[haiku-development] Re: Blog Posting Permission?

On Fri, Oct 17, 2008 at 08:22, Michael Crawford <mdcrawford@xxxxxxxxx> wrote:
[mega-snip]
> I'm an expert with FireWire storage, in particular SCSI and the Serial
> Bus Protocol 2.  My understanding is that while Haiku does have basic
> FireWire support, SBP2 is either not yet implemented or doesn't work
> reliably.
>
> (SBP2 is used to transport SCSI Command Descriptor Blocks (CDBs) over
> the FireWire bus.)
>
> Whatever state Haiku FireWire is in, I expect I can improve it and if
> FireWire storage isn't implemented or doesn't work reliably, I can
> implement it and make it work flawlessly and fast.

Yay! Isn't Firewire lovely?

While you're at it, do you mind implementing
kernel-debug-over-Firewire, like FreeBSD and Mac OS X sport? It
happens that my laptop doesn't have a classic serial interface, while
the desktop happens to have one, and both support Firewire. Granted, I
could use an USB-to-RS232 dongle, but as I'm developing the new thread
scheduler for Haiku, testing and debugging on the SMP machine (which
happens to be the laptop, a C2D) is the better option, as eventually
I'll have to let go of the safety net provided by emulators and UP
hardware. And the dongle isn't really that cheap around here. And I
have a Firewire cable already. :)

(And yes, despite me implementing a whole new cutting edge etc etc
scheduler, doing device drivers is *way* out of my league [and
hopefully this will change one day :)]. Still, implementing a KDoF
facility shouldn't be much different than doing a device driver.)

Implementing it shouldn't be that hard; it's a matter of setting up
the correct registers and then going wild with the DMA hardware to
manipulate the memory of the host machine. You're probably familiar
with this setup already, since you maintain a RAID driver for Mac OS
X, and AFAICT the debugging setup is identical. Darwin and FreeBSD
code should provide good reference and guidance.



Thanks for considering this :)
A.

Other related posts: