        dev->interface = interface;
-       usb_get_intf(interface);

Even though this increments the refcount (which might well be unnecessary), it is also decremented with usb_put_intf. So is there actually a bug here?

And if this is changed, then I expect that both get_intf and put_intf should be removed, and not just the get.

Regards,

Hans

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
drivers/media/usb/stkwebcam/stk-webcam.c +0 -1
        if (hflip != -1)
                dev->vsettings.hflip = hflip;
--

2.25.1

#12 lore raw
Quoted 43 lines · Hans Verkuil

Hi Dongliang,

Quoted 7 lines · Dongliang Mu
Quoted 6 lines · Dongliang Mu
Quoted 4 lines · Dongliang Mu

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

Fix this by removing the execution of usb_get_intf.

Any idea about this patch?

+cc Dan Carpenter, gregkh

There is no reply in this thread in one month. Can someone give some feedback on this patch?

I saw that it was marked as Obsoleted in patchwork, but I might have confused this patch with similar, but not identical, patches for this driver.

I've moved the state back to New.

Comments follow below:

Quoted 16 lines
Quoted 16 lines
Quoted 16 lines
Fixes 0aa77f6c2954 ("[media] move the remaining USB drivers to drivers/media/usb")
---
 drivers/media/usb/stkwebcam/stk-webcam.c | 1 -
 1 file changed, 1 deletion(-)
drivers/media/usb/stkwebcam/stk-webcam.c +0 -0
index a45d464427c4..5bd8e85b9452 100644
@@ -1311,7 +1311,6 @@ static int stk_camera_probe(struct usb_interface *interface,
 
Quoted 7 lines
Quoted 3 lines
Quoted 3 lines

drivers/media/usb/stkwebcam/stk-webcam.c static int stk_camera_probe(struct usb_interface *interface, · line 1314

Quoted 3 lines
drivers/media/usb/stkwebcam/stk-webcam.c +0 -1
        dev->udev = udev;
        dev->interface = interface;
-       usb_get_intf(interface);

Even though this increments the refcount (which might well be unnecessary), it is also decremented with usb_put_intf. So is there actually a bug here?

Yes, if the increment and decrement of refcount are balanced, it's fine.

But this probe function only increases the refcount, but forgets to decrease the refcount.

Quoted 14 lines

And if this is changed, then I expect that both get_intf and put_intf should be removed, and not just the get.

Regards,

Hans

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
drivers/media/usb/stkwebcam/stk-webcam.c +0 -1
        if (hflip != -1)
                dev->vsettings.hflip = hflip;
--

2.25.1

#13 lore raw
In reply to Dongliang Mu
Quoted 31 lines · Dongliang Mu
Quoted 9 lines · Dongliang Mu
Quoted 5 lines · Dongliang Mu

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Fix this by removing the execution of usb_get_intf.

Any idea about this patch?

+cc Dan Carpenter, gregkh

There is no reply in this thread in one month. Can someone give some feedback on this patch?

Quoted 15 lines
Quoted 15 lines
Fixes 0aa77f6c2954 ("[media] move the remaining USB drivers to drivers/media/usb")
---
 drivers/media/usb/stkwebcam/stk-webcam.c | 1 -
 1 file changed, 1 deletion(-)
drivers/media/usb/stkwebcam/stk-webcam.c +0 -0
index a45d464427c4..5bd8e85b9452 100644
@@ -1311,7 +1311,6 @@ static int stk_camera_probe(struct usb_interface *interface,
 
Quoted 3 lines
Quoted 3 lines

drivers/media/usb/stkwebcam/stk-webcam.c static int stk_camera_probe(struct usb_interface *interface, · line 1314

Quoted 3 lines
drivers/media/usb/stkwebcam/stk-webcam.c +0 -1
        dev->udev = udev;
        dev->interface = interface;
-       usb_get_intf(interface);

The patch is wrong. We're storing a reference to "interface".

	dev->interface = interface;

So we need to boost the refcount of interface. Pavel Skripkin was on the right patch but you need to add a:

	usb_put_intf(interface);

to the stk_camera_disconnect() function as you sort of mentioned. That's the correct fix.

regards,
dan carpenter

#14 lore raw
Quoted 42 lines · Dan Carpenter
Quoted 31 lines · Dongliang Mu
Quoted 9 lines · Dongliang Mu
Quoted 5 lines · Dongliang Mu

stk_camera_probe mistakenly execute usb_get_intf and increase the refcount of interface->dev.

Fix this by removing the execution of usb_get_intf.

Any idea about this patch?

+cc Dan Carpenter, gregkh

There is no reply in this thread in one month. Can someone give some feedback on this patch?

Quoted 15 lines
Quoted 15 lines
Fixes 0aa77f6c2954 ("[media] move the remaining USB drivers to drivers/media/usb")
---
 drivers/media/usb/stkwebcam/stk-webcam.c | 1 -
 1 file changed, 1 deletion(-)
drivers/media/usb/stkwebcam/stk-webcam.c +0 -0
index a45d464427c4..5bd8e85b9452 100644
@@ -1311,7 +1311,6 @@ static int stk_camera_probe(struct usb_interface *interface,
 
Quoted 3 lines
Quoted 3 lines

drivers/media/usb/stkwebcam/stk-webcam.c static int stk_camera_probe(struct usb_interface *interface, · line 1314

Quoted 3 lines
drivers/media/usb/stkwebcam/stk-webcam.c +0 -1
        dev->udev = udev;
        dev->interface = interface;
-       usb_get_intf(interface);

The patch is wrong. We're storing a reference to "interface".

        dev->interface = interface;
Quoted 7 lines

So we need to boost the refcount of interface. Pavel Skripkin was on the right patch but you need to add a:

        usb_put_intf(interface);

to the stk_camera_disconnect() function as you sort of mentioned. That's the correct fix.

Thanks for your explanation, Dan. It's really helpful.

I sent the inquiry email in this thread because I did not receive the notification of patchwork to mark my patch as obsolete and did not notice Pavel had sent one patch before.

Now, this patch is marked as obsolete. Let's ignore it now.

Quoted 3 lines

regards,
dan carpenter