Hi,
> > > Hopefully this patch includes all the of feedback from the last one.
> > > This patch adds a character device driver to allow user space access
> > > to the aranym natfeats nfstderr.
Great (even though, in the light of Geert's explanation of /dev/console,
one wonders if console=/dev/tty console=/dev/lp wouldn't have achieved
the same :-) - IIRC Petr explained that /dev/lp goes to stderr as well).
> > BTW, I am amazed by simplicity of the patch and am wondering if it is
> > similarly simple to write a disk driver (to avoid the IDE emulation in
> > ARAnyM).
>
> I don't think it's quite as simple.
Not nearly as simple - you won't get away with a single fops, for
starters. OTOH we don't need to fudge around with interrupts and could
just pretend a disk request returns right away, meaning to call the
end_request() function from the request function directly (if that's a
no-no for some reason, call it via a task queue right away).
Looking at drivers/block/z2ram.c, you'll get away with two fops (open,
close) and one do_request function that does the actual work (i.e. pass
blocks to read/write to the emulator).
To be honest, I always thought having IDE emulation in ARAnyM was neat,
and i would love to have SCSI emulation as well, for kernel debugging. But
for raw performance, a natfeat block driver would be best. If Petr can
provide the natfeat calls to write x blocks at offset y in the disk image
(plus odds and ends like get_capacity) I'll provide the rest.
> I wonder if the uml hostfs already in the kernel wouldn't be a good
> starting place. I believe you can boot from it and everything.
That'll export (part of) the host filesystem to the m68k kernel, i.e.
you'll write to ext3 or whatever directly, not to a disk image. Won't help
with testing stuff like "partitioning" the disk image.
Michael
-
To unsubscribe from this list: send the line "unsubscribe linux-m68k" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
