[PATCH] net/9p: Fix buffer overflow in USB transport layer

[PATCH] net/9p: Fix buffer overflow in USB transport layer

[Yuhao Jiang]

A buffer overflow vulnerability exists in the USB 9pfs transport layer
where inconsistent size validation between packet header parsing and
actual data copying allows a malicious USB host to overflow heap buffers.

The issue occurs because:
- usb9pfs_rx_header() validates only the declared size in packet header
- usb9pfs_rx_complete() uses req->actual (actual received bytes) for memcpy

This allows an attacker to craft packets with small declared size (bypassing
validation) but large actual payload (triggering overflow in memcpy).

Add validation in usb9pfs_rx_complete() to ensure req->actual does not
exceed the buffer capacity before copying data.

Reported-by: Yuhao Jiang <danisjiang@gmail.com>
Fixes: a3be076dc174 ("net/9p/usbg: Add new usb gadget function transport")
Cc: stable@vger.kernel.org
Signed-off-by: Yuhao Jiang <danisjiang@gmail.com>
---
 net/9p/trans_usbg.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/net/9p/trans_usbg.c b/net/9p/trans_usbg.c
index 6b694f117aef..047a2862fc84 100644
--- a/net/9p/trans_usbg.c
+++ b/net/9p/trans_usbg.c
@@ -242,6 +242,15 @@ static void usb9pfs_rx_complete(struct usb_ep *ep, struct usb_request *req)
 	if (!p9_rx_req)
 		return;
 
+	/* Validate actual received size against buffer capacity */
+	if (req->actual > p9_rx_req->rc.capacity) {
+		dev_err(&cdev->gadget->dev,
+			"received data size %u exceeds buffer capacity %zu\n",
+			req->actual, p9_rx_req->rc.capacity);
+		p9_req_put(usb9pfs->client, p9_rx_req);
+		return;
+	}
+
 	memcpy(p9_rx_req->rc.sdata, req->buf, req->actual);
 
 	p9_rx_req->rc.size = req->actual;
-- 
2.43.0

Re: [PATCH] net/9p: Fix buffer overflow in USB transport layer

[asmadeus]

Yuhao Jiang wrote on Mon, Jun 16, 2025 at 09:25:39PM +0800:
> A buffer overflow vulnerability exists in the USB 9pfs transport layer
> where inconsistent size validation between packet header parsing and
> actual data copying allows a malicious USB host to overflow heap buffers.
> 
> The issue occurs because:
> - usb9pfs_rx_header() validates only the declared size in packet header
> - usb9pfs_rx_complete() uses req->actual (actual received bytes) for memcpy
> 
> This allows an attacker to craft packets with small declared size (bypassing
> validation) but large actual payload (triggering overflow in memcpy).
> 
> Add validation in usb9pfs_rx_complete() to ensure req->actual does not
> exceed the buffer capacity before copying data.

Thanks for this check!

Did you reproduce this or was this static analysis found?
(to knowi if you tested wrt question below)

> Reported-by: Yuhao Jiang <danisjiang@gmail.com>
> Fixes: a3be076dc174 ("net/9p/usbg: Add new usb gadget function transport")
> Cc: stable@vger.kernel.org
> Signed-off-by: Yuhao Jiang <danisjiang@gmail.com>
> ---
>  net/9p/trans_usbg.c | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/net/9p/trans_usbg.c b/net/9p/trans_usbg.c
> index 6b694f117aef..047a2862fc84 100644
> --- a/net/9p/trans_usbg.c
> +++ b/net/9p/trans_usbg.c
> @@ -242,6 +242,15 @@ static void usb9pfs_rx_complete(struct usb_ep *ep, struct usb_request *req)
>  	if (!p9_rx_req)
>  		return;
>  
> +	/* Validate actual received size against buffer capacity */
> +	if (req->actual > p9_rx_req->rc.capacity) {
> +		dev_err(&cdev->gadget->dev,
> +			"received data size %u exceeds buffer capacity %zu\n",
> +			req->actual, p9_rx_req->rc.capacity);
> +		p9_req_put(usb9pfs->client, p9_rx_req);

I still haven't gotten around to setting up something to test this, and
even less the error case, but I'm not sure a single put is enough --
p9_client_cb does another put.
Conceptually I think it's better to mark the error and move on
e.g. (not even compile tested)
```
	int status = REQ_STATUS_RCVD;

	[...]

	if (req->actual > p9_rx_req->rc.capacity) {
		dev_err(...)
		req->actual = 0;
		status = REQ_STATUS_ERROR;
	}
	
	memcpy(..)

        p9_rx_req->rc.size = req->actual;

        p9_client_cb(usb9pfs->client, p9_rx_req, status);
        p9_req_put(usb9pfs->client, p9_rx_req);

	complete(&usb9pfs->received);
```
(I'm not sure overriding req->actual is allowed, might be safer to use
an intermediate variable like status instead)

What do you think?

