WhatEverGreen Support Topic

[Andres ZeroCross]

@vit9696 i used Whatevegreen for my RX580.
1. CSM = Never / Disable
2. Boot in UEFI
3. NO inject ATI, No RadeonDeInit and other properties at Config.plist

But, i can't change model name of RX 580. Model name is always "Radeon RX 580". I know it's not related to performance, but at Github of WhateverGreen is written that i can change the model with SSDT/ DSDT. But it's not work, i am pretty sure the DSDT / SSDT code is good, i can change the model name of my IGPU or my previous graphics card (NVIDIA GTX970). For temporary solution, i use clover kext to patch to AMD9500Controller.kext to change model name, and it works. Sorry for bad english

[vit9696]

Try changing this line:

https://github.com/acidanthera/WhateverGreen/blob/d61b1723637b4f4242c2a4e45b7e9310467c1b19/WhateverGreen/kern_rad.cpp#L682

to

if (*static_cast<uint32_t *>(bytes) == ' DMA' || *static_cast<uint32_t *>(bytes) == ' ITA' || *static_cast<uint32_t *>(bytes) == 'edaR') {

 

[Andres ZeroCross]

2 minutes ago, vit9696 said:

Try changing this line:

https://github.com/acidanthera/WhateverGreen/blob/d61b1723637b4f4242c2a4e45b7e9310467c1b19/WhateverGreen/kern_rad.cpp#L682

to

 

if (*static_cast<uint32_t *>(bytes) == ' DMA' || *static_cast<uint32_t *>(bytes) == ' ITA' || *static_cast<uint32_t *>(bytes) == 'edaR') {

 

 

 

Cool,, now i feel headache 

5 minutes ago, vit9696 said:

Try changing this line:

https://github.com/acidanthera/WhateverGreen/blob/d61b1723637b4f4242c2a4e45b7e9310467c1b19/WhateverGreen/kern_rad.cpp#L682

to

 

if (*static_cast<uint32_t *>(bytes) == ' DMA' || *static_cast<uint32_t *>(bytes) == ' ITA' || *static_cast<uint32_t *>(bytes) == 'edaR') {

 

 

But for information i try to change all of "Radeon RX 580" to "MSI RX 580 Armor OC" in this file https://github.com/acidanthera/WhateverGreen/blob/d61b1723637b4f4242c2a4e45b7e9310467c1b19/WhateverGreen/kern_model.cpp then i build the kext and use it. But, the model name still "Radeon RX 580". I think the model name is from "AMD9500Controller.kext"

[vit9696]

That is what I meant, currently model names provided by AMD9500Controller.kext start with "Radeon" instead of "AMD" or "ATI". And WhateverGreen did not expect it.

I committed the change to master. Report if there are any issues.

[Sweet Twok]

28 minutes ago, vit9696 said:

That is what I meant, currently model names provided by AMD9500Controller.kext start with "Radeon" instead of "AMD" or "ATI". And WhateverGreen did not expect it.

I committed the change to master. Report if there are any issues.

What can you say about my problem?

 

[vit9696]

No idea. You are not the first one to report this, but I am not too interested, and even if I was there is not enough data to explore. If you were using WhateverGreen previously, then you had a discrete GPU I believe, but you have a connector-full IGPU frame. Most likely you need to reread the manuals (e.g. the ones on life).

[Andres ZeroCross]

8 hours ago, vit9696 said:

That is what I meant, currently model names provided by AMD9500Controller.kext start with "Radeon" instead of "AMD" or "ATI". And WhateverGreen did not expect it.

I committed the change to master. Report if there are any issues.

 

The latest commit to WhateverGreen is around 5 days ago for "Fixed warning about legacy processors (e.g. Xeon)". So i used the latest commit of WhateverGreen when use it. But, still model name from the kexts

Sorry,, my mistake. Try to rebuilt now

Edited October 17, 2018 by Andres ZeroCross

[Andres ZeroCross]

7 hours ago, vit9696 said:

No idea. You are not the first one to report this, but I am not too interested, and even if I was there is not enough data to explore. If you were using WhateverGreen previously, then you had a discrete GPU I believe, but you have a connector-full IGPU frame. Most likely you need to reread the manuals (e.g. the ones on life).

 

Nice works,, my patch of SSDT about model name of RX 580 is work flawlessly with latest build WhateverGreen

Thanks

But long ago when i use R9 280X, i can use this "VRAM, totalMB" to change my VRAM of GPU (Cosmetic) and still can use "VRAM,totalMB" for Nvdia Card. But now it won't work. No problem,, no big deal. I can live without it

Edited October 17, 2018 by Andres ZeroCross

[Sweet Twok]

On 10/16/2018 at 10:38 PM, vit9696 said:

No idea. You are not the first one to report this, but I am not too interested, and even if I was there is not enough data to explore. If you were using WhateverGreen previously, then you had a discrete GPU I believe, but you have a connector-full IGPU frame. Most likely you need to reread the manuals (e.g. the ones on life).

