Re: [PATCH v2 1/2] KEYS: fix signature length passed to KEYCTL_PKEY_VERIFY
From: Rodolfo Giometti
Date: Thu Nov 19 2026 - 05:25:22 EST
On 19/11/2026 00:19, Florian Bauckholt via B4 Relay wrote:
> From: Florian Bauckholt <florian.bauckholt@xxxxxxxx>
>
> Commit cb481e59ea6c ("KEYS: fix overflow in keyctl_pkey_params_get_2()")
> made keyctl_pkey_params_get_2() set params->out_len to the maximum
> output size of the operation instead of the size provided by the caller.
> It does so for KEYCTL_PKEY_VERIFY as well, but verify has no output
> buffer: out_len shares a union with in2_len, the length of the
> signature.
>
> As a result keyctl_pkey_verify() ignores the signature length passed by
> the caller and always reads max_sig_size bytes from the signature
> pointer. A shorter signature is extended with whatever follows it in
> the caller's memory, or the call fails with -EFAULT if that memory is
> not readable.
>
> The x962 ECDSA template currently ignores data following a DER-encoded
> signature, which hides the problem there. A truncated RSA signature,
> however, is accepted if the bytes following it happen to match the ones
> that were cut off, as reported by Rodolfo Giometti [1].
>
> Use the length provided by the caller, which has already been checked
> against max_sig_size.
>
> Fixes: cb481e59ea6c ("KEYS: fix overflow in keyctl_pkey_params_get_2()")
> Cc: stable@xxxxxxxxxxxxxxx
> Link: https://lore.kernel.org/all/20260919221742.3709658-1-benquike@xxxxxxxxx/
> Link: https://lore.kernel.org/all/0e20f799-e389-40d2-aaee-2f174748faae@xxxxxxxxxxxx/ [1]
> Reviewed-by: Jarkko Sakkinen <jarkko@xxxxxxxxxx>
> Tested-by: Stefan Berger <stefanb@xxxxxxxxxxxxx>
> Signed-off-by: Florian Bauckholt <florian.bauckholt@xxxxxxxx>
> ---
> security/keys/keyctl_pkey.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/security/keys/keyctl_pkey.c b/security/keys/keyctl_pkey.c
> index e7121523b9c9..722521a014ab 100644
> --- a/security/keys/keyctl_pkey.c
> +++ b/security/keys/keyctl_pkey.c
> @@ -175,7 +175,7 @@ static int keyctl_pkey_params_get_2(const struct keyctl_pkey_params __user *_par
> goto out_up_key_sem;
> }
>
> - params->out_len = info.max_sig_size;
> + params->in2_len = uparams.in2_len;
> break;
> default:
> ret = -EOPNOTSUPP;
>
I ran test3008 of craf 2000 times on a UML kernel of current mainline
(666539eecdf4) with this change applied: no failures, against 4 out of
2000 without it.
Tested-by: Rodolfo Giometti <giometti@xxxxxxxxxxxx>
Ciao,
Rodolfo
--
GNU/Linux Solutions e-mail: giometti@xxxxxxxxxxxx
Linux Device Driver giometti@xxxxxxxx
Embedded Systems phone: +39 349 2432127
UNIX programming