OpenCore General Discussion

[eSaF]

1 hour ago, AlvinGenius said:

 

@eSaF Thanks for your reply

 

So, no way to enable only audio from AMD Radeon HD 8690A.?

 

Even while using this screen as HDMI In from another PC, the audio isn't working in the other PC windows and it's AMD audio ...!!

I don't know but I will see what I can find and get back to you.  

 

 PS - Have a look here I think it includes your machine - https://github.com/mbrula/hackintosh - Good luck.

Edited October 15, 2020 by eSaF

[ghost8282]

Anybody having issues accessing recovery in big sur beta 10?

My case is that with prev-lang:kbd=aXQtSVQ6MjIz (italian lang+italian keyboard) trying to access recovery, after the apple logo and the progress bar I have a grey screen (the top menu "Language Chooser" rapidly flashes and then disappears).

If I empty the prev-lang:kbd in config.plist the language chooser menu appears correctly, if I choose "Italiano" again grey screen; choosing "English" works.

Also the italian keyboard layout choosen from the top right menu after the language selector seems to have issues, some keys are not mapped correctly.

 

It seems an apple bug.

[maxb2000]

Why sometimes is required to disable SecureBootModel in order to update macOS Big Sur betas? 

[ghost8282]

5 minutes ago, maxb2000 said:

Why sometimes is required to disable SecureBootModel in order to update macOS Big Sur betas? 

Because it boots from the snapshot, and it is not sometimes, but always (I think from beta 7-8).

Edited October 16, 2020 by ghost8282

[Planet X]

1 hour ago, ghost8282 said:

Because it boots from the snapshot, and it is not sometimes, but always (I think from beta 7-8).

Will that change again with the final release? 

[ghost8282]

22 minutes ago, Planet X said:

Will that change again with the final release? 

who knows?

[ghost8282]

@Andrey1970 @Download-Fritz

I found some issue in building edk2 on macos with xcode/clang.

I prefer to build myself OVMF_CODE.fd and OVMF_VARS.fd for my vm and just recently I found some strange behavior with my files.

Relevant bug I opened is here:
https://bugzilla.tianocore.org/show_bug.cgi?id=3006

 

To summarize xcode/clang (tested on catalina and big sur) has some issue with the RELEASE version (tested 202005 and 202008): my fd files are somehow "corrupted" when clang optimize the build in the RELEASE version.

DEBUG version and linux built versions (both RELEASE and DEBUG) have no issues.

 

Since in opencore Configuration.pdf there are instructions about how to build edk2 (3.3 Contribution) the RELEASE built version with XCODE5 may have some issue?

Not sure if and how to commit this?Is it relevant for opencore?

Edited October 16, 2020 by ghost8282

[mhaeuser]

@ghost8282 We recently had an issue where XCODE5 builds did not include HII sections where declared but CLANG38 builds of the same codebase did. I don't think that is the issue, just saying: Yes, there may be differences between toolchains. Honestly, I'm a bit tired of EDK II by now.

 

For debugging this issue, I can only think of two things.

1) You can try tightening optimisation step-by-step (i.e. O1, O2, ...) and once it breaks, try to find the offending module and bin-diff

2) You can try to find the offending code section by e.g. serial logs and compare the disassembly to the code for inadequate optimisations

 

EDK II has a lot of illegal casts and alignment issues, I could very well see some SSE-based optimisation or whatever breaking things.

Edited October 16, 2020 by Download-Fritz

[vit9696]

I do not believe SSE is enabled when targeting EFI in Xcode. It is quite a bit complicated to discover it in such a complex piece OVMF is.

 

Basically you have a firmware that is built by a package, which is built from a number of drivers. These drivers consist of libraries. Libraries are built once for the whole package and are then linked into the firmware. So even though you can adjust the compilation flags of the drivers and the libraries separately, one library will always affect all the drivers, and 80% of driver code come from the libraries.

