118 replies to this topic

[unaligned]

I am not educated on this issue, but: are you using a smbios.plist in /Extra? I think you should (iMac 9,1 or similar)

Have you tried Mu|tibeast - does it offer a app-store hack?

No, I have not tried Mu|tibeast, but it might be an idea. Though I don't know if it has an App-Store hack. I do not have a smbios file - that is one of the last things I've to do. I'll try those two, see if it helps. I know it's obscure, but do you have any experience with Corona on hackintosh and/or do you know anything about app signing for iOS on a non-Apple Mac?

I am now succesfully logged into the app store, after extracting the Ethernet card's EFI string w. EFI studio, adding the following code into org.chameleon.Boot.plist

<key>device-properties</key>
<string>4b00000001000000010000003f0000000100000002010c00d041030a0400000001010600041c0101060000007fff0400160000006200750069006c0074002d0069006e0000000500000001</string>

restart, app store, login

Before it wouildn't work ("Device not recognized")

Yeah, this is what I've endlessly tried, and it has not helped.

[BugsB]

we're getting closer. Make sure you are using org.chameleon.Boot.plist not the old com..

forget about MB, just checked, it does not offer such option

yes, you DO need a correct smbios. Then you neither need the CPU disabler kext in E/E

I at this point do not yet have exp. w. app signing etc, but I equipped a friend with a h'tosh who's coding a game that also needs to work on iOS, so we'll have to get into that at some point, but not yet. With the correct smbios and an EFI string here and there it is pretty much impossible for Apple to tell a hack from a Mac. I know of a h'tosh on the US mainland that ran OS X Server as a Mac Mini (with original Mac Mini smbios settings) running Apple's hardware diagnostics, communicating w. Apple ALL THE TIME (the Mini was needed somewhere else). Apple never complained ..

Chose a pre-config iMac 9,1 SMBIOS w. Chameleon Wizard, generate a random serial, safe. Verify that above EFI string is really in the org.chameleon.Boot.plist. reboot. test app strore login.

[unaligned]

Still no go. I downloaded and used Chameleon Wizard to generate my org.chameleon.Boot.plist and my smbios.plist with a random serial, and used EFI Studio to get my <key> and <string> to put in the boot plist and the App Store still isn't recognizing my device.

EDIT:
Nevermind, I got it going now. My problem was that for my EFI key for my device, I had

<key>PciRoot(0x4)/Pci(0x1c,0x4)/Pci(0x0,0x0)</key>

instead of

<key>device-properties</key>

Or it may have been that I changed

IOBuiltIn

from

false

to

true

in my NetworkInterfaces.plist. It was one of those two - or both.

Alright... making progress. Now if only my sound control panel would show HDMI out as an option instead of just Line Out (Green Rear). Oh, and get file sharing with my W7 boxes to work... none of them will connect. I keep getting an error that says 'the parameter is incorrect' for some reason and they will never allow me to connect to the drives.

[BugsB]

very cool.

File sharing: do you have SMB activated under Options? However it did not work perfectly for me either right away. I had to go into sharing, disable, reenable. Eventually I created an ALIAS of my shared windows folder via Finder onto the Desktop. Now, when I double click that, the shared folder on my Windows PC always opens right away.

Also make sure you have, in Windows, created an account with the exact same name as the name of the Mac user account on the 745. Also I'm pretty sure the account *must* have a PW ..

Audio via HDMI: I don't use that so can't say. However there's threads in the forum about that exact topic. I remember reading that this can be enabled in VoodooHDA (via step-by-step guide)

Just for completion I added the package for the 745. Should be of help to some (regardless of SFF, DFF etc.).

EDIT: DSDT is only error auto-corrected by DSDT patcher and *only* contains DTGP as added method, NOTHING else

EDIT 2012-05-09: NEW VERSION 2.1:
- removed SleepEnabler for 10.7.4 compatibility
- added RTC patch script

older changes (credits also to machinist and Vlad):
- added correct PCIRootUID 4 to increase chances that ATI Cards load w. correct Framebuffer (working w. ATI Saphire Radeon HD 5450 1GB DDR3 DVI/ HDMI/ VGA)
- deleted 'Device Property' key & string which interfered w. Framebuffer loading
- replaced smbios.plist with one from Mac Mini 4,1 so App Store login works
- added 'darkwake=0' so computer and screen now wake-up from sleep with one signal from kb or mouse (rather than going back into sleep mode after only a single user signal)

