Re: [Syllable-kernel] Strace patch

[Kristian V. D. V. <va...@li...>]

On Sunday 30 January 2005 21:38, Jake Hamby wrote:
> Kristian Van Der Vliet wrote:
>> Heres a tiny patch to fix errno printing in strace.  Untested but obvious
>> :)
>
> I'll put it in the batch with Daniel's scheduling cleanup patches that
> will become Build 0003.  Right now the version in CVS is Build 0002
> under my new versioning scheme, which includes Arno's big cleanup patch
> of 12/30/2004 and Daniel's cleanup patches of 1/27.

Cool.

> On my own system, I had to add "disable_smp=true" because the SMP stuff
> isn't working on my Pentium 4 HT now that the ACPI code finds the
> simulated SMP. 

If we don't get the kernel stable by 0.5.6 we'll add disable_smp=true to the 
default parameters.

> The error I get is that the init process gives a memory 
> access violation in getpid(), which is implemented in libc by returning
> the thread ID from a structure pointed to by the GS segment register.
> So perhaps the GS register is not getting initialized properly during
> the boot_ap_processors() phase.

This would make sense.  init is the first user space process to run, which is 
a good indication of the problem.  If GS was munged it's possible that it 
doesn't affect the (1:1 mapping) kernel memory space but it certainly would 
mess up the user space mappings, so we get segfaults in user threads but not 
earlier on in kernel threads.

> The good news is that Arno's patches fixed the v86 code and now the PS/2
> mouse driver and APM poweroff code are working again.

I know about the APM problems but when did PS/2 not work?  Is this the SiS 
chipset PS/2 bug or something else?

-- 
Vanders
http://www.syllable.org
http://www.liqwyd.com