Could you give a complete list of changes compared to the WhateverGreen version 1.1.7?
I would like to understand what causes such a problem.
Everything was fine when WhateverGreen + IntelGraphicsFixUp worked.

[vit9696]

Consider changing your IGPU ig-platform-id to connector less. WhateverGreen got AGDP patches, and they are likely to conflict with active IGPU and discrete GPU. Use 0x3E910003 (03 00 91 3E).

[Sweet Twok]

1 hour ago, vit9696 said:

Consider changing your IGPU ig-platform-id to connector less. WhateverGreen got AGDP patches, and they are likely to conflict with active IGPU and discrete GPU. Use 0x3E910003 (03 00 91 3E).

Thank you, but the fact is that I already have it done.
 

[Sweet Twok]

Could you advise something else?

[Andres ZeroCross]

Could you advise something else?

May i know the problem detail?? Just for my knowledge

Sent from my Mi Note 2 using Tapatalk

[Sweet Twok]

37 minutes ago, Andres ZeroCross said:

May i know the problem detail?? Just for my knowledge

Yes of course.
---

 

[Andres ZeroCross]

49 minutes ago, Sweet Twok said:

Yes of course.
---

 

 

Spoiler

 

So is this delay too??? I just installed this software, but never use it

[Sweet Twok]

10 minutes ago, Andres ZeroCross said:

So is this delay too??? I just installed this software, but never use it

It does not matter. The main thing is that after upgrading to the new version of Green, it began like this. It turns out that something does not work correctly, and can affect not only the program Adobe.

[roastable]

Hey all,

 

It was mentioned once before in this thread, but I thought I might give a little more insight into the rotation bug; attached is a video of it in action right from the login screen.

 

Using v1.1.7 + Shiki circumvents this and is what I was using on High Sierra. With the update to Mojave however, it still works albeit with only one screen being detected so I finally gave in to upgrading. Doesn't seem to have an impact on performance or anything.

 

If there's any more information I can provide, please let me know!

[Joni_78]

Is -rad4200 deprecated? I noticed boot behavior on latest kext has changed back to what it was without this flag.

[87zrjojo]

Laptop:MSI GL62M 7REX-1650

CPU: intel core i7-7700HQ

Graphics: intel graphic HD 630

 

I have problem with black screen after booting on Mojave, when injecting AAPL,ig-platform-id 0x00001b59 using whatevergreen. Then I spoof KabyLake to SkyLake using 0x00001b19,FakeCPUID 0x0506e3 and SMBIOS Mbp13,1, the screen comes up, but likely whatevergreen not support KabyLake spoof, it doesn't work with inject things. Would you please improve it? so i can use KabyLake spoof with HD630 graphics.

Edited October 24, 2018 by 87zrjojo

[muryk]

On 10/19/2018 at 6:16 PM, roastable said:

It was mentioned once before in this thread, but I thought I might give a little more insight into the rotation bug; attached is a video of it in action right from the login screen.

 

 

Same issue. It happened as of "Initial merge of IntelGraphicsFixup, NvidiaGraphicsFixup, Shiki" WG commit. Have post-boot rotation issue on my portrait mode (90°) oriented secondary display (attached to DVI) . The primary display is attached to DP, card is GA RX 560. Screen image is oriented as landscape while mouse pointer moves properly (portrait). Had to create a start-up Apple Script to switch display orientation to 0° then to 90° back. Tried to change connector priority (because the DVI is a primary connector for RX 560 - I thought it might affect the result), but this did not help. 

[roastable]

1 hour ago, muryk said:

Same issue. It happened as of "Initial merge of IntelGraphicsFixup, NvidiaGraphicsFixup, Shiki" WG commit. Have post-boot rotation issue on my portrait mode (90°) oriented secondary display (attached to DVI) . The primary display is attached to DP, card is GA RX 560. Screen image is oriented as landscape while mouse pointer moves properly (portrait). Had to create a start-up Apple Script to switch display orientation to 0° then to 90° back. Tried to change connector priority (because the DVI is a primary connector for RX 560 - I thought it might affect the result), but this did not help. 

I got a response this morning from a user on reddit who provided me with kext patch that worked!
https://www.reddit.com/r/hackintosh/comments/9pktrv/whatevergreen_rotation_bug/e8ctrue

[vit9696]

Sigh… I have already told you to try disabling AGDP patches, but I guess nobody understood it.

