Clover General discussion

[Riley Freeman]

After set duallink =0 in config.plist, please clover log.

 

Also tell me that most of features is working or not like before.

 

Everything seems to be working fine compared to my previous DSDT injection.

 

With DualLink=0 in config:

14:333  0:000  Intel HD Graphics 4000 [8086:0166] :: PciRoot(0x0)\Pci(0x2,0x0)
14:333  0:000    Resolution: Width=1366 Height=768
14:333  0:000    Use override Custom EDID
14:333  0:000    AAPL00,DualLink = 0
14:333  0:000    FakeID Intel GFX: not set
14:334  0:000    ig-platform-id: not set
14:334  0:000    Beginning Intel GFX auto-detection with ACPI injection
14:334  0:000    Found FakeID Intel GFX = 0x01668086
14:334  0:000    Found ig-platform-id = 0x01660003
14:334  0:000  Intel GFX revision  = 0x9
14:334  0:000   detected codec: 10EC:0269
14:334  0:000  stringlength = 3142
14:334  0:000  CurrentMode: Width=1366 Height=768

[Sherlocks]

i am ok to convert this format! i look forward to this fixes in the main Clover branch.

@tluck

 

I want to know duallink issue in hd3000. Need more clearfiy check.

Gujiangjiang report me duallink issue.

 

In latest RC3, you can boot hd 3000 with no screen issue? Dualink 1 or 0 in config.plist

 

I want clover log and ioreg file. Its important me

 

Thanks in advance.

 

Duallink dynamic

-sandybride need check

-ivy checked by user

-Haswell checked by user

-broadwell checked by me

-Skylake checked by me

-Kabylake - maybe works like skylake. Because use skylake framebuffer

 

 

 

나의 LG-F410S 의 Tapatalk에서 보냄

[tluck]

@sherlocks

 

yes with your RC3 no problems on HD 3000

 

see post

No, i don't set DualLink in the config.plist now - removed.

Inject Custom EDID

Inject Intel True

4:943  0:000  === [ RestSetup macOS ] ===================================
4:943  0:000  EdidDiscovered size=0
4:943  0:000  Intel HD Graphics 3000 [8086:0126] :: PciRoot(0x0)\Pci(0x2,0x0)
4:943  0:000    Resolution: Width=1600 Height=900
4:943  0:000    Use IntelHDMI
4:943  0:000    Use override Custom EDID
4:943  0:000    AAPL00,DualLink = 1
4:943  0:000    FakeID Intel GFX: not set
4:943  0:000    snb-platform-id: not set
4:943  0:000    Beginning Intel GFX auto-detection with ACPI injection
4:943  0:000    Found snb-platform-id = 0x00010000
4:943  0:000  Intel GFX revision  = 0x9
4:943  0:000   setting specified layout-id=12 (0xC)
4:943  0:000  stringlength = 2714
4:944  0:000  CurrentMode: Width=1600 Height=900
4:944  0:000  Beginning FSInjection

[Funky frank]

Sherlocks, you seem to be very experienced with graphics drivers, too.  So I have a question:

 

Using an ATI HD7750, everything works fine (except wakeup of course). If I now enable the Intel HD4600, too, it will lead into distorted graphics, e.g. on h264 html5 videos. Both devices are detected in OpenCL benchmark (ATI is the display device), and I even can run the OpenCL benchmark. Still in a bunch of places, there will be graphics distortion.

 

Do you have any idea what could be the cause of this?  Maybe my mainboard bios? Did you know this problem, or is it very uncommon.

 

I would like to enable the Intel HD4600 as an additional OpenCL accelerator...

 

Thanks for any info, ideas, tipps!

[Sherlocks]

@sherlocks

 

yes with your RC3 no problems on HD 3000

 

see post

No, i don't set DualLink in the config.plist now - removed.