 

The only "working" solution I can think of is building the drivers separately (i.e. their DEBUG versions that are known to work) and then replace their counterparts via binary blobs. Here is an example inf loading the HFS+ driver: https://github.com/acidanthera/OpenCorePkg/blob/master/Legacy/BinDrivers/HfsPlus.inf. You just put the driver relative to this inf file and then include it in the firmware: https://github.com/acidanthera/OpenCorePkg/blob/0.6.2/OpenDuetPkg.fdf#L119.

 

By generating such .inf files with a script you should be able to quickly build the overrides for the whole OVMF DXE volume and put them here one by one: https://github.com/acidanthera/audk/blob/cbccf995920a28071f5403b847f29ebf8b732fa9/OvmfPkg/OvmfPkgX64.fdf#L202. Once you figure out which driver or which set of drivers is at fault (try starting with a simple binary search: try that replacing all the DXE drivers with DEBUG overrides makes the RELEASE version work, then try replacing a half of them, and so on), you should be able to compare the two more precisely. Here we could be of some help I guess.

[xKaoSx]

Hi everbody.

 

I am install macos big sur beta 1-9 not problem but beta update 10 finish and not open macos. 

 

System:

Asus TUF H310 Plus Gaming

İntel i5-9400F

Rx 570VR

16 GB DDR4 Ram

 

I am try OC 0.6.2 bu not effect and update my EFI

 

EFI

OC 0.6.3

WEG 1.4.4

VirtualSMC 1.1.8

İntelMausi 1.0.5

Lilu 1.4.9

AppleALC 1.5.4

USBInjectALL 0.7.5

 

Boot args : -v debug=0x100 keepsyms=1 dart=0 vsmcgen=1

 

Edited October 16, 2020 by xKaoSx

[En0r1m]

Hello guys,
installed Linux Manjaro on a partition
and in my EFI Folder with OC 0.6.2 I have a Manjaro folder with grubx64.efi, and in the OC boot menu what is shown to me is an EFI icon and by booting it I can access Manjaro, but some way that does not appear the EFI partition, but the Linux Manjaro? I hope it was clear here.

[Mike Ranger]

Hi... with the latest Developer Beta 10, after the the reboot during installation I get the following KP:

 

