sleep - what are the odds of it working sometime?

I'm looking to pick up a netbook to run Leopard, and it's going to be either an AA1 or an MSI Wind. I like the look of the AA1 better (especially in blue) and it's cheaper, but since this will be primarily for couch surfing I really would like to be able to use sleep, and sleep DOES work on the Wind. What is the problem with sleep on the AA1? I thought most of these netbooks used the same processor and chipset, so there must be something on the AA1 that's preventing sleep from working....

 

I would say something to do with graphics, but no idea how to check.
Rory

 

I've seen reports that sometimes it has to do with the SSD, but I'm guessing quite a lot of people with AA1s don't have SSD and sleep doesn't work. Don't the Wind and AA1 have the same onboard graphics?

 

Yeah but there are 4 revisions of the GMA950 so they could be different. Ours is 0x27ae.
I am trying out something now.... Long shot but might save us using the kexts.
Rory

 

I'm pretty sold on the Aspire One over the MSI and I have confidence in the community to solve the sleep problem Seems like $380 is a good price for 6 cell with 160gb HD.

 

Well tonight I am actively trying to fix sleep.
The console.log and system.log are useless and give no indication as to why it restarts when trying to sleep.
Tried various drivers, and have found updated drivers for the GMA 950 chipset.
Still think the graphics is to blame. Removing the graphics drivers makes it sortof stop but not go off. So clearly something else is to blame too.
Rory

 

Good luck sir!

 

Anything from dmesg? Or getting deeper into it dtrace but on which process?
I've had a look at man syslogd and putting my desktop into debug mode doesn't seem to do anything :?:

Disclaimer: I Am Not A Programmer, but can I be any use sending you logs, traces, if you tell me what to look for?
Thanks for the comment above about 4 revisions of GMA950. Ppl forget that goes on in a manufacturing world...

 

To comment on one request, tonight I managed to use the patch27ae.command script from the MSI wind forums to patch a 10.5.5 GMA950 driver. Not sure what is different from my attempts to do this by hand, but it does work. And it puts a known driver on the system, which is better than the old patched binary that may have other sleep problems.

The wind does have sleep working. Do we know what is different?

Rory - The removing graphics driver is the only data point or do you have others?

jeffs

 

Thats the only thing so far. I think its inbuilt kexts because if you literally just install then try and sleep, it does the same as if you install all kexts except graphics. what I need is a log which tells me what caused sleep to fail
Rory

 

guys,

i spent many hours working on sleep on this one. it is due to the graphics drivers, specifically the framebuffer. im fairly certain it is due to the binary patching of this file. as there is no current way to get hw accelerated graphics working w/o this file being binary patched for use and for our dev ids, i think getting sleep working is a long shot. If you were going to focus your energy somewhere i would say it would be on using a plist only kext file to get the graphics working.

if you are serious about spending time on this, jump on irc.osx86.hu in the xnu-test / xnu-chat rooms. this is where the big names in the osx86 / hackintosh world discuss / work out issues.

fyi, i have my machine able to go to sleep but not able to wake up currently. i am losely trying to get this working but its not #1 prio atm.

dave

 

Thanks dfiore. I will listen to the IRC and get my feet wet there. I have not seen much macos or reverse engineering in my days. I'm used to having source and knowing how the chip works :?

Is the theory the binary patch causes a bug or incorrect code path, or something fundamental about patched binaries in OSX?

 

I have a MSI Wind U100.

I looked at the graphics card in the "about this Mac" section on the computer. They are the same.

I have this file called MSIdrivers or something. I think there is a sleep driver on there. I am going to try it out on my computer.

Here is what I found on the "about this mac"

Device ID: 0x27ae
Revision ID: 0x0003

Hope that helps.

 

luke, the hardware is simalar with the wind, but not the same. i dont have a wind so im not entirely sure on the differences, it may even be something as simple as a bios setting (thing sleep state , s1 , s3 ect)

jeffs, yes that is exactly it, the driver is not fully developed, just a hack to make it work with the existing hardware, thus causing a kernel panic.

i have given this some thought, and i think i have an idea on how we can try to make this work. instead of replacing the apple drivers with patched versions of our own, why not rename the driver bundles (kexts, and plist declarations in side them) and put them side by side with the mac ones. Im not entirely sure this would work but if the issue is that sleep, amongst others is looking for the factory drivers, this could work.

i think we should grab a pack of all of the latest drivers, audio, video, cardreader, disabler, and decrypter and rename them all not to interfere with the apple spec ones. once we do this, we may find that no other work will be required to make sleep work, and as an added benefit apple updates wont break our machines so long as they dont change at the core how the original drivers work.

thoughts?

 

Makes sense.. If I wasnt on such a slow pipe I would look into trying to get the original GMA 950 kexts out of the 10.5.6 Update and trying it. Not sure how easy it will be... Hope it works...

 

I already have all the kexts from 10.5.6, they werent installed on update, but i extracted them and installed them. all are newer versions than any i have seen elswhere but still no sleep. does the same as it does with the kexts in my pack. tries to sleep and fails and reboots
R

 

rory,

did you binary edit the framebuffer and gma kexts to work with the aspire or did they work out of the box?

also are you booting retail (w efi partition) or editing S/L/E

dave

 

Just to help motivate you guys.. The Wind isn't the only Netbook that sleeps, the Mini9 sleeps great too. They initially had trouble waking up from sleep, but it turned out to be fixed by changing to another brand of SSD (the RunCore drives wake up just fine).
Some of the Eee's sleep too, but there seems to be about a dozen different versions of that Netbook, I don't know how they get anything done like that.
The video drivers we are using are linked here:
http://forum.insanelymac.com/index.php? ... 4363&st=25
Second page of that huge thread, PeteRock posted some video drivers that sleep just fine.

We nailed down our SSD being the problem by opening terminal, do a "tail -f /var/logs/system.log"
and then putting it to sleep, when it comes back up, you might get lucky and see what kind of error messages are coming through.
Good luck guys. and have a Merry Christmas.

 

The patch27ae.command script does as suggested and moves the two GMA kexts to 0x27ae<name> and leaves the originals intact. This is a good construct for upgrades too.

One of the issues is I think everyone has a slight different install set-up. The only kext I have overwritten now is AppleHDA.kext

Install technique is boot-132 retail CD to MBR disk with munky EFI dsdt boot. I'm at 10.5.5 still. Hopefully the details don't matter too much.

The symptom I see is
10.5.5 kernel - system hangs, then eventually reboots to the BIOS screen
voodoo - system panics with a fault at 0x14. So a NULL pointer deference in the IOKit "dgraph_add_dependency".

I will try to isolate to the graphics driver like others have suggested. I think my next step is to try to set-up network GDB, apple & voodoo symbols. I hope the realtek driver supports network gdb.

[edit] My system w/o the two graphics drivers does not get far into the sleep process. The desktop is still active, but the system still thinks it is sleeping so you have to hard reboot. Reboot command does nothing. So it goes farther with the graphics kext isntalled :-(

As pointed out, I do suspect BIOS settings differences given the other netbooks can be made to work with patched binaries.

 

I broke down and ordered my AAO yesterday so I'll finally get to play around with this stuff. I appreciate all of the work you guys are trying to do here, and will help out any way I can once mine arrives.