Attached Files

•  Dell Optiplex 745 SFF.zip   520.5KB   156 downloads

[Vlad]

I have the same sapphire radeon hd 5450 with 1gb ddr3 and LP
i added the line you specified in the issue there but the card is still unrecognized

here is the source code + compiled files : http://www.mediafire...gpp6edmugjtdt0k maybe i'm wrong somewhere


If you managed to get the videocard work please post here how you did the trick because i did not get it to work fully - the dvd player crashes

[BugsB]

thank you for compiling, Vlad, very much appreciated! One less thing I have to get into ..

I confirm, trunk/sym/i386/boot does STILL NOT invoke the correct FB for the 5450 (GE=yes). Indeed no FB is selectable as before, and naturally DVD player never launches. Yes, you added the code correctly (I checked)

Looking at the other two 5450's in ati.c, I have been wondering about their different DID.. Is it possible that the GPU itself has a different ID than the grfx card ?? I think it would be very much worth a try if you changed our 5450's DID also to 0x68F9 in ati.c, and try that compile

[Vlad]

as far as i can see ErmaC integrated the device id and the subsystem id into his build of chameleon and i see that the park, maddison and other fammilies are comming into his buid:
http://forge.voodoop...6/libsaio/ati.c

so check it out here and tell me if it's working because i not have my hackintosh here to test it and i'm very courious .
Another thing: the subsytem id that you posted here was wrong. The right version is: 0x3000174b , the vendor id (Sapphire - 174b) has to be in the end of declaration of subsystem_id

Waiting your feedback

UPDATE: The device has been there since commit 1905 so tehnically speeking if you download this: ErmaC Trunk sub-Branch [PKG r1908] ==> ErmaC-Chameleon-r1908.pkg.zip 2.56MB 147 downloads Mountain Lion BOOT SUPPORT it should work just fine

[BugsB]

as far as i can see ErmaC integrated the device id and the subsystem id into his build of chameleon and i see that the park, maddison and other fammilies are comming into his buid:
http://forge.voodoop...6/libsaio/ati.c

I see //␉CHIP_FAMILY_PARK,␊, thanks for digging this out, but that needs to be corrected because there is no such thing as a family 'Park' - the family is 'MANHATTAN', with the member GPUs Park, Broadway, Madison.

so check it out here and tell me if it's working because i not have my hackintosh here to test it and i'm very courious .

no it's not, regardless of all sorts of settings in the boot.plist (GE on/off, ATIConfigs, PCIroot, Computer type bla bla bla), and regardless of additional injectors present or not in S/L/E...

Another thing: the subsytem id that you posted here was wrong. The right version is: 0x3000174b , the vendor id (Sapphire - 174b) has to be in the end of declaration of subsystem_id

thank you for pointing thsi out! Why can't they make it e-z and user-friednly, sigh - it would be easy to have a function in Chameleon to turn the values around, either during compilation or at run-time. But no, geeks need to let the rest of the world feel who's geek, and who's not ..

I corrected in my request!

The line in ErmaC's ati.c reads:

{ 0x68E1,␉0x3000174B, CHIP_FAMILY_CEDAR,␉␉"ATI Radeon HD 5450M",␉␉␉␉kEulemur␉},␊

however we're not dealing with the mobile but with the Desktop version here, so it should rather read

{ 0x68e1,␉0x3000174b, CHIP_FAMILY_CEDAR,␉␉"ATI Radeon HD 5450",␉␉␉␉kEulemur␉},␊

no idea if that makes a difference. Our 745's system type is set to iMac, which *might* be the reason that OS X refuses to set *any* FB (for a GPU it detects as mobile in a Desktop system).

so if you feel like compiling once more, I'd recommend ..HD 5450.. (w/o the M). However we might really have to wait until the MANHATTAN chip family has gotten integrated into Chameleon ....