Inject Custom EDID

Inject Intel True

4:943  0:000  === [ RestSetup macOS ] ===================================4:943  0:000  EdidDiscovered size=04:943  0:000  Intel HD Graphics 3000 [8086:0126] :: PciRoot(0x0)\Pci(0x2,0x0)4:943  0:000    Resolution: Width=1600 Height=9004:943  0:000    Use IntelHDMI4:943  0:000    Use override Custom EDID4:943  0:000    AAPL00,DualLink = 14:943  0:000    FakeID Intel GFX: not set4:943  0:000    snb-platform-id: not set4:943  0:000    Beginning Intel GFX auto-detection with ACPI injection4:943  0:000    Found snb-platform-id = 0x000100004:943  0:000  Intel GFX revision  = 0x94:943  0:000   setting specified layout-id=12 (0xC)4:943  0:000  stringlength = 27144:944  0:000  CurrentMode: Width=1600 Height=9004:944  0:000  Beginning FSInjection

I noticed something. Thank you

 

 

Sherlocks, you seem to be very experienced with graphics drivers, too. So I have a question:

 

Using an ATI HD7750, everything works fine (except wakeup of course). If I now enable the Intel HD4600, too, it will lead into distorted graphics, e.g. on h264 html5 videos. Both devices are detected in OpenCL benchmark (ATI is the display device), and I even can run the OpenCL benchmark. Still in a bunch of places, there will be graphics distortion.

 

Do you have any idea what could be the cause of this? Maybe my mainboard bios? Did you know this problem, or is it very uncommon.

 

I would like to enable the Intel HD4600 as an additional OpenCL accelerator...

 

Thanks for any info, ideas, tipps!

What is your setting? Try RC3 and report log like tluck's report.

 

I cant help ATI. Only need to test HD4600. Already hd 4600(desktop and laptop) tested by bronxteck.

 

나의 LG-F410S 의 Tapatalk에서 보냄

[Allrounder]

Is there a problem with the DSDT injection in r4049?

Every time I place a DSDT I get a kernel panic.

 

This happens even when I use the origin DSDT...

[Sherlocks]

hello.

i did all cases.

2:097 0:000 Intel HD Graphics 520 [8086:1916] :: PciRoot(0x0)\Pci(0x2,0x0)
2:097 0:000 Resolution: Width=1920 Height=1080
2:097 0:000 AAPL00,override-no-connect: added
2:097 0:000 AAPL00,DualLink = 1
2:097 0:000 FakeID Intel GFX: not set
2:097 0:000 ig-platform-id: not set
2:097 0:000 Beginning Intel GFX auto-detection with ACPI Injection
2:097 0:000 Found FakeID Intel GFX = 0x19168086
2:097 0:000 Found ig-platform-id = 0x19160000
2:097 0:000 Intel GFX revision = 0x7

first clover support framebuffer ig-platorm-id 0xXXXX(deviceid)0000 except haswell graphic and ivy to boot osx. both graphics were proved by other users.

then you can config framebuffer in config.plist with your kexttopatch.

because some user report me that if use 000X frambuffer, happen sleep issue and other issue. maybe it causes complicated problem and misunderstand.

Here is RC4.

please test with clover log.

test method.
duallink 0 and 1
get log
and report.

after check all, i will upload source file in bug/report thread.

Summary
Clover Intel GFX auto-detection Goal(if you don't have ig-platform-id or fakeid or know both)
- can boot without black screen issue and fail case compared to clover's old gma.c source. can't cover all. because according to graphic buffer, need patches. so support basic(boot, 0xXXXX0000)
- potential duallink issue fix according to resolution.
- have options like skylake glitches fix and other intel graphic's basic options(ex. hd3000 need os_info to get windows and etc in snow leopard.)
- arranged all intel graphic device-id and buffer information.

 

sorry my english is not very good

thanks in advance.

 

added. updated haswell in RC5.

