[PATCH v4 0/6] lib/base64: add generic encoder/decoder, migrate users

[Andy Shevchenko]

On Tue, Nov 04, 2025 at 09:03:26AM +0000, David Laight wrote:
> On Mon, 3 Nov 2025 19:07:24 +0800
> Kuan-Wei Chiu <visitorckw at gmail.com> wrote:
> > On Mon, Nov 03, 2025 at 11:24:35AM +0100, Andy Shevchenko wrote:

...

> > Since I believe many people test and care about W=1 builds, I think we
> > need to find another way to avoid this warning? Perhaps we could
> > consider what you suggested:
> > 
> > #define BASE64_REV_INIT(val_plus, val_comma, val_minus, val_slash, val_under) { \
> > 	[ 0 ... '+'-1 ] = -1, \
> > 	[ '+' ] = val_plus, val_comma, val_minus, -1, val_slash, \
> > 	[ '0' ] = 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, \
> > 	[ '9'+1 ... 'A'-1 ] = -1, \
> > 	[ 'A' ] = 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, \
> > 		  23, 24, 25, 26, 27, 28, 28, 30, 31, 32, 33, 34, 35, \
> > 	[ 'Z'+1 ... '_'-1 ] = -1, \
> > 	[ '_' ] = val_under, \
> > 	[ '_'+1 ... 'a'-1 ] = -1, \
> > 	[ 'a' ] = 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, \
> > 		  49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, \
> > 	[ 'z'+1 ... 255 ] = -1 \
> > }
> 
> I've a slightly better version:
> 
> #define INIT_62_63(ch, ch_62, ch_63) \
> 	[ ch ] = ch == ch_62 ? 62 : ch == ch_63 ? 63 : -1
> 
> #define BASE64_REV_INIT(ch_62, ch_63) { \
> 	[ 0 ... '0' - 6 ] = -1, \
> 	INIT_62_63('+', ch_62, ch_63), \
> 	INIT_62_63(',', ch_62, ch_63), \
> 	INIT_62_63('-', ch_62, ch_63), \
> 	INIT_62_63('.', ch_62, ch_63), \
> 	INIT_62_63('/', ch_62, ch_63), \
> 	[ '0' ] = 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, \
> 	[ '9' + 1 ... 'A' - 1 ] = -1, \
> 	[ 'A' ] = 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, \
> 		  23, 24, 25, 26, 27, 28, 28, 30, 31, 32, 33, 34, 35, \
> 	[ 'Z' + 1 ... '_' - 1 ] = -1, \
> 	INIT_62_63('_', ch_62, ch_63), \
> 	[ '_' + 1 ... 'a' - 1 ] = -1, \
> 	[ 'a' ] = 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, \
> 		  49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, \
> 	[ 'z' + 1 ... 255 ] = -1 \
> }
> 
> that only requires that INIT_62_63() be used for all the characters
> that are used for 62 and 63 - it can be used for extra ones (eg '.').
> If some code wants to use different characters; the -1 need replacing
> with INIT_62_63() but nothing else has to be changed.
> 
> I used '0' - 6 (rather than '+' - 1 - or any other expression for 0x2a)
> to (possibly) make the table obviously correct without referring to the
> ascii code table.

Still it's heavily depends on the values of '+,-./_' as an index that
makes it not so flexible.

Moreover this table is basically a dup of the strings in the first array.
Which already makes an unnecessary duplication. That's why I prefer to
see a script (one source of data) to generate the header or something like
this to have the tables and strings robust against typos.

The above is simply an unreadable mess.

-- 
With Best Regards,
Andy Shevchenko