Re: [PATCH] mac80211: aes-cmac: remove VLA usage

[Gustavo A. R. Silva]

On 03/21/2018 08:58 AM, Johannes Berg wrote:

On Wed, 2018-03-21 at 08:57 -0500, Gustavo A. R. Silva wrote:


SHA_DESC_ON_STACK is currently being used in multiple places. But, yeah,
I think we can define multiple macros of the same kind and adjust to the
characteristics of each the component.

How big do you think tfm can get?


I have no idea, I guess you'll have to take that with Herbert.

johannes



I see. I'll contact him then.

Thanks for the feedback.
--
Gustavo

Re: [PATCH] mac80211: aes-cmac: remove VLA usage

[Gustavo A. R. Silva]

On 03/21/2018 08:48 AM, Johannes Berg wrote:

On Wed, 2018-03-21 at 08:42 -0500, Gustavo A. R. Silva wrote:

In preparation to enabling -Wvla, remove VLAs and replace them
with dynamic memory allocation instead.

The use of stack Variable Length Arrays needs to be avoided, as they
can be a vector for stack exhaustion, which can be both a runtime bug
or a security flaw. Also, in general, as code evolves it is easy to
lose track of how big a VLA can get. Thus, we can end up having runtime
failures that are hard to debug.

Also, fixed as part of the directive to remove all VLAs from
the kernel: https://lkml.org/lkml/2018/3/7/621

Signed-off-by: Gustavo A. R. Silva 
---
  net/mac80211/aes_cmac.c | 36 
  1 file changed, 24 insertions(+), 12 deletions(-)

diff --git a/net/mac80211/aes_cmac.c b/net/mac80211/aes_cmac.c
index 2fb6558..c9444bf 100644
--- a/net/mac80211/aes_cmac.c
+++ b/net/mac80211/aes_cmac.c
@@ -27,30 +27,42 @@ static const u8 zero[CMAC_TLEN_256];
  void ieee80211_aes_cmac(struct crypto_shash *tfm, const u8 *aad,
const u8 *data, size_t data_len, u8 *mic)
  {
-   SHASH_DESC_ON_STACK(desc, tfm);
+   struct shash_desc *shash;
u8 out[AES_BLOCK_SIZE];
  
-	desc->tfm = tfm;

+   shash = kmalloc(sizeof(*shash) + crypto_shash_descsize(tfm),
+   GFP_KERNEL);
+   if (!shash)
+   return;


Honestly, this seems like a really bad idea - you're now hitting
kmalloc for every TX/RX frame here.

SHA_DESC_ON_STACK() should just be fixed to not need a VLA, but take
some sort of maximum, I guess?



SHA_DESC_ON_STACK is currently being used in multiple places. But, yeah, 
I think we can define multiple macros of the same kind and adjust to the 
characteristics of each the component.


How big do you think tfm can get?

Thanks
--
Gustavo

Re: [PATCH] mac80211: aes-cmac: remove VLA usage

[Johannes Berg]

On Wed, 2018-03-21 at 08:57 -0500, Gustavo A. R. Silva wrote:
> 
> SHA_DESC_ON_STACK is currently being used in multiple places. But, yeah, 
> I think we can define multiple macros of the same kind and adjust to the 
> characteristics of each the component.
> 
> How big do you think tfm can get?

I have no idea, I guess you'll have to take that with Herbert.

johannes

Re: [PATCH] mac80211: aes-cmac: remove VLA usage

[Johannes Berg]

On Wed, 2018-03-21 at 08:42 -0500, Gustavo A. R. Silva wrote:
> In preparation to enabling -Wvla, remove VLAs and replace them
> with dynamic memory allocation instead.
> 
> The use of stack Variable Length Arrays needs to be avoided, as they
> can be a vector for stack exhaustion, which can be both a runtime bug
> or a security flaw. Also, in general, as code evolves it is easy to
> lose track of how big a VLA can get. Thus, we can end up having runtime
> failures that are hard to debug.
> 
> Also, fixed as part of the directive to remove all VLAs from
> the kernel: https://lkml.org/lkml/2018/3/7/621
> 
> Signed-off-by: Gustavo A. R. Silva 
> ---
>  net/mac80211/aes_cmac.c | 36 
>  1 file changed, 24 insertions(+), 12 deletions(-)
> 
> diff --git a/net/mac80211/aes_cmac.c b/net/mac80211/aes_cmac.c
> index 2fb6558..c9444bf 100644
> --- a/net/mac80211/aes_cmac.c
> +++ b/net/mac80211/aes_cmac.c
> @@ -27,30 +27,42 @@ static const u8 zero[CMAC_TLEN_256];
>  void ieee80211_aes_cmac(struct crypto_shash *tfm, const u8 *aad,
>   const u8 *data, size_t data_len, u8 *mic)
>  {
> - SHASH_DESC_ON_STACK(desc, tfm);
> + struct shash_desc *shash;
>   u8 out[AES_BLOCK_SIZE];
>  
> - desc->tfm = tfm;
> + shash = kmalloc(sizeof(*shash) + crypto_shash_descsize(tfm),
> + GFP_KERNEL);
> + if (!shash)
> + return;

Honestly, this seems like a really bad idea - you're now hitting
kmalloc for every TX/RX frame here.

SHA_DESC_ON_STACK() should just be fixed to not need a VLA, but take
some sort of maximum, I guess?

johannes