Clover General discussion

[Matgen84]

36 minutes ago, Jief_Machak said:

I've refactored boot options as a XStringArray, which allow to get rid of AddLoadOption and RemoveLoadOption

If someone have the impression that something changed with the boot options, come to me 

 

@Badruzeus and linux compiler : compilation under linux doesn't work yet See @Pene posts (https://www.insanelymac.com/forum/topic/341986-c-proposition/?page=29&tab=comments#comment-2719199). There is a unresolved problem with C++ static initialiser section. Will come soon (when I find the time !).

 

Thanks for your effort @Jief_Machak 

 

I don't see "Clover Options" in GUI, only "Options". 

 

 

 

 

[Pene]

4 hours ago, Badruzeus said:

(Really sorry for late reply)

I just built r5114 from latest commit under Debian 10 with GCC53, but on my test it still produces similar issue; once I see boot6 / boot7 I get Red Screen of Death immediately. It' s bit weird that last debug.log line printed about "Anime" while I use "Catalina" theme which contains no anime resource. Thanks.

CloverBuild-AOH-Deb10-042620.txt.zip

debug.log.crash-AOH.zip

Yes, as Jief said, you can try to make manually the change I describe in that post. It's not in master yet, as it is still in the works - but I wondered if the change I describe there solves the problem for you as well, it's really not complicated to do it manually and try.

Edited April 26, 2020 by Pene

[Pene]

2 hours ago, LAbyOne said:

@Slice

Hi sorry to bother you again,

i've found a little glitch in graphics 

 

when selecting options menu, the first menu item, as per screenshots, shows indeed the icon

 

but when selecting a submenu the icons is gone...

 

and on others its half rendered

 

Sorry can't give you more infos about it as i never noticed before

appeared on this release r5114 (2ccc95513))

 

I tried and I cannot reproduce that.

Did you try with a recent build? There are many issues fixed since r5114.

In any case, attach your theme and I'll take a look. And also specify which resolution you are using.

[Jief_Machak]

@Pene You commit your work if you properly #ifdef'd in globals_ctor, it won't impact the others.

[Pene]

1 minute ago, Jief_Machak said:

@Pene You commit your work if you properly #ifdef'd in globals_ctor, it won't impact the others.

The change that makes it work is only in GccBase.lds, nothing changed in global_ctor. I think you might have missed my last post in the C++ thread

[Jief_Machak]

@Pene Yes. Good, it means I don't have to setup a whole linux virtual platform ! Time saved 

 

You committed that ?

[Pene]

Just now, Jief_Machak said:

@Pene Yes. Good, it means I don't have to setup a whole linux virtual platform ! Time saved 

 

You committed that ?

No, I wanted to hear your opinion about moving to to .data first 

[Jief_Machak]

I can't see any problem with that.

And I think that it's common to create a section with one pointer for every static initialisation needed. Even Microsoft does it this way and god knows they like to be different.

I surprise that the init section is called init_array on linux and .ctors on mac with the same compiler, though.

Good job, thanks.

[Pene]

11 minutes ago, Jief_Machak said:

I can't see any problem with that.

And I think that it's common to create a section with one pointer for every static initialisation needed. Even Microsoft does it this way and god knows they like to be different.

I surprise that the init section is called init_array on linux and .ctors on mac with the same compiler, though.

Good job, thanks.

OK, committed.

@Badruzeus you can try now from master, no need to change anything.

[LAbyOne]

2 hours ago, Pene said:

 

I tried and I cannot reproduce that.

Did you try with a recent build? There are many issues fixed since r5114.

In any case, attach your theme and I'll take a look. And also specify which resolution you are using.

 

I'm running the latest commit from master.

Resolution is 1920x1080

It happen with all themes i tested. (on my repo) the one you see is Zhuyin, you can get it from within Clover.app

 

Thanks

[Badruzeus]

2 hours ago, Pene said:

OK, committed.

@Badruzeus you can try now from master, no need to change anything.

Fantastic! It works as promised; built under Debian 10 with GCC53 then just booting fine w/o problem at all (and gotta bit smaller *.efi binaries from Linux as usual, actually about 24-32KB only LOL). Many thanks @Pene and 
@Jief_Machak , @Slice and Others.