RC5.zip

[tluck]

@sherlocks - RC5 testing

 

1) T460 - 

Graphics:

<key>Inject</key>
  <dict>
   <key>Intel</key>
   <true/>
  </dict>

Graphics section - just has Inject Intel True
6:081  0:000  Intel HD Graphics 520 [8086:1916] :: PciRoot(0x0)\Pci(0x2,0x0)
6:081  0:000    Resolution: Width=1920 Height=1080
6:081  0:000    Custom Intel GFX properties injected, continue
6:081  0:000    AAPL00,DualLink = 1
6:081  0:000    FakeID Intel GFX: not set
6:081  0:000    ig-platform-id: not set
6:081  0:000    Beginning Intel GFX auto-detection with ACPI Injection
6:081  0:000    Found FakeID Intel GFX = 0x19168086
6:081  0:000    Found ig-platform-id = 0x19160000
6:081  0:000  Intel GFX revision  = 0x7
6:081  0:000   RCBA disabled; cannot use it
6:081  0:000   setting specified layout-id=28 (0x1C)
6:081  0:000  stringlength = 2028
6:081  0:000  CurrentMode: Width=1920 Height=1080
6:081  0:000  Beginning FSInjection

2) T420 - HD 3000 1600x900

retest - still good.

4:992  0:000  Intel HD Graphics 3000 [8086:0126] :: PciRoot(0x0)\Pci(0x2,0x0)
4:992  0:000    Resolution: Width=1600 Height=900
4:992  0:000    IntelHDMI: used
4:992  0:000    AAPL00,override-no-connect: added
4:992  0:000    AAPL00,DualLink = 1
4:992  0:000    FakeID Intel GFX: not set
4:992  0:000    snb-platform-id: not set
4:992  0:000    Beginning Intel GFX auto-detection with ACPI Injection
4:992  0:000    Found FakeID Intel GFX = 0x01268086
4:992  0:000    Found snb-platform-id = 0x00010000
4:992  0:000  Intel GFX revision  = 0x9
4:993  0:000   setting specified layout-id=12 (0xC)
4:993  0:000  stringlength = 2842
4:993  0:000  CurrentMode: Width=1600 Height=900
4:993  0:000  Beginning FSInjection

[Sherlocks]

@sherlocks - RC5 testing

 

1) T460 -

Graphics:

<key>Inject</key>

<dict>

<key>Intel</key>

<true/>

</dict>

Graphics section - just has Inject Intel True6:081  0:000  Intel HD Graphics 520 [8086:1916] :: PciRoot(0x0)\Pci(0x2,0x0)6:081  0:000    Resolution: Width=1920 Height=10806:081  0:000    Custom Intel GFX properties injected, continue6:081  0:000    AAPL00,DualLink = 16:081  0:000    FakeID Intel GFX: not set6:081  0:000    ig-platform-id: not set6:081  0:000    Beginning Intel GFX auto-detection with ACPI Injection6:081  0:000    Found FakeID Intel GFX = 0x191680866:081  0:000    Found ig-platform-id = 0x191600006:081  0:000  Intel GFX revision  = 0x76:081  0:000   RCBA disabled; cannot use it6:081  0:000   setting specified layout-id=28 (0x1C)6:081  0:000  stringlength = 20286:081  0:000  CurrentMode: Width=1920 Height=10806:081  0:000  Beginning FSInjection

2) T420 - HD 3000 1600x900

retest - still good.

4:992  0:000  Intel HD Graphics 3000 [8086:0126] :: PciRoot(0x0)\Pci(0x2,0x0)4:992  0:000    Resolution: Width=1600 Height=9004:992  0:000    IntelHDMI: used4:992  0:000    AAPL00,override-no-connect: added4:992  0:000    AAPL00,DualLink = 14:992  0:000    FakeID Intel GFX: not set4:992  0:000    snb-platform-id: not set4:992  0:000    Beginning Intel GFX auto-detection with ACPI Injection4:992  0:000    Found FakeID Intel GFX = 0x012680864:992  0:000    Found snb-platform-id = 0x000100004:992  0:000  Intel GFX revision  = 0x94:993  0:000   setting specified layout-id=12 (0xC)4:993  0:000  stringlength = 28424:993  0:000  CurrentMode: Width=1600 Height=9004:993  0:000  Beginning FSInjection

Thank you very much.

Did you get glitches in hd 3000 on old clover? I want to know @gijiangjiang report me glitches. If you have, keep test if you can.

