[PATCH] ALSA: usb-audio: Fix a potential memory leak in scarlett2_init_notify()

[PATCH] ALSA: usb-audio: Fix a potential memory leak in scarlett2_init_notify()

[Christophe JAILLET]

If usb_alloc_coherent() or usb_urb_ep_type_check() fail, we should release
the resources previously allocated.

Fixes: ff49d1df79ae ("ALSA: usb-audio: USB MIDI 2.0 UMP support")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
---
 sound/usb/midi2.c | 17 ++++++++++++++---
 1 file changed, 14 insertions(+), 3 deletions(-)

diff --git a/sound/usb/midi2.c b/sound/usb/midi2.c
index a27e244650c8..4109c82adff6 100644
--- a/sound/usb/midi2.c
+++ b/sound/usb/midi2.c
@@ -302,7 +302,8 @@ static int alloc_midi_urbs(struct snd_usb_midi2_endpoint *ep)
 		ctx->urb = usb_alloc_urb(0, GFP_KERNEL);
 		if (!ctx->urb) {
 			dev_err(&ep->dev->dev, "URB alloc failed\n");
-			return -ENOMEM;
+			err = -ENOMEM;
+			goto err_free_all;
 		}
 		ctx->ep = ep;
 		buffer = usb_alloc_coherent(ep->dev, len, GFP_KERNEL,
@@ -310,7 +311,8 @@ static int alloc_midi_urbs(struct snd_usb_midi2_endpoint *ep)
 		if (!buffer) {
 			dev_err(&ep->dev->dev,
 				"URB buffer alloc failed (size %d)\n", len);
-			return -ENOMEM;
+			err = -ENOMEM;
+			goto err_free_cur_urb;
 		}
 		if (ep->interval)
 			usb_fill_int_urb(ctx->urb, ep->dev, ep->pipe,
@@ -322,13 +324,22 @@ static int alloc_midi_urbs(struct snd_usb_midi2_endpoint *ep)
 		if (err < 0) {
 			dev_err(&ep->dev->dev, "invalid MIDI EP %x\n",
 				endpoint);
-			return err;
+			goto err_free_cur_dma;
 		}
 		ctx->urb->transfer_flags = URB_NO_TRANSFER_DMA_MAP;
 		ep->num_urbs++;
 	}
 	ep->urb_free = ep->urb_free_mask = GENMASK(ep->num_urbs - 1, 0);
 	return 0;
+
+err_free_cur_dma:
+	usb_free_coherent(ep->dev, len, buffer, ctx->urb->transfer_dma);
+err_free_cur_urb:
+	usb_free_urb(ctx->urb);
+	ctx->urb = NULL;
+err_free_all:
+	free_midi_urbs(ep);
+	return err;
 }
 
 static struct snd_usb_midi2_endpoint *
-- 
2.34.1

Re: [PATCH] ALSA: usb-audio: Fix a potential memory leak in scarlett2_init_notify()

[Takashi Iwai]

On Sun, 03 Sep 2023 15:06:00 +0200,
Christophe JAILLET wrote:
> 
> If usb_alloc_coherent() or usb_urb_ep_type_check() fail, we should release
> the resources previously allocated.

Those are freed in the caller side, start_input_streams() instead.


