OpenCore General Discussion

[Stefanalmare]

6 hours ago, eSaF said:

I concur - We'll see what the next 'Master' release brings.

Fix it in 292b7c9.

[Stefanalmare]

I can't compile anymore Lilu and all the Lilu's suite with Xcode 13.3. What I do wrong?

[Guest 5T33Z0]

@eSaF OCAT uses whatever is in the opemncore package when updating. And since the issue occurs in the ocvalidate of the latest built, the error occurs.

 

To fix iit, you can replace the OCValidate file with the working one or wait til the next OC package with the fixed OC Vaidate.

• in OCAT, go to EDIT >  Open Database Directory,
• A Finder window opens containing the OpenCore Files
• DevDatabase cntains the files for the nightly builds
• In DevDatabase > mac is OCValidate

 

BTW: I get no errors when using the lastest build: 292b7c9

 

Edited April 9, 2022 by 5T33Z0

[ghost8282]

14 hours ago, Stefanalmare said:

I can't compile anymore Lilu and all the Lilu's suite with Xcode 13.3. What I do wrong?

I confirm this, something changed in xcode 13.3

The issue seems to be in kern_qsort file:

loop:	SWAPINIT(a, es);

-> Performing pointer subtraction with a null pointer may have undefined behavior

 

a is in fact defined as void pointer:

qsort(void *a, size_t n, size_t es, int (*cmp)(const void *, const void *))

 

[Stefanalmare]

14 minutes ago, ghost8282 said:

I confirm this, something changed in xcode 13.3

The issue seems to be in kern_qsort file:

loop:	SWAPINIT(a, es);

-> Performing pointer subtraction with a null pointer may have undefined behavior

 

a is in fact defined as void pointer:

qsort(void *a, size_t n, size_t es, int (*cmp)(const void *, const void *))

 

Same.

[Anto65]

Released OpenCore   ver.  0.8.0 💥

 

https://github.com/acidanthera/OpenCorePkg/releases/tag/0.8.0  

[deeveedee]

Developers - thank you to those who stepped up and delivered OC 0.8.0.  The upgrade from OC 0.7.9 was effortless (I'm still performing manual updates without a configurator - just my preference).  To perform the upgrade, I applied the changes here.

[MifJpnAlphaPlus]

Hello everyone.
Perhaps it is a problem because Japanese users do not know about it.
However, the administrator of a well-known Japanese forum has pointed out the following problem

In Sample.plist NVRAM / Add / 7C436110-AB2A-4BBB-A880-FE41995C9F82, the entry #INFO (prev-lang: kbd) is not taken to mean commented out, causing it to not start.

　Normally, I would fix a new OpenCore with the results of Ocvalidate, not from Sample, but the administrator there does not seem to have done so.

　What do you think? I feel like it would be nice if I could add a few comments, but I also think it's a bit nosy.

Thank you

 

[deeveedee]

Does anyone know what this 0C 0.8.1 commit actually fixes for the ExtendBTFeatureFlags quirk?

[miliuco]

@deeveedee

It supposedly improves this quirk on recent macOS but I don't know exactly on what.

[Guest 5T33Z0]

@deeveedee According to the documentation, this quirk enables full functionaliy of Bluetooth including Continuity (I guess on modules which don't work out of the box) so that Airdrop and Handoff work (again).

[deeveedee]

@miliuco and @5T33Z0 - thank you for the responses - I appreciate that you made the effort.  I didn't word my question well.  I'd like to know what was broken in ExtendBTFeatures quirk that required the fix in this commit.

[aben]

22 hours ago, deeveedee said:

Does anyone know what this 0C 0.8.1 commit actually fixes for the ExtendBTFeatureFlags quirk?

I'm no developer but my understanding, from looking at the code, is that another (Version 3) BTFeatureFlag patch (OC quirk introduced with v0.6.2 which had replaced BT4LEContinuityFixup) was added to now help patch identifier: com.apple.iokit.IOBluetoothFamily's kernel base/function: __ZN17IOBluetoothDevice25setDeviceSupportedFeatureEj (most likely a new (or existing) class of function found in newer macOS releases that is responsible for proper BT functionality) which, I believe, would be considered an improvement in the case certain BT features (like Watch Unlock) were broken for users on newer macOS release due to this function not being patched yet.

Also, the changelog's wording for the commit was later corrected to improved instead of fixed.

Edited April 26, 2022 by aben

[Guest 5T33Z0]

Today the replacement GPU for my dead GTX760 arrived: Saphire RX580 Nitro+

The Adrenaline App says this card uses 256 MB of rezisable BAR.

Since my previous GPU didn't support this feature, do I have to change anything in the BIOS or Config - or both? I always found these settings confusing since there are 2 options in OC: ResizeGPUBars and ResizeAppleGPUBars.

Thanks.

[deeveedee]

@eSaF and @aben - all good guesses.  Thanks.

 

Had I known that the commit only "improved" the kext instead of "fixing" it, I'd never have asked. LOL.  Much clearer.  Now I just need to figure out which macOS versions are "newer."

