Re: [PATCH] Huge changes in mm.c

[Vincent Pelletier <subdino2004@yahoo.fr>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Yoshinori K. Okuji wrote:
> Basically, I do not want huge changes, unless you have a good reason to do it. 

One of the reasons that made me study mm.c it that I get an unaligned
pointer when I type "rescue".
Built from current cvs & tested with :

cvsco && ./configure && make && ./grub-mkimage -v -d . -o core.img *.mod
&& dd if=core.img of=/dev/fd0 bs=512 seek=1 && qemu -fda /dev/fd0

Tested with debian's gcc 4 and 3.4 .

> Please do not reinvent a wheel only because you do not understand it.

I thought it wasn't reliable because of the error mentioned higher. The
problem is that for now my implementation gets more errors... :'(
I found it really difficult to understand (mainly the alignment things)
I thought that on a maintainer point of view it could be good to rewrite
it with separate alignment functions.
I also thought it would be fun to code, but in the end it was more of a
challenge "I can't believe I can't do that myself !" when it was
crashing again and again.

> At a quick glance, one trivial bug is that GRUB_MM_ALIGN_LOG2 is set to 8 on 
> 64-bit systems. This must be 5.

Works 8|...

Vincent Pelletier
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)

iD8DBQFC07gnFEQoKRQyjtURAq1VAJ9EgPXFR7Hl3FKGNnszLtY+VjpXmgCfWByJ
DiZHotfnwNIRDW7FSgsqK5s=
=tZKb
-----END PGP SIGNATURE-----

	

	
		
___________________________________________________________________________ 
Appel audio GRATUIT partout dans le monde avec le nouveau Yahoo! Messenger 
T