Jump to content
30960 posts in this topic

Recommended Posts

5 hours ago, Jief_Machak said:

What was crashing GCC, in the end ?

this :


  //void* memset(void* dst, int ch, UINTN count) __attribute__ ((used));

or the lto flag ?

 

Which version of GCC ? 9 ?

Supported LTO compression algorithms: zlib
gcc version 10.2.0 (GCC) 

LTO optimized codes at link time by erasing non-used procedures. In this case it is memset and attribute((used)) is just ignored.

But I have to comment out this line because of compilation error.

GCC crashes at link time when it encounter a procedure that is already optimized out. The problem is we have very complex system with several stages to compile and link.

I am not fully understanding what is happen here but I encountered in the past with such problems and understood that LTO should be disabled for libraries and protocols while it is very good for main module Cloverx64.efi.

2 hours ago, Slice said:

Supported LTO compression algorithms: zlib

gcc version 10.2.0 (GCC)

I have the exact same one, one no seg fault. Compiled fine with lto and attribute used.

 

Could you re-create the error, and compile with V=1 to get all command from makefile to see which exact command failed ?

V is for Verbose. It's a new I've committed recently. Cmd could be : "V=1 ./ebuild.sh {arguments}", or "V=1 ./buildme"

  • Like 1

@Jief_MachakThank you so much for fixing the USBFixOwnership commit https://github.com/CloverHackyColor/CloverBootloader/commit/5d91bbe25305104192160413b68f94955e7cb03a. It solves my problems, and now I can boot into Mojave & Catalina & Big Sur with no problems ^^. I should notice that earlier since USB Ownership will exactly cause my early boot problem. 

 

Another question: since we use OC to inject kernel extensions, FSInject.efi should be deprecated as it's the legacy Clover kext load efi driver?

Edited by stevezheng
  • Like 1
  • Confused 1
46 minutes ago, maclinuxG4 said:

 

sur ?

I am not sure what do you mean.

 

12 minutes ago, Jief_Machak said:

I don't know what that is. I never used it, even before OC integration, I think.

I found Slice's explanation here. 

It seems FSInject.efi won't be needed anymore for the OC integration.

  • Like 1
2 hours ago, Jief_Machak said:

I don't know what that is. I never used it, even before OC integration, I think.

@Jief_Machak

so could talk to slice and check for each driver in UEFI, with OC integration for a list (need / not needed)

 

this list is below, excepted openrutime.efi that is of course need:

ApfsDriverLoader.efi

AppleimageCodec.efi

AppleKeyaggregator.efi

AppelUITheme.efi

AptionInputFix.efi

Fat.efi

firmwarevolume.efi

FSinject.efi

HashServiceFix.efi

PartitionDxe.efi

SMCHelper.efi

UsbKbDxe.efi

VBoxHfs.efi

Edited by maclinuxG4

for 5126 and Catalina I have ApfsDriverLoader, AudioDxe, FSInject, HFSPlus, OpenRuntime, VirtualSmc

for 5122 and HS I have ApfsDriverLoader, AudioDxe, DataHubDxe, FSInject, HFSPlus, OCQuirks, OpenRuntime, VirtualSmc

 

BTW, just asking, do we have to compile 5123 + Fix TgtBridge by ourselves? :)

@Jief_Machak

4 hours ago, stevezheng said:

@Jief_MachakThank you so much for fixing the USBFixOwnership commit https://github.com/CloverHackyColor/CloverBootloader/commit/5d91bbe25305104192160413b68f94955e7cb03a. It solves my problems, and now I can boot into Mojave & Catalina & Big Sur with no problems ^^. I should notice that earlier since USB Ownership will exactly cause my early boot problem. 

 

Another question: since we use OC to inject kernel extensions, FSInject.efi should be deprecated as it's the legacy Clover kext load efi driver?

1. I also think that many users report like "5122 works, 5123-5126 no" can be explained by FixUsbOwnership. Thanks Jief.

2. I afraid FSInject.efi was working up to 10.7.5 when we loaded kexts separately from kernel.

  • Like 4
2 hours ago, Slice said:

"5122 works, 5123-5126 no"

Agreed (it's 5124-5126, because 5123 is still not OC integrated).

 

 

@everyone If you had problems to boot, to boot USB installer, try again with this last commit. CloverX64-5d91bbe-jief.zip

 

3 hours ago, Henry2010 said:

BTW, just asking, do we have to compile 5123 + Fix TgtBridge by ourselves? :)

Release updated https://github.com/CloverHackyColor/CloverBootloader/releases/tag/5123.1

 

Edited by Jief_Machak
  • Like 4
  • Thanks 1
  • Haha 1
22 hours ago, Jief_Machak said:

I have the exact same one, one no seg fault. Compiled fine with lto and attribute used.

 

Could you re-create the error, and compile with V=1 to get all command from makefile to see which exact command failed ?