Edited April 26, 2022 by deeveedee

[miliuco]

@5T33Z0

You must enable ReBAR in Bios, usually it is near Above 4G Decoding, if Above 4G is disabled, Resizable Bar is hidden.

OpenCore config.plist: ResizeAppleGpuBars = 0 (to be able to boot macOS with ReBAR enabled in BIOS) and ResizeGpuBars = -1 (to prevent macOS from passing its ReBAR settings to Windows).

Nothing more.

My old XFX RX 580 had the same scores in Windows and macOS in benchmarks with or without ReBAR.

 

Edited April 26, 2022 by miliuco

[aben]

28 minutes ago, deeveedee said:

@eSaF and @aben - all good guesses.  Thanks.

 

Had I known that the commit only "improved" the kext instead of "fixing" it, I'd never have asked. LOL.  Much clearer.  Now I just need to figure out which macOS versions are "newer."

Of course, unless stated or reported, one can only make a calculated guess at best here since there is no pre-defined/easy way of finding out what exactly are the contents or feature-set of this particular macOS BT function that was patched unless one has reported or experienced running into an issue for eg: a BT feature like Watch Unlock, when using a particular Apple Watch model, was all of a sudden acting broken on a newer macOS build that was updated on a hackintosh system fitted with a particular BT module that is reliant on OC’s current ExtendedBTFeatureFlag patches for the Watch Unlock functionality. Only in such situations, can the affected user actually find out if any particular unknown BT function, that requires patches, is being called by the OS via debugging ofcourse. Therefore, directly consulting with the user/developer who helped create the patch is the only way to actually know what exactly was the particular patch for and the macOS version this function was found as well.
Again, my best guess here with “newer” would be macOS 12 😂

[Guest 5T33Z0]

48 minutes ago, miliuco said:

@5T33Z0

You must enable ReBAR in Bios, usually it is near Above 4G Decoding, if Above 4G is disabled, Resizable Bar is hidden.

OpenCore config.plist: ResizeAppleGpuBars = 0 (to be able to boot macOS with ReBAR enabled in BIOS) and ResizeGpuBars = -1 (to prevent macOS from passing its ReBAR settings to Windows).

Nothing more.

My old XFX RX 580 had the same scores in Windows and macOS in benchmarks with or without ReBAR.

 

Thanks.I did follow your instructions. Unfortunately, the GPU kept hanging at the Apple Logo while the system still booted to the desktop in the background. Since the card has 256 MB I also tried 8, but this didn't work either. I've disabled ReBAR again for now. I have no No patience to fiddle around with that stuff at the moment.

[miliuco]

@5T33Z0

That bug suggests misconfiguration in config.plist.

 

According to the OpenCore developers, ResizeAppleGpuBars has to be 0 (ReBAR enabled in BIOS) or -1 (ReBAR disabled in BIOS). No other value is recommended because macOS does not boot well.

 

ResizeGpuBars is always -1 which means that OpenCore does not pass to Windows ReBAR settings when Windows boots from OpenCore.

 

I did the test of booting macOS with the RX 580 GPU and ReBAR enabled and it booted fine on condition of having ResizeAppleGpuBars = 0.

 

But note 2 things:
- macOS does not benefit from this option, the quirks were added so that Windows users could have ReBAR enabled without having to disable it every time they run macOS
- Windows benefits from this option but in my opinion only with powerful GPUs, specifically 6800 and 6900.

[Guest 5T33Z0]

@miliuco

I can only set ReBAR to "Auto" in BIOS. The rest of the values for the config were entered as suggested. But since it doesn't seem beneficial to use ReBAR on the RX580 I leave it disabled.

[Ellybz]

@5T33Z0

Here are my settings running a quad Monitor setup, booting OSX (Monterey & Mojave ), Linux & Windows. Not a single glitch. 

 

Above 4G Decoding : Disabled ( No difference of performance noted when engaged. Some rx580 users reported issues when enabled )

ResizeAppleGpuBars : -1

ResizeGpuBars = -1

agdpmod=pikera 

Edited April 26, 2022 by Ellybz

[Guest 5T33Z0]

agdpmod=pikera is for navi gpu's but the rx580 is a polaris card.

[miliuco]

@Ellybz

I haven't noted difference when Above 4G enabled, as you. If Above 4G is disabled, ReBAR can't be enabled. When ReBAR disabled, the 2 GpuBars quirks must be -1.

 

@5T33Z0

My old RX 580 also worked better (or so it seemed to me) with agdpmod=pikera although this boot arg is intended for 5000 and 6000 series.

Edited April 26, 2022 by miliuco

[Ellybz]

25 minutes ago, 5T33Z0 said:

agdpmod=pikera is for navi gpu's but the rx580 is a polaris card.

With multiple monitors, my 38" Monitor stays black at the OSX login without it. 

Edited April 26, 2022 by Ellybz

[Guest 5T33Z0]

Interesting. I'll see if there's any difference with agdpmod=pikera