S3 wakeup code

S3 wakeup code

[Florian Lohoff]

[-- Attachment #1: Type: text/plain, Size: 540 bytes --]


Hi,
i just tried 2.6.0-test1 and to my expectation the S3 resume stuff still
does not work on the Sony Vaio PCG-C1MHP aka Picturebook. It turns on -
Bios set the powerled correct but then - nothing - Nothing on the
screen, no backlight, no disk activity.

Is there any way to dump the Windows wakeup code ? 
Is that machine specific ?
Might it be useful ?

Flo
-- 
Florian Lohoff                  flo-BCn6idZOOBwdnm+yROfE0A@public.gmane.org             +49-171-2280134
                        Heisenberg may have been here.

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: S3 wakeup code

[Pavel Machek]

Hi!

> i just tried 2.6.0-test1 and to my expectation the S3 resume stuff still
> does not work on the Sony Vaio PCG-C1MHP aka Picturebook. It turns on -
> Bios set the powerled correct but then - nothing - Nothing on the
> screen, no backlight, no disk activity.
> 
> Is there any way to dump the Windows wakeup code ? 
> Is that machine specific ?

Wakeup code should *not* be machine-specific, but video card
save/restore is obviously video-card specific (and you basically can't
get docs). Write beeping code in 8086 assembly and start debuggin'
;-))).

								Pavel

-- 
Horseback riding is like software...
...vgf orggre jura vgf serr.


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100006ave/direct;at.asp_061203_01/01

Re: S3 wakeup code

[Knut Neumann]

Am Mit, 2003-07-23 um 18.24 schrieb Pavel Machek:
> > i just tried 2.6.0-test1 and to my expectation the S3 resume stuff still
> > does not work on the Sony Vaio PCG-C1MHP aka Picturebook. It turns on -
> > Bios set the powerled correct but then - nothing - Nothing on the
> > screen, no backlight, no disk activity.

Hmmm...if I understand that right its (still) the very same behaviour
with my Z600TEK and I remember someone else having that type of problem.
Now I once did some debugging and...nothing. As far as I can tell it
does not jump into wakeup code. I never could get any further on that.
Though I might have done something wrong - is there any foolproof method
on checking wether the system jumps to the wakeup address?

-Knut

>  
> > Is there any way to dump the Windows wakeup code ? 
> > Is that machine specific ?
> 
> Wakeup code should *not* be machine-specific, but video card
> save/restore is obviously video-card specific (and you basically can't
> get docs). Write beeping code in 8086 assembly and start debuggin'
> ;-))).
> 
> 								Pavel





-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Florian Lohoff]

[-- Attachment #1: Type: text/plain, Size: 1060 bytes --]

On Thu, Jul 24, 2003 at 12:18:09PM +0200, Knut Neumann wrote:
> 
> Hmmm...if I understand that right its (still) the very same behaviour
> with my Z600TEK and I remember someone else having that type of problem.
> Now I once did some debugging and...nothing. As far as I can tell it
> does not jump into wakeup code. I never could get any further on that.
> Though I might have done something wrong - is there any foolproof method
> on checking wether the system jumps to the wakeup address?
> 

Yea - Would be my request too - Currently S3 is top on my wishlist. I
did a lot of m68k assembly but i tryed to stay as far away from x86
assembly as i could. So it would be nice if the more experienced ones
would provide some #define DEBUG where we would hear interesting sounds
or blinking keyboard leds or parallel/serial output if certain steps are
handled successfully.

Flo
-- 
Florian Lohoff                  flo-BCn6idZOOBwdnm+yROfE0A@public.gmane.org             +49-171-2280134
                        Heisenberg may have been here.

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: S3 wakeup code

[Pavel Machek]

Hi!

> > Hmmm...if I understand that right its (still) the very same behaviour
> > with my Z600TEK and I remember someone else having that type of problem.
> > Now I once did some debugging and...nothing. As far as I can tell it
> > does not jump into wakeup code. I never could get any further on that.
> > Though I might have done something wrong - is there any foolproof method
> > on checking wether the system jumps to the wakeup address?
> > 
> 
> Yea - Would be my request too - Currently S3 is top on my wishlist. I
> did a lot of m68k assembly but i tryed to stay as far away from x86
> assembly as i could. So it would be nice if the more experienced ones
> would provide some #define DEBUG where we would hear interesting sounds
> or blinking keyboard leds or parallel/serial output if certain steps are
> handled successfully.