V is for Verbose. It's a new I've committed recently. Cmd could be : "V=1 ./ebuild.sh {arguments}", or "V=1 ./buildme"

Reproduced without any problem V=1 ./ebuild.sh -t GCC53

Uncomment line with attribute((used));

Comment out the line with -fno-lto

"/Users/sergey/src/opt/local/cross/bin/x86_64-clover-linux-gnu-gcc-ar" -cr /Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/rEFIt_UEFI/refit/OUTPUT/CLOVERX64.lib  @/Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/rEFIt_UEFI/refit/OUTPUT/object_files.lst
[SLINK] CLOVERX64
"/Users/sergey/src/opt/local/cross/bin/x86_64-clover-linux-gnu-gcc" -o /Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/rEFIt_UEFI/refit/DEBUG/CLOVERX64.dll -nostdlib -Wl,-n,-q,--gc-sections,-z,common-page-size=0x40,--defsym=PECOFF_HEADER_SIZE=0x240,--script=/Users/sergey/src/CloverBootloader/BaseTools/Scripts/GccBase.lds,--entry,_ModuleEntryPoint,-u,_ModuleEntryPoint,-Map,/Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/rEFIt_UEFI/refit/DEBUG/CLOVERX64.map,-melf_x86_64,--oformat=elf64-x86-64,-pie -fshort-wchar -fno-builtin -fno-strict-aliasing -Wall -Werror -Wno-array-bounds -ffunction-sections -fdata-sections -include AutoGen.h -fno-common -DSTRING_ARRAY_NAME=CLOVERX64Strings -m64 -fno-stack-protector -mabi=ms "-DEFIAPI=__attribute__((ms_abi))" -maccumulate-outgoing-args -mno-red-zone -Wno-address -mcmodel=small -fpie -fno-asynchronous-unwind-tables -Wno-address -Wno-unused-const-variable -Wno-unused-but-set-variable -Wl,--start-group   @/Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/rEFIt_UEFI/refit/OUTPUT/static_library_files.lst -Wl,--end-group
"/Users/sergey/src/opt/local/cross/bin/x86_64-clover-linux-gnu-objcopy"  /Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/rEFIt_UEFI/refit/DEBUG/CLOVERX64.dll
[DLINK] CLOVERX64
/Users/sergey/src/opt/local/cross/bin/x86_64-clover-linux-gnu-objcopy --only-keep-debug /Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/rEFIt_UEFI/refit/DEBUG/CLOVERX64.dll /Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/rEFIt_UEFI/refit/DEBUG/CLOVERX64.debug
/Users/sergey/src/opt/local/cross/bin/x86_64-clover-linux-gnu-objcopy  /Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/rEFIt_UEFI/refit/DEBUG/CLOVERX64.dll
cp -f /Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/rEFIt_UEFI/refit/DEBUG/CLOVERX64.debug /Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/CLOVERX64.debug
[OBJCOPY] CLOVERX64
"GenFw" -e UEFI_APPLICATION -o /Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/rEFIt_UEFI/refit/OUTPUT/CLOVERX64.efi /Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/rEFIt_UEFI/refit/DEBUG/CLOVERX64.dll
make: *** [/Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/rEFIt_UEFI/refit/OUTPUT/CLOVERX64.efi] Segmentation fault: 11


build.py...
 : error 7000: Failed to execute command
	make tbuild [/Users/sergey/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/rEFIt_UEFI/refit]


build.py...
 : error F002: Failed to build module
	/Users/sergey/src/CloverBootloader/rEFIt_UEFI/refit.inf [X64, GCC53, RELEASE]

- Failed -
Build end time: 21:58:29, Nov.05 2020
Build total time: 00:02:17

iMac2017:CloverBootloader sergey$ 

Try split the problem...

  • Thanks 1
11 hours ago, stevezheng said:

@Jief_MachakThank you so much for fixing the USBFixOwnership commit https://github.com/CloverHackyColor/CloverBootloader/commit/5d91bbe25305104192160413b68f94955e7cb03a. It solves my problems, and now I can boot into Mojave & Catalina & Big Sur with no problems ^^. I should notice that earlier since USB Ownership will exactly cause my early boot problem.

 

Hello @Slice and @Jief_Machak I can also see in the Clover releases list that the last build before OpenCore integration was released as 5123.1:

https://github.com/CloverHackyColor/CloverBootloader/releases/tag/5123.1

 

Again, for those wishing to stay with up to Catalina, this is great help that would keep Clover less complicated to use (I admit I have a hard time understanding the efforts for Big Sur although always appreciative).

 

@Jief_Machak is this build affected by the bug @stevezheng refers to? My Clover config does not use USB's FixOwnership = true and this is the reason I did not detect any bug.

 

Thank you again both for your help, time, and effort and everyone else for contributing!

Edited by MacKonsti
  • Like 1
7 hours ago, Slice said:

1. I also think that many users report like "5122 works, 5123-5126 no" can be explained by FixUsbOwnership. Thanks Jief.

2. I afraid FSInject.efi was working up to 10.7.5 when we loaded kexts separately from kernel.

So @Slice you say that for recent macOS versions, we don't need FSInject.efi ?

