[PATCH 1/1] mwifiex: Fix possible buffer overflows in mwifiex_config_scan

[PATCH 1/1] mwifiex: Fix possible buffer overflows in mwifiex_config_scan

[Xiaohui Zhang]

From: Zhang Xiaohui <ruc_zhangxiaohui@163.com>

mwifiex_config_scan() calls memcpy() without checking
the destination size may trigger a buffer overflower,
which a local user could use to cause denial of service
or the execution of arbitrary code.
Fix it by putting the length check before calling memcpy().

Signed-off-by: Zhang Xiaohui <ruc_zhangxiaohui@163.com>
---
 drivers/net/wireless/marvell/mwifiex/scan.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/wireless/marvell/mwifiex/scan.c b/drivers/net/wireless/marvell/mwifiex/scan.c
index c2a685f63..b1d90678f 100644
--- a/drivers/net/wireless/marvell/mwifiex/scan.c
+++ b/drivers/net/wireless/marvell/mwifiex/scan.c
@@ -930,6 +930,8 @@ mwifiex_config_scan(struct mwifiex_private *priv,
 				    "DIRECT-", 7))
 				wildcard_ssid_tlv->max_ssid_length = 0xfe;
 
+			if (ssid_len > 1)
+				ssid_len = 1;
 			memcpy(wildcard_ssid_tlv->ssid,
 			       user_scan_in->ssid_list[i].ssid, ssid_len);
 
-- 
2.17.1

Re: [PATCH 1/1] mwifiex: Fix possible buffer overflows in mwifiex_config_scan

[Kalle Valo]

Xiaohui Zhang <ruc_zhangxiaohui@163.com> writes:

> From: Zhang Xiaohui <ruc_zhangxiaohui@163.com>
>
> mwifiex_config_scan() calls memcpy() without checking
> the destination size may trigger a buffer overflower,
> which a local user could use to cause denial of service
> or the execution of arbitrary code.
> Fix it by putting the length check before calling memcpy().
>
> Signed-off-by: Zhang Xiaohui <ruc_zhangxiaohui@163.com>
> ---
>  drivers/net/wireless/marvell/mwifiex/scan.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/drivers/net/wireless/marvell/mwifiex/scan.c b/drivers/net/wireless/marvell/mwifiex/scan.c
> index c2a685f63..b1d90678f 100644
> --- a/drivers/net/wireless/marvell/mwifiex/scan.c
> +++ b/drivers/net/wireless/marvell/mwifiex/scan.c
> @@ -930,6 +930,8 @@ mwifiex_config_scan(struct mwifiex_private *priv,
>  				    "DIRECT-", 7))
>  				wildcard_ssid_tlv->max_ssid_length = 0xfe;
>  
> +			if (ssid_len > 1)
> +				ssid_len = 1;
>  			memcpy(wildcard_ssid_tlv->ssid,
>  			       user_scan_in->ssid_list[i].ssid, ssid_len);

min_t()?

-- 
https://patchwork.kernel.org/project/linux-wireless/list/

https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches

Re: [PATCH 1/1] mwifiex: Fix possible buffer overflows in mwifiex_config_scan

[Julian Calaby]

Hi Xiaohui,

On Wed, Dec 9, 2020 at 12:07 AM Xiaohui Zhang <ruc_zhangxiaohui@163.com> wrote:
>
> From: Zhang Xiaohui <ruc_zhangxiaohui@163.com>
>
> mwifiex_config_scan() calls memcpy() without checking
> the destination size may trigger a buffer overflower,
> which a local user could use to cause denial of service
> or the execution of arbitrary code.
> Fix it by putting the length check before calling memcpy().
>
> Signed-off-by: Zhang Xiaohui <ruc_zhangxiaohui@163.com>
> ---
>  drivers/net/wireless/marvell/mwifiex/scan.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/drivers/net/wireless/marvell/mwifiex/scan.c b/drivers/net/wireless/marvell/mwifiex/scan.c
> index c2a685f63..b1d90678f 100644
> --- a/drivers/net/wireless/marvell/mwifiex/scan.c
> +++ b/drivers/net/wireless/marvell/mwifiex/scan.c
> @@ -930,6 +930,8 @@ mwifiex_config_scan(struct mwifiex_private *priv,
>                                     "DIRECT-", 7))
>                                 wildcard_ssid_tlv->max_ssid_length = 0xfe;
>
> +                       if (ssid_len > 1)
> +                               ssid_len = 1;
>                         memcpy(wildcard_ssid_tlv->ssid,
>                                user_scan_in->ssid_list[i].ssid, ssid_len);

Can ssid_len ever be 0 here?

