Re: [0xffff] Blocks filled with 0

[pancake]

I have imported your patches into the repository:

  http://news.nopcode.org/miau/pvc.cgi?prj=0xFFFF&leaf=3eb71906

Thanks. I have been using gcc3 and looks that it's not very good catching
stupid bugs like these ones. Now I'll use -pedantic -std=c99 :P

On Mon, 19 Nov 2007 17:49:07 -0300
"Osvaldo Santana Neto" <[EMAIL PROTECTED]> wrote:
>
> > FIASCO is not a good format for firmwares, checksumming is poor,
> > extensibility is limited to the flasher tool and it's proprietary,
> 
> Yes, I agree with you. BTW, what algorithm is used for checksumming the 
> pieces?

It's a xorpair on the whole piece. This is in C:

unsigned short do_hash(usho *b, int len)
{
        usho result = 0;
        for(len>>=1;len--;b=b+1)
                result^=b[0];
        return result;
}

(...)
        do {    ret   = fread(&buf, 1, BSIZE, fd);
                if (ret == -1)
                        return 0;
                hash ^= do_hash((usho *)&buf, ret);
        } while(ret);
(...)

This is extracted from src/hash.c, you can calculate this by using 0xFFFF or
hasher (from radare (from radare.nopcode.org)):

  $ 0xFFFF -H rootfs.jffs2
  $ hasher -qv -a xorpair -f rootfs.jffs2

> > I'm looking forward for testing Mamona asap, it looks like an
> > interesting project with great advantages in front of the current
> > official one.
> 
> Yes and you are already contributing with us :)
> 
> I've forgot to tell you that I create a patch to compile 0xFFFF with
> gcc-4.1. The build is breaking with this version of GCC (Rodrigo Vivi
> is talking with you about this and I'm CCing him). I don't know if the
> patches really fix the problem and I dindn't make extensive tests with
> it.

Thanks for your patches. I have submitted an initial patch to Rodrigo
few hours ago, and now the head of the version control is fixed too:

 http://news.nopcode.org/pvcroot/0xFFFF-0.3.1.tar.gz

I will replace the 0.3 with this one after some more testing.


  --pancake
_______________________________________________
0xffff mailing list
0xffff@lists.nopcode.org
https://lists.nopcode.org/mailman/listinfo/0xffff