Clover General discussion

[Cyberdevs]

2 minutes ago, apianti said:

Pre-Sierra versions should say OS X, Sierra and after should say macOS. That is what is happening? And no duplicates?

Yup I know that Apple changed OS X to macOS since macOS Sierra and well in the GUI for Yosemite and El Capitan it says OS X and for Sierra it's also showing OS X just for High Sierra it says macOS.

[Sherlocks]

18 minutes ago, apianti said:

Pre-Sierra versions should say OS X, Sierra and after should say macOS. That is what is happening? And no duplicates?

old and last clover did not divide this name part of OS X and macOS. now, use OS X install

consider? i will test if need

Edited April 5, 2018 by Sherlocks

[Cyberdevs]

Just now, Sherlocks said:

old and last clover did not divide this part OS X and macOS. now, use OS X install

Yes in one of the screenshots I sent you there was macOS install for Sierra as well but now it only showing "Boot OS X Install..." that's weird.

I tested all versions you sent me but I can't reproduce the same results.

[apianti]

27 minutes ago, Sherlocks said:

old and last clover did not divide this part OS X and macOS. now, use OS X install

Yeah, I don't know what you mean by this, it most certainly has the distinction. I said that as well when you committed the code. It should say macOS for Sierra not OS X.

[Sherlocks]

2 minutes ago, apianti said:

Yeah, I don't know what you mean by this, it most certainly has the distinction. I said that as well when you committed the code. It should say macOS for Sierra not OS X.

yes. but i didn't touch name

https://sourceforge.net/p/cloverefiboot/code/4022/tree/rEFIt_UEFI/entry_scan/loader.c#l1158

it's same in old source. now i'm considering name.

[apianti]

1 hour ago, Sherlocks said:

yes. but i didn't touch name

https://sourceforge.net/p/cloverefiboot/code/4022/tree/rEFIt_UEFI/entry_scan/loader.c#l1158

it's same in old source. now i'm considering name.

The problem with that assumption is that previous versions used that location, one of the updates to Sierra moved it from there to the location used in the line above that. But all versions of sierra are branded macOS not OS X. So you need to make a determination what version it is. However that's way too much work and those lines are dumb anyway, they should be replaced with this code:

for (Index = 0; Index < OSXInstallerPathsCount; ++Index) {
  if (AddLoaderEntry(OSXInstallerPaths[Index], NULL, L"macOS Install", Volume, NULL, OSTYPE_OSX_INSTALLER, 0)) {
    break;
  }
}

 I still think that you have some situations where the installer won't appear if it's at the default location only with .IAPhysicalMedia present. I see no place where that is resolved, is that not the original issue? Also in custom entry, the path shouldn't be skipped if it's the second stage, it should be added as the installer type if appropriate.

EDIT: I'm not sure maybe it does work but wow it could be a whole lot better by just checking the file once later. Removing all that stuff, use the code above to detect installers. Then use this code to detect the default entry:

if (!FileExists(Volume->RootDir, L"\\.IAPhysicalMedia")) {
  if (EFI_ERROR(GetRootUUID(Volume)) || isFirstRootUUID(Volume)) {
    AddLoaderEntry(MACOSX_LOADER_PATH, NULL, L"macOS", Volume, NULL, OSTYPE_OSX, 0);
  }
} else if (Index >= OSXInstallerPathsCount) {
  AddLoaderEntry(MACOSX_LOADER_PATH, NULL, L"macOS Install", Volume, NULL, OSTYPE_OSX_INSTALLER, 0);
}

EDIT2: Just some typos.

EDIT3: Actually my logic was wrong so changed the if.

Edited April 5, 2018 by apianti

[Shaneee]

Any of the devs got a thought on this? https://www.insanelymac.com/forum/forums/topic/284656-clover-general-discussion/?do=findComment&comment=2607766

[apianti]

23 hours ago, Shaneee said:

Issues with AMD CPUs again. I'm still using r4392 as updating to the release on SF or above ends with the X64 Exception Type error we had before. Has the code for AMD changed at all or is it something else. All tests of Clover were booting the same system with only r4392 booting successfully. Slice sorted it before in r4114.

Here's the debug log.

 

3 hours ago, Shaneee said:

Any of the devs got a thought on this?

The log is closed so it's something that is happening after boot. It's really hard to pin point without a better idea of what works and which revision is the one that actually stops working. Are you able to build yourself? Because if you could get at least within a few revisions, that would be most useful. Also did you not get an exception screen? Because sometimes that can be helpful, maybe.

