Re: [PATCH 1/6 linux-next] FS/AFFS: replace min() with casting/constant suffix by min_t

[Andrew Morton <akpm@linux-foundation.org>]

On Tue,  3 Feb 2015 19:07:52 +0100 Fabian Frederick <fabf@skynet.be> wrote:

> This patch prepares for AFFSNAMEMAX definition
> 
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Signed-off-by: Fabian Frederick <fabf@skynet.be>
> ---
>  fs/affs/amigaffs.c | 2 +-
>  fs/affs/dir.c      | 2 +-
>  fs/affs/namei.c    | 4 ++--
>  3 files changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/fs/affs/amigaffs.c b/fs/affs/amigaffs.c
> index 511ab6b..7a6de60 100644
> --- a/fs/affs/amigaffs.c
> +++ b/fs/affs/amigaffs.c
> @@ -508,7 +508,7 @@ affs_check_name(const unsigned char *name, int len, bool notruncate)
>  int
>  affs_copy_name(unsigned char *bstr, struct dentry *dentry)
>  {
> -	int len = min(dentry->d_name.len, 30u);
> +	int len = min_t(u64, dentry->d_name.len, 30);

	int = min_t(u64, u32, int)

Doesn't seem to make sense?

>  	*bstr++ = len;
>  	memcpy(bstr, dentry->d_name.name, len);
> diff --git a/fs/affs/dir.c b/fs/affs/dir.c
> index a682892..580b958 100644
> --- a/fs/affs/dir.c
> +++ b/fs/affs/dir.c
> @@ -114,7 +114,7 @@ inside:
>  				break;
>  			}
>  
> -			namelen = min(AFFS_TAIL(sb, fh_bh)->name[0], (u8)30);
> +			namelen = min_t(u8, AFFS_TAIL(sb, fh_bh)->name[0], 30);
>  			name = AFFS_TAIL(sb, fh_bh)->name + 1;
>  			pr_debug("readdir(): dir_emit(\"%.*s\", ino=%u), hash=%d, f_pos=%llx\n",
>  				 namelen, name, ino, hash_pos, ctx->pos);
> diff --git a/fs/affs/namei.c b/fs/affs/namei.c
> index de84f4d..4a04c2d 100644
> --- a/fs/affs/namei.c
> +++ b/fs/affs/namei.c
> @@ -71,7 +71,7 @@ __affs_hash_dentry(struct qstr *qstr, toupper_t toupper, bool notruncate)
>  		return i;
>  
>  	hash = init_name_hash();
> -	i = min(qstr->len, 30u);
> +	i = min_t(u64, qstr->len, 30);

same here

>  	for (; i > 0; name++, i--)
>  		hash = partial_name_hash(toupper(*name), hash);
>  	qstr->hash = end_name_hash(hash);
> @@ -175,7 +175,7 @@ affs_hash_name(struct super_block *sb, const u8 *name, unsigned int len)
>  	toupper_t toupper = affs_get_toupper(sb);
>  	int hash;
>  
> -	hash = len = min(len, 30u);
> +	hash = len = min_t(unsigned, len, 30);

	int = min_t(unsigned, u32, int)

It's a mess!  It would be better to do something like

	u32 hash;

	hash = min(len, 30U);

Use of min_t is a pretty reliable sign that the types have been screwed
up.  Let's try to unscrew the type choices rather than working around
earlier mistakes.