89 replies to this topic

[rr7]

Sure, it was previously mentioned.

It says hd4000 kext disks to load. Cannot find kext IOGraphicsFamily.kext, something about dependency.
Then cannot alloc class gen7accel.

It also complains in the same way about applehda kext.

In safe mode it doesn't complain, as it doesn't try and load those kexts. Blut Capri still blanks out.

Surely x86platform plugin should load?

[eep357]

Do you mean AppleACPIPlatform? It is one of the dependencies for IOGraphicsFamily. I don't know of any "x86Platform" files other than disassemblers like as part of a SDK

[Akai92]

OMG I get my HD 4000 working on probook 4540s. All I did is reset bios to default and disable discrete graphic card.

I am using efi string id 10660003 and dsdt patch, but not sure what made it work. I also installed legacyagpm.kext. Thank you guys for giving me useful information. Now I am so happy. Ready to get other things working.

[rr7]

eep357, on 28 August 2012 - 07:50 AM, said:

Do you mean AppleACPIPlatform? It is one of the dependencies for IOGraphicsFamily. I don't know of any "x86Platform" files other than disassemblers like as part of a SDK

Take a look at the rMBP ioreg I attached, the X86PlatformPlugin sits below acpiaplatform.

It explains it better than I can

[eep357]

OK, it lives inside IOPlatformFamilyPlugin.kext. Of plugin inside which include that one, only saw one with device ID's in Plist, AppleSMCPDRC.kext

[rr7]

Is there a method of trace/debugging these dependency issues?

All looks a bit mystic right now and uk not sure how to get past it

[eep357]

usually one of the dependent kext either is not loaded, or is a version # that is incompatible, appleHDA is dependent on IOGraphicsFamily, so is why it not loading. run kextstat in terminal and each kext listed will have a # before it which is ID #, followed by a 2nd # which I believe is how many other things are dependent on it. after name of kext is version #, then the ID #'s of it's dependencies

[rr7]

apparently all dependance is satisfied. so i have no idea.

i think, really, i'll have to drop this. maybe look again in a few updates time. :\

[Pillendreher]

I can confirm that id 01660004 is a good starting point for Laptop-users who get blank Internal-Displays with other IDs. Without any patches or additional kexts, I got graphics-acceleration with that id

In Systeminfo, it says 512mb Ram though..going to post a screenshot later.

[rr7]

i picked up another type of AIO (ECS-G11) for a bit of a test, and the bloody bios doesnt let me set AHCI so i cant boot it. how annoying is that?!

[eep357]

odd that newer PC would not have option yet default to IDE compatibility mode. Did you try with AHCIPortInjector?

[rr7]

eep357, on 14 September 2012 - 07:46 AM, said:

odd that newer PC would not have option yet default to IDE compatibility mode. Did you try with AHCIPortInjector?

apparantely H61 chipsets dont support it, funny that the main AIO i've been working on seems to have it enabled okay. i've asked for an unlock bios option from the R&D guys, maybe they'll come through.

meanwhile, no, i havent tried AHCIPortInjector, is there any info on it? usage, which devices need adding to plist etc?

[eep357]

Even in IDE mode osx is bootable with correct kexts, on ipad via 3g atm so can't upload, but all info on both is in the faq section about still waiting on root device

[rr7]

Thanks, ill give it a go in an hour or so....

[rr7]

Although I couldn't make the ahci port injector load (it went panic) I was able to hack the dev ID into ioatafamily.kext and got it to boot that way.

It'll do for testing.

Interestingly, I couldn't make the g11 AIO work either, blank screen on all platform Id types. And this time, no display got detected.

Might wait a while until I can get another processor for it as switching between the two chassis is becoming a pain.

Maybe even the CPU is a problem. I seem to be the only person on the planet using a 3475S!

[eep357]

Could have something to do with it, don't even see those for sale by themselves and I think it's the only i5 with HD4000

[rr7]

not given up, just out of ideas.

small factoid though: whatever port is configured in the connector table, the first one (@0) is always given port number 0x0 in ioreg.

real port numbers i've been able to find are as follows.

0205 is 0x5
0304 is 0x6
0406 is 0x7
0503 is 0x8
0602 is 0x8 (vga apparently)
0100 is also 0x8 (not sure this is even a real port type, but this crashes the HD4000 kext if its not the first entry on the list.)

none of this seems important though.

i'd like to know, what does FBMem: 1 or FBMem: 2 refer to, and why?, how is this different from pipes?

also, there is a property called AAPL,DisplayPipe, and on my system the display is detected on port 0x7 always and it shows display pipe 00 00 00 00, all others are ff ff 00 00, on the IOreg i dumped from the retina macbook pro, the AAPL,DisplayPipe is set to 02 00 00 00

i dont see how i can change this in any way, as i cant seem to influence the properties of the framebuffers.

right now, on the whole, by and large, all things considered...i still have no idea whats going on.

[eep357]

Take a look at Fabio (ErmaC) and Slice's newly posted Clover tutorial. I have used it at a few different stages of development. Aside from it's ability to patch kexts durning boot, Fabio also gives a pretty good explanation about connector tables. Just another thing to play with and it shouldn't hurt anything to try

[rr7]

eep357, on 22 September 2012 - 09:58 AM, said:

Take a look at Fabio (ErmaC) and Slice's newly posted Clover tutorial. I have used it at a few different stages of development. Aside from it's ability to patch kexts durning boot, Fabio also gives a pretty good explanation about connector tables. Just another thing to play with and it shouldn't hurt anything to try

i wiped the install and went from scratch, just using clover as the bootloader. it doesnt do anything i havent already tried, and i'm not sure i'd like it for a final solution, BUT, its a great toy for when you're experiementing!

the HD3000 connector info is great, i've been using the original post for that for some time. the info is for HD3000 though, and it seems that while the connector codes might well be the same, the starting hex values are not. so i know no more than we've already worked out on this thread.

i tried this on the other AIO i have for testing too, and that doesnt play ball at all, just as before. none of the ports seem active.

it doesnt look like there are any current drivers with source code for linux yet that we can look at either.

[rr7]

i've gotten hold of another thin-itx motherboard to go in one of the all in one chassis i have.

firstly, good news, it boots into gfx with acceleration. so the CPU is fine. interestingly, this one seems to report as being connected as hdmi, even though its LVDS. it doesnt need anything other than the platform id in order to work.

i intent to try this board on the other chassus to prove the panel is able to display.

all boards i have are using H61 chipset.

until i can continue however, i've hit another hurdle here that i simple cant work out yet. this new board, when it boots, it mostly hangs before it skips to the desktop, the last message i normally see is about am IO bluetooth being complete. the error isnt related to that though, its just the last message i normally see.

i've googled, and it seems a common problem, but there doesnt seem to be a common solution.
when i did notice is, that if i used the OEMSMBIOS plugin with fakesmc, this problem went away, then when i installed 10.8.2 update, it crashed, so i removed it, and now the problem is back.