thanks,

Takashi

> 
> Fixes: ff49d1df79ae ("ALSA: usb-audio: USB MIDI 2.0 UMP support")
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> ---
>  sound/usb/midi2.c | 17 ++++++++++++++---
>  1 file changed, 14 insertions(+), 3 deletions(-)
> 
> diff --git a/sound/usb/midi2.c b/sound/usb/midi2.c
> index a27e244650c8..4109c82adff6 100644
> --- a/sound/usb/midi2.c
> +++ b/sound/usb/midi2.c
> @@ -302,7 +302,8 @@ static int alloc_midi_urbs(struct snd_usb_midi2_endpoint *ep)
>  		ctx->urb = usb_alloc_urb(0, GFP_KERNEL);
>  		if (!ctx->urb) {
>  			dev_err(&ep->dev->dev, "URB alloc failed\n");
> -			return -ENOMEM;
> +			err = -ENOMEM;
> +			goto err_free_all;
>  		}
>  		ctx->ep = ep;
>  		buffer = usb_alloc_coherent(ep->dev, len, GFP_KERNEL,
> @@ -310,7 +311,8 @@ static int alloc_midi_urbs(struct snd_usb_midi2_endpoint *ep)
>  		if (!buffer) {
>  			dev_err(&ep->dev->dev,
>  				"URB buffer alloc failed (size %d)\n", len);
> -			return -ENOMEM;
> +			err = -ENOMEM;
> +			goto err_free_cur_urb;
>  		}
>  		if (ep->interval)
>  			usb_fill_int_urb(ctx->urb, ep->dev, ep->pipe,
> @@ -322,13 +324,22 @@ static int alloc_midi_urbs(struct snd_usb_midi2_endpoint *ep)
>  		if (err < 0) {
>  			dev_err(&ep->dev->dev, "invalid MIDI EP %x\n",
>  				endpoint);
> -			return err;
> +			goto err_free_cur_dma;
>  		}
>  		ctx->urb->transfer_flags = URB_NO_TRANSFER_DMA_MAP;
>  		ep->num_urbs++;
>  	}
>  	ep->urb_free = ep->urb_free_mask = GENMASK(ep->num_urbs - 1, 0);
>  	return 0;
> +
> +err_free_cur_dma:
> +	usb_free_coherent(ep->dev, len, buffer, ctx->urb->transfer_dma);
> +err_free_cur_urb:
> +	usb_free_urb(ctx->urb);
> +	ctx->urb = NULL;
> +err_free_all:
> +	free_midi_urbs(ep);
> +	return err;
>  }
>  
>  static struct snd_usb_midi2_endpoint *
> -- 
> 2.34.1
> 

Re: [PATCH] ALSA: usb-audio: Fix a potential memory leak in scarlett2_init_notify()

[Christophe JAILLET]

Le 03/09/2023 à 16:23, Takashi Iwai a écrit :
> On Sun, 03 Sep 2023 15:06:00 +0200,
> Christophe JAILLET wrote:
>>
>> If usb_alloc_coherent() or usb_urb_ep_type_check() fail, we should release
>> the resources previously allocated.
> 
> Those are freed in the caller side, start_input_streams() instead.

Thanks for the fast review.

Hmpm, If IIUC, resources allocated *before* the ending "ep->num_urbs++" 
still need to be freed here, otherwise free_midi_urbs() in the caller 
will not free them.

Do you agree?

If yes, I can send v2 which would look like:
	usb_alloc_urb()
	if (err)
		return -ENOMEM

	usb_alloc_coherent()
	if (err) {
		usb_free_urb()
		urb = NULL
		return -ENOMEM
	}
	
	 usb_urb_ep_type_check()
	if (err) {
		usb_free_coherent()
		usb_free_urb()
		urb = NULL
		return -err
	}

Or, if yuo prefer, with an error handling path just like below, but 
without the final free_midi_urbs() + a comment explaining that the 
caller does this part of job instead.

CJ

> 
> 
> thanks,
> 
> Takashi
> 
>>
>> Fixes: ff49d1df79ae ("ALSA: usb-audio: USB MIDI 2.0 UMP support")
>> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
>> ---
>>   sound/usb/midi2.c | 17 ++++++++++++++---
>>   1 file changed, 14 insertions(+), 3 deletions(-)
>>
>> diff --git a/sound/usb/midi2.c b/sound/usb/midi2.c
>> index a27e244650c8..4109c82adff6 100644
>> --- a/sound/usb/midi2.c
>> +++ b/sound/usb/midi2.c
>> @@ -302,7 +302,8 @@ static int alloc_midi_urbs(struct snd_usb_midi2_endpoint *ep)
>>   		ctx->urb = usb_alloc_urb(0, GFP_KERNEL);
>>   		if (!ctx->urb) {
>>   			dev_err(&ep->dev->dev, "URB alloc failed\n");
>> -			return -ENOMEM;
>> +			err = -ENOMEM;
>> +			goto err_free_all;
>>   		}
>>   		ctx->ep = ep;
>>   		buffer = usb_alloc_coherent(ep->dev, len, GFP_KERNEL,
>> @@ -310,7 +311,8 @@ static int alloc_midi_urbs(struct snd_usb_midi2_endpoint *ep)
>>   		if (!buffer) {
>>   			dev_err(&ep->dev->dev,
>>   				"URB buffer alloc failed (size %d)\n", len);
>> -			return -ENOMEM;
>> +			err = -ENOMEM;
>> +			goto err_free_cur_urb;
>>   		}
>>   		if (ep->interval)
>>   			usb_fill_int_urb(ctx->urb, ep->dev, ep->pipe,
>> @@ -322,13 +324,22 @@ static int alloc_midi_urbs(struct snd_usb_midi2_endpoint *ep)
>>   		if (err < 0) {
>>   			dev_err(&ep->dev->dev, "invalid MIDI EP %x\n",
>>   				endpoint);
>> -			return err;
>> +			goto err_free_cur_dma;
>>   		}
>>   		ctx->urb->transfer_flags = URB_NO_TRANSFER_DMA_MAP;
>>   		ep->num_urbs++;
>>   	}
>>   	ep->urb_free = ep->urb_free_mask = GENMASK(ep->num_urbs - 1, 0);
>>   	return 0;
>> +
>> +err_free_cur_dma:
>> +	usb_free_coherent(ep->dev, len, buffer, ctx->urb->transfer_dma);
>> +err_free_cur_urb:
>> +	usb_free_urb(ctx->urb);
>> +	ctx->urb = NULL;
>> +err_free_all:
>> +	free_midi_urbs(ep);
>> +	return err;
>>   }
>>   
>>   static struct snd_usb_midi2_endpoint *
>> -- 
>> 2.34.1
>>
> 