WG currently automatically disables AGDP (AppleGraphicsDevicePolicy) support on all the boards with discrete GPUs, because it leads to AMD visual glitches in Mojave, breaks multimonitor support on both AMD and NVIDIA, and may result in blackscreen with NVIDIA on several models.

Indeed, it may result in some features being disabled, and for this reason there is a way to configure this via agdpmod boot argument.

 

By default it works as if you wrote "agdpmod=pikera,vit9696" but you may write agdmod=none to get the behaviour of the original kext.

The values are listed here: https://github.com/acidanthera/WhateverGreen#boot-arguments

 

Also watch out for CFG_USE_AGDC property, see details in the bottom of:

https://github.com/acidanthera/WhateverGreen/blob/master/Manual/FAQ.Radeon.en.md

[muryk]

2 hours ago, vit9696 said:

Sigh… I have already told you to try disabling AGDP patches, but I guess nobody understood it.

WG currently automatically disables AGDP (AppleGraphicsDevicePolicy) support on all the boards with discrete GPUs, because it leads to AMD visual

...

By default it works as if you wrote "agdpmod=pikera,vit9696" but you may write agdmod=none to get the behaviour of the original kext

Thank you very much vit9696! "agpmod=pikera" boot arg option did the trick! (other combinations like "none", "pikera,vit9696", "cfgmap" just disable secondary display (made it black))

 

Thank you roastable for sharing your solution also

2 hours ago, roastable said:

I got a response this morning from a user on reddit who provided me with kext patch that worked!

I have not tried this kext patching recipe but will keep it in mind as a possible alternative

[yamahahornist]

Hopefully someone can help me out with this issue. I have a dell precision 7510 mobile laptop; it has the ability to use igpu + amd gpu; just igpu; or just amd gpu. This allows for some interesting configurations in mac; where I am either use the igpu as primary or the amd gpu as primary. I want to use the amd gpu as primary but I get a gradient (low color bit rate) whenever I use it. My laptop uses eDP as the primary connection to the laptop screen. When I use the iGPU the colors are fine and the EDID is fine. When I use the AMD gpu the SAME EDID is used by it causes issues. I have used the edid.sh script within the whatevergreen master folder and that changes two values within my EDID. The problem I am having is the EDID won't inject (through SSDT) unless I set the connector type as DVI, but this causes a black screen. I used VNC to screen share in and view my IOREG where the display is detected and the new EDID is injected. Whatevergreen detects the screen defaultly as LVDS, but the EDID wont inject with LVDS; but I get display, just low color depth. :C

While I use the iGPU it is also detected as LVDS but its totally fine.. 

 

My gpu is a AMD w5170m Firepro card; also known as the AMD R9 375M; and its a cape verde (6000 controller) card; which requires the radpg=15 for power gating, or it boots to a yellow screen.

 

I have contemplated changing out the card to a AMD wx4150 MXM type A card; and I hope that would fix the issue; but I also think the issue is related to the laptop using eDP, which I know Apple doesn't use.

 

Any help would be appreciated... If I could get the EDID to inject over LVDS it would work.

[alex.daoud]

Hey all, I need some help getting QE/CI working on a recent upgrade to Mojave.

 

The system is a Dell Venue 11 Pro which is a bit of strange machine. It is a tablet machine Haswell i3 with integrated HD4300. I got it working on Yosemite a few years ago using the old FakePCIID.kext and FakePCIID_HD4400_HD4600.kext .in the following thread:  

The bizarre thing about this machine is that the only way I was able to get QE/CI working was to fake an ID of 0x0A16 and a (desktop!) platform-id of 0x0D220003. I have attempted to do a similar thing with WhateverGreen but it doesn't work. I have tried injecting many different platform-id's and fake IDs to no avail. The Azure kexts simply do not load. Would anyone have any ideas on what to do here?

 

Edit: I've made a little bit of progress here: using a fake ID of 0x0412 and the desktop ID 0x0D220003, the laptop gets stuck during boot (either in verbose or the progress bar just gets stuck). However, I can connect to the laptop via screen share from my other machine and the system is working with full QE/CI. I am guessing there is an issue that using a desktop ID causes here. Note that not a single mobile ID works and I guess its because this laptop has an eDP screen.

 

I have attached a debug archive for the current install on this system. I have also attached an IOReg from the older Yosemite install showing that it had QE/CI. Unfortunately this IOReg is all I have in terms of debug information from the old install.

 

Thanks in advance!

 

debug_10132.zip

Venue Yosemite install.ioreg

Edited October 26, 2018 by alex.daoud