 

Test list

Arrandale - not test

 

Sandybrigde - hd3000laptop(gujiangjiang, tluck). Done

 

Ivybridge-hd4000laptop(bronxteck, Riley Freeman)

 

Haswell-hd4400laptop(bronxteck),hd4600desktop(bronexteck). Remain some cards

 

Broadwell-

Hd5300tablet(Sherlocks). remain some cards

 

Skylake-

Hd520laptop(Sherlocks, tluck) hd530laptop(gujiangjiang). Remain some cards

 

Kabylake-not test

 

 

 

 

나의 LG-F410S 의 Tapatalk에서 보냄

[tluck]

@sherlocks - no glitches on HD 3000 for a long time

 

but i also inject a custom EDID and and is in Display Overrides that have

- timing fixes lines/glitches.

- set to use RGB only mode in the EDID - so there weird colored fonts. 

[mingo76]

i've a dell 7140 with hd5300, if you need i could test

[Philip Petev]

@Slice: found a minor issue with ebuild.sh. The script will fail to build Clover on the very first run (the Clover and edk2 sources have just been downloaded), if TOOLCHAIN is set to XCODE8. Currently, the script checks if the mtoc binary is properly installed for TOOLCHAIN, set to XCODE32 and XCODE5, but not if it's set to XCODE8, so the script doesn't extract and install the mtoc binary properly in this case. In order to fix it, line 542:

XCLANG|XCODE32|XCODE5) checkXcode ;;

should become:

XCLANG|XCODE*) checkXcode ;;

This way the script will handle all present and future XCODE profiles.

[Sherlocks]

i've a dell 7140 with hd5300, if you need i could test

Please test. I need broadwell debug with clover log. Works or not.

 

Thanks

 

나의 LG-F410S 의 Tapatalk에서 보냄

[JahStories]

Just to report Sherlocks, don't know if you can fix it, but clover is not correctly applying patches...

There is something wrong...

The only way to fix I've found is manually patch the kexts with hex fiend, install them and after a reboot and the reinstallation of the vanilla kext, clover sucessfully apply the patches.

But after every update I've to repeat the process...

It happens with:

Don't disable gfx0 patch (AppleGraphicsDevicePolicy.kext)

Disable minStolenSize (AppleIntelSKLGraphicsFramebuffer.kext)

 

Hope you can do something, thanks!

[mingo76]

Please test. I need broadwell debug with clover log. Works or not.

Thanks

나의 LG-F410S 의 Tapatalk에서 보냄

I set only intel inject... All is working, full res, can't provide log now sorry

 

...but can attach pic

[Sherlocks]

Just to report Sherlocks, don't know if you can fix it, but clover is not correctly applying patches...

There is something wrong...

The only way to fix I've found is manually patch the kexts with hex fiend, install them and after a reboot and the reinstallation of the vanilla kext, clover sucessfully apply the patches.

But after every update I've to repeat the process...

It happens with:

Don't disable gfx0 patch (AppleGraphicsDevicePolicy.kext)

Disable minStolenSize (AppleIntelSKLGraphicsFramebuffer.kext)

 

Hope you can do something, thanks!

DVMT problem that i already done to make source and tested it.

https://github.com/BarbaraPalvin/IntelGraphicsDVMTFixup

 

But remain to optimize it. I focus on gma.c source now, after finish, i will check.

 

What is your IGPU? please test and result to report me with clover log by using pm.

I set only intel inject... All is working, full res, can't provide log now sorrytmp_13241-IMG_20170417_2041401031811606.jpg

 

...but can attach pic

Library/Logs/CloverEFI and please take log.

 

 

If you cant enter CloverEFI, copy CloverEFI and paste where you want place.

 

Then please upload files.

 

Added. I can see now your pic.

 

Its good thank you

 

 

 

 

나의 LG-F410S 의 Tapatalk에서 보냄

[mingo76]

 

DVMT problem that i already done to make source and tested it.
https://github.com/BarbaraPalvin/IntelGraphicsDVMTFixup

But remain to optimize it. I focus on gma.c source now, after finish, i will check.

What is your IGPU? please test and result to report me with clover log by using pm.
Library/Logs/CloverEFI and please take log.

If you cant enter CloverEFI, copy CloverEFI and paste where you want place.

Then please upload files.



