Re: [PATCH] Only use MSDOS-Partitions by default on X86

[David Woodhouse <dwmw2@infradead.org>]

On Wed, 2003-05-07 at 08:22, Jens Axboe wrote:
> > That's a little short of what I was intending. Ideally we stick 'struct
> > request', 'struct buffer_head' and 'struct bio' inside #ifdef
> > CONFIG_BLK_DEV, then kill all the dead code which uses them.
> 
> struct request can be a goner with my patch, the others not really.
> request is really a block private property, so it's easy to kill off.
> You are going for the really minimal approach, basically ruling out lots
> of filesystems and requiring major surgery all around. While I can see
> that make sense for an embedded kernel, I'm having a hard time
> envisioning this as something mergable :-)

Last time I looked, it wasn't that bad until I got mired in VM code.
I haven't looked at that since we got CONFIG_SWAP, but I'm fairly sure
the VM bits will be a lot nicer now too.

> > mtdblock.c cleanup noted with interest -- I'll play with that shortly;
> > thanks. Note that you don't actually need flash hardware, you can load
> > the 'mtdram' device which fakes it with vmalloc-backed storage instead.
> > Not too useful for powerfail-testing but for mounting something like
> > ext2 on mtdblock on mtdram it's fine.
> 
> I'm attaching an updated version, I don't think it's safe to use atomic
> kmaps across the do_cached_read/write.

It's not -- the flash read/write/erase functions may sleep.

> Also, I want bio_endio() to increment the sector position of the bio as
> well. Makes for a nicer api, and the sector var in mtdblock would then
> be killable.

OK. Let me know when you're done and I'll fix up FTL and NFTL
accordingly too.

-- 
dwmw2