Re: TSS switching

[Marcus Brinkmann]

On Sun, Oct 07, 2001 at 07:05:52PM -0400, Roland McGrath wrote:
> > Although the user tss is correctly (AFAICS) setup, it doesn't come
> > effective:  The program is killed with Illegal Instruction.  I am at my wits
> > end, as the code seems to be correct and I don't know how to debug it
> > further.
> 
> I could speculate about it or suggesting methods of debugging.
> But you know, why don't we just punt this whole implementation?

Well, it's okay for "us" to do that, but I didn't foresee this problem.  I
thought that fixing the current code with minimal changes would enable me to
understand the code better and try the new implementation.  I have learned a
lot, but as I don't understand why the code as it is doesn't work, it seems
I don't know yet enough.  It's okay to not fix the broken code and put the
new implementation there immediately, but I would like to have at least an
idea why the code doesn't work "as is".

[snipped explanation of what linux does]

> So let's do that.  It's easy and you probably don't have as much weirdo
> Intel magic to debug than with tss switching.

Yes, I thought that doing it this way eventually is better.  I just prefer
having a choice over "I do it this way because I can't get the other way to
work and don't know why" :).

BTW, can we avoid copying the bitmap if we switch from one thread to another
thread in the same task?  Or, a related question, when does a change in the
bitmap become effective, immediately or at the next switch to the thread?
This is a part of the current implementation I don't see.  Linux simply
diddles the task (thread?! [1]) bitmap and the kernel tss at the same time.

Thanks,
Marcus

[1] "do it in the per-thread copy and in the TSS ..." 
(arch/i386/kernel/ioport.c)

-- 
`Rhubarb is no Egyptian god.' Debian http://www.debian.org brinkmd@debian.org
Marcus Brinkmann              GNU    http://www.gnu.org    marcus@gnu.org
Marcus.Brinkmann@ruhr-uni-bochum.de
http://www.marcus-brinkmann.de