나의 LG-F410S 의 Tapatalk에서 보냄

 

no log in library path... I generate it with clover configurator... is it ok for you?

bootlog.log.txt

[Sherlocks]

  

 

no log in library path... I generate it with clover configurator... is it ok for you?

I cant see your pic before. I can see now your log.

 

Normally, width of resolution use 1366 or 1600 or 1920 or 2560 or 3844.

 

Your log provides 1024. Good info. properly works now that we can see log flow. HD5300 also was tested by me. Difference between you and me have differ resolution. Me 1920 you 1024.

 

After get log and check from arrandale and kabylake, i will upload source

 

Thank you. Have a great day.

 

나의 LG-F410S 의 Tapatalk에서 보냄

[PMheart]

Just to report Sherlocks, don't know if you can fix it, but clover is not correctly applying patches...

There is something wrong...

The only way to fix I've found is manually patch the kexts with hex fiend, install them and after a reboot and the reinstallation of the vanilla kext, clover sucessfully apply the patches.

But after every update I've to repeat the process...

It happens with:

Don't disable gfx0 patch (AppleGraphicsDevicePolicy.kext)

Disable minStolenSize (AppleIntelSKLGraphicsFramebuffer.kext)

 

Hope you can do something, thanks!

 

Hi.

As for AppleGraphicsDevicePolicy, you can try this patch instead and test if it works.

As for AppleIntelSKLGraphicsFramebuffer, you have to rebuild kernel cache (Use 0x12345678 ig-platform-id to boot without graphics acceleration and run `kextcache` to rebuild it.) before Clover can patch it properly due to some reasons which unfortunately I can't remember.

(By the way, I guess ForceKextToLoad may give you a hand? i.e. Adding \System\Library\Extensions\AppleIntelSKLGraphicsFramebuffer.kext to ForceKextToLoad item.)

[tluck]

DVMT problem that i already done to make source and tested it.

https://github.com/BarbaraPalvin/IntelGraphicsDVMTFixup

 

 

bravo!

 

 

As for AppleIntelSKLGraphicsFramebuffer, you have to rebuild kernel cache (Use 0x12345678 ig-platform-id to boot without graphics acceleration and run `kextcache` to rebuild it.) before Clover can patch it properly due to some reasons which unfortunately I can't remember.

 

@PMheart - it seems that sherlocks built a awesome patch mechanism to eliminate these steps of

1) inject fake ID.

2) Load patched Skylake binary (that works around DVMT KP)

3) and then rebuild caches - so Clover can patch the original on-the-fly.

[JahStories]

Ok thx for answering I'll try and post a report!

[PMheart]

bravo!

 

@PMheart - it seems that sherlocks built a awesome patch mechanism to eliminate these steps of

1) inject fake ID.

2) Load patched Skylake binary (that works around DVMT KP)

3) and then rebuild caches - so Clover can patch the original on-the-fly.

Ah sounds great! Well, I guess ForceKectToLoad may help us, or inject a patched binary via InjectKext.

But if the patch has been done, then no need to inject illegal graphics ID to prevent AppleIntelSKLGraphicsFrameBuffer from loading I guess. It won't lead to a KP if we jmp it.

 

 

Sent from my iPhone 7 using Tapatalk

[tluck]

@PMheart - well it is a catch 22 problem.

 

to use kextpatch it must be in cache.

to get in cache it must be loaded

you can not (force)load a kext if it will KP

you must patch it to stop KP

 

so we had to do this little square dance 

[PMheart]

@PMheart - well it is catch 22 problem.

 

to use kextpatch it must be in cache.

to get in cache it must be loaded

you can not (force)load a kext if it will KP

you must patch it to stop KP

 

so we had do this little square dance 

Oops, I see.

Will forcing AppleIntelSKLGraphicsFramebuffer to load via ForceKextsToLoad, then patching it and booting system make no sense? If so, that's really unfortunate...

But if I remember correctly, KextToPatch can also patch a kext that we've injected via InjectKexts, which is AppleIntelSKLGraphicsFramebuffer.kext from /S/L/E in this case.

[Slice]

Yes, Clover will patch kexts forced to load.