 

Hi @vector sigma, there's no Option on 'buildme' script for removing previous (bad / obsolete) "Build" dir? I usually, beside clean BaseTools also do 'rm -rf Build' or... Plz correct me; is this step required? Thanks.

 

debug.log_r5114_cppfixed_deb10_AOH.zip

Edited April 26, 2020 by Badruzeus

[vector sigma]

32 minutes ago, Badruzeus said:

Hi @vector sigma, there's no Option on 'buildme' script for removing previous (bad / obsolete) "Build" dir? I usually, beside clean BaseTools also do 'rm -rf Build' or... Plz correct me; is this step required? Thanks.

I guess once the Build dir is deleted your good to go. BaseTools shouldn't be a problem I guess, but a clean will not be bad anyway.

[Pene]

1 hour ago, LAbyOne said:

I'm running the latest commit from master.

Resolution is 1920x1080

It happen with all themes i tested. (on my repo) the one you see is Zhuyin, you can get it from within Clover.app

 

Thanks

OK, I think I see the cause. Committed a fix to master, try again.

[LAbyOne]

1 hour ago, Pene said:

OK, I think I see the cause. Committed a fix to master, try again.

 

thanks, just booted from it : r5114 (SHA: e210ec2d0)

Submenus are good now, but, don't know why, curiously going further ( on 3rd menu choice) the glitch still there

i.e.

Options/System Prameters/Boot_args-Flags

[Pene]

41 minutes ago, LAbyOne said:

 

thanks, just booted from it : r5114 (SHA: e210ec2d0)

Submenus are good now, but, don't know why, curiously going further ( on 3rd menu choice) the glitch still there

i.e.

Options/System Prameters/Boot_args-Flags

Hmm, I can't reproduce that now even with your theme. Do you take some special steps to make that bug appear? Or just entering the menu location you specified directly?

[LAbyOne]

1 hour ago, Pene said:

Hmm, I can't reproduce that now even with your theme. Do you take some special steps to make that bug appear? Or just entering the menu location you specified directly?

no special steps, just entered the menu directly

 

PS

not that would make any difference, but i'm building Clover with XCODE8 not GCC53.

specifying because its weird, if it works byyour side why shouldn't here too?

 

My bad, just after replying to your post, just for the sake of it, I rebuilt Clover from master

and YOU WERE RIGHT!

Glitch not there anymore, thanks!

Edited April 27, 2020 by LAbyOne
i'm a coconut

[Pene]

47 minutes ago, LAbyOne said:

no special steps, just entered the menu directly

 

PS

not that would make any difference, but i'm building Clover with XCODE8 not GCC53.

specifying because its weird, if it works byyour side why shouldn't here too?

 

My bad, just after replying to your post, just for the sake of it, I rebuilt Clover from master

and YOU WERE RIGHT!

Glitch not there anymore, thanks!

Good

But it's for the best, as while navigating the menus I did notice some other glitches, so I just pushed another commit.

[Matgen84]

Hi @Slice @Pene @Jief_Machak

 

Thanks for all commits  I just boot from r5114 commit 84f7f87

 

I can't find "Clover Boot Options" in GUI: Is it normal because of changes or I make a mistake somewhere.
 

Sorry for my stupid question: about Options/System Parameters/Boot_args-Flags, where can i find explanations on boot-flags, some are unknown to me. Please.

 

EDIT: it's longer to display Clover GUI and boot on a real hard disk (Black screen, users settings, test hardware, GUI and after select OS, black screen before booting). 
 

 

 

 

 

 

 

Edited April 27, 2020 by Matgen84

[Pene]

3 hours ago, Matgen84 said:

I can't find "Clover Boot Options" in GUI: Is it normal because of changes or I make a mistake somewhere.

I can't reproduce that so please post your config.plist and a boot.log

[Matgen84]

23 minutes ago, Pene said:

I can't reproduce that so please post your config.plist and a boot.log

 

The "Clover Boot Options" button come back: I use r5114 (84f7f87a5). Despite everything, I attach my config.plist and preboot.log. If you can take a look.