Hey, i386 assembly is not that hard, and if you are serious about
debugging S3 wakeup, you'll need to know it, anyway.

I don't remember code for beeping... To send data to the paralel port,

	movb	$0x12, %al
	movw	$0x378, %dx
	outb	%al, %dx

should do the trick (of course, you need to preserve eax/edx if they
contain something important). Beeping used to be done in case of panic
in some versions of kernel.

*And* someone actually posted patch to do beeping in acpi wakeup,
search the archives. [It would be nice if you could transform that
patch to something that could be used by #ifdef DEBUG and submitted me
a patch).
								Pavel
-- 
When do you have a heart between your knees?
[Johanka's followup: and *two* hearts?]


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Knut Neumann]

Am Do, 2003-07-24 um 13.00 schrieb Pavel Machek:
> I don't remember code for beeping... To send data to the paralel port,
> 
> 	movb	$0x12, %al
> 	movw	$0x378, %dx
> 	outb	%al, %dx

Ok. I actually got myself up and created a parallel port plug with Leds
- works fine with test program. Nothing on resume/wakeup (I put the
above code into first place in wakeup.S, is that supposed to work?)...so
obviously the computers now for sure never jumps to wakeup address, so
either somethings wrong with that wakeup address or with
pre-jump-to-wakeup-address-stuff. Any idea?

-Knut

-- 
Knut Neumann <knut.neumann-4bfl1RV3iZDOEhgYWvzSCYQuADTiUCJX@public.gmane.org>
Physikalische Grundpraktika - Heinrich-Heine Universitaet Duesseldorf
Raum 25.33.01.63  -   Universitaetsstrasse 1   -  D-40225 Duesseldorf
fon: +49-211-81-11314                           fax: +49-211-81-13105











-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Pavel Machek]

Hi!

> > I don't remember code for beeping... To send data to the paralel port,
> > 
> > 	movb	$0x12, %al
> > 	movw	$0x378, %dx
> > 	outb	%al, %dx
> 
> Ok. I actually got myself up and created a parallel port plug with Leds
> - works fine with test program. Nothing on resume/wakeup (I put the
> above code into first place in wakeup.S, is that supposed to work?)...so
> obviously the computers now for sure never jumps to wakeup address, so
> either somethings wrong with that wakeup address or with
> pre-jump-to-wakeup-address-stuff. Any idea?

1st try to putting this code followed by 

1:	jmp 1b

into, say, powerdown code (or maybe suspend_save_state?) to make sure
the code works as expected. Then send me diff -u wakeup.S.old
wakeup.S. You might want to add that jmp there, too. Not sure if
parport should be powered up at that point, but if not even parport is
powered, I don't know how to debug that.
								Pavel
-- 
Horseback riding is like software...
...vgf orggre jura vgf serr.


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Knut Neumann]

Hi,

Am Mi, 2003-07-30 um 20.22 schrieb Pavel Machek:
> the code works as expected. Then send me diff -u wakeup.S.old

Attached. Code works. Still nothing after resume though - I rememeber
someone debugging S3 via parport/serial so I dont think its still
powered down. 

Actually changing _set_wakeup_vector ((acpi_physical_address)
acpi_wakeup_vector)) to _set_wakeup_vector ((acpi_physical_address) 0))
gives the very same result on resume. 

Thanks 
-Knut

-- 
Knut Neumann <knut.neumann-4bfl1RV3iZDOEhgYWvzSCYQuADTiUCJX@public.gmane.org>
Physikalische Grundpraktika - Heinrich-Heine Universitaet Duesseldorf
Raum 25.33.01.63  -   Universitaetsstrasse 1   -  D-40225 Duesseldorf
fon: +49-211-81-11314                           fax: +49-211-81-13105











-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Pavel Machek]

Hi!

> Am Mi, 2003-07-30 um 20.22 schrieb Pavel Machek:
> > the code works as expected. Then send me diff -u wakeup.S.old
> 
> Attached. Code works. Still nothing after resume though - I rememeber
> someone debugging S3 via parport/serial so I dont think its still
> powered down. 

Sorry, I do not see anything attached.