[BugsB]

It looks more and more that, in order to get OS X to set the FB correctly for this Amazon 1GB Saphire 5450, we'd have to follow artur-pt's step-by-step guide = do a bunch of testing and editing (probably incl. hex-editing some kext's binary)

[Vlad]

I see //␉CHIP_FAMILY_PARK,␊, thanks for digging this out, but that needs to be corrected because there is no such thing as a family 'Park' - the family is 'MANHATTAN', with the member GPUs Park, Broadway, Madison.


no it's not, regardless of all sorts of settings in the boot.plist (GE on/off, ATIConfigs, PCIroot, Computer type bla bla bla), and regardless of additional injectors present or not in S/L/E...

thank you for pointing thsi out! Why can't they make it e-z and user-friednly, sigh - it would be easy to have a function in Chameleon to turn the values around, either during compilation or at run-time. But no, geeks need to let the rest of the world feel who's geek, and who's not ..

I corrected in my request!

The line in ErmaC's ati.c reads:

{ 0x68E1,␉0x3000174B, CHIP_FAMILY_CEDAR,␉␉"ATI Radeon HD 5450M",␉␉␉␉kEulemur␉},␊

however we're not dealing with the mobile but with the Desktop version here, so it should rather read

{ 0x68e1,␉0x3000174b, CHIP_FAMILY_CEDAR,␉␉"ATI Radeon HD 5450",␉␉␉␉kEulemur␉},␊

no idea if that makes a difference. Our 745's system type is set to iMac, which *might* be the reason that OS X refuses to set *any* FB (for a GPU it detects as mobile in a Desktop system).

so if you feel like compiling once more, I'd recommend ..HD 5450.. (w/o the M). However we might really have to wait until the MANHATTAN chip family has gotten integrated into Chameleon ....

i will compile it as soon i will be in front of my "mac" -- sunday maybe

[machinist]

BugsB et al:

I feel your frustration. I suspect yours is a PCIRootUID issue.
Chameleon r1897 reports "our" card as a 5430M Cedar while r1908 displays it as a 5450M Cedar. It's purely cosmetic and makes no difference in performance.
Whether it's in a laptop or desktop, it's still a 5450M "Park". It's a "Mobility Series" AMD chip.
However, using the "Baboon" fb improves performance by every metric in every test. Save for weaknesses in opengl extensions 1.2 through 1.4, the 5450 performs almost as well as it does in Windows. Haven't edited every fb monkey personality yet for testing, but "Eulemur" is the worst by far thus far. Maybe one of the other "monkeys" will surprise me and "outdo" "Baboon"?
This Sapphire card however, does overclock solidly to 750/850 in Windows without any additional cooling, and I'm using it at 800/850 right now with no issues, though I've a small fan trained on the heat sink. Once I'm sure it's solid, I'll flash it in Windows so it will run this way in OS X. Cinebench results improve dramatically, from ~9fps to 13+fps, which of course is not "great," but it is a $14.99 card, not $149. I for one like to squeeze every fps of performance from my hardware.
Try using PCIRootUID=1.
Hope you get your rig working.

[BugsB]

@ machinist: thanks for your input! Tried Baboon via

<key>GraphicsEnabler</key>
<string>Yes</string>
<key>AtiConfig</key>
<string>Baboon</string>

added PCIRootUID=1 as follows:

<key>Kernel Flags</key>
<string>-v PCIRootUID=1</string>

Isn't this redundant since I have before already tried

<key>PCIRoot</key>
<string>1</string>

?

Anyway, result: no improvement (ErmaC r1908 boot which seems most promising since it contains this card's values, supposedly ..)! DVD Player still does not launch, ioreg|grep ATY: no FB; SysInfo PCICards: no cards installed ..

When and where did you purchase your Saphire 5450 1GB DDR3? Would you please provide a link? I got the 100292DDR3L (Amazon) on February 10, 2012. I had the exact same BS before with an Nvidia card, and when we compared our card's ROMS we realized that they were *DIFFERENT*!

You can find my 5450's ROM here. I guess you know how to hex/bin compare files. If not just upload your ROM and I'll go ahead.

Thanks again!

[machinist]

BugsB:

I'm certain our cards are the same.

- Make sure device id 0x68E11002 is present in ATIRadeonX3000 and ATI5000Controller kexts. Continue to say "Yes" to the GraphicsEnabler.

