Re: [PATCH] nfsd: Replace open-coded conversion of bytes to hex

[Eric Biggers <ebiggers@kernel.org>]

On Mon, Aug 04, 2025 at 10:01:49AM -0400, Chuck Lever wrote:
> On 8/3/25 5:24 PM, Eric Biggers wrote:
> > Since the Linux kernel's sprintf() has conversion to hex built-in via
> > "%*phN", delete md5_to_hex() and just use that.  Also add an explicit
> > array bound to the dname parameter of nfs4_make_rec_clidname() to make
> > its size clear.  No functional change.
> > 
> > Signed-off-by: Eric Biggers <ebiggers@kernel.org>
> > ---
> >  fs/nfsd/nfs4recover.c | 18 ++----------------
> >  1 file changed, 2 insertions(+), 16 deletions(-)
> > 
> > diff --git a/fs/nfsd/nfs4recover.c b/fs/nfsd/nfs4recover.c
> > index 2231192ec33f5..54f5e5392ef9d 100644
> > --- a/fs/nfsd/nfs4recover.c
> > +++ b/fs/nfsd/nfs4recover.c
> > @@ -90,26 +90,12 @@ static void
> >  nfs4_reset_creds(const struct cred *original)
> >  {
> >  	put_cred(revert_creds(original));
> >  }
> >  
> > -static void
> > -md5_to_hex(char *out, char *md5)
> > -{
> > -	int i;
> > -
> > -	for (i=0; i<16; i++) {
> > -		unsigned char c = md5[i];
> > -
> > -		*out++ = '0' + ((c&0xf0)>>4) + (c>=0xa0)*('a'-'9'-1);
> > -		*out++ = '0' + (c&0x0f) + ((c&0x0f)>=0x0a)*('a'-'9'-1);
> > -	}
> > -	*out = '\0';
> > -}
> > -
> >  static int
> > -nfs4_make_rec_clidname(char *dname, const struct xdr_netobj *clname)
> > +nfs4_make_rec_clidname(char dname[HEXDIR_LEN], const struct xdr_netobj *clname)
> >  {
> >  	struct xdr_netobj cksum;
> >  	struct crypto_shash *tfm;
> >  	int status;
> >  
> > @@ -131,11 +117,11 @@ nfs4_make_rec_clidname(char *dname, const struct xdr_netobj *clname)
> >  	status = crypto_shash_tfm_digest(tfm, clname->data, clname->len,
> >  					 cksum.data);
> >  	if (status)
> >  		goto out;
> >  
> > -	md5_to_hex(dname, cksum.data);
> > +	sprintf(dname, "%*phN", 16, cksum.data);
> 
> Hello Eric,
> 
> Can the raw "16" be replaced with "HEXDIR_LEN / 2" ?
> 
> 
> >  	status = 0;
> >  out:
> >  	kfree(cksum.data);
> >  	crypto_free_shash(tfm);
> > 
> > base-commit: 186f3edfdd41f2ae87fc40a9ccba52a3bf930994

It's the same 16 as was there before.  It really should be
MD5_DIGEST_SIZE, but there are a few more things to clean up regarding
that.  I didn't want to bloat the scope of this patch.

I sent a new series with this patch unchanged, but the digest size stuff
cleaned up in another patch.

- Eric