linux-kernel - Re: [PATCH] ALSA: usb-audio: Fix a potential memory leak in scarlett2_init

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <8cde2320-517f-3a38-8c3f-f807791c6c52@wanadoo.fr>
Date:   Sun, 3 Sep 2023 17:04:47 +0200
From:   Christophe JAILLET <christophe.jaillet@...adoo.fr>
To:     Takashi Iwai <tiwai@...e.de>
Cc:     Jaroslav Kysela <perex@...ex.cz>, Takashi Iwai <tiwai@...e.com>,
        linux-kernel@...r.kernel.org, kernel-janitors@...r.kernel.org,
        alsa-devel@...a-project.org
Subject: Re: [PATCH] ALSA: usb-audio: Fix a potential memory leak in
 scarlett2_init_notify()

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@...adoo.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
>>
>