On two of my machines, even video is up and running when we enter
acpi_wakeup_vector. Machines differ in these little details :-(.

> Actually changing _set_wakeup_vector ((acpi_physical_address)
> acpi_wakeup_vector)) to _set_wakeup_vector ((acpi_physical_address) 0))
> gives the very same result on resume. 

								Pavel

-- 
When do you have a heart between your knees?
[Johanka's followup: and *two* hearts?]


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Knut Neumann]

Am Don, 2003-07-24 um 12.45 schrieb Florian Lohoff:
> Yea - Would be my request too - Currently S3 is top on my wishlist. I
> did a lot of m68k assembly but i tryed to stay as far away from x86
> assembly as i could. So it would be nice if the more experienced ones
> would provide some #define DEBUG where we would hear interesting sounds
> or blinking keyboard leds or parallel/serial output if certain steps are
> handled successfully.

I received some tips from Ole Rohne some time ago, which were:

* Insert lcall $0xffff,$0 or machin_restart in various places in    
  wakeup.S and upwards and see if your machine reboots on S3 resume
* If you have a parallel port, make a dongle with eight LEDs and trace
  your way with OUTB until it hangs

Unfortunately, none of this did anything on resume. We had some more
discussions (see archives for Ole Myren Rohne or my name or S3 vaio),
which ended up with the conclusion, that the wakeup vector does not get
called on resume. I played around with pre-suspend code a little to see
if that changes anything, but it didnt. So I stopped looking into it -
though I would be happy to have S3 running.

I think I will put up a webpage with all the information I have
collected so far. Maybe that will help someone else to track the problem
down.

-Knut




-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Florian Lohoff]

[-- Attachment #1: Type: text/plain, Size: 859 bytes --]

On Thu, Jul 24, 2003 at 01:18:56PM +0200, Knut Neumann wrote:
> I received some tips from Ole Rohne some time ago, which were:
> 
> * Insert lcall $0xffff,$0 or machin_restart in various places in    
>   wakeup.S and upwards and see if your machine reboots on S3 resume

Ok - i tried this - Doesnt work - Machine seems not to execute the
wakeup code at all.

And now ?

I read a bit in the ACPI Specs - It seems there are a 32Bit and a 64Bit 
pointer to the wakeup routine. The 32Bit pointer says "superseeded by
"X_Firmware_Waking_Vector" which seems not to be set in
drivers/acpi/hardware/hwsleep.c:acpi_set_firmware_waking_vector

Might that be one of the causes ?

Flo
-- 
Florian Lohoff                  flo-BCn6idZOOBwdnm+yROfE0A@public.gmane.org             +49-171-2280134
                        Heisenberg may have been here.

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: S3 wakeup code

[Pavel Machek]

Hi!

> > I received some tips from Ole Rohne some time ago, which were:
> > 
> > * Insert lcall $0xffff,$0 or machin_restart in various places in    
> >   wakeup.S and upwards and see if your machine reboots on S3 resume
> 
> Ok - i tried this - Doesnt work - Machine seems not to execute the
> wakeup code at all.
> 
> And now ?
> 
> I read a bit in the ACPI Specs - It seems there are a 32Bit and a 64Bit 
> pointer to the wakeup routine. The 32Bit pointer says "superseeded by
> "X_Firmware_Waking_Vector" which seems not to be set in
> drivers/acpi/hardware/hwsleep.c:acpi_set_firmware_waking_vector

What specs is that? As wakeup code starts in *real* mode, I don't see
how 64bit pointer is going to be usefull...

> Might that be one of the causes ?

I don't think so. lcall $0xffff, $0 is not guaranteed to
reboot. Machine is in pretty weird state at that point.
 

-- 
When do you have a heart between your knees?
[Johanka's followup: and *two* hearts?]


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 suspend code

[Mark Santcroos]

Are there any Linux users with a Dell Latitude C640?

S3 suspend returns immediately, and is therefor not really usefull ;-)

I'm using FreeBSD myself, but have checked this awhile go in Linux too,
and had the same problem.

It would be nice if people can confirm if this is still the case.

If not, are there people that have advise how I can fix this problem.

Like I said, the suspend returns immediately, but it does succeed.
It seems that it gets a wakeup signal immediately though. 

Any help appreciated.