(remove any injectors or inserters)

- Try PCIRootUID=1 and PCIRootUID=0. <key>PCIRoot</key> is not the same as <key>PCIRootUID</key> and would not have and will not work. You did it right the first time, but with a wrong command. I use the Property List Editor found in Apple's free Developer package, but if you simply enter, precisely, PCIRootUID=X when your chameleon volume choice presents, it reads as a boot argument. Your PCIRootUID value may be 0 or 1; try both. If you'd like, upload your boot.plist and I'll review it for you?

- Add Wait=Yes to your boot.plist so you can see which fb loads.

It's not the card. No offense, but you're likely making a very simple error somewhere, as when you previously entered PCIRoot for PCIRootUID, or reversed the order of the subsysid and subsys. For an alternate rom to be read, for example 1002_68E1_3000174b.rom, it must be in that specific format. Though it doesn't matter in this instance whether "E" or "b" is in caps or not, it does when you enter UseAtiROM=Yes, or AtiConfig=Eulemur/Baboon/Hoolock/OtherMonkey.

Good luck.

[Vlad]

I've recompiled the ermac 1918 boot loader and after loading it with chameleon wizard in bdmesg.log i have:

Framebuffer @0xD0000000  MMIO @0xFE9E0000 I/O Port @0x0000DC00 ROM Addr @0xFEA00000
ATI card POSTed, reading VBIOS from legacy space
(AtiConfig) Framebuffer set to: Baboon
Nr of ports set to framebuffer's default: 3
ATI Cedar Saphire Radeon HD 5450 1024MB (Baboon) [1002:68e1] (subsys [174b:3000]):: PciRoot(0x0)/Pci(0x1,0x0)/Pci(0x0,0x0)

and after running the ioreg|grep ATY i have this

ioreg|grep ATY
	| |   |   | +-o ATY,ATY,RadeonFramebuffer@0  <class AtiFbStub, id 0x1000002a6$
	| |   |   | | +-o ATY_ATY,RadeonFramebuffer  <class ATIFramebuffer, id 0x1000$
	| |   |   | +-o ATY,ATY,RadeonFramebuffer@1  <class AtiFbStub, id 0x1000002a7$
	| |   |   | | +-o ATY_ATY,RadeonFramebuffer  <class ATIFramebuffer, id 0x1000$
	| |   |   | +-o ATY,ATY,RadeonFramebuffer@2  <class AtiFbStub, id 0x1000002a8$
	| |   |   | | +-o ATY_ATY,RadeonFramebuffer  <class ATIFramebuffer, id 0x1000$

this is without running any of pcirootuid and without using any rom
as a conclusion: the dvd is not working but chess is working and anyother 3d game: i tested fifa 11

edit: i forgot to mention: the card is viewed as ati radeon 5000

BugsB:

I'm certain our cards are the same.

- Make sure device id 0x68E11002 is present in ATIRadeonX3000 and ATI5000Controller kexts. Continue to say "Yes" to the GraphicsEnabler.

(remove any injectors or inserters)

- Try PCIRootUID=1 and PCIRootUID=0. <key>PCIRoot</key> is not the same as <key>PCIRootUID</key> and would not have and will not work. You did it right the first time, but with a wrong command. I use the Property List Editor found in Apple's free Developer package, but if you simply enter, precisely, PCIRootUID=X when your chameleon volume choice presents, it reads as a boot argument. Your PCIRootUID value may be 0 or 1; try both. If you'd like, upload your boot.plist and I'll review it for you?

- Add Wait=Yes to your boot.plist so you can see which fb loads.

It's not the card. No offense, but you're likely making a very simple error somewhere, as when you previously entered PCIRoot for PCIRootUID, or reversed the order of the subsysid and subsys. For an alternate rom to be read, for example 1002_68E1_3000174b.rom, it must be in that specific format. Though it doesn't matter in this instance whether "E" or "b" is in caps or not, it does when you enter UseAtiROM=Yes, or AtiConfig=Eulemur/Baboon/Hoolock/OtherMonkey.

