[PATCH] wifi: mac80211_hwsim: drop short frames

[PATCH] wifi: mac80211_hwsim: drop short frames

[Johannes Berg]

From: Johannes Berg <johannes.berg@intel.com>

Frames that don't even have addr1 are clearly not valid,
drop those early in the netlink/wmediumd path here.

Reported-by: syzbot+b2645b5bf1512b81fa22@syzkaller.appspotmail.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
---
 drivers/net/wireless/virtual/mac80211_hwsim.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/drivers/net/wireless/virtual/mac80211_hwsim.c b/drivers/net/wireless/virtual/mac80211_hwsim.c
index f446fd0e8cd0..8f1e4420ed1e 100644
--- a/drivers/net/wireless/virtual/mac80211_hwsim.c
+++ b/drivers/net/wireless/virtual/mac80211_hwsim.c
@@ -5626,14 +5626,15 @@ static int hwsim_cloned_frame_received_nl(struct sk_buff *skb_2,
 	frame_data_len = nla_len(info->attrs[HWSIM_ATTR_FRAME]);
 	frame_data = (void *)nla_data(info->attrs[HWSIM_ATTR_FRAME]);
 
+	if (frame_data_len < offsetofend(typeof(*hdr), addr1) ||
+	    frame_data_len > IEEE80211_MAX_DATA_LEN)
+		goto err;
+
 	/* Allocate new skb here */
 	skb = alloc_skb(frame_data_len, GFP_KERNEL);
 	if (skb == NULL)
 		goto err;
 
-	if (frame_data_len > IEEE80211_MAX_DATA_LEN)
-		goto err;
-
 	/* Copy the data */
 	skb_put_data(skb, frame_data, frame_data_len);
 
-- 
2.41.0

Re: [PATCH] wifi: mac80211_hwsim: drop short frames

[Jeff Johnson]

On 8/15/2023 9:16 AM, Johannes Berg wrote:
> From: Johannes Berg <johannes.berg@intel.com>
> 
> Frames that don't even have addr1 are clearly not valid,
> drop those early in the netlink/wmediumd path here.
> 
> Reported-by: syzbot+b2645b5bf1512b81fa22@syzkaller.appspotmail.com
> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
> ---
>   drivers/net/wireless/virtual/mac80211_hwsim.c | 7 ++++---
>   1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/net/wireless/virtual/mac80211_hwsim.c b/drivers/net/wireless/virtual/mac80211_hwsim.c
> index f446fd0e8cd0..8f1e4420ed1e 100644
> --- a/drivers/net/wireless/virtual/mac80211_hwsim.c
> +++ b/drivers/net/wireless/virtual/mac80211_hwsim.c
> @@ -5626,14 +5626,15 @@ static int hwsim_cloned_frame_received_nl(struct sk_buff *skb_2,
>   	frame_data_len = nla_len(info->attrs[HWSIM_ATTR_FRAME]);
>   	frame_data = (void *)nla_data(info->attrs[HWSIM_ATTR_FRAME]);
>   
> +	if (frame_data_len < offsetofend(typeof(*hdr), addr1) ||

curious why addr1. if the frame ends after addr1 then don't you have a 
problem in mac80211_hwsim_rx() when it passes hdr->addr2 to 
ieee80211_find_sta_by_link_addrs()?


> +	    frame_data_len > IEEE80211_MAX_DATA_LEN)
> +		goto err;
> +
>   	/* Allocate new skb here */
>   	skb = alloc_skb(frame_data_len, GFP_KERNEL);
>   	if (skb == NULL)
>   		goto err;
>   
> -	if (frame_data_len > IEEE80211_MAX_DATA_LEN)
> -		goto err;
> -
>   	/* Copy the data */
>   	skb_put_data(skb, frame_data, frame_data_len);
>   

Re: [PATCH] wifi: mac80211_hwsim: drop short frames

[Johannes Berg]

On Tue, 2023-08-15 at 11:20 -0700, Jeff Johnson wrote:
> On 8/15/2023 9:16 AM, Johannes Berg wrote:
> > From: Johannes Berg <johannes.berg@intel.com>
> > 
> > Frames that don't even have addr1 are clearly not valid,
> > drop those early in the netlink/wmediumd path here.
> > 
> > Reported-by: syzbot+b2645b5bf1512b81fa22@syzkaller.appspotmail.com
> > Signed-off-by: Johannes Berg <johannes.berg@intel.com>
> > ---
> >   drivers/net/wireless/virtual/mac80211_hwsim.c | 7 ++++---
> >   1 file changed, 4 insertions(+), 3 deletions(-)
> > 
> > diff --git a/drivers/net/wireless/virtual/mac80211_hwsim.c b/drivers/net/wireless/virtual/mac80211_hwsim.c
> > index f446fd0e8cd0..8f1e4420ed1e 100644
> > --- a/drivers/net/wireless/virtual/mac80211_hwsim.c
> > +++ b/drivers/net/wireless/virtual/mac80211_hwsim.c
> > @@ -5626,14 +5626,15 @@ static int hwsim_cloned_frame_received_nl(struct sk_buff *skb_2,
> >   	frame_data_len = nla_len(info->attrs[HWSIM_ATTR_FRAME]);
> >   	frame_data = (void *)nla_data(info->attrs[HWSIM_ATTR_FRAME]);
> >   
> > +	if (frame_data_len < offsetofend(typeof(*hdr), addr1) ||
> 
> curious why addr1.
> 

I figured that was the shortest legal 802.11 frame, e.g. an ACK frame.

> if the frame ends after addr1 then don't you have a 
> problem in mac80211_hwsim_rx() when it passes hdr->addr2 to 
> ieee80211_find_sta_by_link_addrs()?

But of course you're right about that, so we should just make sure we
have at least the full struct ieee80211_hdr_3addr here, I guess.

johannes