panic(cpu 0 caller 0xffffff80192bd8ea): "Rooting from the live fs of a sealed volume is not allowed on a RELEASE build\n"@/AppleInternal/BuildRoot/Library/Caches/com.apple.xbs/Sources/apfs/apfs-1677.40.51.201.1/kext/apfs_vfsops.c:2011
Backtrace (CPU 0), Frame : Return Address
0xffffffa08fc6ae20 : 0xffffff80162b666d mach_kernel : _handle_debugger_trap + 0x3dd
0xffffffa08fc6ae70 : 0xffffff80163f8f93 mach_kernel : _kdp_i386_trap + 0x143
0xffffffa08fc6aeb0 : 0xffffff80163e95ca mach_kernel : _kernel_trap + 0x55a
0xffffffa08fc6af00 : 0xffffff801a00e924 as.vit9696.VirtualSMC : __ZN18VirtualSMCProvider10kernelTrapI22x86_saved_state_1010_tEEvPT_Pm + 0x454
0xffffffa08fc6af80 : 0xffffff801625ba2f mach_kernel : _return_from_trap + 0xff
0xffffffa08fc6afa0 : 0xffffff80162b5f0d mach_kernel : _DebuggerTrapWithState + 0xad
0xffffffa08fc6b0c0 : 0xffffff80162b61f8 mach_kernel : _panic_trap_to_debugger + 0x268
0xffffffa08fc6b130 : 0xffffff8016ab792a mach_kernel : _panic + 0x54
0xffffffa08fc6b1a0 : 0xffffff80192bd8ea com.apple.filesystems.apfs : _apfs_vfsop_mount + 0x339b
0xffffffa08fc6b990 : 0xffffff80192c6d70 com.apple.filesystems.apfs : _apfs_vfsop_mountroot + 0x3d
0xffffffa08fc6b9c0 : 0xffffff801653cb20 mach_kernel : _vfs_mountroot + 0x140
0xffffffa08fc6bb40 : 0xffffff80167f9883 mach_kernel : _set_rootvnode + 0x3c53
0xffffffa08fc6be70 : 0xffffff80162e3e83 mach_kernel : _max_valid_stack_address + 0xdf3
0xffffffa08fc6bfa0 : 0xffffff801625b13e mach_kernel : _call_continuation + 0x2e
      Kernel Extensions in backtrace:
         as.vit9696.VirtualSMC(1.1.8)[7BBAC837-0013-398A-BBEE-6BF8DD81C31B]@0xffffff8019fff000->0xffffff801a025fff
            dependency: as.vit9696.Lilu(1.4.9)[AC8A0C7F-6FCC-3F6A-9A82-96789EEDE7B5]@0xffffff8019f79000->0xffffff8019ffcfff
            dependency: com.apple.iokit.IOACPIFamily(1.4)[F42E4E8D-BBE9-37B7-A3A4-9D9641BBB848]@0xffffff80186aa000->0xffffff80186abfff
         com.apple.filesystems.apfs(1677.40.51)[82FE49D0-2A08-3128-B5A7-D0D3BF99D28D]@0xffffff8019243000->0xffffff80193a6fff
            dependency: com.apple.driver.AppleEFINVRAM(2.1)[AB65643D-1BE4-35C1-B566-2898844DADE0]@0xffffff8017634000->0xffffff801763dfff
            dependency: com.apple.driver.AppleEffaceableStorage(1.0)[7A75BFCC-F40A-3991-8D3D-BA843F37A7F8]@0xffffff8017647000->0xffffff801764cfff
            dependency: com.apple.iokit.CoreAnalyticsFamily(1)[B69C1AF1-494E-3321-B1FE-1EC42C6F2A5E]@0xffffff8017a82000->0xffffff8017a88fff
            dependency: com.apple.iokit.IOStorageFamily(2.1)[63B7DD4D-C3CC-311C-9FD7-30032779F1F8]@0xffffff8018d0f000->0xffffff8018d20fff
            dependency: com.apple.kec.corecrypto(1.0)[B1A179FD-27C2-3F5B-A97B-267774ECA9DF]@0xffffff80193d4000->0xffffff8019460fff
            dependency: com.apple.security.AppleImage4(3.0.0)[9F4F723F-52EE-3B0F-AB02-9755CCEB7DF7]@0xffffff80176a9000->0xffffff80176b8fff

Process name corresponding to current thread: kernel_task
Boot args: keepsyms=1 debug=0x100 npci=0x2000 -v agdpmod=pikera -bsdmgroot chunklist-security-epoch=0 -chunklist-no-rev2-dev

Mac OS version:
Not yet set

Kernel version:
Darwin Kernel Version 20.1.0: Wed Oct  7 21:17:40 PDT 2020; root:xnu-7195.40.137.201.1~1/RELEASE_X86_64
Kernel UUID: D5AE727A-F8F2-31EE-B874-4E907B0C5115
KernelCache slide: 0x0000000016000000
KernelCache base:  0xffffff8016200000
Kernel slide:      0x0000000016010000
Kernel text base:  0xffffff8016210000
__HIB  text base: 0xffffff8016100000
System model name: iMacPro1,1 (Mac-7BA5B2D9E42DDD94)
System shutdown begun: NO
Panic diags file unavailable, panic occurred prior to initialization
Hibernation exit count: 0

System uptime in nanoseconds: 9406301602
Last Sleep:           absolute           base_tsc          base_nano
  Uptime  : 0x00000002346808ff
  Sleep   : 0x0000000000000000 0x0000000000000000 0x0000000000000000
  Wake    : 0x0000000000000000 0x000017a143d7e380 0x0000000000000000

Not really sure what is going on. Beta 9 was working fine.

