linux-kernel - Re: [PATCH v1 1/1] base64: Unroll the tables initialisers

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <aQm2366wMJ1K1fp7@smile.fi.intel.com>
Date: Tue, 4 Nov 2025 10:18:39 +0200
From: Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
To: David Laight <david.laight.linux@...il.com>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
	Guan-Chun Wu <409411716@....tku.edu.tw>,
	Kuan-Wei Chiu <visitorckw@...il.com>, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v1 1/1] base64: Unroll the tables initialisers

On Mon, Nov 03, 2025 at 10:18:57PM +0000, David Laight wrote:
> On Mon,  3 Nov 2025 20:05:10 +0100
> Andy Shevchenko <andriy.shevchenko@...ux.intel.com> wrote:
> 
> > Currently the initialisers of the tables have duplicate indices.
> > This prevents from building with `make W=1`.
> > 
> > To address the issue, unroll the table initialisers with generated
> > arrays by the following Python excerpt:
> > 
> > CHARS = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789"
> > 
> > def gen_table(ch62, ch63):
> >     table = [ 0xff ] * 256
> >     for idx, char in enumerate(CHARS):
> >         table[ord(char)] = idx
> >     table[ord(ch62)] = 62
> >     table[ord(ch63)] = 63
> > 
> >     for i in range(0, len(table), 8):
> >         print (f"\t{', '.join(f"0x{c:02x}" for c in table[i:i+8])},\t/* {i:-3d} - {i+7:-3d} */")

...

> > +	[BASE64_STD] = {
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/*   0 -   7 */
> 
> You need to use -1 not 0xff.

Whu? The s8 type is pretty much 8-bit, care to explain the point?

> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/*   8 -  15 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/*  16 -  23 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/*  24 -  31 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/*  32 -  39 */
> 
> Nothing wrong with [ 0 ... 39 ] = -1,

Have you tried? It's unreadable if we use ranges.

> > +		0xff, 0xff, 0xff, 0x3e, 0xff, 0xff, 0xff, 0x3f,	/*  40 -  47 */
> > +		0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3a, 0x3b,	/*  48 -  55 */
> > +		0x3c, 0x3d, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/*  56 -  63 */
> > +		0xff, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06,	/*  64 -  71 */
> > +		0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e,	/*  72 -  79 */
> > +		0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16,	/*  80 -  87 */
> > +		0x17, 0x18, 0x19, 0xff, 0xff, 0xff, 0xff, 0xff,	/*  88 -  95 */
> > +		0xff, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20,	/*  96 - 103 */
> > +		0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28,	/* 104 - 111 */
> > +		0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e, 0x2f, 0x30,	/* 112 - 119 */
> > +		0x31, 0x32, 0x33, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 120 - 127 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 128 - 135 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 136 - 143 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 144 - 151 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 152 - 159 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 160 - 167 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 168 - 175 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 176 - 183 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 184 - 191 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 192 - 199 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 200 - 207 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 208 - 215 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 216 - 223 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 224 - 231 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 232 - 239 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 240 - 247 */
> > +		0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,	/* 248 - 255 */
> > +	},

...

> This is impenetrable crap...

Send your version for the review, please, we still have a build issue here.
But I definitely do not like current state of affairs.

-- 
With Best Regards,
Andy Shevchenko