Clover General discussion

[Guest 5T33Z0]

I would just focus on the mandatory drivers required to boot a current UEFI system with a recent version of macOS.

Legacy users are stuck with r5123.1 anyway and most of the know what to do anyway since it's more challenging.

 

And delete the "Docs" Folder or updated it so it's actually useful. Nobody needs Install Instructions for OSX 10.7/8

 

These are the kind of small but important things that give Clover a bad reputation.

[LAbyOne]

INFO: @Slice

As per today new GCC 12.1.0 update is available

Heads up for future builds

this release fails building Clover 

halting the process on UsbBusDxe driver

 

/Users/labyone/src/CloverBootloader/Drivers/UsbBusDxe/UsbBus.c: In function 'UsbIoBulkTransfer':
/Users/labyone/src/CloverBootloader/Drivers/UsbBusDxe/UsbBus.c:278:13: error: 'UsbHcBulkTransfer' accessing 80 bytes in a region of size 8 [-Werror=stringop-overflow=]
  278 |   Status  = UsbHcBulkTransfer (
      |             ^~~~~~~~~~~~~~~~~~~
  279 |               Dev->Bus,
      |               ~~~~~~~~~
  280 |               Dev->Address,
      |               ~~~~~~~~~~~~~
  281 |               Endpoint,
      |               ~~~~~~~~~
  282 |               Dev->Speed,
      |               ~~~~~~~~~~~
  283 |               EpDesc->Desc.MaxPacketSize,
      |               ~~~~~~~~~~~~~~~~~~~~~~~~~~~
  284 |               BufNum,
      |               ~~~~~~~
  285 |               &Data,
      |               ~~~~~~
  286 |               DataLength,
      |               ~~~~~~~~~~~
  287 |               &Toggle,
      |               ~~~~~~~~
  288 |               Timeout,
      |               ~~~~~~~~
  289 |               &Dev->Translator,
      |               ~~~~~~~~~~~~~~~~~
  290 |               UsbStatus
      |               ~~~~~~~~~
  291 |               );
      |               ~
/Users/labyone/src/CloverBootloader/Drivers/UsbBusDxe/UsbBus.c:278:13: note: referencing argument 7 of type 'void *[10]'
In file included from /Users/labyone/src/CloverBootloader/Drivers/UsbBusDxe/UsbBus.h:46,
                 from /Users/labyone/src/CloverBootloader/Drivers/UsbBusDxe/UsbBus.c:16:
/Users/labyone/src/CloverBootloader/Drivers/UsbBusDxe/UsbUtility.h:200:1: note: in a call to function 'UsbHcBulkTransfer'
  200 | UsbHcBulkTransfer (
      | ^~~~~~~~~~~~~~~~~
At top level:
cc1: note: unrecognized command-line option '-Wno-incompatible-ms-struct' may have been intended to silence earlier diagnostics
cc1: all warnings being treated as errors
make: *** [/Users/labyone/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/Drivers/UsbBusDxe/UsbBusDxe/OUTPUT/UsbBus.obj] Error 1


build.py...
 : error 7000: Failed to execute command
	make tbuild [/Users/labyone/src/CloverBootloader/Build/Clover/RELEASE_GCC53/X64/Drivers/UsbBusDxe/UsbBusDxe]


build.py...
 : error F002: Failed to build module
	/Users/labyone/src/CloverBootloader/Drivers/UsbBusDxe/UsbBusDxe.inf [X64, GCC53, RELEASE]

- Failed -
Build end time: 20:03:08, May.06 2022
Build total time: 00:03:08

investigating further..

[Guest 5T33Z0]

BTW: that System ID / Hardware-ID endiannes flipping isssue is still not resolved. I've build Clover from source with Clover Suite Builder and it still flipped the pairs of digits as explained in my previous post.

[Slice]

@LAbyOne

I can test new gcc. Is there any changes in build_gcc script except gcc version?

 

10 hours ago, 5T33Z0 said:

BTW: that System ID / Hardware-ID endiannes flipping isssue is still not resolved. I've build Clover from source with Clover Suite Builder and it still flipped the pairs of digits as explained in my previous post.

Flipped compared to 5146? I see no this. The issue with smUUID I resolved. But CustonUUID entered in text form so it should be in right endianess.

[Guest 5T33Z0]

Please copy over Data from OpenCore Config into Clover, reboot and see what happens to System UUID and Hardware ID in hackintool when iusing the latest build. Maybe then you understand.