Mark

-- 
Mark Santcroos                    RIPE Network Coordination Centre
http://www.ripe.net/home/mark/    New Projects Group/TTM


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 suspend code

[Pavel Machek]

Hi!

> Like I said, the suspend returns immediately, but it does succeed.
> It seems that it gets a wakeup signal immediately though. 

Like it turns into sleep mode then wakes up? You probably have some
device that is enabled to wake the machine up, which should not really
do that... Check 2.6.0-t1.
								Pavel
-- 
When do you have a heart between your knees?
[Johanka's followup: and *two* hearts?]


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 suspend code

[Mark Santcroos]

On Fri, Jul 25, 2003 at 12:41:41AM +0200, Pavel Machek wrote:
> > Like I said, the suspend returns immediately, but it does succeed.
> > It seems that it gets a wakeup signal immediately though. 
> 
> Like it turns into sleep mode then wakes up? You probably have some
> device that is enabled to wake the machine up, which should not really
> do that... Check 2.6.0-t1.

I've put a different disk in my laptop and installed Linux on it.
With 2.6.0-t1 it does the right thing. Also 2.5.75 works. 
I'm going do to a binary search to find the specific change that fixed it.

Mark

-- 
Mark Santcroos                    RIPE Network Coordination Centre
http://www.ripe.net/home/mark/    New Projects Group/TTM


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 suspend code

[Mark Santcroos]

Hi,

On Sun, Jul 27, 2003 at 01:41:29AM +0200, Mark Santcroos wrote:
> On Fri, Jul 25, 2003 at 12:41:41AM +0200, Pavel Machek wrote:
> > > Like I said, the suspend returns immediately, but it does succeed.
> > > It seems that it gets a wakeup signal immediately though. 
> > 
> > Like it turns into sleep mode then wakes up? You probably have some
> > device that is enabled to wake the machine up, which should not really
> > do that... Check 2.6.0-t1.
> 
> I've put a different disk in my laptop and installed Linux on it.
> With 2.6.0-t1 it does the right thing. Also 2.5.75 works. 
> I'm going do to a binary search to find the specific change that fixed it.

It looks like I had confused my earlier tests with something else, because
also old Linux kernels do it "correctly".

My machine suspends as early as with 2.5.6 (the first kernel with
/proc/acpi/sleep), note that it doesn't resume correctly up until 2.5.30 or 
so, but that doesn't really matter.

Do you know if there is any way to find out what device or bus triggers
the wakeup? A state stored somewhere or so.

Thanks

Mark

-- 
Mark Santcroos                    RIPE Network Coordination Centre
http://www.ripe.net/home/mark/    New Projects Group/TTM


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 suspend code

[Pavel Machek]

Hi!

> > > > Like I said, the suspend returns immediately, but it does succeed.
> > > > It seems that it gets a wakeup signal immediately though. 
> > > 
> > > Like it turns into sleep mode then wakes up? You probably have some
> > > device that is enabled to wake the machine up, which should not really
> > > do that... Check 2.6.0-t1.
> > 
> > I've put a different disk in my laptop and installed Linux on it.
> > With 2.6.0-t1 it does the right thing. Also 2.5.75 works. 
> > I'm going do to a binary search to find the specific change that fixed it.
> 
> It looks like I had confused my earlier tests with something else, because
> also old Linux kernels do it "correctly".
> 
> My machine suspends as early as with 2.5.6 (the first kernel with
> /proc/acpi/sleep), note that it doesn't resume correctly up until 2.5.30 or 
> so, but that doesn't really matter.
> 
> Do you know if there is any way to find out what device or bus triggers
> the wakeup? A state stored somewhere or so.
> 

I do not know how to tell which device is that. Older kernels simply disabled all wakeups
altogether (that change should be easy to spot, it is somewhere really close to entering sleep),
which is wrong but worked for a lot of people. This change was taken/reverted few times, iirc.
-- 
				Pavel
Written on sharp zaurus, because my Velo1 broke. If you have Velo you don't need...



