-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Thu, Apr 12, 2007 at 10:10:28PM -0400, John Anthony Kazos Jr. wrote:
> I did a few simple independent tests between 'char *foo = "foo"' and 'char
> foo[] = "foo"' and they confirmed what the KJ docs say, that the latter
> results in a smaller object. So I did a test with <fs/partitions/check.c>.
>
> static char *make_block_name(struct gendisk *disk)
> {
> - static char *block_str = "block:";
> + static char block_str[] = "block:";
>
> <fs/partitions/check.o> goes from 8149 bytes to 8137 bytes. 'readelf'
> shows that "block:" got moved from .rodata.str1.1 to .data and a
> relocation token got removed that pointed into .rodata.str1.1, saving 12
> bytes.
>
> <fs/built-in.o> goes from 1337360 to 1337320, saving 40 bytes.
>
> <vmlinux>, however, stays exactly the same, at 4520468 bytes. And
> <arch/i386/boot/bzImage> actually increases from 1702520 to 1072552,
> losing 32 bytes.
The linker script aligns most sections at page boundaries (i.e.: 4k).
If you save 40 bytes in one of the objects, the chance that you shrink
the particular section over a 4k boundary is pretty slim.
Erik
- --
They're all fools. Don't worry. Darwin may be slow, but he'll
eventually get them. -- Matthew Lammers in alt.sysadmin.recovery
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iD8DBQFGH0mZ/PlVHJtIto0RAltpAKCDSKea1gmjwlKD0+6RrMTjQxCgAgCfQQOC
P/sZFzJbbnczwliJObptjDE=
=tCAs
-----END PGP SIGNATURE-----
--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
Please read the FAQ at http://kernelnewbies.org/FAQ