Re: [PATCH] ALSA: usb-audio: Fix a potential memory leak in scarlett2_init_notify()

[Takashi Iwai]

On Sun, 03 Sep 2023 17:04:47 +0200,
Christophe JAILLET wrote:
> 
> Le 03/09/2023 à 16:23, Takashi Iwai a écrit :
> > On Sun, 03 Sep 2023 15:06:00 +0200,
> > Christophe JAILLET wrote:
> >> 
> >> If usb_alloc_coherent() or usb_urb_ep_type_check() fail, we should release
> >> the resources previously allocated.
> > 
> > Those are freed in the caller side, start_input_streams() instead.
> 
> Thanks for the fast review.
> 
> Hmpm, If IIUC, resources allocated *before* the ending
> "ep->num_urbs++" still need to be freed here, otherwise
> free_midi_urbs() in the caller will not free them.
> 
> Do you agree?
> 
> If yes, I can send v2 which would look like:
> 	usb_alloc_urb()
> 	if (err)
> 		return -ENOMEM
> 
> 	usb_alloc_coherent()
> 	if (err) {
> 		usb_free_urb()
> 		urb = NULL
> 		return -ENOMEM
> 	}
> 	
> 	 usb_urb_ep_type_check()
> 	if (err) {
> 		usb_free_coherent()
> 		usb_free_urb()
> 		urb = NULL
> 		return -err
> 	}
> 
> Or, if yuo prefer, with an error handling path just like below, but
> without the final free_midi_urbs() + a comment explaining that the
> caller does this part of job instead.

Indeed.  The fix would be rather a oneliner like below, though:

--- a/sound/usb/midi2.c
+++ b/sound/usb/midi2.c
@@ -265,7 +265,7 @@ static void free_midi_urbs(struct snd_usb_midi2_endpoint *ep)
 
 	if (!ep)
 		return;
-	for (i = 0; i < ep->num_urbs; ++i) {
+	for (i = 0; i < NUM_URBS; ++i) {
 		ctx = &ep->urbs[i];
 		if (!ctx->urb)
 			break;

That was the intended behavior of free_midi_urbs().


Takashi

Re: [PATCH] ALSA: usb-audio: Fix a potential memory leak in scarlett2_init_notify()

[Christophe JAILLET]

Le 03/09/2023 à 18:37, Takashi Iwai a écrit :
> On Sun, 03 Sep 2023 17:04:47 +0200,
...

> Indeed.  The fix would be rather a oneliner like below, though:

Looks much better than mine :)

I let you send the patch, it is your solution.



Just for my understanding, how is snd_ump_ops used, especially .open?
I've not been able to figure out where it was called.

In alloc_midi_urbs(), if usb_alloc_coherent() fails, then 
ctx->urb->transfer_buffer could be anything because usb_fill_xxx_urb() 
is not called.
So there could be an edge case where your fix could still be incomplete.

For the start_input_streams() caller, this is fine, because the 
corresponding memory is kzalloc()'ed in start_input_streams() at some 
point, but I've not been able to check for snd_usb_midi_v2_open().

