linux-kernel - [RESEND] SHASH_DESC_ON

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 23 Mar 2018 14:09:46 -0500
From:   "Gustavo A. R. Silva" <gustavo@...eddedor.com>
To:     Herbert Xu <herbert@...dor.apana.org.au>
Cc:     "David S. Miller" <davem@...emloft.net>,
        linux-crypto@...r.kernel.org, linux-kernel@...r.kernel.org,
        "Gustavo A. R. Silva" <gustavo@...eddedor.com>
Subject: [RESEND] SHASH_DESC_ON_STACK macro

Hi Herbert,

There is an ongoing effort to remove all VLAs from the code base [1] and 
while working on that I came across the following macro at 
include/crypto/hash.h:154:

#define SHASH_DESC_ON_STACK(shash, ctx)                           \
         char __##shash##_desc[sizeof(struct shash_desc) +         \
                 crypto_shash_descsize(ctx)] CRYPTO_MINALIGN_ATTR; \
         struct shash_desc *shash = (struct shash_desc *)__##shash##_desc

Currently, this macro is being used in 46 different places.

I wonder how big can tfm->descsize can get?

Do you think it is feasible to replace the call to crypto_shash_descsize 
with a constant value and get rid of 46 VLA warnings?

I have sent some patches to replace the use of this macro with dynamic 
memory allocation instead, but it seems that this is not a suitable 
solution for all cases due to performance issues.

[1] https://lkml.org/lkml/2018/3/7/621

I'd really appreciate any feedback.
Thanks
--
Gustavo