-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Florian Lohoff]

[-- Attachment #1: Type: text/plain, Size: 968 bytes --]

On Thu, Jul 24, 2003 at 10:04:11PM +0200, Pavel Machek wrote:
> > 
> > I read a bit in the ACPI Specs - It seems there are a 32Bit and a 64Bit 
> > pointer to the wakeup routine. The 32Bit pointer says "superseeded by
> > "X_Firmware_Waking_Vector" which seems not to be set in
> > drivers/acpi/hardware/hwsleep.c:acpi_set_firmware_waking_vector
> 
> What specs is that? As wakeup code starts in *real* mode, I don't see
> how 64bit pointer is going to be usefull...
> 

ACPI 2.0 Page 128 - I think these are extensions to support ia64 - But
forget about that - used debug code - its using the 32bit pointer.

And i was misreading the code - Basically it puts a "generic" pointer
to the right Firmware vector inthe gbl_common_fACS and writes to the
previously defined position.

Flo
-- 
Florian Lohoff                  flo-BCn6idZOOBwdnm+yROfE0A@public.gmane.org             +49-171-2280134
                        Heisenberg may have been here.

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: S3 wakeup code

[Ole Myren Rohne]

Florian Lohoff <flo-BCn6idZOOBwdnm+yROfE0A@public.gmane.org> writes:

> > * Insert lcall $0xffff,$0 or machin_restart in various places in    
> >   wakeup.S and upwards and see if your machine reboots on S3 resume
> 
> Ok - i tried this - Doesnt work - Machine seems not to execute the
> wakeup code at all.

IIRC, at one point linux wrote the kernel VM address to the firmware
wakeup vector, typically 0xc0000000 + hardware address. Now, there is
of course only one way to interpret that as a real-mode seg:off
address, but it did strike me as weird. I once checked that the
leading 0xc is irrelevant on my P2120, but YMMV.

I would also check that the suspend/resume doesn't choke somewhere in
32-bit land by jumping over the call to acpi_enter_sleep_state in
wakeup.S.

Ole





-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Florian Lohoff]

[-- Attachment #1: Type: text/plain, Size: 1147 bytes --]

On Fri, Jul 25, 2003 at 02:25:36PM +0200, Ole Myren Rohne wrote:
> 
> IIRC, at one point linux wrote the kernel VM address to the firmware
> wakeup vector, typically 0xc0000000 + hardware address. Now, there is
> of course only one way to interpret that as a real-mode seg:off
> address, but it did strike me as weird. I once checked that the
> leading 0xc is irrelevant on my P2120, but YMMV.

Hmmm - I didnt have a look at that - I was assuming that alloc_bootmem
returned a physical address. In 2.5 it seems we dont write out the
allocated address on boot as we did in 2.4 (hint, hint)

> I would also check that the suspend/resume doesn't choke somewhere in
> 32-bit land by jumping over the call to acpi_enter_sleep_state in
> wakeup.S.

I kind of had to debug this on this machine - No legacy hardware nice
for debugging (Parallel, Serial etc). Even the speaker is switched
through the sound card so when the chip is still powered down - no
sound.

Flo
-- 
Florian Lohoff                  flo-BCn6idZOOBwdnm+yROfE0A@public.gmane.org             +49-171-2280134
                        Heisenberg may have been here.

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: S3 wakeup code

[Knut Neumann]

Am Fre, 2003-07-25 um 14.50 schrieb Florian Lohoff:
> Hmmm - I didnt have a look at that - I was assuming that alloc_bootmem
> returned a physical address. In 2.5 it seems we dont write out the
> allocated address on boot as we did in 2.4 (hint, hint)

Without a look and everything I remember that S3 once at least worked
with disk and stuff resuming - no video though. I actually though it
stopped working in course of 2.5 - but might have been the change from
2.4 to 2.5. You wouldnt mind testing with 2.4, would you?

-Knut



-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Florian Lohoff]

[-- Attachment #1: Type: text/plain, Size: 874 bytes --]

On Fri, Jul 25, 2003 at 03:10:38PM +0200, Knut Neumann wrote:
> Am Fre, 2003-07-25 um 14.50 schrieb Florian Lohoff:
> > Hmmm - I didnt have a look at that - I was assuming that alloc_bootmem
> > returned a physical address. In 2.5 it seems we dont write out the
> > allocated address on boot as we did in 2.4 (hint, hint)
> 
> Without a look and everything I remember that S3 once at least worked
> with disk and stuff resuming - no video though. I actually though it
> stopped working in course of 2.5 - but might have been the change from
> 2.4 to 2.5. You wouldnt mind testing with 2.4, would you?