CJ

> 
> --- a/sound/usb/midi2.c
> +++ b/sound/usb/midi2.c
> @@ -265,7 +265,7 @@ static void free_midi_urbs(struct snd_usb_midi2_endpoint *ep)
>   
>   	if (!ep)
>   		return;
> -	for (i = 0; i < ep->num_urbs; ++i) {
> +	for (i = 0; i < NUM_URBS; ++i) {
>   		ctx = &ep->urbs[i];
>   		if (!ctx->urb)
>   			break;
> 
> That was the intended behavior of free_midi_urbs().
> 
> 
> Takashi
> 

Re: [PATCH] ALSA: usb-audio: Fix a potential memory leak in scarlett2_init_notify()

[Takashi Iwai]

On Sun, 03 Sep 2023 21:42:55 +0200,
Christophe JAILLET wrote:
> 
> Le 03/09/2023 à 18:37, Takashi Iwai a écrit :
> > On Sun, 03 Sep 2023 17:04:47 +0200,
> ...
> 
> > Indeed.  The fix would be rather a oneliner like below, though:
> 
> Looks much better than mine :)
> 
> I let you send the patch, it is your solution.
> 
> 
> 
> Just for my understanding, how is snd_ump_ops used, especially .open?
> I've not been able to figure out where it was called.

It's called via rawmidi open (the snd_ump_endpoint is a sort of child
class of snd_rawmidi).

> In alloc_midi_urbs(), if usb_alloc_coherent() fails, then
> ctx->urb->transfer_buffer could be anything because usb_fill_xxx_urb()
> is not called.
> So there could be an edge case where your fix could still be incomplete.

Each URB is allocated in the loop via usb_alloc_urb(), and it does
zero-initialize the object, hence the buffer is supposed to be NULL
until it's set up via usb_fill_xxx().


thanks,

Takashi

> For the start_input_streams() caller, this is fine, because the
> corresponding memory is kzalloc()'ed in start_input_streams() at some
> point, but I've not been able to check for snd_usb_midi_v2_open().
> 
> CJ
> 
> > 
> > --- a/sound/usb/midi2.c
> > +++ b/sound/usb/midi2.c
> > @@ -265,7 +265,7 @@ static void free_midi_urbs(struct snd_usb_midi2_endpoint *ep)
> >     	if (!ep)
> >   		return;
> > -	for (i = 0; i < ep->num_urbs; ++i) {
> > +	for (i = 0; i < NUM_URBS; ++i) {
> >   		ctx = &ep->urbs[i];
> >   		if (!ctx->urb)
> >   			break;
> > 
> > That was the intended behavior of free_midi_urbs().
> > 
> > 
> > Takashi
> > 
> 

Re: [PATCH] ALSA: usb-audio: Fix a potential memory leak in scarlett2_init_notify()

[Takashi Iwai]

On Mon, 04 Sep 2023 16:08:15 +0200,
Takashi Iwai wrote:
> 
> On Sun, 03 Sep 2023 21:42:55 +0200,
> Christophe JAILLET wrote:
> > 
> > Le 03/09/2023 à 18:37, Takashi Iwai a écrit :
> > > On Sun, 03 Sep 2023 17:04:47 +0200,
> > ...
> > For the start_input_streams() caller, this is fine, because the
> > corresponding memory is kzalloc()'ed in start_input_streams() at some
> > point, but I've not been able to check for snd_usb_midi_v2_open().

Oh I overlooked that point.  Yes, it's a missing call, although the
memory leaks as free_midi_urbs() is called also at the destructor,
free_midi2_endpoint(), too.  But it's definitely better to call at the
error path, too.  Will fix it up together and submit the proper fix
patch.


thanks,

Takashi



> > 
> > CJ
> > 
> > > 
> > > --- a/sound/usb/midi2.c
> > > +++ b/sound/usb/midi2.c
> > > @@ -265,7 +265,7 @@ static void free_midi_urbs(struct snd_usb_midi2_endpoint *ep)
> > >     	if (!ep)
> > >   		return;
> > > -	for (i = 0; i < ep->num_urbs; ++i) {
> > > +	for (i = 0; i < NUM_URBS; ++i) {
> > >   		ctx = &ep->urbs[i];
> > >   		if (!ctx->urb)
> > >   			break;
> > > 
> > > That was the intended behavior of free_midi_urbs().
> > > 
> > > 
> > > Takashi
> > > 
> > 
>