If it can't, should we just set ssid_len to 1 unconditionally?

If it can, should we just skip the memcpy as it won't do anything?

Thanks,

-- 
Julian Calaby

Email: julian.calaby@gmail.com
Profile: http://www.google.com/profiles/julian.calaby/

Re: [PATCH 1/1] mwifiex: Fix possible buffer overflows in mwifiex_config_scan

[Peter Seiderer]

Hello Zhang,

On Tue,  8 Dec 2020 20:45:23 +0800, Xiaohui Zhang <ruc_zhangxiaohui@163.com> wrote:

> From: Zhang Xiaohui <ruc_zhangxiaohui@163.com>
>
> mwifiex_config_scan() calls memcpy() without checking
> the destination size may trigger a buffer overflower,
> which a local user could use to cause denial of service
> or the execution of arbitrary code.
> Fix it by putting the length check before calling memcpy().
>
> Signed-off-by: Zhang Xiaohui <ruc_zhangxiaohui@163.com>
> ---
>  drivers/net/wireless/marvell/mwifiex/scan.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/drivers/net/wireless/marvell/mwifiex/scan.c b/drivers/net/wireless/marvell/mwifiex/scan.c
> index c2a685f63..b1d90678f 100644
> --- a/drivers/net/wireless/marvell/mwifiex/scan.c
> +++ b/drivers/net/wireless/marvell/mwifiex/scan.c
> @@ -930,6 +930,8 @@ mwifiex_config_scan(struct mwifiex_private *priv,
>  				    "DIRECT-", 7))
>  				wildcard_ssid_tlv->max_ssid_length = 0xfe;
>
> +			if (ssid_len > 1)
> +				ssid_len = 1;

Why do your believe the available size is only '1'? A SSID is expected
to be of size IEEE80211_MAX_SSID_LE/32 and the wildcard_ssid_tlv pointer
is casted from tlv_pos (some lines above) which is a pointer/index into
scan_cfg_out->tlv_buf...

And the define (line 44) indicates there should be enough space for a SSID:

  42 /* Memory needed to store a max number/size WildCard SSID TLV for a firmware
  43         scan */
  44 #define WILDCARD_SSID_TLV_MAX_SIZE  \
  45         (MWIFIEX_MAX_SSID_LIST_LENGTH *                                 \
  46                 (sizeof(struct mwifiex_ie_types_wildcard_ssid_params)   \
  47                         + IEEE80211_MAX_SSID_LEN))

For sure something to improve here instead of using a confusing 'u8 ssid[1]'
in struct mwifiex_ie_types_wildcard_ssid_params...

Regards,
Peter

>  			memcpy(wildcard_ssid_tlv->ssid,
>  			       user_scan_in->ssid_list[i].ssid, ssid_len);
>

Re: [PATCH 1/1] mwifiex: Fix possible buffer overflows in mwifiex_config_scan

[Brian Norris]

(Note: this is version 1; there's a later version posted, which does
not have a v2 tag...)

https://lore.kernel.org/linux-wireless/20201208150951.35866-1-ruc_zhangxiaohui@163.com/

On Sat, Jan 9, 2021 at 7:11 AM Peter Seiderer <ps.report@gmx.net> wrote:
> On Tue,  8 Dec 2020 20:45:23 +0800, Xiaohui Zhang <ruc_zhangxiaohui@163.com> wrote:
> > From: Zhang Xiaohui <ruc_zhangxiaohui@163.com>
> > mwifiex_config_scan() calls memcpy() without checking
> > the destination size may trigger a buffer overflower,
> > which a local user could use to cause denial of service
> > or the execution of arbitrary code.
> > Fix it by putting the length check before calling memcpy().
> >
> > Signed-off-by: Zhang Xiaohui <ruc_zhangxiaohui@163.com>
> > ---
> >  drivers/net/wireless/marvell/mwifiex/scan.c | 2 ++
> >  1 file changed, 2 insertions(+)
> >
> > diff --git a/drivers/net/wireless/marvell/mwifiex/scan.c b/drivers/net/wireless/marvell/mwifiex/scan.c
> > index c2a685f63..b1d90678f 100644
> > --- a/drivers/net/wireless/marvell/mwifiex/scan.c
> > +++ b/drivers/net/wireless/marvell/mwifiex/scan.c
> > @@ -930,6 +930,8 @@ mwifiex_config_scan(struct mwifiex_private *priv,
> >                                   "DIRECT-", 7))
> >                               wildcard_ssid_tlv->max_ssid_length = 0xfe;
> >
> > +                     if (ssid_len > 1)
> > +                             ssid_len = 1;
>
> Why do your believe the available size is only '1'? A SSID is expected
> to be of size IEEE80211_MAX_SSID_LE/32 and the wildcard_ssid_tlv pointer
> is casted from tlv_pos (some lines above) which is a pointer/index into
> scan_cfg_out->tlv_buf...

