Re: [Qemu-devel] Re: Boot Failure (CDROM boot failure code : 0004)

[Juergen Keil]

Volkan YAZICI writes:
> 
> On Tue, 17 Jun 2008, Volkan YAZICI <address@hidden> writes:
> >   root:~# diff \
> >   > /root/scd0.iso.md5sum
> >   > /home/windows/windows-server-2003-standard-x86-64-sp2.iso.md5sum
> >   1c1
> >   < b5d670b7360dc43af8157a50de43bac1  scd0.iso
> >   ---
> >   > cfcf3b24d9b44e75826259c9e914bf40  
windows-server-2003-standard-x86-64-sp2.iso
> 
> More interesting results:
> 
>   windows:~# md5sum windows-server-2003-standard-x86-64-sp2-dd-bs-*
>   b5d670b7360dc43af8157a50de43bac1  
windows-server-2003-standard-x86-64-sp2-dd-bs-1024.iso
>   b253aa547f8a7269b6589caa001bd566  
windows-server-2003-standard-x86-64-sp2-dd-bs-2048.iso
>   64a9d318690bfe223b0e3c5c15b392e3  
windows-server-2003-standard-x86-64-sp2-dd-bs-32768.iso
>   b5d670b7360dc43af8157a50de43bac1  
windows-server-2003-standard-x86-64-sp2-dd-bs-512.iso
> 
> Quite funny. (By the way, all of them are of same size.) Should I trust
> that bs={1024,512} produce the right result?  Anyway, neither of them
> solve the problem. Any ideas?

That is unexpected.  I'd expect that you get exactly the same
bits no matter what block size >= 2048 (and power of 2) you use.

Some drives (or the OS) also allow block sizes of 1024 or
512 bytes, but you should still get exactly the same bits.


Smells like a bug, either with the optical drive, or some
driver, filesystem, or OS issue.


What happens when you use "cmp -l" on a pair of the above
iso images which have not have the same md5 checksum?  

At what file offset does it find differences?  Only on the
last 2-3 blocks of the file, perhaps?

Does cmp -l only find a few differences per iso image pair?
What are the differences?