[Shaneee]

 

47 minutes ago, apianti said:

 

The log is closed so it's something that is happening after boot. It's really hard to pin point without a better idea of what works and which revision is the one that actually stops working. Are you able to build yourself? Because if you could get at least within a few revisions, that would be most useful. Also did you not get an exception screen? Because sometimes that can be helpful, maybe.

Somewhere between r4392 and r4411 is where it stopped working. Yeah there's the x64 Exception Type error. I can take a pic of this if needed. I can't build the source as I get a Segmentation Fault 11 error near the end of building.

[Sherlocks]

7 hours ago, Shaneee said:

 

Somewhere between r4392 and r4411 is where it stopped working. Yeah there's the x64 Exception Type error. I can take a pic of this if needed. I can't build the source as I get a Segmentation Fault 11 error near the end of building.

https://github.com/Dids/clover-builder/releases?after=v2.4k_r4406

Here is each builds. Install each builds step by step

You have to find wrong point from revision between 4392 and 4411.

[Slice]

The problem is rev 4395

VOID
 EFIAPI
 OnExitBootServices(IN EFI_EVENT Event, IN VOID *Context)
 {
+  UINT64 msr = 0;
+
+  msr = AsmReadMsr64(MSR_PKG_CST_CONFIG_CONTROL); //0xE2
+//  AsciiPrint("MSR 0xE2 on Exit BS %08x\n", msr);
+

AMD CPU will crash at the MSR 0xE2

I will correct this in 4428

[Shaneee]

Thanks @Slice r4428 is working on AMD once again 

[Pavo]

Anyone else having issues with using KernelXCPM on 2x Xeon E5 2670 v2 Ivy Bridge CPUs? Mine will not boot at all with using the KernelXCPM flag.

[apianti]

4 hours ago, Pavo said:

Anyone else having issues with using KernelXCPM on 2x Xeon E5 2670 v2 Ivy Bridge CPUs? Mine will not boot at all with using the KernelXCPM flag.

I have no idea why it wouldn't work there are MacPros with multiple CPU packages, however I doubt that they are any that use XCPM since they all predate Ivy Bridge. So I'm betting that XCPM probably only works with one CPU package, maybe? Does it work without XCPM? Because I see no reason why you need XCPM.

[Pavo]

Just now, apianti said:

I have no idea why it wouldn't work there are MacPros with multiple CPU packages, however I doubt that they are any that use XCPM since they all predate Ivy Bridge. So I'm betting that XCPM probably only works with one CPU package, maybe? Does it work without XCPM? Because I see no reason why you need XCPM.

Yeah it works without XCPM, its possible that its because I have 2x CPUs like you said. Oh well getting around 42k GB score right now anyways.

[apianti]

21 hours ago, Pavo said:

Yeah it works without XCPM, its possible that its because I have 2x CPUs like you said. Oh well getting around 42k GB score right now anyways.

You know you can patch the state data in the PM kexts to give you better performance if that's what you want. Just be aware that most likely since there is a MacPro with a very similar CPU that it probably has pretty good information already to counter balance work performance and power management. There's usually a point where you use much more energy to get much less performance, it would be hard to determine this on your own without the technical specification documents of the exact CPU (to which I doubt you have access), although you can compare your ACPI CPU PM tables to get somewhat valid values.

[zxv]

Any chance we could have the official Clover repo relocated away from SourceForce, given that SF is now down or otherwise dysfunctional 75% of the time?

[Zenith432]

1 hour ago, zxv said:

Any chance we could have the official Clover repo relocated away from SourceForce, given that SF is now down or otherwise dysfunctional 75% of the time?

It is probably a local problem.  I do not experience such misery, but SF svn repository is usually sluggish compared to github git.

[apianti]

10 hours ago, zxv said:

Any chance we could have the official Clover repo relocated away from SourceForce, given that SF is now down or otherwise dysfunctional 75% of the time?

I have literally never had one problem and I use it mostly daily. I beginning to wonder if it's some sort of geolocation issue, that you can't get a reliable packet path to/from the servers....

8 hours ago, Zenith432 said:

It is probably a local problem.  I do not experience such misery, but SF svn repository is usually sluggish compared to github git.

