Re: There are smaller ways to encode a CRC32 table...

2008-02-01 Thread Ian Campbell 

On Thu, 2008-01-31 at 23:53 -0500, George Spelvin wrote:
> The code to fill it in is smaller than the table itself.
> Is it worth complicating things with some INIT code to reduce
> the stored image size?  (The table is not compressible.)
[snip]

Thanks but since the code is only used when building the image I'm not
too worried which way is used and I've tested the table based one ;-).

> And BTW, storing the inverse of the CRC only catches trailing (after the CRC)
> all-zero padding.  If this is not a problem, it's not necessary, although you
> still might want to do it just for consistency.  This inversion changes the
> CRC of the entire image (body + CRC) from all-zero to a fixed non-zero value.
> (To be precise, to the (non-inverted) CRC of 0x.)

I didn't know the precise details of why you might invert it, thanks for
the info.

Ian
-- 
Ian Campbell
Current Noise: Reverend Bizarre - Sodoma Sunrise

I am a man: nothing human is alien to me.
-- Publius Terentius Afer (Terence)

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Re: There are smaller ways to encode a CRC32 table...

2008-01-31 Thread H. Peter Anvin 

George Spelvin wrote:

The code to fill it in is smaller than the table itself.
Is it worth complicating things with some INIT code to reduce
the stored image size?  (The table is not compressible.)


I think it matters not at all either way.

-hpa

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

There are smaller ways to encode a CRC32 table...

2008-01-31 Thread George Spelvin 
The code to fill it in is smaller than the table itself.
Is it worth complicating things with some INIT code to reduce
the stored image size?  (The table is not compressible.)

#define CRC32POLY   0xedb88320  /* CRC32 polynomial, little-endian */

static uint32_t crctab32[256];

void
crc32init(void)
{
unsigned i, j;
uint32_t crc = 1;

crctab32[0] = 0;

for (i = 128; i; i >>= 1) {
crc = (crc >> 1) ^ ((crc & 1) ? CRC32POLY : 0);
for (j = 0; j < 256; j += 2*i)
crctab32[i+j] = crc ^ crctab32[j];
}
}

The above code basically computes the CRCs of the bytes 0x80, 0x40, ... 0x01,
and applies the identity crctab32[i^j] = crctab32[i] ^ crctab32[j].

And BTW, storing the inverse of the CRC only catches trailing (after the CRC)
all-zero padding.  If this is not a problem, it's not necessary, although you
still might want to do it just for consistency.  This inversion changes the
CRC of the entire image (body + CRC) from all-zero to a fixed non-zero value.
(To be precise, to the (non-inverted) CRC of 0x.)
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/