68 replies to this topic

[rayap]

This topic regularly crept in the CMOS Resets After Sleep thread. So I thought it's about time to move the discussion to a new thread.
In my setup with EP45-DS4P, with SnowLeopard, Power Button and WOL Wakes also brings up the display within a second or two. But the 'Wake Reason=' in the log is empty.  
However, with Lion the display fails to come up and it just sleeps again within a few seconds. Nevertheless,  a tap within the few seconds, on my trackpad or keyboard/mouse brings the display to life. Later checking the unsuccessful wakes in the log show the 'Wake Reason = ?'
On the other hand, my wife's setup on a 945GCM-S2L with GMA 950 graphics, has no such problems in Snow Leopard or Lion. As a matter of fact the kernel log even show the Wake Reason as 'power-button (User)'.

Otherwise, all forms of Sleep (including power-button sleep) and Wake (USB & Bluetooth) works normally.

Over the last couple of weeks, testing out various mods by comparing the two dsdt's proved unsuccessful. I may have to move my HD4850 to my wife's setup to see it the ATI card is the problem - but must wait for her to go on a short holiday. Or maybe use a analogue connected monitor- yet searching for the adaptor.

Those of you with GMA950 graphics can confirm if Lion wakes up by the Power Button successfully.

Edited OP on 23 July, 2011 - Include WOL.

After post#26
Power-button wakes work succesfully on GA-945GCM-S2L with GT220 graphics too.

After post#34
Overcomes topic's problem with darkwake=0 kernel parameter.

[miqlas]

Have You patched the power button in Your DSDT?

[rayap]

If, unpatched even power-button sleep will not function let alone power-button wake.
Will be useful to see your signature.

[rayap]

In this post by antic, replacing the USB Keyboard by a Bluetooth Keyboard reportedly solved this problem.
Tested with Apple Wireless Keyboard and Trackpad with no USB Devices attached - other than a Bluetooth Dongle - problem still persists.
Additionally, tried changing framebuffer for HD4850 between Motmot (2 display working) and default Radeon (1 display only working), but of no avail.

[nissefar]

rayap, on Jul 11 2011, 12:53 PM, said:

Additionally, tried changing framebuffer for HD4850 between Motmot (2 display working) and default Radeon (1 display only working), but of no avail.

I have the same problem regardless if I use a GTX260 or a HD6850. Additionally, I think there are several people who do not have the problem who do not have a GMA 950.

This is probably more likely to be a motherboard specific problem, requiring a DSDT fix or the like. My knowledge of DSDT patching is pretty weak though.

I think getting to know why "Wake Reason" is empty would be an important clue, and I've tried to figure out where the "Wake Reason" is determined, but I've not had much luck. I downloaded the XNU kernel sources and poked around, but only found one mention about it in a header file.

More likely is it that "Wake Reason" is determined in a kext module such as the Intel Power Management kext.

[nissefar]

Okay, the relevant Kext is AppleACPIPlatform.

In the binary I found the string Wake reason: %s.

Is that Kext closed source?

[rayap]

As per my Opening Post, in the DS4P (where power-button wake fails)  I noticed  AppleSMCClient is missing in ioreg whereas on the S2L, this is loaded. Could i seek feedback, if those who are successful in Power Button Wakes in Lion confirm their AppleSMcClient is loaded or if otherwise.

And, could some expert throw light on the function of AppleSMCClient and how to enable it.

Thanks.

[nissefar]

Have you tried if the SL AppleSMCClient works in Lion and if that would make a difference?

[antipop2323]

rayap, on Jul 17 2011, 01:38 AM, said:

As per my Opening Post, in the DS4P (where power-button wake fails)  I noticed  AppleSMCClient is missing in ioreg whereas on the S2L, this is loaded. Could i seek feedback, if those who are successful in Power Button Wakes in Lion confirm their AppleSMcClient is loaded or if otherwise.

And, could some expert throw light on the function of AppleSMCClient and how to enable it.

Thanks.

I have the same issue with power button wakes and it going back to sleep without an additional click, yet AppleSMCClient is loaded in my ioreg.  Incidentally, it's not only my power button, but also waking from magic mouse/magic trackpad, it requires two clicks as opposed to one (with one the system wakes with no monitor and then shortly goes back to sleep, as would just pressing the power button).  One thing to note though, both my wired apple aluminum keyboard and wireless wake with one click, as well as my wired mouse.  So I just have this problem with power button, magic mouse, and magic trackpad.