Yeah, I agree I think it is also a location issue. But svn is slower than git in general, also clover commits are sometimes really bad, like there are some where like fifty files were changed then reverted later in a different commit, so those types of commits really eat into the repo since it doesn't have as robust of a storage/transfer mechanism as git. And for christ sakes guys, I'm going as fast as I can trying to get v3 going, it will be a git repo, then I'll mirror it on github as well I guess. Since you seem to not be giving me much of a choice.... Nah it's probably best to have it mirrored just in case, lol.

[Pavo]

12 minutes ago, apianti said:

I have literally never had one problem and I use it mostly daily. I beginning to wonder if it's some sort of geolocation issue, that you can't get a reliable packet path to/from the servers....

Yeah, I agree I think it is also a location issue. But svn is slower than git in general, also clover commits are sometimes really bad, like there are some where like fifty files were changed then reverted later in a different commit, so those types of commits really eat into the repo since it doesn't have as robust of a storage/transfer mechanism as git. And for christ sakes guys, I'm going as fast as I can trying to get v3 going, it will be a git repo, then I'll mirror it on github as well I guess. Since you seem to not be giving me much of a choice.... Nah it's probably best to have it mirrored just in case, lol.

I don't mind svn but I agree SourceForge is starting to become very unreliable. Github would be a good choice to move to and to start out with v3 in my opinion.

[apianti]

2 minutes ago, Pavo said:

I don't mind svn but I agree SourceForge is starting to become very unreliable. Github would be a good choice to move to and to start out with v3 in my opinion.

I have not witnessed any reliability issues from SF besides like one or two days during their server migration, only others saying that are having issues. Yet I don't really see any of these issues ever being asked about.... It will probably stay on SF and maybe get a github mirror. IDK, I don't really want to split the project as the eventuality will come that v2 will need to be converted to legacy firmware emulation only. There are some problems with that though, not changing the SMBIOS to be like Apple's and the default boot paths. Is it worth not just using DUET as the firmware emulator at that point? Unsure but not really that important right now...

[zxv]

13 hours ago, Zenith432 said:

It is probably a local problem.  I do not experience such misery, but SF svn repository is usually sluggish compared to github git.

Probably right. Getting errors like this when attempting to update edk2:

svn: E170013: Unable to connect to a repository at URL 'svn://svn.code.sf.net/p/edk2/code/trunk/edk2/IntelFrameworkModulePkg'

svn: E210002: Network connection closed unexpectedly

Didn't happen prior to their recent server move.

Edited April 8, 2018 by zxv

[chris1111]

I built Clover since 2015 with the Developper command on the Clover Wiki, as far as I know I have always used svn without any problem except when Source Forge became Off with the latest changes. I also build my nasm with the command line. The important thing is not to go quickly but correctly

Edited April 8, 2018 by chris1111

[Maniac10]

OptionRomPkg:
checking..available, continuing..
Updating '.':
svn: E170013: Unable to connect to a repository at URL 'svn://svn.code.sf.net/p/edk2/code/trunk/edk2/OptionRomPkg'
svn: E210002: Network connection closed unexpectedly

An error was encountered syncing the repository:
------------------------------

Updating '.':
svn: E170013: Unable to connect to a repository at URL 'svn://svn.code.sf.net/p/edk2/code/trunk/edk2/OptionRomPkg'
svn: E210002: Network connection closed unexpectedly

------------------------------
Can be temporarily, retry later or seek for help.

Lately I see these errors or massive slowdowns almost every time I want to update something from SF. The location theory might be right.

[apianti]

3 hours ago, Maniac10 said:

OptionRomPkg:
checking..available, continuing..
Updating '.':
svn: E170013: Unable to connect to a repository at URL 'svn://svn.code.sf.net/p/edk2/code/trunk/edk2/OptionRomPkg'
svn: E210002: Network connection closed unexpectedly

An error was encountered syncing the repository:
------------------------------

Updating '.':
svn: E170013: Unable to connect to a repository at URL 'svn://svn.code.sf.net/p/edk2/code/trunk/edk2/OptionRomPkg'
svn: E210002: Network connection closed unexpectedly

------------------------------
Can be temporarily, retry later or seek for help.

Lately I see these errors or massive slowdowns almost every time I want to update something from SF. The location theory might be right.

Ok, I think it's two parts. One, location to the nearest SF server, and two, the EDK2 svn repo is probably constantly checking to see if it needs synced with the GH git repo and causes any current activity to be cut off. All of these errors are to do with the EDK2 svn repo, which as I said I don't use - I use the GH git repo, which is why I'm not experiencing the slow downs.