Solution for intel3000? Maybe. [Test required]

[masso912]

I've messed up some info.plist of the Kext/bundle for intel3000 and now i've not seen a single glitch since 3 hour.
If you would like to test it out, these are the files.

(Note that some change i made are only for the MacbookPro8,1 smbios).

I started running all the apps one by one, this is the result.
No glitches so far

(since i left open Facebook, i censured names and other thing, so don't worry about white boxes).

EDIT: This is my boot.plist
I have graphic enabler non activated and the EFI string is just added because i wanted to test if it was possible to rollback the driver with those of snow leopard, but since i'm testing this solution with lion driver probably you can delete it. Adding it cause no issue.

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>EthernetBuiltIn</key>
<string>Yes</string>
<key>GUI</key>
<string>Yes</string>
<key>Quiet Boot</key>
<string>No</string>
<key>SystemType</key>
<string>2</string>
<key>UseKernelCache</key>
<string>Yes</string>
<key>device-properties</key>
<string>600000000100000001000000540000000100000002010c00d041030a000000000101060000027fff04001e0000004100410050004c002c006f0073002d0069006e0066006f0000001800000030490111111108000001f01f0100000010070000</string>
</dict>
</plist>
</plist>

 

[blueking]

Can you please detail what changes you made?

 

[masso912]

Can you please detail what changes you made?

Yes, actually i don't even remember everything.
I start next by Kext...

AppleVaDriver.bundle
Every graphic Kexts for intel has a "XXXXVA.bundle" (xxx name of driver).
In fact, it exist also a SNBVA.bundle.
In these info.plist there are settings for every mac.
In the AppleVaDriver.bundle there are a lot of intel based mac.
The Macbook 8,1 misses, so i added this key

<key>MacBookPro8,1</key>
<dict>
<key>bDualGPU</key>
<true/>
<key>boostIG</key>
<true/>
</dict>

I putted dual gpu to true because i was testing if ati would have worked, but it's not.
This change probably don't fix nothing.

AppleIntelSNBVA.bundle

Like on previous VA driver, there's the section with mac models.
But this time there's only one code.

<dict>
<key>TSNBCRB1,1</key>
<dict>
<key>boostIG</key>
<true/>
<key>preferredBGRA</key>
<true/>
</dict>
</dict>

TSNBCRB1,1 is probably a Mac Debug code. I left unchanged this (it should be like this by default),
because i remember that when i changed it to false in "preferredBGRA" and "boostIG"
Flash player crashes on youtube video, either on firefox and safari, so i reverted back.


AppleIntelSNBGraphicsFB.kext

Here i changed the .plist comparing it to the Snow leopard one.
The most interesting value was "GPUInterruptHandling".
there are 3 controllers listed on the driver, probably the last one is the right one.
I changed the 3 controllers making them similar to snow leopard. (too long to write everything, check the snow leopard
driver and mine).

This is the result in IOreg for this frame buffer.

AppleIntelHDGraphicFB.kext

Here i tried to change the GPUInterruptHandling to 1, and maybe is the cause of the flash crash that i mentioned above.
(i tested the changes contemporaneously, so i really don't know if boostIG was the problem or this).
I reverted back to 0 and set the "PowerState" to 0.

Why i changed this knowing that our frame buffer is the GEN6 and this is GEN575?
Take a look at ioreg:

Every Gen6 occurrence in Gen6Accellerator is matched with a 575 one.

Maybe i changed something else in other Kext, but i really don't remember.

 

[xequtor]

for a test we need a WAY to test some flash web site where the problems (line/glitches) occurs but
not on a mac or working hack ProBook.

So if someone can post a link to a site with good test will be very helpful!

 

[masso912]

for a test we need a WAY to test some flash web site where the problems (line/glitches) occurs but
not on a mac or working hack ProBook.

So if someone can post a link to a site with good test will be very helpful!

http://www.syfy.com/tinman/oz/

 

[masso912]

A single partial line has occurred, but it went off instantly :?
Still testing...

(also testing http://waterlife.nfb.ca/)

 

[xequtor]

for a test we need a WAY to test some flash web site where the problems (line/glitches) occurs but
not on a mac or working hack ProBook.

So if someone can post a link to a site with good test will be very helpful!

http://www.syfy.com/tinman/oz/

no lines here Safari and Firefox parallel started, safari in Full screen also no line

 

[BigDonkey]

@masso912
What version are you running?

I've got plenty of lines on 10.7.0 and almost non existent on 10.7.3. Latest Lion revisions is big improvement over previous ones.
My lines occur after the fresh boot and I know immediately that I'm going to have problems with graphics. On a "good" boot I can go for hours or days without single artifact.

I wish there was proven method to create display artifacts otherwise tests will be very time consuming.

 

[masso912]

@masso912
What version are you running?

I've got plenty of lines on 10.7.0 and almost non existent on 10.7.3. Latest Lion revisions is big improvement over previous ones.
My lines occur after the fresh boot and I know immediately that I'm going to have problems with graphics. On a "good" boot I can go for hours or days without single artifact.

I wish there was proven method to create display artifacts otherwise tests will be very time consuming.

I am running 10.7.3 and sometimes i get lines very randomly.
I am testing this mod and i had 1 single line that popped out and disappeared

 

[xequtor]

@masso912
What version are you running?

I've got plenty of lines on 10.7.0 and almost non existent on 10.7.3. Latest Lion revisions is big improvement over previous ones.
My lines occur after the fresh boot and I know immediately that I'm going to have problems with graphics. On a "good" boot I can go for hours or days without single artifact.

I wish there was proven method to create display artifacts otherwise tests will be very time consuming.

that's what i meant with artifacts testing.