Re: [PATCH v1] imap/auth_cram.c: Don't use strfcpy() where we don't want to generate a C string

Alejandro Colomar via Mutt-dev <[email protected]> Mon, 20 Apr 2026 11:52:35 +0200

Newsgroups	gmane.mail.mutt.devel
Message-ID	<aeX3F8ZcY07CxOWk@debian>

Hi Kevin,

On 2026-04-20T07:04:02+0800, Kevin J. McCarthy wrote:
> On Sun, Apr 19, 2026 at 08:24:01PM +0200, Alejandro Colomar via Mutt-dev wrote:
> > strfcpy() ensures it creates a string.  But that's not what we want.
> > Here, we just want raw bytes in the output.  It's only the input which
> > is a string (password).
> > 
> > strncpy(3) is quite appropriate for this specific use.  It's a function
> > that takes a C string as input, and fills a fixed-size buffer with bytes
> > from it, zeroing the unused remainder part of the buffer.
> > 
> > Because of this use of strncpy(3), we can remove the memset(3) calls.
> > They're now entirely redundant.  (The other branch fills the entire
> > buffer, so it was only meaningful in the branch we're touching.)
> 
> It's early and my brain isn't fully awake.  But I believe this is incorrect.
> The other branch only fills to MD5_DIGEST_LEN.  Removing the memset() in
> that case, and copying the entire secret buffer, instead of secret_len,
> would break the algorithm.

Oh, I had misread and confused MD_DIGEST_LEN by MD5_BLOCK_LEN, and
thought there was only one size being used everywhere.  Yup; please
discard my patch.

I'll have another look today.


Cheers,
Alex

> 
> -- 
> Kevin J. McCarthy
> GPG Fingerprint: 8975 A9B3 3AA3 7910 385C  5308 ADEF 7684 8031 6BDA



-- 
<https://www.alejandro-colomar.es>

signature.asc (application/pgp-signature, 833 B)