Good luck.

my boot.plist is :

<?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>AtiConfig</key>
<string>Baboon</string>
<key>DSDT</key>
<string>/Extra/DSDT.aml</string>
<key>GenerateCStates</key>
<string>Yes</string>
<key>GeneratePStates</key>
<string>Yes</string>
<key>GraphicsEnabler</key>
<string>Yes</string>
<key>Kernel</key>
<string>mach_kernel</string>
<key>Kernel Flags</key>
<string>darkwake=0</string>
<key>Legacy Logo</key>
<string>Yes</string>
<key>Theme</key>
<string>LoginToLion</string>
<key>Timeout</key>
<string>5</string>
<key>UseKernelCache</key>
<string>Yes</string>
</dict>
</plist>

and thanks for your support it's really appreciated

Attached Files

•  i386.zip   966.56KB   16 downloads

[machinist]

Vlad, your boot.plist looks fine. Add PCIRootUID=1 (or 0, whichever works) and that should be good.
Chameleon is seeing and identifying your 5450, but not matching it up properly to your board. The ioreg indicates you are loading the generic RadeonFb. Try the PCIRootUID argument and see if that flies?

[BugsB]

machinist, tried/ verified everything you wrote - even erased all other IDs out of the two ATI ketxs - no go .. :( exact same situation as Vlad ..

I see two possibilities:

a) something in the 745's DSDT is blocking the 5450

b) your 5450 is indeed slightly different

Would you be so kind and extract and upload your ROM? Until we compare it can't be excluded. If they indeed are NOT the same, I'll load it via UseAtiROM=Yes.

BTW, you write EITHER UseAtiROM=Yes OR AtiConfig - does one *overwrite* the other ?? If so, do you know which is dominant?

EDIT: can't we just FORCE Baboon somewhere in the ATIFramebuffer kext ??

[Vlad]

Vlad, your boot.plist looks fine. Add PCIRootUID=1 (or 0, whichever works) and that should be good.
Chameleon is seeing and identifying your 5450, but not matching it up properly to your board. The ioreg indicates you are loading the generic RadeonFb. Try the PCIRootUID argument and see if that flies?

i have tried adding the PCIRootUID and the card is stil recognized as 5000 series - and the default frame buffer is loaded

interesting is that the acceleration is still available even if the GE is no

i cannot believe that the same card is working for you and for me not , can you post please your configuration for making it work: what frame buffer you used, what parameters you specially set in your boot.plist. I tried booting with baboon and setting PCIRootUID to different values but it turned out with kernel panic in every situation

[machinist]

BugsB, Vlad, et al:

Attached is my bin/rom extracted in Win7(64) by trixx, sapphire's oc utility. It is NOT a zip file, so simply rename it to bin or rom. Can't be bothered compressing a ~60kb file. Notably, it is a couple kb smaller then extracts from other tools. Bought my card through Newegg a couple/few weeks ago. I think BugsB "Amazon" card was sourced through them at that time?

UseAtiROM is a Yes/No command whereas AtiConfig summons the Monkeys.

A fb can't be forced on a card the system doesn't recognize. And the rom only acquaints the system with the card's specs. If it could override the system's read of the card we'd all be using the function to oc our gfx.

Previous testing was with 10.6.8. With 10.7.3 Uakari is a strong contender to displace Baboon as my fb of choice. I'll be monkeying about with it for a while I suspect? I'm using Uakari right now.

Vlad, perhaps you might want to not use the kernel cache in your boot.plist? While it speeds booting, it won't automatically reflect changes made in the system.

BugsB, there are more then those two options.

5450M was recognized in 10.7+ without adding in device ids, so I suspect they are already in the appropriate kexts. GE is necessary, of course.

I'll run some trials with different configs and see what's the least one needs do to get the 5450M recognized.

Attached Files

•  Park5450M.zip   58KB   13 downloads

[Vlad]

ok trying to see if it works


LATER:
maybe i'm stupid this is what i did:
-renamed the file you posted such as: 1002_68e1_3000174b.rom
-put it in extra folder
-changed config: UseAtiRom = Yes
-changed config: AtiConfig = Baboon
-changed GE=yes

when i'm booting again the boot loader is seeing the rom file, read's it and hangs booting the os

what i'm doing wrong?, what should i do?

Edited by Vlad, 10 April 2012 - 09:40 PM.

[machinist]

Keep trying? Sometimes a reboot or two is necessary for the changes to take effect.

Remove "Baboon" first and see if it still hangs, then remove the rom call.

Run
ioreg -l | grep -15 "AppleACPIPCI" | grep "_UID"
to discover your pciuid, if you haven't already?

Don't know that you are doing anything "wrong," but these are a few of the things you can try?