S4/swsusp works like a charm for me on 2.4 - S3 is not available 
for 2.4 i thought ?

Flo
-- 
Florian Lohoff                  flo-BCn6idZOOBwdnm+yROfE0A@public.gmane.org             +49-171-2280134
                        Heisenberg may have been here.

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: S3 wakeup code

[Florian Lohoff]

[-- Attachment #1: Type: text/plain, Size: 839 bytes --]

On Fri, Jul 25, 2003 at 02:25:36PM +0200, Ole Myren Rohne wrote:
> IIRC, at one point linux wrote the kernel VM address to the firmware
> wakeup vector, typically 0xc0000000 + hardware address. Now, there is
> of course only one way to interpret that as a real-mode seg:off
> address, but it did strike me as weird. I once checked that the
> leading 0xc is irrelevant on my P2120, but YMMV.

I tried to make a &0xfffff on setting of the address - no success.

> I would also check that the suspend/resume doesn't choke somewhere in
> 32-bit land by jumping over the call to acpi_enter_sleep_state in
> wakeup.S.

That one works without any problem ...

Flo
-- 
Florian Lohoff                  flo-BCn6idZOOBwdnm+yROfE0A@public.gmane.org             +49-171-2280134
                        Heisenberg may have been here.

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: S3 wakeup code

[David G Hamblen]

Hi,

My problem is the backlight on a Dell Inspiron 8100 with the nVidia VGA
controller. Everything seems to work as expected in all modes detected
(S0, S1, S3, S4bios, S4, S5), except that the backlight stays on after
"echo [1,3] > /proc/acpi/sleep".  S1 just refrigerates everything, and
leaves the display on, resume works fine.

S3 refrigerates everything and suspends, leaving the display white.
Resuming turns the backlight off.  I can retrieve the backlight with the
LCD/CRT button.

Both S4's behave properly (full power off/restore with the power button).

What part of the code should I look at to attack this problem (ACPI
tables, linux/arch/i386/kernel/acpi/*, or linux/drivers/acpi/*)?  DSDT
compiles ok with the iasl assembler/disassembler, and Windows XP goes into
standby properly; so I assume that the ACPI tables can be driven by the
OS.  Perhaps something along the lines of the asus_acpi code?

	dave


On Wed, 23 Jul 2003, Pavel Machek wrote:

> Hi!
>
> > i just tried 2.6.0-test1 and to my expectation the S3 resume stuff still
> > does not work on the Sony Vaio PCG-C1MHP aka Picturebook. It turns on -
> > Bios set the powerled correct but then - nothing - Nothing on the
> > screen, no backlight, no disk activity.
> >
> > Is there any way to dump the Windows wakeup code ?
> > Is that machine specific ?
>
> Wakeup code should *not* be machine-specific, but video card
> save/restore is obviously video-card specific (and you basically can't
> get docs). Write beeping code in 8086 assembly and start debuggin'
> ;-))).
>
> 								Pavel



-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[M. Warner Losh]

In message: <Pine.LNX.4.44.0307250755460.2827-100000-zmH7q5lE+0tBz13eCZem0Q@public.gmane.org>
            David G Hamblen <dave-RfZrpLcwWZ/QT0dZR+AlfA@public.gmane.org> writes:
: S3 refrigerates everything and suspends, leaving the display white.
: Resuming turns the backlight off.  I can retrieve the backlight with the
: LCD/CRT button.

I've seen exactly the same problem under FreeBSD which has supported
S3 state for a long time.  I've seen the problem going back at least 9
months.

Warner


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Mark Santcroos]

On Fri, Jul 25, 2003 at 07:25:04AM -0600, M. Warner Losh wrote:
> In message: <Pine.LNX.4.44.0307250755460.2827-100000-zmH7q5lE+0tBz13eCZem0Q@public.gmane.org>
>             David G Hamblen <dave-RfZrpLcwWZ/QT0dZR+AlfA@public.gmane.org> writes:
> : S3 refrigerates everything and suspends, leaving the display white.
> : Resuming turns the backlight off.  I can retrieve the backlight with the
> : LCD/CRT button.
> 
> I've seen exactly the same problem under FreeBSD which has supported
> S3 state for a long time.  I've seen the problem going back at least 9
> months.

