2021-07-11 14:18:05

by Len Baker

[permalink] [raw]
Subject: [PATCH] rtw88: Fix out-of-bounds write

In the rtw_pci_init_rx_ring function the "if (len > TRX_BD_IDX_MASK)"
statement guarantees that len is less than or equal to GENMASK(11, 0) or
in other words that len is less than or equal to 4095. However the
rx_ring->buf has a size of RTK_MAX_RX_DESC_NUM (defined as 512). This
way it is possible an out-of-bounds write in the for statement due to
the i variable can exceed the rx_ring->buff size.

Fix it using the ARRAY_SIZE macro.

Cc: [email protected]
Addresses-Coverity-ID: 1461515 ("Out-of-bounds write")
Fixes: e3037485c68ec ("rtw88: new Realtek 802.11ac driver")
Signed-off-by: Len Baker <[email protected]>
---
drivers/net/wireless/realtek/rtw88/pci.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/wireless/realtek/rtw88/pci.c b/drivers/net/wireless/realtek/rtw88/pci.c
index e7d17ab8f113..b9d8c049e776 100644
--- a/drivers/net/wireless/realtek/rtw88/pci.c
+++ b/drivers/net/wireless/realtek/rtw88/pci.c
@@ -280,7 +280,7 @@ static int rtw_pci_init_rx_ring(struct rtw_dev *rtwdev,
}
rx_ring->r.head = head;

- for (i = 0; i < len; i++) {
+ for (i = 0; i < ARRAY_SIZE(rx_ring->buf); i++) {
skb = dev_alloc_skb(buf_sz);
if (!skb) {
allocated = i;
--
2.25.1


2021-07-12 01:44:18

by Pkshih

[permalink] [raw]
Subject: RE: [PATCH] rtw88: Fix out-of-bounds write


> -----Original Message-----
> From: Len Baker [mailto:[email protected]]
> Sent: Sunday, July 11, 2021 10:17 PM
> To: Yan-Hsuan Chuang; Kalle Valo; David S. Miller; Jakub Kicinski
> Cc: Len Baker; Stanislaw Gruszka; Brian Norris; [email protected]; [email protected];
> [email protected]; [email protected]
> Subject: [PATCH] rtw88: Fix out-of-bounds write
>
> In the rtw_pci_init_rx_ring function the "if (len > TRX_BD_IDX_MASK)"
> statement guarantees that len is less than or equal to GENMASK(11, 0) or
> in other words that len is less than or equal to 4095. However the
> rx_ring->buf has a size of RTK_MAX_RX_DESC_NUM (defined as 512). This
> way it is possible an out-of-bounds write in the for statement due to
> the i variable can exceed the rx_ring->buff size.
>
> Fix it using the ARRAY_SIZE macro.
>
> Cc: [email protected]
> Addresses-Coverity-ID: 1461515 ("Out-of-bounds write")
> Fixes: e3037485c68ec ("rtw88: new Realtek 802.11ac driver")
> Signed-off-by: Len Baker <[email protected]>
> ---
> drivers/net/wireless/realtek/rtw88/pci.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/net/wireless/realtek/rtw88/pci.c b/drivers/net/wireless/realtek/rtw88/pci.c
> index e7d17ab8f113..b9d8c049e776 100644
> --- a/drivers/net/wireless/realtek/rtw88/pci.c
> +++ b/drivers/net/wireless/realtek/rtw88/pci.c
> @@ -280,7 +280,7 @@ static int rtw_pci_init_rx_ring(struct rtw_dev *rtwdev,

I think "if (len > TRX_BD_IDX_MASK)" you mentioned is

if (len > TRX_BD_IDX_MASK) {
rtw_err(rtwdev, "len %d exceeds maximum RX entries\n", len);
return -EINVAL;
}

This statement is used to ensure the length doesn't exceed hardware capability.

To prevent the 'len' argument from exceeding the array size of rx_ring->buff, I
suggest to add another checking statement, like

if (len > ARRAY_SIZE(rx_ring->buf)) {
rtw_err(rtwdev, "len %d exceeds maximum RX ring buffer\n", len);
return -EINVAL;
}

But, I wonder if this a false alarm because 'len' is equal to ARRAY_SIZE(rx_ring->buf)
for now.


> }
> rx_ring->r.head = head;
>
> - for (i = 0; i < len; i++) {
> + for (i = 0; i < ARRAY_SIZE(rx_ring->buf); i++) {
> skb = dev_alloc_skb(buf_sz);
> if (!skb) {
> allocated = i;
> --
> 2.25.1

--
Ping-Ke

2021-07-16 15:09:53

by Len Baker

[permalink] [raw]
Subject: Re: [PATCH] rtw88: Fix out-of-bounds write

On Mon, Jul 12, 2021 at 11:38:43AM -0700, Brian Norris wrote:
> On Sun, Jul 11, 2021 at 6:43 PM Pkshih <[email protected]> wrote:
> > > -----Original Message-----
> > > From: Len Baker [mailto:[email protected]]
> > >
> > > In the rtw_pci_init_rx_ring function the "if (len > TRX_BD_IDX_MASK)"
> > > statement guarantees that len is less than or equal to GENMASK(11, 0) or
> > > in other words that len is less than or equal to 4095. However the
> > > rx_ring->buf has a size of RTK_MAX_RX_DESC_NUM (defined as 512). This
> > > way it is possible an out-of-bounds write in the for statement due to
> > > the i variable can exceed the rx_ring->buff size.
> > >
> > > Fix it using the ARRAY_SIZE macro.
> > >
> > > Cc: [email protected]
> > > Addresses-Coverity-ID: 1461515 ("Out-of-bounds write")
>
> Coverity seems to be giving a false warning here. I presume it's
> taking the |len| comparison as proof that |len| might be as large as
> TRX_BD_IDX_MASK, but as noted below, that's not really true; the |len|
> comparison is really just dead code.

I agree.

> > > Fixes: e3037485c68ec ("rtw88: new Realtek 802.11ac driver")
> > > Signed-off-by: Len Baker <[email protected]>
>
> > To prevent the 'len' argument from exceeding the array size of rx_ring->buff, I
> > suggest to add another checking statement, like
> >
> > if (len > ARRAY_SIZE(rx_ring->buf)) {
> > rtw_err(rtwdev, "len %d exceeds maximum RX ring buffer\n", len);
> > return -EINVAL;
> > }
>
> That seems like a better idea, if we really need to patch anything.

I think it is reasonable to protect any potencial overflow (for example, if
this function is used in the future with a parameter greater than 512). It
is better to be defensive in this case :)

> > But, I wonder if this a false alarm because 'len' is equal to ARRAY_SIZE(rx_ring->buf)
> > for now.
>
> Or to the point: rtw_pci_init_rx_ring() is only ever called with a
> fixed constant -- RTK_MAX_RX_DESC_NUM (i.e., 512) -- so the alleged
> overflow cannot happen.
>
> Brian

I will send a v2 for review.

Thanks,
Len