linux-kernel - Re: [PATCH V3] crypto: Mark intermediary memory as clean

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 for Android: free password hash cracker in your pocket

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

Message-ID: <2401599.22IY78Rhhi@tauon>
Date: Mon, 18 Aug 2025 15:32:43 +0200
From: Stephan Mueller <smueller@...onox.de>
To: Edward Adam Davis <eadavis@...com>
Cc: davem@...emloft.net, eadavis@...com, herbert@...dor.apana.org.au,
 linux-crypto@...r.kernel.org, linux-kernel@...r.kernel.org,
 syzbot+e8bcd7ee3db6cb5cb875@...kaller.appspotmail.com,
 syzkaller-bugs@...glegroups.com
Subject: Re: [PATCH V3] crypto: Mark intermediary memory as clean

Am Montag, 18. August 2025, 15:24:17 Mitteleuropäische Sommerzeit schrieb 
Edward Adam Davis:

Hi Edward,

> This is not a leak! The stack memroy is hashed and fed into the
> entropy pool. We can't recover the original kernel memory from it.
> 
> Reported-by: syzbot+e8bcd7ee3db6cb5cb875@...kaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=e8bcd7ee3db6cb5cb875
> Signed-off-by: Edward Adam Davis <eadavis@...com>

Thank you for the patch. Just for the records:

- the intermediary buffer could be initialized to 0 without any effect on the 
Jitter RNG, because all it wants is actually the execution of the Keccak 
operation as part of crypto_shhash_finup.

- the intermediary buffer is inserted into the Jitter RNG state to ensure that 
the compiler cannot optimize away the loop if the intermediary buffer would 
not be used at all

- the intermediary buffer is not credited with any entropy as we only want the 
Keccak operation

- by keeping the intermediary uninitialized, the Jitter RNG may get some 
variations from the uninitialized buffer so that its internal state may 
benefit from it

That said, I am fine with this current patch. But if there is still lingering 
concern, I am equally fine to have it initialized to zero.

Thanks a lot
Stephan

> ---
> V1 -> V2: mark it as unpoison
> V2 -> V3: replace to sizeof, minimize the possibilities where
> inconsistencies can occur
> 
>  crypto/jitterentropy-kcapi.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/crypto/jitterentropy-kcapi.c b/crypto/jitterentropy-kcapi.c
> index 1266eb790708..4020a6e41b0e 100644
> --- a/crypto/jitterentropy-kcapi.c
> +++ b/crypto/jitterentropy-kcapi.c
> @@ -117,6 +117,7 @@ int jent_hash_time(void *hash_state, __u64 time, u8
> *addtl, pr_warn_ratelimited("Unexpected digest size\n");
>  		return -EINVAL;
>  	}
> +	kmsan_unpoison_memory(intermediary, sizeof(intermediary));
> 
>  	/*
>  	 * This loop fills a buffer which is injected into the entropy pool.


Ciao
Stephan