This needs to be handled by the video driver afaik.

Mark

-- 
Mark Santcroos                    RIPE Network Coordination Centre
http://www.ripe.net/home/mark/    New Projects Group/TTM


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[David G Hamblen]

On Fri, 25 Jul 2003, Mark Santcroos wrote:

> On Fri, Jul 25, 2003 at 07:25:04AM -0600, M. Warner Losh wrote:
> > In message: <Pine.LNX.4.44.0307250755460.2827-100000-zmH7q5lE+0tBz13eCZem0Q@public.gmane.org>
> >             David G Hamblen <dave-RfZrpLcwWZ/QT0dZR+AlfA@public.gmane.org> writes:
> > : S3 refrigerates everything and suspends, leaving the display white.
> > : Resuming turns the backlight off.  I can retrieve the backlight with the
> > : LCD/CRT button.
> >
> > I've seen exactly the same problem under FreeBSD which has supported
> > S3 state for a long time.  I've seen the problem going back at least 9
> > months.
>
> This needs to be handled by the video driver afaik.

I'm using the vesafb driver; I also tried with non-framebuffer VGA console
with the same behavior.  I'm not running XF86 for these experiments.
Should I look into the vesafb code?


	dave



-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Mark Santcroos]

On Fri, Jul 25, 2003 at 09:39:34AM -0400, David G Hamblen wrote:
> > > : S3 refrigerates everything and suspends, leaving the display white.
> > > : Resuming turns the backlight off.  I can retrieve the backlight with the
> > > : LCD/CRT button.
> > >
> > > I've seen exactly the same problem under FreeBSD which has supported
> > > S3 state for a long time.  I've seen the problem going back at least 9
> > > months.
> >
> > This needs to be handled by the video driver afaik.
> 
> I'm using the vesafb driver; I also tried with non-framebuffer VGA console
> with the same behavior.  I'm not running XF86 for these experiments.
> Should I look into the vesafb code?

I don't think there is a generic solution for this. We need specific code 
for every video driver that needs it.

That's my educated guess though.

Mark

-- 
Mark Santcroos                    RIPE Network Coordination Centre
http://www.ripe.net/home/mark/    New Projects Group/TTM


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Pavel Machek]

Hi!

> > > : S3 refrigerates everything and suspends, leaving the display white.
> > > : Resuming turns the backlight off.  I can retrieve the backlight with the
> > > : LCD/CRT button.
> > >
> > > I've seen exactly the same problem under FreeBSD which has supported
> > > S3 state for a long time.  I've seen the problem going back at least 9
> > > months.
> >
> > This needs to be handled by the video driver afaik.
> 
> I'm using the vesafb driver; I also tried with non-framebuffer VGA console
> with the same behavior.  I'm not running XF86 for these experiments.
> Should I look into the vesafb code?

I believe you need to use specific kernel fb driver, and write
suspend/resume code for it :-(.
							Pavel
-- 
When do you have a heart between your knees?
[Johanka's followup: and *two* hearts?]


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Mark Santcroos]

On Fri, Jul 25, 2003 at 05:31:10PM +0200, Pavel Machek wrote:
> I believe you need to use specific kernel fb driver, and write
> suspend/resume code for it :-(.

I did it once for my ATI Radeon 7500, I will look it up and post it here.

Mark

-- 
Mark Santcroos                    RIPE Network Coordination Centre
http://www.ripe.net/home/mark/    New Projects Group/TTM


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[David G Hamblen]

On Fri, 25 Jul 2003, Mark Santcroos wrote:

> On Fri, Jul 25, 2003 at 05:31:10PM +0200, Pavel Machek wrote:
> > I believe you need to use specific kernel fb driver, and write
> > suspend/resume code for it :-(.
>
> I did it once for my ATI Radeon 7500, I will look it up and post it here.

Stuff like this might help.  I haven't looked closely, but grepping thru
linux/drivers/video shows a few drivers (including atiradeonfb.c) that
have code relating to backlight power.

Dave



-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Pavel Machek]

Hi!

