From: Gábor Lénárt (lgb_at_lgb.hu)
Date: 2007-09-20 16:20:32
On Thu, Sep 20, 2007 at 04:26:50PM +0300, Marko Mäkelä wrote:
> I hope that I can be proven wrong about the difficulty of getting
> additional drivers accepted to Linus' Linux kernel tree. It is a pity
> that implementing and deploying low-level system programs is so hard in
> the age of MMUs and multi-tasking.
Well, in the age multitasking and MMU it's harder to implement correct
timing etc, because you haven't got exclusive access to the hardware; don't
know the exact memory access timing (related to the MMU mapping which
involves TLB, not counting the overhead of virtual memory, exceptions, disk
I/Os etc). Actually we're moving away from direct hardware access, just
think about more and more abstraction layer between user space apps and
hardware: today's new mode is virtualization, hypervisors etc. For strict
ctritical timing purposes the so called "RTOS" (real time OS) is needed.
With a plain 6510 is much more easy to program the hardware :)
Sorry, I don't know OpenCBM well, but it would be also interesting idea to
think about OpenCBM implementation, like device drivers in user space
(http://lwn.net/Articles/66829/). It's the same idea as FUSE (filesystem in
user space). Sure, because of direct hw access, some kernel part may be
needed. This is about the idea that not everything should be placed in
kernel space, some things would fit better into userspace (just think about
special filesystems like sshfs and gmailfs: without FUSE they would be
almost impossible to do in kernel space).
Sorry, if this part is off topic here as well, please inform me, however
implementation of OpenCBM sounds much more hot topic than generating
flamewar about Linux (it was my fault sorry).
--
- Gábor
Message was sent through the cbm-hackers mailing list
Archive generated by hypermail pre-2.1.8.