I pointed out this discrepancy already, taking a slightly different approach:

https://lore.kernel.org/linux-wireless/CA+ASDXPVu5S0Vm0aOcyqLN090u3BwA_nV358YwkpXuU223Ug9g@mail.gmail.com/

> And the define (line 44) indicates there should be enough space for a SSID:
>
>   42 /* Memory needed to store a max number/size WildCard SSID TLV for a firmware
>   43         scan */
>   44 #define WILDCARD_SSID_TLV_MAX_SIZE  \
>   45         (MWIFIEX_MAX_SSID_LIST_LENGTH *                                 \
>   46                 (sizeof(struct mwifiex_ie_types_wildcard_ssid_params)   \
>   47                         + IEEE80211_MAX_SSID_LEN))

Ah, good catch. So this may not be a true overflow issue at all, even
if it's confusing and bad code. The "problem" is that this piece of
the struct is variable-length, and unless we're going to dynamically
resize/reallocate the whole buffer, we have to assume the initial
allocation was large enough (and per your note, it is).

> For sure something to improve here instead of using a confusing 'u8 ssid[1]'
> in struct mwifiex_ie_types_wildcard_ssid_params...

Yep.

Brian

[PATCH 1/1] mwifiex: Fix possible buffer overflows in mwifiex_config_scan

[Xiaohui Zhang]

From: Zhang Xiaohui <ruc_zhangxiaohui@163.com>

mwifiex_config_scan() calls memcpy() without checking
the destination size may trigger a buffer overflower,
which a local user could use to cause denial of service
or the execution of arbitrary code.
Fix it by putting the length check before calling memcpy().

Signed-off-by: Zhang Xiaohui <ruc_zhangxiaohui@163.com>
---
 drivers/net/wireless/marvell/mwifiex/scan.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/wireless/marvell/mwifiex/scan.c b/drivers/net/wireless/marvell/mwifiex/scan.c
index c2a685f63..34293fd80 100644
--- a/drivers/net/wireless/marvell/mwifiex/scan.c
+++ b/drivers/net/wireless/marvell/mwifiex/scan.c
@@ -931,7 +931,7 @@ mwifiex_config_scan(struct mwifiex_private *priv,
 				wildcard_ssid_tlv->max_ssid_length = 0xfe;
 
 			memcpy(wildcard_ssid_tlv->ssid,
-			       user_scan_in->ssid_list[i].ssid, ssid_len);
+			       user_scan_in->ssid_list[i].ssid, min_t(u32, ssid_len, 1));
 
 			tlv_pos += (sizeof(wildcard_ssid_tlv->header)
 				+ le16_to_cpu(wildcard_ssid_tlv->header.len));
-- 
2.17.1

Re: [PATCH 1/1] mwifiex: Fix possible buffer overflows in mwifiex_config_scan

[Brian Norris]

On Tue, Dec 8, 2020 at 7:14 AM Xiaohui Zhang <ruc_zhangxiaohui@163.com> wrote:
>
> From: Zhang Xiaohui <ruc_zhangxiaohui@163.com>
>
> mwifiex_config_scan() calls memcpy() without checking
> the destination size may trigger a buffer overflower,
> which a local user could use to cause denial of service
> or the execution of arbitrary code.
> Fix it by putting the length check before calling memcpy().

^^ That's not really what you're doing any more, for the record. But
then, describing "what" is not really the point of a commit message
(that's what the code is for), so maybe that's not that important.

> Signed-off-by: Zhang Xiaohui <ruc_zhangxiaohui@163.com>
> ---
>  drivers/net/wireless/marvell/mwifiex/scan.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/net/wireless/marvell/mwifiex/scan.c b/drivers/net/wireless/marvell/mwifiex/scan.c
> index c2a685f63..34293fd80 100644
> --- a/drivers/net/wireless/marvell/mwifiex/scan.c
> +++ b/drivers/net/wireless/marvell/mwifiex/scan.c
> @@ -931,7 +931,7 @@ mwifiex_config_scan(struct mwifiex_private *priv,
>                                 wildcard_ssid_tlv->max_ssid_length = 0xfe;
>
>                         memcpy(wildcard_ssid_tlv->ssid,
> -                              user_scan_in->ssid_list[i].ssid, ssid_len);
> +                              user_scan_in->ssid_list[i].ssid, min_t(u32, ssid_len, 1));

This *looks* like it should be wrong, because SSIDs are clearly longer
than 1 byte in many cases, but you *are* right that this is what the
struct is defined as:

struct mwifiex_ie_types_wildcard_ssid_params {
...
    u8 ssid[1];
};

This feels like something that could use some confirmation from
NXP/ex-Marvell folks if possible, but if not that, at least some
creative testing. Did you actually test this patch, to make sure
non-wildcard scans still work?

Also, even if this is correct, it seems like it would be more correct
to use 'sizeof(wildcard_ssid_tlv->ssid)' instead of a magic number 1.

Brian

>
>                         tlv_pos += (sizeof(wildcard_ssid_tlv->header)
>                                 + le16_to_cpu(wildcard_ssid_tlv->header.len));
> --
> 2.17.1
>

Re: [PATCH 1/1] mwifiex: Fix possible buffer overflows in mwifiex_config_scan

[Kalle Valo]

Brian Norris <briannorris@chromium.org> writes:

> On Tue, Dec 8, 2020 at 7:14 AM Xiaohui Zhang <ruc_zhangxiaohui@163.com> wrote:
>>
>> From: Zhang Xiaohui <ruc_zhangxiaohui@163.com>
>>
>> mwifiex_config_scan() calls memcpy() without checking
>> the destination size may trigger a buffer overflower,
>> which a local user could use to cause denial of service
>> or the execution of arbitrary code.
>> Fix it by putting the length check before calling memcpy().
>
> ^^ That's not really what you're doing any more, for the record. But
> then, describing "what" is not really the point of a commit message
> (that's what the code is for), so maybe that's not that important.
>
>> Signed-off-by: Zhang Xiaohui <ruc_zhangxiaohui@163.com>
>> ---
>>  drivers/net/wireless/marvell/mwifiex/scan.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/net/wireless/marvell/mwifiex/scan.c b/drivers/net/wireless/marvell/mwifiex/scan.c
>> index c2a685f63..34293fd80 100644
>> --- a/drivers/net/wireless/marvell/mwifiex/scan.c
>> +++ b/drivers/net/wireless/marvell/mwifiex/scan.c
>> @@ -931,7 +931,7 @@ mwifiex_config_scan(struct mwifiex_private *priv,
>>                                 wildcard_ssid_tlv->max_ssid_length = 0xfe;
>>
>>                         memcpy(wildcard_ssid_tlv->ssid,
>> -                              user_scan_in->ssid_list[i].ssid, ssid_len);
>> +                              user_scan_in->ssid_list[i].ssid, min_t(u32, ssid_len, 1));
>
> This *looks* like it should be wrong, because SSIDs are clearly longer
> than 1 byte in many cases, but you *are* right that this is what the
> struct is defined as:
>
> struct mwifiex_ie_types_wildcard_ssid_params {
> ...
>     u8 ssid[1];
> };
>
> This feels like something that could use some confirmation from
> NXP/ex-Marvell folks if possible, but if not that, at least some
> creative testing. Did you actually test this patch, to make sure
> non-wildcard scans still work?
>
> Also, even if this is correct, it seems like it would be more correct
> to use 'sizeof(wildcard_ssid_tlv->ssid)' instead of a magic number 1.

Xiaohui, please respond to Brian's comments. If you ignore review
comments I have a hard time trusting your patches.

Also when you submit a new version you should mark it as v2. See more in
the wiki link below.

-- 
https://patchwork.kernel.org/project/linux-wireless/list/

https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches