Re: Partitions whith length % 4096 != 0

[address@hidden]

Thanks for your ideas.
Unfortunatly there not usable for my problem, (reasons
listed under), so
I'll do the simplest workaround imaginable : round the
partition size down
to 4096 bytes in newfs_hfs. My code already handle HFS
filesystems smaller
than the size of the partition they're in. I'll check that
kernel implementations
do this as well (at least for Linux, maybe even Darwin :p).
If they don't, I'll
have to find a new idea / send a kernel patch, but the last
solution would
prevent users of current kernels to use my solution.

>(1) you might need to use that read/write-last-sector
ioctl.  (dd should
>do this - so use dd?).  See libparted/linux.c,
_write_lastoddsector().
>(Note: there is a patch pending against this stuff)
It seems that theses ioctl are not in 2.4.25 and 2.4.26
anymore...
there seem to be quite strange things happening in Linux...

>(2) set the blocksize to 512 bytes?  I don't think there's
an ioctl to
>do this, but you could hack one in for your tests (yuck!).
 The function
>you need is: set_blocksize() in fs/block_dev.c.
It would prevent other users to make an hfs filesystem...
>       That said, I would like to run your tests too.

>(3) use 2.6.x.
One day I'll do that :)
But anyway we need a way to create hfs / hfs+ / hfsx file
systems under Linux, and it would be good to be able to do
that under 2.4 versions.

Cheers,
Guillaume.
-----

******ADSL Tiscali, le Haut débit au meilleur prix ******
Avec Tiscali, profitez de l'ADSL au meilleur prix partout en France !
Pour profiter de cette offre exceptionnelle, cliquez ici : 
http://register.tiscali.fr/adsl
Sous réserve d'éligibilité à l'ADSL.