Regards, Mike

[eSaF]

1 minute ago, Mike Ranger said:

Hi... with the latest Developer Beta 10, after the the reboot during installation I get the following KP:

Did you make the necessary changes in the config.plist e.g Security/SecureBootModel/ change this from Default to Disabled also insure car-active-config is fully enabled (00000000) also won't hurt to run 'FirstAid' in Disk Utility on all BS volumes reboot and clean NVRAM then see how it goes. Good luck.

[Mike Ranger]

@eSaF: many thanks, that worked out perfectly.

Best, Mike

[ghost8282]

On 10/16/2020 at 2:40 PM, Download-Fritz said:

1) You can try tightening optimisation step-by-step (i.e. O1, O2, ...) and once it breaks, try to find the offending module and bin-diff

 

EDITED

It seems both debug and release are built with -Os, so it seems it's not an optimization issue.

 

Invalid info

 

All I can say is that files work properly with clang 9.0.0 and 12.0.0 and do not work properly with clang 7.0.0

 

Personal note: the world needs more Download-Fritz and vit9696! it should be a better world!!It's clearly seen in opencore, not only "it just works" as most users are finding, order and precision are all over the code and I appreciate it a lot.

Edited October 17, 2020 by ghost8282

[vit9696]

Something is wrong with this description I am afraid. You should not be able to build any binary at all without running make -C BaseTools. Maybe you had some older tools compiled in some other directory out of the tree (i.e. outside of edk2 directory), and they were used instead of the ones you checked out?

[ghost8282]

2 hours ago, vit9696 said:

Something is wrong with this description I am afraid. You should not be able to build any binary at all without running make -C BaseTools. Maybe you had some older tools compiled in some other directory out of the tree (i.e. outside of edk2 directory), and they were used instead of the ones you checked out?

 

I think no, if I look at the log made with clang 7.0.0, it seems basetools are automatically compiled from edk2 directory when I run "nice make -C "$EDK_TOOLS_PATH" -j $(getconf _NPROCESSORS_ONLN)", but maybe I'm not reading the log correctly.

This doesn't happen with other toolchains.

Moreover, if I had already installed BaseTools somewhere why only clang 7.0.0 has problem?

I mean, I used clang 9.0.0 before clang 7.0.0 and clang 9.0.0 threw error about compiling BaseTools first.

For me it's a mistery

 

log-macOs.txt.zip

Edited October 17, 2020 by ghost8282

[mhaeuser]

11 minutes ago, ghost8282 said:

This doesn't happen with other toolchains.

What do you mean by "doesn't happen"? The command does nothing when executed? Nothing logged either? What happens when you do not include "-j $(getconf _NPROCESSORS_ONLN)"? Does $EDK_TOOLS_PATH evaluate properly?

 

I wonder whether multithreaded BaseTools building is broken, but vit uses it...

Edited October 17, 2020 by Download-Fritz

[ghost8282]

2 hours ago, Download-Fritz said:

What do you mean by "doesn't happen"?

 

Sorry, I mean that with clang 9 and 12 I must run "make -C BaseTools" before "source edksetup.sh" / "nice make -C "$EDK_TOOLS_PATH" -j $(getconf _NPROCESSORS_ONLN)".

 

With clang 7: no errors about BaseTools (no command "make -C BaseTools" needed), so at the beginning I was not running "make -C BaseTools" and from what I can understand BaseTools are compiled with "nice make -C "$EDK_TOOLS_PATH" -j $(getconf _NPROCESSORS_ONLN)".

 

I'm starting from the beginning again saving all the logs now.

 

Update: I modify my post above with correct informations and I'm going to upload the logs in a next post.

 

Edited October 17, 2020 by ghost8282

[ghost8282]

Tried 3 tests with more calm and step by step instructions:

all I can say is that clang 7.0.0 is not generating properly the files, so my post above was totally wrong with invalid steps (I edited it).

 

Test 1: clang 12.0.0 - Files OK

Test 2: clang 7.0.0 - Files NOT ok

Test 3: clang 7.0.0 - Files NOT ok

 

Everytime I deleted edk2 folder and started new.

