Re: RT patch acceptance

[Andrea Arcangeli]

On Wed, Jun 01, 2005 at 02:59:13PM -0700, Bill Huey wrote:
> I forgot. You basically turn it into one single big system wide mutex and
> and deal pathological cases as it turns up. Doing this is optional and
> if you can get away with letting the cli/sti function stay in place, then
> it's less work for us to handle.

Do you worry about "less work" after all this new stuff added? I'd
understand "less work" for a simple and safe solution like rtlinux/RTAI,
but going down your path, isn't looking for "less work" or "simpler".

The advantage you get with preempt-RT is a chance to run syscalls with
higher prio by sharing the same syscall code of the non-RT case, but
it'd be little advantage if you then have to lose in reliability.

The argument that only a subset of drivers is used isn't very valid,
people will just assume it to be hard-RT and they could build hardware
with random drivers thinking that they will get the gurantee. I
understand it's ok with you since you're able to evaluate the RT-safety
of every driver you use, but I sure prefer "ruby hard" solutions that
don't require looking into drivers to see if they're RT-safe.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/