[nissefar]

What does the kernel log say if you press your magic mouse or trackpad just once? "wake reason: ?" maybe?  Then you know you have the same problem.

[rayap]

nissefar, on Jul 17 2011, 10:34 AM, said:

Have you tried if the SL AppleSMCClient works in Lion and if that would make a difference?

AppleSMCClient is not loaded so i do not know if it makes a difference. In post #9 antipop2323 says, in his case it is loaded but problem persists.

@antipop2323 - In my DS4P case , tested Apple AL Keyboard, Magic Trackpad, Magic Mouse and Wired Mouse all work with one click and HID Tickle appears automatically in less than 200 msecs. As for Power-Buttom and WOL, only wakes with second click or just touch of trackpad, mouse or button and HID Tickle appears accordingly; if re-sleep has not already happened. As for S2L, only WOL requires second click.

[antipop2323]

nissefar, on Jul 17 2011, 07:40 AM, said:

What does the kernel log say if you press your magic mouse or trackpad just once? "wake reason: ?" maybe?  Then you know you have the same problem.

Power Button Without Addition Click:
http://pastebin.com/Rd9Kt5r6

Magic Trackpad 1 Click:
http://pastebin.com/UHJtV7Mr

Wireless Keyboard 1 Click
http://pastebin.com/MXNgSk7r

I'm not convinced that the power button wake fails because there isn't a properly identified wake reason, clearly it is partially waking up with the power button... for some reason, there is no HID Tickle and the graphics are suppressed, which you'll also note happens with 1 click of the Magic Tracpad, even though it has a wake reason of Wake reason: IGBE HDEF USE2...

[TouchMyBox]

I have this, also, when awaking, my speedstep stops working. :\

[nissefar]

I haven't noticed if this is the case for me. What app can I monitor my speedstep state with again? Some time since I did it last after I got it working in SL.

[TouchMyBox]

I'm using istat menus

It tells you a rough reading of your CPU volage, speedstep for me means 0.89 V when idle. After waking it stays at 1.26 V

[antipop2323]

this is pretty much my only hackintosh specific lion issue, but am not sure where we go from here... perhaps if anyone without the graphics suppression issue with power button wakes without additional mouse/keyboard strokes could post their DSDT we could compare...

[nissefar]

This is my only issue as well.

I'm just wondering how not more people could be affected by this, as it must be something that plagues at least anyone with the same motherboards as ours. I figured the EP45-UD3(XX) motherboard to quite popular around here.

[rayap]

As per my OP, had a chance to install the HD4850 into the 945GCM-S2L but failed to get it working satisfactorily within that day. Reverted to GMA950 graphics. After stripping the S2L-dsdt including the WAK and PTS methods, sleeps and wakes in SL were AOK, while in Lion only WOL lead to no display and re-sleep. Also WOL can always boot if system is on soft-off.
As for the DS4P, no change or improvement, and WOL when system is soft-off failed to boot most of the time.
Must check in Windows the LAN wake properties, as both boards have the same RTL 8111C.

[nissefar]

It must have something to do with the new low power wake state introduced in Lion.

On genuine Macs (I have a 2007 Mini and a 2010 MBA), if you access network services or just do WOL the systems will wake with "Graphics Suppressed". The systems stay awake and do not go instantly back to sleep, so bonjour advertised network services can be accessed. They can then be woken fully with mouse/keyboard ("HID Tickle").

If I do "pmset -g log" on my EP45-UD3LR, it lists only wakes where it knows the reason, ie. USB EHC UHC. Since it can't recognize PWB1 or PEX wakes, I think powerd perhaps will reinitiate sleep instantly.

So I think either we need to get these wake reasons recognized (I have no idea if that is determined in DSDT), or perhaps it would be possible to downgrade powerd to the SL version (not the greatest solution).


Anyway which motherboards or not this is a problem for isn't dependent on if it includes GMA950 or not... I guess most GMA950 including boards just got a more Intel/Apple friendly power mangement/DSDT/Bios. Remember that there are Macs that do not have an Intel graphics card.

powerd sources released: http://opensource.ap...e/mac-os-x-107/

[rayap]

Installed a cheap nVidia GT220 in the 945GCM-S2L setup. Status is same as with GMA950 graphics (Lion 11A511). Power button wakes are successful but for WOL system re-sleeps.