Thanks,
-- 
Dominique Martinet | Asmadeus

Re: [PATCH] net/9p: Fix buffer overflow in USB transport layer

[Danis Jiang]

On Tue, Jun 17, 2025 at 7:00 AM <asmadeus@codewreck.org> wrote:
>
> Yuhao Jiang wrote on Mon, Jun 16, 2025 at 09:25:39PM +0800:
> > A buffer overflow vulnerability exists in the USB 9pfs transport layer
> > where inconsistent size validation between packet header parsing and
> > actual data copying allows a malicious USB host to overflow heap buffers.
> >
> > The issue occurs because:
> > - usb9pfs_rx_header() validates only the declared size in packet header
> > - usb9pfs_rx_complete() uses req->actual (actual received bytes) for memcpy
> >
> > This allows an attacker to craft packets with small declared size (bypassing
> > validation) but large actual payload (triggering overflow in memcpy).
> >
> > Add validation in usb9pfs_rx_complete() to ensure req->actual does not
> > exceed the buffer capacity before copying data.
>
> Thanks for this check!
>
> Did you reproduce this or was this static analysis found?
> (to knowi if you tested wrt question below)

I found this by static analysis.

>
> > Reported-by: Yuhao Jiang <danisjiang@gmail.com>
> > Fixes: a3be076dc174 ("net/9p/usbg: Add new usb gadget function transport")
> > Cc: stable@vger.kernel.org
> > Signed-off-by: Yuhao Jiang <danisjiang@gmail.com>
> > ---
> >  net/9p/trans_usbg.c | 9 +++++++++
> >  1 file changed, 9 insertions(+)
> >
> > diff --git a/net/9p/trans_usbg.c b/net/9p/trans_usbg.c
> > index 6b694f117aef..047a2862fc84 100644
> > --- a/net/9p/trans_usbg.c
> > +++ b/net/9p/trans_usbg.c
> > @@ -242,6 +242,15 @@ static void usb9pfs_rx_complete(struct usb_ep *ep, struct usb_request *req)
> >       if (!p9_rx_req)
> >               return;
> >
> > +     /* Validate actual received size against buffer capacity */
> > +     if (req->actual > p9_rx_req->rc.capacity) {
> > +             dev_err(&cdev->gadget->dev,
> > +                     "received data size %u exceeds buffer capacity %zu\n",
> > +                     req->actual, p9_rx_req->rc.capacity);
> > +             p9_req_put(usb9pfs->client, p9_rx_req);
>
> I still haven't gotten around to setting up something to test this, and
> even less the error case, but I'm not sure a single put is enough --
> p9_client_cb does another put.
> Conceptually I think it's better to mark the error and move on
> e.g. (not even compile tested)
> ```
>         int status = REQ_STATUS_RCVD;
>
>         [...]
>
>         if (req->actual > p9_rx_req->rc.capacity) {
>                 dev_err(...)
>                 req->actual = 0;
>                 status = REQ_STATUS_ERROR;
>         }
>
>         memcpy(..)
>
>         p9_rx_req->rc.size = req->actual;
>
>         p9_client_cb(usb9pfs->client, p9_rx_req, status);
>         p9_req_put(usb9pfs->client, p9_rx_req);
>
>         complete(&usb9pfs->received);
> ```
> (I'm not sure overriding req->actual is allowed, might be safer to use
> an intermediate variable like status instead)
>
> What do you think?
>
> Thanks,
> --
> Dominique Martinet | Asmadeus

Yes, I think your patch is better, my initial patch forgot p9_client_cb.