[Slice]

56 minutes ago, 5T33Z0 said:

Please copy over Data from OpenCore Config into Clover, reboot and see what happens to System UUID and Hardware ID in hackintool when iusing the latest build. Maybe then you understand.

I have no working copy of OpenCore.

So you say that this operation is successful with Clover 5146 and failed with latest commit?

[Guest 5T33Z0]

@Slice YES. That is what I have been trying to tell you since a couple of days now. As explained in detail in this post

 

 

Edited May 7, 2022 by 5T33Z0

[Slice]

5 minutes ago, 5T33Z0 said:

@Slice YES. That is what me have been trying to tell you since a couple of days now. As explained in detail in this post

 

 

Test, please

rEFIt_UEFI/Settings/ConfigPlist/Config_SystemParameters.h: line 78

    g.takeValueFrom(CustomUUID.value()); ->     g.takeValueFromBE(CustomUUID.value());

[Guest 5T33Z0]

1 minute ago, Slice said:

Test, please

rEFIt_UEFI/Settings/ConfigPlist/Config_SystemParameters.h: line 78

    g.takeValueFrom(CustomUUID.value()); ->     g.takeValueFromBE(CustomUUID.value());

 

I don't know enough about coding to do this. Maybe @LAbyOne can test this.

[Slice]

1 hour ago, 5T33Z0 said:

 

I don't know enough about coding to do this. Maybe @LAbyOne can test this.

Open the file by TextEdit and write two letters in line 78. Then compile as you usually do. 

[MifJpnAlphaPlus]

1 hour ago, Slice said:

Test, please

rEFIt_UEFI/Settings/ConfigPlist/Config_SystemParameters.h: line 78

    g.takeValueFrom(CustomUUID.value()); ->     g.takeValueFromBE(CustomUUID.value());

@Slice

Thank you for writing.
It just didn't work with smUUID.

I couldn't imagine where the meaningful smUUID data was going through the path. (Because it was not reflected in Hackintool even if it was swapped at the patching in the middle.)

Spoiler

But for CustomUUID, the takeValueFromBE method worked fine.

Spoiler

Thank you

 

[Guest 5T33Z0]

@Slice Great News! After changing that line in rEFIt_UEFI/Settings/ConfigPlist/Config_SystemParameters.h to "g.takeValueFromBE(CustomUUID.value());" and compiling Clover everything is back to normal again!

 

SMBIOS before updating Clover:

 

SMBIOS in Hackintool after update and reboot:

 

No more mismatch, no more Apple-ID authorization requests when switching Bootmanagers. I'm so happy that this is resolved now! Thank you very much! 😋

 

Crazy how two letters in a line of code can make a difference.

 

Edited May 7, 2022 by 5T33Z0

[Slice]

@LAbyOne

I tested gcc12 and came to conclusions:

1. Thanks for new warnings and errors. They must be eliminated. The error with UsbBus has origin from MdeModulePkg which is still wrong. Let OpenCore use it ahead.

2. Clover compiled by gcc12 is not working. Crash at "construct_globals_objects(gImageHandle);". Attention to @Jief_Machak. I am not able to correct this.

So I switch back to gcc11.

1 hour ago, MifJpnAlphaPlus said:

@Slice

Thank you for writing.
It just didn't work with smUUID.

I couldn't imagine where the meaningful smUUID data was going through the path. (Because it was not reflected in Hackintool even if it was swapped at the patching in the middle.)

  Reveal hidden contents

But for CustomUUID, the takeValueFromBE method worked fine.

  Reveal hidden contents

Thank you

 

Same here

SMBIOSPlist.h: line 674

      g.takeValueFromBE(SmUUID.value());

[Guest 5T33Z0]

@Slice I think I got excited too soon.

 

After testing I moved  the newly compiled files to the Internal EFI folder and then the the issue returned. When testing, I was booting the new version from the Flash Drive but via Bootloader chooser on the SSD. And the system's EFI folder still contained the current release version of Clover r5146. My bad. Clearly a mistake in testing on my end.

 

So now, after booting from Disk with the new compiled version, the error returns:

 

 

I guess for now, I just flip the pairs in Clover so that they come out correct.

 

 

Edited May 7, 2022 by 5T33Z0

[Slice]

45 minutes ago, 5T33Z0 said:

@Slice I think I got excited too soon.

