tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: revivesa status 2008/07/09

In article <>,
Andrew Doran  <> wrote:
>- it works only on a handful of architectures, eg x86.

Because md support has not been written. Is there an arch in which it cannot
be made to work?

>- in most tests its performance is demonstrably inferior to 1:1.

Except in the uniprocessor case, which was the issue that Bucky had...

>- it's completely unreliable, even opening the machine to DOS attacks.

This is a problem.

>- it has architectural, code quality and code maintenance issues.

Again this is a problem.

>- it completely lacks any kind of real-time support.

Because it has not been written.

>In its current form SA threading is a regressive proposition. Even if all
>the remaining issues are addressed, what benefits would it bring over and
>above 1:1 threading?

It is interesting as a research threading platform. The N:M threading model
is ~impossible to get working correctly. As long as it does not krud the kernel
code, it would be made a kernel option. Perhaps even a loadable module.


Home | Main Index | Thread Index | Old Index