Re: acpi5 20040715

[YONETANI Tomokazu <qhwt+dfly@xxxxxxxxxx>]

On Mon, Sep 27, 2004 at 03:09:20PM -0000, Andreas Hauser wrote:
> qhwt+dfly wrote @ Mon, 27 Sep 2004 23:06:11 +0900:
> > Hi.
> > On Mon, Sep 27, 2004 at 07:36:31AM +0200, Jeroen Ruigrok/asmodai wrote:
> > > -On [20040926 23:52], Andreas Hauser (andy@xxxxxxxxxxxxxxx) wrote:
> > > >tried to get my acpi to do S1 and thought maybe the new intel release helps.
> > > >It did not :(
> > 
> > Andreas, please make sure that S1 is actually supported on your machine (see
> > sysctl hw.acpi.supported_sleep_state, and if so, what happens when you
> > run this command as root:
> > # acpiconf -s1
> > Also, did you see any acpi-related warnings in /var/run/dmesg.boot?
> 
> It's an Acer Aspire 1500.
> There are no BIOS upgrades i know of.
> It does not show S1 in hw.acpi.supported_sleep_state, but i think it should
> be able to do it. S3 and S4 are shown but make it hang.
> 
> acpiconf -s1:
> acpi0: AcpiGetSleepTypeData failes - AE_NOT_FOUND
> 
> other than that no warnings.

Hmm, that's a bit strange. When AcpiGetSleepTypeData() returns AE_NOT_FOUND
it should print "Sleep state S1 not supported by BIOS". It's following
code in acpi.c that's showing the message. If you happen to have
    device acpica
(or device acpica5) in your kernel config, please try removing it.

1695     switch (state) {
1696     case ACPI_STATE_S1:
1697     case ACPI_STATE_S2:
1698     case ACPI_STATE_S3:
1699     case ACPI_STATE_S4:
1700         status = AcpiGetSleepTypeData((UINT8)state, &TypeA, &TypeB);
1701         if (status == AE_NOT_FOUND) {
1702             device_printf(sc->acpi_dev,
1703                           "Sleep state S%d not supported by BIOS\n", state);
1704             break;
1705         } else if (ACPI_FAILURE(status)) {
1706             device_printf(sc->acpi_dev, "AcpiGetSleepTypeData failed - %s\n",
1707                           AcpiFormatException(status));
1708             break;
1709         }


> I fixed (as good as i knew to) my dsdt and load the fixed in loader.conf but
> it didn't help. I tried the new intel code as next step. Now i'm looking
> at the code. Maybe try fbsd 5.1 as some say they had acer hw working (sos@ e.g.).

I recommend 5.3BETA as it should have some more fixes in ACPI code
that DragonFly doesn't have yet.

> Also it seems like the ACPI debugging infrastructure is not complete.
> The one thing i found interesting was that acpidump -t only shows comments.

It's expected, and you get the same result on FreeBSD-5.x?

> What i was really looking after though was being able to manipulate the
> CPU speed (like i can in linux).
> 
> My via epia (another machine) e.g. shows:
> hw.acpi.cpu.max_speed: 2
> hw.acpi.cpu.current_speed: 2
> hw.acpi.cpu.performance_speed: 2
> hw.acpi.cpu.economy_speed: 1
> 
> So i guess some infrastructure for that is already available.

My mail server is also an EPIA ME6000 running DragonFly, has the same lines
as yours, and it can throttle as yours does. But not all machine can throttle.
Please run the command

# acpidump -t -d -o andy-aspire1500.dsdt > andy-aspire1500.asl

and send me andy-aspire1500.asl privately or put it somewhere where I can
download it and I can take a look.