> My problem is the backlight on a Dell Inspiron 8100 with the nVidia VGA
> controller. Everything seems to work as expected in all modes detected
> (S0, S1, S3, S4bios, S4, S5), except that the backlight stays on after
> "echo [1,3] > /proc/acpi/sleep".  S1 just refrigerates everything, and
> leaves the display on, resume works fine.
> 
> S3 refrigerates everything and suspends, leaving the display white.
> Resuming turns the backlight off.  I can retrieve the backlight with the
> LCD/CRT button.
> 
> Both S4's behave properly (full power off/restore with the power button).
> 
> What part of the code should I look at to attack this problem (ACPI
> tables, linux/arch/i386/kernel/acpi/*, or linux/drivers/acpi/*)?  DSDT
> compiles ok with the iasl assembler/disassembler, and Windows XP goes into
> standby properly; so I assume that the ACPI tables can be driven by the
> OS.  Perhaps something along the lines of the asus_acpi code?

Video is hard to do... You should write specific suspend/resume
support for your video card. Bad luck if you are using vesafb :-(.

> > Wakeup code should *not* be machine-specific, but video card
> > save/restore is obviously video-card specific (and you basically can't
> > get docs). Write beeping code in 8086 assembly and start debuggin'
> > ;-))).


-- 
When do you have a heart between your knees?
[Johanka's followup: and *two* hearts?]


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Karol Kozimor]

Thus wrote David G Hamblen:
> standby properly; so I assume that the ACPI tables can be driven by the
> OS.  Perhaps something along the lines of the asus_acpi code?

The asus_acpi module uses the ASUS-specific DSDT methods to handle the
backlight switching. I have quite the same problem with S1 on my machine,
but I get around it by doing echo 0 > /proc/acpi/asus/lcd; echo 1 >
/proc/acpi/sleep; echo 1 > /proc/acpi/asus/lcd. You may want to look for
the relevant methods in your DSDT and write a kernel interface to call
them, so that you could then activate it prior to suspending. I suppose
that's exactly what the Windows drivers do. You may also try something
video-specific, i.e. radeontool.

BTW: Pavel: will there be a place in the new PM interface for functions
that drivers could bind onto, i.e. something like a generic
disable_backlight(), assigned to
{asus,dell,toshiba,nvidia,radeon,etc.}_disable_backlight()? I think this
could solve some important problems.

Best regards,

-- 
Karol 'sziwan' Kozimor
sziwan-DETuoxkZsSqrDJvtcaxF/A@public.gmane.org


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[David G Hamblen]

On Fri, 25 Jul 2003, Karol Kozimor wrote:

> Thus wrote David G Hamblen:
> > standby properly; so I assume that the ACPI tables can be driven by the
> > OS.  Perhaps something along the lines of the asus_acpi code?
>
> The asus_acpi module uses the ASUS-specific DSDT methods to handle the
> backlight switching. I have quite the same problem with S1 on my machine,
> but I get around it by doing echo 0 > /proc/acpi/asus/lcd; echo 1 >
> /proc/acpi/sleep; echo 1 > /proc/acpi/asus/lcd. You may want to look for
> the relevant methods in your DSDT and write a kernel interface to call
> them, so that you could then activate it prior to suspending. I suppose
> that's exactly what the Windows drivers do. You may also try something
> video-specific, i.e. radeontool.
>
> BTW: Pavel: will there be a place in the new PM interface for functions
> that drivers could bind onto, i.e. something like a generic
> disable_backlight(), assigned to
> {asus,dell,toshiba,nvidia,radeon,etc.}_disable_backlight()? I think this
> could solve some important problems.

The /linux/drivers/skeletonfb.c file has some potential sample hooks for
this kind of thing.  Several of the drivers in that directory have code
drive the various power states.  When I get a chance, I might try hacking
the vesafb driver.  I don't know if I can get any hints out of the nv
driver for XF86.


Dave




-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01

Re: S3 wakeup code

[Pavel Machek]

Hi!

> BTW: Pavel: will there be a place in the new PM interface for functions
> that drivers could bind onto, i.e. something like a generic
> disable_backlight(), assigned to
> {asus,dell,toshiba,nvidia,radeon,etc.}_disable_backlight()? I think this
> could solve some important problems.

I guess you should create sysfs node for your video card, and hook
your methods there. If you can't place it in the sysfs hierarchy place
it into system devices...

BTW ACPI lid refresh code should be also moved to sysfs like that...

								Pavel
-- 
When do you have a heart between your knees?
[Johanka's followup: and *two* hearts?]


-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01