Re: [ACPI] Toshiba Satellite P10 + ati-drivers (fglrx)

[Jaco K. <ja...@kr...>]

Pavel Machek wrote:
> On So 24-12-05 15:14:16, Jaco Kroon wrote:
>>Pavel Machek wrote:
>>>>Now, with APIC, but no IO-APIC there is a bug notice during resume.
>>>>Going to type it over (from dmesg):
>>>>
>>>>Back to C!
>>>>Debug: sleeping function called from invalid context at mm/slab.c:2472
>>>>in_atomic():0, irqs_disabled():1
>>>>__might_sleep+0x9e/0xa6
>>>>acpi_os_allocate+0x15/0x26
>>>>kmem_cache_alloc+0x95/0xad
>>>>acpi_ut_callocate+0x37/0x79
>>>
>>>I know, it is scary, but it is also known and harmless. You can ignore
>>>this one.
>>
>>Now _that_ made me raise an eyebrow.  Even more than the fact that my
>>DSDT insists on checking for the version of Microsoft Windows I'm
>>running!
> 
> See archives. Len even had a fix at some point, but it was too ugly
> IIRC.
> 

Oh ok.  No problem then.

New weird error.  This may also be harmless though.  Just first a few
notes, I'm back onto 2.6.14.3 kernel (fglrx doesn't work with 2.6.15-rc?
- it just crashes the whole system), and I really need that.

With fglrx I can suspend to disk and resume.  However, I can't
suspend-to-ram.  Well, suspend works and it comes back up fine but as
soon as I switch back to X it hard-crashes the machine.  Now, I'm
currently between resume and switching to X, and the last few lines from
dmesg looks as follows:

Restarting tasks... done
    ACPI-0284: *** Error: Region EmbeddedController(3) has no handler
    ACPI-0508: *** Error: Method execution failed [\_SB_.EPWR.PCLK]
(Node c14d3200), AE_NOT_EXIST
    ACPI-0508: *** Error: Method execution failed
[\_SB_.CPI0.LPC0.EC0_._Q1E] (Node c14ceee0), AE_NOT_EXIST

Now I have a very weird feeling down in my gut that I did not see this
before and it's probably due to the fglrx module in my kernel.  And
interrupt 11, btw, is for the screen (it gets shown now in
/proc/interrupts after I restored most of my kernel).

Now, initially when I switch back to X I get the correct image on the
bottom third of the screen with the top two thirds seemingly corrupted.
 This is pretty normal and is identical to what I normally see when
switching between consoles and X.  This lasts for about half a second
before it freezes up and causes a hard-lock.  Normally the screen just
resets itself at the time it now locks up.

Anyhow, not too serious, overall I'm pretty happy with the progress made
during this week.  suspend-to-disk == much better than no suspend at all
(at least now I can stop my notebook halfway through some busy task to
shut it down, transport it and resume elsewhere).

Then again, I'll play around with the kernel options for fglrx and see
whether they don't perhaps make a difference.

Jaco
-- 
There are only 10 kinds of people in this world,
  those that understand binary and those that don't.
http://www.kroon.co.za/