Clover General discussion

[apianti]

Does XCODE 10.1 build previous revisions to r4761 without issue? Or does it not build a valid boot firmware at all? Are the different build rules for XCODE 10.1 causing the problem?

[Planet X]

On 11/29/2018 at 1:48 AM, scj312 said:

For some reason when I have an APFS volume present, Clover takes a very long time to display the GUI (30+ seconds). I am using ApfsDriverLoader and there is no delay when booting without an APFS volume present. In the preboot log it looks like [InitScreen] line is 47 seconds. Any ideas?

preboot.log

Do you use a NVMe drive? The same happened to me. I use apfs.efi from 10.13.0 again as it boots way faster. But please somebody have a look into this? NVMe drives should also boot faster isn't it?

[Slice]

14 hours ago, fusion71au said:

 

I believe compilation with XCODE 10.1 is the problem.  From your debug log, you are compiling Clover with XCODE 10.1 (with toolchain set to XCODE8) on your system... 

 

Build with: [Args: -mc --no-usb -D NO_GRUB_DRIVERS_EMBEDDED -t XCODE8 | -D DISABLE_USB_SUPPORT -D NO_GRUB_DRIVERS_EMBEDDED --conf=/Users/pkovacs/src/edk2/Conf -D USE_BIOS_BLOCKIO -D USE_LOW_EBDA -a X64 -b RELEASE -t XCODE8 -n 9 | OS: 10.13.6 | XCODE: 10.1]

 

I already mentioned in this post that other XCODE versions eg 8.2.1, 8.3.3, 9.2, or GCC5.3, generate working boot6 and boot7 for my legacy desktop (system 2 in signature) but not XCODE 10.1 ---> boot hanging on stuck underscore "_" symbol.  Also interesting to note that the officially released Clover r4741 was compiled with XCODE 8.3.3 but r4769 (which also fails for your system) with XCODE 10.1.

 

 

 

No, still bad boot6,7 if Clover r4760 compiled with XCODE 10.1 (toolchain set to XCODE8, edk2 r27956).  I think the problem/bug is intrinsic to XCODE 10.1.

At this moment I can include boot6 and boot7 files from good 4741 revision into next releases until the issue will be resolved.

[pkdesign]

19 hours ago, fusion71au said:

 

I believe compilation with XCODE 10.1 is the problem.  From your debug log, you are compiling Clover with XCODE 10.1 (with toolchain set to XCODE8) on your system... 

 

Build with: [Args: -mc --no-usb -D NO_GRUB_DRIVERS_EMBEDDED -t XCODE8 | -D DISABLE_USB_SUPPORT -D NO_GRUB_DRIVERS_EMBEDDED --conf=/Users/pkovacs/src/edk2/Conf -D USE_BIOS_BLOCKIO -D USE_LOW_EBDA -a X64 -b RELEASE -t XCODE8 -n 9 | OS: 10.13.6 | XCODE: 10.1]

 

I already mentioned in this post that other XCODE versions eg 8.2.1, 8.3.3, 9.2, or GCC5.3, generate working boot6 and boot7 for my legacy desktop (system 2 in signature) but not XCODE 10.1 ---> boot hanging on stuck underscore "_" symbol.  Also interesting to note that the officially released Clover r4741 was compiled with XCODE 8.3.3 but r4769 (which also fails for your system) with XCODE 10.1.

 

 

 

No, still bad boot6,7 if Clover r4760 compiled with XCODE 10.1 (toolchain set to XCODE8, edk2 r27956).  I think the problem/bug is intrinsic to XCODE 10.1.

Ah, i see. What an odd problem.

Can anyone build current Clover with Xcode 8 and I can try on my machine?

[apianti]

20 hours ago, fusion71au said:

No, still bad boot6,7 if Clover r4760 compiled with XCODE 10.1 (toolchain set to XCODE8, edk2 r27956).  I think the problem/bug is intrinsic to XCODE 10.1.

 

Are you sure there is not a problem with the build rules for XCODE 10.1? I don't know if you mean that the problem is an intrinsic function or that it's inherent to XCODE 10.1. If you mean the former, can you narrow down exactly what intrinsic is causing the problem since really all that an intrinsic should be doing is generating an instruction or set of instructions in the same way, this behavior shouldn't really change. For either, the most likely scenario is that optimization is deciding that some needed instructions are not helpful for some reason, have you tried completely disabling optimization?

 

EDIT: Didn't realize you said the toolchain is set to XCODE8, are there no toolchain build rules for XCODE10.1? There may need to be different decisions made, so there should probably be XCODE10 toolchain.

Edited November 30, 2018 by apianti

[fusion71au]

On 12/1/2018 at 2:47 AM, pkdesign said:

Ah, i see. What an odd problem.

Can anyone build current Clover with Xcode 8 and I can try on my machine?

 

Attached Clover r4784 compiled with XCODE 8.2.1 in 10.11.6: Clover_v2.4k_r4784.zip

 

Alternatively, you can use GCC instead of XCODE to build Clover eg if you use the Build_Clover.command script, edit the BuildCloverConfig.txt file like below...

 

GNU=GCC53
Build_Tool=GNU

 

On 12/1/2018 at 4:35 AM, apianti said:

EDIT: Didn't realize you said the toolchain is set to XCODE8, are there no toolchain build rules for XCODE10.1? There may need to be different decisions made, so there should probably be XCODE10 toolchain.

 Unfortunately, this expertise is beyond my pay grade...calling @Zenith432 for help?