 

Now, Clover is more longer: test hardware, user settings and of Corus scan entries before Clover GUI, after select OS there is a black screen before boot (verbose mode). Do you know why?

 

config.plist

preboot.log

 

 

 

Edited April 27, 2020 by Matgen84

[Pene]

49 minutes ago, Matgen84 said:

 

The "Clover Boot Options" button come back: I use r5114 (84f7f87a5). Despite everything, I attach my config.plist and preboot.log. If you can take a look.

 

Now, Clover is more longer: test hardware, user settings and of Corus scan entries before Clover GUI, after select OS there is a black screen before boot (verbose mode). Do you know why?

 

config.plist

preboot.log

Not sure how it came back, there were no changes today to that.

Anyway, attach a preboot.log with a previous version that you say loads faster, and I'll try to look where's the difference.

 

EDIT: Or if you mean that it is displaying "test hardware" etc. that it did not display before, that is probably because you set a timeout option in config.plist, that is when it is displayed.

Edited April 27, 2020 by Pene

[Matgen84]

52 minutes ago, Pene said:

Not sure how it came back, there were no changes today to that.

Anyway, attach a preboot.log with a previous version that you say loads faster, and I'll try to look where's the difference.

 

EDIT: Or if you mean that it is displaying "test hardware" etc. that it did not display before, that is probably because you set a timeout option in config.plist, that is when it is displayed.

 

Maybe, I make a mistake somewhere but I don't have previous version in EFI-Folder Backups for r5114.  I build the latest commit and report. Thanks.

 

EDIT: @Pene  there is maybe a build problem before. I compile and install latest r5114 commit (de2bb5f95): no "test hardware", no "users settings". Clover is fast as before.

 

It seems I have a issue with EFI-Backups folder: the new backup erase the previous one.

 

I need to do more test  

 

Request: Is it possible that clover install log show this below, for example, like preboot.log
Installer version: r5114 EFI bootloader master, commit (de2bb5f95)

 

Sorry for my poor English.

 

 

Edited April 27, 2020 by Matgen84

[vector sigma]

On 4/24/2020 at 9:46 PM, MacNB said:

Thanks for updating your system and testing.

You're probably in a better influencing position than me to report it  ?

I'm back to xcpm (with pleasure of the temperatures of my lappy ) as I find another location that needs to be patched in 10.15.5, can I have your kernel because I want to see if compatible with 10.15.4?

 

EDIT

I saw users now uses a cpu fake id to enable xcpm and .. maybe it's no longer needed.

 

EDIT

Any one has the 10.15.4 kernel is welcome as well

Edited April 27, 2020 by vector sigma

[Jief_Machak]

I went back to the PoolPrint total extermination I've started a while ago. Today, 66 PoolPrint removed, 175 left. Getting there...

This time, it was mainly path, a lot of them in the loader mechanism. The one that look for efi to boot, deducting partition list.

So if you have some icons gone and can't boot on a partition : blame me... and tell me.

 

Be sure to try before using this Clover (Well, you should always do that, but now isn't the time to be lazy).

 

 

3 hours ago, Matgen84 said:

Now, Clover is more longer:

Check if the delay is in "construct_globals_objects()". If it is, it could be because we now have quite a lot XString globals. Even more since my today's commits. Currently, these object are constructed at runtime (in construct_globals_objects()). I will soon improv litteral XString operator that will construct these object at compile time. Can save some time.

 

If Clover is longer than before, but not in "construct_globals_objects()", I'm sorry but I don't know.

Edited April 27, 2020 by Jief_Machak

[MacNB]

25 minutes ago, vector sigma said:

I'm back to xcpm (with pleasure of the temperatures of my lappy ) as I find another location that needs to be patched in 10.15.5, can I have your kernel because I want to see if compatible with 10.15.4?

 

EDIT

I saw users now uses a cpu fake id to enable xcpm and .. maybe it's no longer needed.

 

EDIT

Any one has the 10.15.4 kernel is welcome as well

 

Happy to help. See attached 10.15.4 (19E287) kernel.

kernel.zip