 

After testing I moved  the newly compiled files to the Internal EFI folder and then the the issue returned. When testing, I was booting the new version from the Flash Drive but via Bootloader chooser on the SSD. And the system's EFI folder still contained the current release version of Clover r5146. My bad. Clearly a mistake in testing on my end.

 

So now, after booting from Disk with the new compiled version, the error returns:

 

 

I guess for now, I just flip the pairs in Clover so that they come out correct.

 

 

I committed my changes, be carefull to git pull to not have a conflict with the existing patches.

[Guest 5T33Z0]

I think that fixed it.

[MifJpnAlphaPlus]

2 hours ago, Slice said:

@LAbyOne

I tested gcc12 and came to conclusions:

1. Thanks for new warnings and errors. They must be eliminated. The error with UsbBus has origin from MdeModulePkg which is still wrong. Let OpenCore use it ahead.

2. Clover compiled by gcc12 is not working. Crash at "construct_globals_objects(gImageHandle);". Attention to @Jief_Machak. I am not able to correct this.

So I switch back to gcc11.

Same here

SMBIOSPlist.h: line 674

      g.takeValueFromBE(SmUUID.value());

Thank you so much @Slice

Everything has been fixed.

Spoiler

Spoiler

 

[Guest 5T33Z0]

This might be interesting: Since my "new" used GPU arrived, I did run some tests in Geekbench and compared Clover vs OpenCore since I got configured identical (as far as this is possible). And in direct comparison, Clover always came out on top. Results here

Edited May 8, 2022 by 5T33Z0

[Slice]

30 minutes ago, 5T33Z0 said:

This might be interesting: Since my "new" used GPU arrived, I did run some tests in Geekbench and compared Clover vs OpenCore since I got configured identical (as far as this is possible). And in direct comparison, Clover always came out on top. Results here

It may be different bus speed calculations.

[Guest 5T33Z0]

@Slice You are right. I entered sysctl hw.busfrequency in Terminal. For Clover it returned 100 mHz and in OpenCore it returned 400 mHz.

[Slice]

6 hours ago, 5T33Z0 said:

@Slice You are right. I entered sysctl hw.busfrequency in Terminal. For Clover it returned 100 mHz and in OpenCore it returned 400 mHz.

The correct value should not be round. It must be like 99.973MHz. It provides maximum CPU speed because small deviations causes hangs.

[rafale77]

On 5/7/2022 at 7:45 PM, 5T33Z0 said:

This might be interesting: Since my "new" used GPU arrived, I did run some tests in Geekbench and compared Clover vs OpenCore since I got configured identical (as far as this is possible). And in direct comparison, Clover always came out on top. Results here

 

Looking at your results and based on my experience with GB5, and maybe it is just my setup, the differences you are seeing are far within the standard deviation of the test and therefore not statistically significant. I wouldn’t conclude that there is any difference at all except maybe on some of the extremes of the Metal test. Have you run multiple consecutive tests? For example GB5 single core on my 10900 ranges from 1280-1400. Multi-Core typically from 9450-9600. I have less experience on the compute test and it seems a little more stable but around the 21K mark I see variations of up to +/-~500 points under the exact same conditions so I would think in the 50K that +/-1200pts would be within noise.

[Guest 5T33Z0]

 @rafale77 In my short experice with the compute tests, I noticed that results deviate more than results from the CPU tests. But everytime I switched to Clover and repeated a test, the result was better. I also did some rendering in Davinci Resolve and when using Clover it was a bit faster – 33 seconds instead of 35. I have to see if I can find some type of bigger test project to render a larger file.

 

@Hervé Are you talking about me or who is the OP you are referring to?

Edited May 9, 2022 by 5T33Z0

[Common_Sense]

I might have seen a bug in Clover, it behaves a bit strange with the partitions that I can select in the boot menu.

 

I cloned an SSD using Super Duper. As expected I got 2 macOS and 2 Recovery partitions in the boot menu.

 

I now go into Windows and re-partition the old SSD to an NTFS drive.

 

In the boot menu, I still have 2 macOS and 2 Recovery partitions.

 

How can it be?

Edited May 10, 2022 by Common_Sense

[Guest 5T33Z0]

@Common_Sense I doubt that this is a Clover issue. Can I see a screenshot of windows diskmanagent for the drive, please? Also check your config.plist and see if it has custom entries in the GUI section and delete those.