Actually, Linux is much better proof of the concept than all the papers,
is not it?
> hands-down. They did note that spl made sense on older machines where
> interrupt routines were, relatively, much longer due to the slower clock
> speeds, but concluded that it didn't make sense on modern, fast CPUs.
Provided these modern fast CPUs serve old slow peripheral devices
or devices, which are insensitive to latency. Do we really leave dumb,
but high speed, serial devices to RTlinux or to that poor netbsd? 8)
Note also, when irq disabling is combined with spinlocks,
latency bounds are difficult to control, if it is possible at all.
Al least, multiply by maximal depth of nested locks and by number of cpus...
By the way, the system used by code, working at BHs, does not differ
of single-level spls essentially, to be honest.
> familiar, doesn't it? :-)
Such papers should be read with good share of sane scepticism
and smiley face, and not be used as proofs. 8)
Alexey
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/