Re: [PATCH] lib/base64: fix -Wconstant-conversion warning in INIT_1()

[David Laight <david.laight.linux@gmail.com>]

On Tue, 3 Nov 2026 12:00:52 -0700
Nathan Chancellor <nathan@kernel.org> wrote:

> Hi Rosalie,
> 
> On Tue, Nov 03, 2026 at 04:34:04PM +0000, Rosalie Fang wrote:
> > When expanding BASE64_REV_INIT() for v in 0x7c..0x7f (124..127), the
> > unselected ternary branch (v) - '0' + 52 evaluates to 128..131, which
> > exceeds SCHAR_MAX (127). Recent Clang versions diagnose implicit integer
> > constant conversions in unevaluated ternary arms when initializing the
> > s8 lookup tables:
> > 
> >   lib/base64.c:58:18: error: implicit conversion from 'int' to 's8'
> >   (aka 'signed char') changes value from 131 to -125
> >   [-Werror,-Wconstant-conversion]
> >      58 |         [BASE64_IMAP] = BASE64_REV_INIT('+', ',')
> >         |                         ^~~~~~~~~~~~~~~~~~~~~~~~~
> >   lib/base64.c:34:42: note: expanded from macro 'INIT_1'
> >      34 |                 : (v) >= '0' && (v) <= '9' ? (v) - '0' + 52 \
> >         |                                              ~~~~~~~~~~^~~~
> > 
> > Explicitly cast the digit-branch expression to s8 to silence the
> > warning.
> > 
> > Fixes: c4eb7ad32eab ("lib/base64: optimize base64_decode() with reverse lookup tables")
> > Assisted-by: LLM
> > Signed-off-by: Rosalie Fang <rosaliefang@google.com>
> > ---
> >  lib/base64.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/lib/base64.c b/lib/base64.c
> > index 325c7332b049..1489c048b1fd 100644
> > --- a/lib/base64.c
> > +++ b/lib/base64.c
> > @@ -31,7 +31,7 @@ static const char base64_tables[][65] = {
> >  #define INIT_1(v, ch_62, ch_63) \
> >  	[v] = (v) >= 'A' && (v) <= 'Z' ? (v) - 'A' \
> >  		: (v) >= 'a' && (v) <= 'z' ? (v) - 'a' + 26 \
> > -		: (v) >= '0' && (v) <= '9' ? (v) - '0' + 52 \
> > +		: (v) >= '0' && (v) <= '9' ? (s8)((v) - '0' + 52) \
> >  		: (v) == (ch_62) ? 62 : (v) == (ch_63) ? 63 : -1
> >  
> >  /*
> > -- 
> > 2.56.0.385.gac671c819a-goog
> >   
> 
> Thanks for the change. I avoided this in a different way, which has
> recently been merged into a tree that should be merged into mainline
> relatively soon:
> 
>   https://git.kernel.org/akpm/mm/c/3af83c11cf5990fda0bfe3959aa6391ff903ffdd

#define INIT_1(v, ch_62, ch_63) \
	[v] = (s8)((v) >= 'A' && (v) <= 'Z' ? (v) - 'A' \
		 : (v) >= 'a' && (v) <= 'z' ? (v) - 'a' + 26 \
		 : (v) >= '0' && (v) <= '9' ? (v) - '0' + 52 \
		 : (v) == (ch_62) ? 62 : (v) == (ch_63) ? 63 : -1)
probably looks best?

> For the record, I do hope that
> 
>   https://github.com/llvm/llvm-project/pull/226775
> 
> eventually gets merged so that these workarounds will be unnecessary but
> I am not sure how that will play out yet.

Someone really doesn't like kernel code!
There are other warnings like 'cast of pointer to/from integer of a different
size' that it would be nice not to have reported in code that will be optimised
away.
At least the kernel gave up on some of the 'type limits' warnings, they
had similar problems (even inside __builtin_choose_expr()).

David

(I'm reading these unthreaded...)