Thanks,
Yuhao Jiang

Re: [PATCH] net/9p: Fix buffer overflow in USB transport layer

[asmadeus]

Danis Jiang wrote on Tue, Jun 17, 2025 at 11:01:40AM +0800:
>>> Add validation in usb9pfs_rx_complete() to ensure req->actual does not
>>> exceed the buffer capacity before copying data.
>>
>> Thanks for this check!
>>
>> Did you reproduce this or was this static analysis found?
>> (to knowi if you tested wrt question below)
> 
> I found this by static analysis.

Ok.

>> I still haven't gotten around to setting up something to test this, and
>> even less the error case, but I'm not sure a single put is enough --
>> p9_client_cb does another put.
>> Conceptually I think it's better to mark the error and move on
>> e.g. (not even compile tested)
>> ```
>>         int status = REQ_STATUS_RCVD;
>>
>>         [...]
>>
>>         if (req->actual > p9_rx_req->rc.capacity) {
>>                 dev_err(...)
>>                 req->actual = 0;
>>                 status = REQ_STATUS_ERROR;
>>         }
>>
>>         memcpy(..)
>>
>>         p9_rx_req->rc.size = req->actual;
>>
>>         p9_client_cb(usb9pfs->client, p9_rx_req, status);
>>         p9_req_put(usb9pfs->client, p9_rx_req);
>>
>>         complete(&usb9pfs->received);
>> ```
>> (I'm not sure overriding req->actual is allowed, might be safer to use
>> an intermediate variable like status instead)
>>
>> What do you think?
> 
> Yes, I think your patch is better, my initial patch forgot p9_client_cb.

Ok, let's go with that then.

Would you like to resend "my" version, or should I do it (and
refer to your patch as Reported-by)?

Also if you resend let's add Mirsad Todorovac <mtodorovac69@gmail.com> too Ccs,
I've added him now.
(Mirsad, please check lore for full context if quote wasn't enough:
 https://lkml.kernel.org/r/20250616132539.63434-1-danisjiang@gmail.com
)


Thanks,
-- 
Dominique Martinet | Asmadeus

Re: [PATCH] net/9p: Fix buffer overflow in USB transport layer

[Danis Jiang]

On Tue, Jun 17, 2025 at 12:12 PM <asmadeus@codewreck.org> wrote:
>
> Danis Jiang wrote on Tue, Jun 17, 2025 at 11:01:40AM +0800:
> >>> Add validation in usb9pfs_rx_complete() to ensure req->actual does not
> >>> exceed the buffer capacity before copying data.
> >>
> >> Thanks for this check!
> >>
> >> Did you reproduce this or was this static analysis found?
> >> (to knowi if you tested wrt question below)
> >
> > I found this by static analysis.
>
> Ok.
>
> >> I still haven't gotten around to setting up something to test this, and
> >> even less the error case, but I'm not sure a single put is enough --
> >> p9_client_cb does another put.
> >> Conceptually I think it's better to mark the error and move on
> >> e.g. (not even compile tested)
> >> ```
> >>         int status = REQ_STATUS_RCVD;
> >>
> >>         [...]
> >>
> >>         if (req->actual > p9_rx_req->rc.capacity) {
> >>                 dev_err(...)
> >>                 req->actual = 0;
> >>                 status = REQ_STATUS_ERROR;
> >>         }
> >>
> >>         memcpy(..)
> >>
> >>         p9_rx_req->rc.size = req->actual;
> >>
> >>         p9_client_cb(usb9pfs->client, p9_rx_req, status);
> >>         p9_req_put(usb9pfs->client, p9_rx_req);
> >>
> >>         complete(&usb9pfs->received);
> >> ```
> >> (I'm not sure overriding req->actual is allowed, might be safer to use
> >> an intermediate variable like status instead)
> >>
> >> What do you think?
> >
> > Yes, I think your patch is better, my initial patch forgot p9_client_cb.
>
> Ok, let's go with that then.
>
> Would you like to resend "my" version, or should I do it (and
> refer to your patch as Reported-by)?
>
> Also if you resend let's add Mirsad Todorovac <mtodorovac69@gmail.com> too Ccs,
> I've added him now.
> (Mirsad, please check lore for full context if quote wasn't enough:
>  https://lkml.kernel.org/r/20250616132539.63434-1-danisjiang@gmail.com
> )
>
>
> Thanks,
> --
> Dominique Martinet | Asmadeus

Sure, you can do it and add me as Reported-by, thanks!