I ask because the latest Clover releases for some months, included that file in the "basic" minimum drivers folder...

In my r5123.1 boot log (via Hackintool) a search for "fsinject" string does produce results. For example:

0:111  0:000  === [ LoadDrivers ] =============================
0:112  0:001  Loading ApfsDriverLoader.efi  status=Success
0:264  0:151   - driver needs connecting
0:264  0:000  Loading AptioMemoryFix.efi  status=Success
0:415  0:151  Loading FSInject.efi  status=Success
0:567  0:151  Loading HFSPlus.efi  status=Success
0:718  0:151   - driver needs connecting
0:718  0:000  Loading VirtualSMC.efi  status=Success
0:871  0:152  2 drivers needs connecting ...
0:871  0:000  PlatformDriverOverrideProtocol not found. Installing ... Success
0:871  0:000  HFS+ driver loaded
0:871  0:000  APFS driver loaded
0:871  0:000  Searching for invalid DiskIo BY_DRIVER connects: not found, all ok
[...]
12:963  0:000  Beginning FSInjection
FSInjectionInstall ...
- Our FSI_SIMPLE_FILE_SYSTEM_PROTOCOL installed on handle: 75311918
FSInjectionInstall ...
- Our FSI_SIMPLE_FILE_SYSTEM_PROTOCOL installed on handle: 75311918
12:967  0:004  Use origin smbios table type 1 guid.
12:973  0:005  UniOSVersion == 10.15.7
12:973  0:000  UniShortOSVersion == 10.15
[...]

When I removed it from my "drivers/UEFI/" folder, my Lenovo upon next reboot crashed during Catalina boot and took a couple of boots to come to desktop :(

I think it was crashed the moment WhateverGreen was loading (I know the boot screen flashing for a moment).

The bdmesg log in Hackintool reports now:

3:126  0:000  Beginning FSInjection
3:126  0:000   - ERROR: gFSInjectProtocolGuid not found!

I am not totally sure if it is safe to remove it?

Edited by MacKonsti
  • Sad 1

Apple released a new supplemental update for 10.15.7 today,

 

now OC 063 and Clover 5126 both need to be updated(?) as the macOS installer says something like it "cannot install" LOL

 

Edit: Security/SecureBootModel is disabled.

Edited by Henry2010
  • Like 2

I have updated my rig in signature to macOS Catalina 19H15 without problem, Clover revision: 5126 (master, commit 9e7b4311c)

 

1454641494_Screenshot2020-11-06at08_53_53.png.b31c7857206f1b7e88b79fc9c537661c.png

 

first update step create a macOS Catalina Install preboot icon .

AMD users must not use it to continue the update because otherwise system hangs.

macOS Catalina install data icon is fine :)

 

@Slice and @Jief_Machak

I just realized current Clover can't give properly SMBIOS based cpu generation.

With old version before "mix" with OpenCore Stuff, Clover can give proper SMBIOS with "blank SMBIOS" property in config.plist.

But now, with blank SMBIOS in config.plist, Clover just give "MacPro3.1".

I hope this good features comeback again,, thank you

 

3 hours ago, fabiosun said:

first update step create a macOS Catalina Install preboot icon .

AMD users must not use it to continue the update because otherwise system hangs.

macOS Catalina install data icon is fine :)

 

all the icons of the various macOS do not appear on the screen, in the second screenshot I moved to the right
specified this so as not to cause confusion, this behavior occurred on all the hacks in which I updated catalina to 19H5

 

1874809745_Schermata2020-11-06alle12_28_59.thumb.png.8fdeeacb785e96da1484ac674e880f89.png

1682255290_Schermata2020-11-06alle12_29_22.thumb.png.e654620bff4beb3d9491d5d61f0fbbb8.png

1388246954_Schermata2020-11-06alle12_26_58.png.0ea0916dfb76016417ec0c9b261b59ed.png

 

  • Like 3
  • Confused 1
4 minutes ago, Henry2010 said:

That's very strange then...

Neither "Install from SSD - Data via Preboot" or "Install from SSD - Data" works.

It's always "macOS could not be installed on your computer"

 

I performed the catalina update to the 19H5 on X570, Z370, Z68 and Z97 and in all this anomaly, perhaps due to the partition scanning algorithm that Clover has at boot; but the update happened everywhere without any difficulty, just choose the right icon ;)

reset nvram with f11 then to the push f3 gui to view all partitions @Henry2010

  • Like 3

Big sur 10.11.01 installed
all is working - Clover 5126

100-series motherboard

config.plist

 

thanks 2 miljpn his efi folder/kext/Config.plist Quirks etc. with changed config.plist

- add IntelMausi.kext instead RealtekRTL8111.kext

- add new USBPorts.kext for wireless Usb Mouse/keyboard with hackintool

 

 

Schermafbeelding 2020-11-07 om 16.13.28.png

Edited by himac59
  • Like 2
  • Thanks 1
  • Haha 1
×
×
  • Create New...