 

By "files NOT ok" I mean I cannot boot without hdmi plugged in (I can't boot only with dvi, I can only boot with hdmi or hdmi+dvi)

 

Commands for tests 1 and 2:

$ git clone https://github.com/tianocore/edk2.git
$ cd edk2
$ git clean -ffdx
$ git reset --hard
$ git submodule deinit --force --all
$ git checkout edk2-stable202008
$ git submodule update --init --force
$ source edksetup.sh
$ nice make -C "$EDK_TOOLS_PATH" -j $(getconf _NPROCESSORS_ONLN)
$ build -a X64 -b RELEASE -p OvmfPkg/OvmfPkgX64.dsc -t XCODE5

Commands for test 3:

$ git clone https://github.com/tianocore/edk2.git
$ cd edk2
$ git clean -ffdx
$ git reset --hard
$ git submodule deinit --force --all
$ git checkout edk2-stable202008
$ git submodule update --init --force
$ make -C BaseTools/
$ source edksetup.sh
$ build -a X64 -b RELEASE -p OvmfPkg/OvmfPkgX64.dsc -t XCODE5

Logs attached.

So, nothing new, need to go with Download-Fritz step 2 or vit9696 suggestions dho!

 

Sorry, maybe I didn't restart libvirtd to use the files generated by clang 7.0.0.

I will make more tests in the next days.

Clang7-test3-log.txt.zip

Clang7-test2-log.txt.zip

Clang12-test1-log.txt.zip

[mhaeuser]

@ghost8282 Interesting, but what if you run make with 7 and build with 12?

[1Revenger1]

1 hour ago, mifjpn said:

I am grateful to the development team for their daily efforts.

There are the following problems.

Start Windows, then restart.

When I hold down the DEL key and reboot, the BIOS screen does not appear.

Black screen only.

The only way to get this back back is to flash the BIOS again.

This issue occurs on the ASUS PRIME H370-A (i7 8700). (BIOS is up to date)

Is there anyone with the same problem?

Thank you.

Try setting Misc->Security->BootProtect to None (likely is Bootstrap right now). Haven't seen this issue occur in H370 before though - usually just Z87/Z97

Edit: https://github.com/acidanthera/bugtracker/issues/1222

I think that's the relevant issue on the bugtracker.

Edited October 18, 2020 by 1Revenger1

[ghost8282]

11 hours ago, Download-Fritz said:

@ghost8282 Interesting, but what if you run make with 7 and build with 12?

 

Files are OK!

 

Test4: make with clang 7.0.0, build with clang 12.0.0

 

With clang 7.0.0:

$ git clone https://github.com/tianocore/edk2.git
$ cd edk2
$ git clean -ffdx
$ git reset --hard
$ git submodule deinit --force --all
$ git checkout edk2-stable202008
$ git submodule update --init --force
$ make -C BaseTools/

Switch to clang 12.0.0:

$ source edksetup.sh
$ build -a X64 -b RELEASE -p OvmfPkg/OvmfPkgX64.dsc -t XCODE5

 

NO gpu issue, tianocore logo not appearing but this was expected.

 

So this means it's the build command having issues with clang 7.0.0

Test4-make-clang7-build-clang12.txt.zip

Edited October 18, 2020 by ghost8282

[Matgen84]

Hi @vit9696 @Download-Fritz

 

I've this message when I compile commit. Can you explain to me, what I'm doing wrong. Please.

From https://github.com/openssl/openssl
   b40498c6e7..6c8149df1f  master     -> origin/master
Could not access submodule 'boringssl'Could not access submodule 'krb5'Could not access submodule 'pyca-cryptography'Fetching submodule MdeModulePkg/Library/BrotliCustomDecompressLib/brotli
Could not access submodule 'research/esaxx'Could not access submodule 'research/libdivsufsort'Fetching submodule MdeModulePkg/Universal/RegularExpressionDxe/oniguruma

 

[mhaeuser]

@Matgen84 If you google the "Could not access submodule" error, multiple people have that issue but there is no clear solution by anyone... looks like a git bug maybe? To be honest, just clone again, I have no idea.