 

 

13 hours ago, apianti said:

 

You can try a test by changing the tools_def.txt file in your UDK2018\EDK2/Conf directory and modifying the XCODE8 definitions near the bottom, either line 6227 or line 7000 depending which branch you are using, change the flag -Os to -O0 to disable optimizations for debug. Then build the debug version and see if that results in a working boot firmware. If so, then some specific optimization needs disabled, if not, then new rules need made for XCODE10. In that case, I'll let Zenith432 figure out what that would be since he's much better at that and I don't have the time to screw around to figure out what the problem is...

 

EDIT: You might also want to try adding -O0 to the the ASM rule a few lines before as well to see if that makes a difference.

 

Disabled optimization during compilation with -O0 flag in tools_def.txt still results in bad boot6,7 using XCODE 10.1.  Looks like new rules for XCODE10 will be required...

 

Edited December 1, 2018 by fusion71au
Tried @apianti's suggestion

[Slice]

Yesterday I released wrong package. Today is reloaded. 

Update your Clover again if any issue.

[apianti]

3 hours ago, fusion71au said:

 Unfortunately, this expertise is beyond my pay grade...calling @Zenith432 for help?

 

You can try a test by changing the tools_def.txt file in your UDK2018\EDK2/Conf directory and modifying the XCODE8 definitions near the bottom, either line 6227 or line 7000 depending which branch you are using, change the flag -Os to -O0 to disable optimizations for debug. Then build the debug version and see if that results in a working boot firmware. If so, then some specific optimization needs disabled, if not, then new rules need made for XCODE10. In that case, I'll let Zenith432 figure out what that would be since he's much better at that and I don't have the time to screw around to figure out what the problem is...

 

EDIT: You might also want to try adding -O0 to the the ASM rule a few lines before as well to see if that makes a difference.

Edited December 1, 2018 by apianti

[Matgen84]

Hi

 

I use Debug kexts function from Clover GUI to diagnose my system. Where I can the log file? Please.

 

Thanks

[Slice]

/EFI/CLOVER/misc/debug.log

[Matgen84]

1 hour ago, Slice said:

/EFI/CLOVER/misc/debug.log

For Debug kexts, there is nothing inside the folder. I try again. Thanks.

[Slice]

2 hours ago, Matgen84 said:

For Debug kexts, there is nothing inside the folder. I try again. Thanks.

You means Debug in section KextPatches? Log is impossible here. You can see the process only on screen.

[Matgen84]

19 minutes ago, Slice said:

You means Debug in section KextPatches? Log is impossible here. You can see the process only on screen.

Exactly. Thanks you so much.

[Matgen84]

Hi,

 

Despite of localizable.strings file, Theme description for Clovy doesn't exist in clover Package from SourceForge, or build with Build_Clover.command and Did script. 

 

I don't understand, why? Some ideas. Please.

[pkdesign]

On 11/30/2018 at 9:52 PM, fusion71au said:

 

Attached Clover r4784 compiled with XCODE 8.2.1 in 10.11.6: Clover_v2.4k_r4784.zip

 

 

Thanks. Will test this tonight or tomorrow.

[pkdesign]

Sorry it took me so long to install. I couldn't mess with my production machine and still trying to figure out why my AMD RX 570 doesn’t work right.

 

Installed this version of Clover and it worked without a hitch. So it must be the Xcode version that causes the issue.

[midi-sama]

Hi,

Does `MatchOS` work in AddProperties section?

[Slice]

1 hour ago, midi-sama said:

Hi,

Does `MatchOS` work in AddProperties section?

No.

[Riley Freeman]

Is this new bootercfg log parameter supposed to stop the boot text spam I've been getting with the latest ApfsLoader? Because on my X79 with r4796 it's still showing up.

 

I had to write the bootercfg log=0 value into nvram manually from the recovery partition because it isn't being written by Clover. Clover is reading it ok though (I checked in the options menu).

 

It wrote the value ok on my laptop but that has a APFS partition so I don't get the boot text. I assume it only shows up when there's no APFS volume or container.

[LockDown]

i use the previous apfsloader to avoid the spam. is it bad?  i dont think so imo.

Edited December 7, 2018 by ellaosx

[Riley Freeman]

I was doing that too but if this is designed to fix it I'd rather go that route.

 

I just tried my Z68 and Clover isn't writing the bootercfg value to nvram there either. Only the laptop managed to do it.

[Slice]

37 minutes ago, Riley Freeman said:

Is this new bootercfg log parameter supposed to stop the boot text spam I've been getting with the latest ApfsLoader? Because on my X79 with r4796 it's still showing up.

 

I had to write the bootercfg log=0 value into nvram manually from the recovery partition because it isn't being written by Clover. Clover is reading it ok though (I checked in the options menu).

 

It wrote the value ok on my laptop but that has a APFS partition so I don't get the boot text. I assume it only shows up when there's no APFS volume or container.

Spam from boot.efi and spam from apfs.efi are different.

[Riley Freeman]

4 minutes ago, Slice said:

Spam from boot.efi and spam from apfs.efi are different.

 

Thanks Slice. I don't remember ever getting any spam from boot.efi that needed to be hidden. Guess I'll roll back apfsloader for now.

 

Still, have you any idea why Clover can't write the value to nvram on my Asus boards? I have SIP fully enabled on all systems but it can write it on the laptop.

[Slice]

I know there are problems with Z370 but usually AptioMemoryFix.efi is enough for most motherboards.

PS. You have to say "SIP fully disabled"?

[Riley Freeman]

5 minutes ago, Slice said:

I know there are problems with Z370 but usually AptioMemoryFix.efi is enough for most motherboards.

PS. You have to say "SIP fully disabled"?

 

No, fully enabled. Everything is on including the nvram protection. I have AptioMemoryFix on the two Asus boards and nvram seems to be working properly (I can write, store and delete values ok).