Exception Type: EXC_CRASH (Code Signature Invalid)
AMDMTLBronzeDriver isn't even loaded.
have you disabled csr ?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
macOS 10.14 Mojave on Unsupported Macs Thread
- Thread starter dosdude1
- WikiPost WikiPost
- Start date
- Sort by reaction score
- Status
AMDMTLBronzeDriver isn't even loaded.
have you disabled csr ?
Has anyone with a MP3,1 gotten the @netkas AMD video patch to work?
I bought a new RX570 to test my SSE4.2 emulator with the AMD video drivers, and I wanted to use the existing patched driver as a baseline for testing. Using the @dosdude1 patcher v1.3.3, I installed Mojave 10.14.5 onto a new SSD, with all relevant patches (other than Legacy Video). That installation boots and operates correctly using the stock ATI 2600XT. Substituting the RX570 for the 2600XT yields a black screen; a cursor appears for a few seconds (during which it is responsive to the mouse), disappears, then reappears about 5-10 seconds later (loop). Power cycle required. With both cards in, and the monitor hooked to the 2600XT, it gets to the first progress bar (white background) and hangs at about 55-60%.
A glance at the logs showed a "Code Signature Invalid" error, so I set my boot-args to "-no_compat_check -v kext-dev-mode=1". Beyond the boot details whizzing by unreadably fast, no real insights there. System boots fine into Mojave 10.14.3 with both cards inserted (but, of course, it can't tell what the RX570 is).
I looked closer at the logs (see attached for one example of many nearly-identical logs): it appears that the SkyLight WindowServer is crashing repeatedly, about every 8-10 seconds, even with kext-dev-mode=1 set. The Termination Reason is CODESIGNING, but the crashed thread threw a trap 14 (page fault). Interestingly, the neither the AMDMTLBronzeDriver nor any other AMD-named file appears in the crash report. For clarity, the crash report is from a vanilla install using dosdude1's patcher, with nothing else installed or configured (including my SSE4.2 emulator).
Anybody have any insights?
EDIT: Yes, SIP is disabled. System boots fine with a 2600XT, this only happens with the RX570 present.
I bought a new RX570 to test my SSE4.2 emulator with the AMD video drivers, and I wanted to use the existing patched driver as a baseline for testing. Using the @dosdude1 patcher v1.3.3, I installed Mojave 10.14.5 onto a new SSD, with all relevant patches (other than Legacy Video). That installation boots and operates correctly using the stock ATI 2600XT. Substituting the RX570 for the 2600XT yields a black screen; a cursor appears for a few seconds (during which it is responsive to the mouse), disappears, then reappears about 5-10 seconds later (loop). Power cycle required. With both cards in, and the monitor hooked to the 2600XT, it gets to the first progress bar (white background) and hangs at about 55-60%.
A glance at the logs showed a "Code Signature Invalid" error, so I set my boot-args to "-no_compat_check -v kext-dev-mode=1". Beyond the boot details whizzing by unreadably fast, no real insights there. System boots fine into Mojave 10.14.3 with both cards inserted (but, of course, it can't tell what the RX570 is).
I looked closer at the logs (see attached for one example of many nearly-identical logs): it appears that the SkyLight WindowServer is crashing repeatedly, about every 8-10 seconds, even with kext-dev-mode=1 set. The Termination Reason is CODESIGNING, but the crashed thread threw a trap 14 (page fault). Interestingly, the neither the AMDMTLBronzeDriver nor any other AMD-named file appears in the crash report. For clarity, the crash report is from a vanilla install using dosdude1's patcher, with nothing else installed or configured (including my SSE4.2 emulator).
Anybody have any insights?
EDIT: Yes, SIP is disabled. System boots fine with a 2600XT, this only happens with the RX570 present.
kext-dev-mode=1 does absolutely nothing on OS X 10.11 and later. If you installed patches using the post-install tool, SIP will be disabled for you automatically. What I'd recommend doing, is for any binary you patched, run "sudo codesign -f -s - /path/to/binary", and see if that helps.
"for any binary you patched [...]" - you're suggesting I do this to the AMDMTLBronzeDriver file that came from your patcher? Or only for the stuff I'm developing (which isn't even in the picture here)? Remember, this is a more-or-less vanilla installation, fresh SSD, Mojave downloaded this morning...
OK, that made a difference. Now, with -v still in boot-args, it eventually progresses to a black text screen, where a two-line message flashes by about every 10 seconds, but I was never able to read it. And instead of a ReportCrash log, I get a WindowServer log (attached) every 10 seconds, saying it failed to create a MetalDevice. (I let it loop about 10 times, just in case it was actually making progress, which it wasn't; all the logs are essentially identical.)
Has anybody else gotten this to work, or am I the first to actually use it? It's conceivable that my RX570 is somehow incompatible, but from what I've seen online, that's unlikely. And with a vanilla OOB installation, it's hard to see what I'm doing wrong...
(EDIT (seven hours later): just had a chance to review recent activity in this thread, and I realize that I'm not the only one - @Ludacrisvp seems to be encountering more or less the same issues. Sorry for not keeping up with all the reading.)
Last edited:
SIP says disabled in the crash log. I’d have to double check with csrutil later but I’m 99% sure that it would say it’s all disabled.
I upgraded our original Apple Broadcom BCM94321MC Bluetooth WiFi card in my 17" MacBookPro4,1 to an Apple Broadcom BCM94322MC card ($9.50 from China at eBay) AFTER I'd installed Mojave 10.14.1 and discovered WiFi and Bluetooth wouldn't work with dosDude1's Mojave Patcher. The new BCM94322MC card has worked great with my Logitech Bluetooth mouse plus 2.4 and 5.0 GHz WiFi in both 10.14.3 and 10.14.5 (I skipped 10.14.4 since it had SO many issues with patched Mojave systems) .
FYI, after I installed the new Broadcom BCM94322MC card, I believe DosDude1 released a post install patch which supposedly makes my original BCM94321MC Bluetooth WiFi card work.
Which BCM943XXMC card do you have?
Hi,
I have a 2010 17" MacBook Pro (i.e. with Nvidia GeForce GT 330M graphics adapter) running a patched Mojave 10.14.5. Everything works fine except when waking from sleep. If I shut the MacBook Pro's lid and let the system sleep (front light on case flashes slowly), and then open the lid, the display comes on for about 3-4 seconds and then goes black. The only way to recover is to restart the system.
I'm running in dark mode and have "reduce transparency" enabled.
I'd appreciate any ideas as to how to rectify the problem?
Thanks for everyone's help. I've searched the thread but can't find same problem on same hardware...
I have a 2010 17" MacBook Pro (i.e. with Nvidia GeForce GT 330M graphics adapter) running a patched Mojave 10.14.5. Everything works fine except when waking from sleep. If I shut the MacBook Pro's lid and let the system sleep (front light on case flashes slowly), and then open the lid, the display comes on for about 3-4 seconds and then goes black. The only way to recover is to restart the system.
I'm running in dark mode and have "reduce transparency" enabled.
I'd appreciate any ideas as to how to rectify the problem?
Thanks for everyone's help. I've searched the thread but can't find same problem on same hardware...
[QUOTE="
Is it FCP 7 OR x? 7 is 32 bit, and unfortunately no longer supported. That's why I've kept a couple older Macs on El Capitan. The functionalities are no longer available today. I bought a 2017 5k iMac, and by the time I had the same functionality my 9,1 has, it cost nearly $3,000. Welcome to adapter and dongle hell.[/QUOTE]
FCP is 10.4.6.
Is it FCP 7 OR x? 7 is 32 bit, and unfortunately no longer supported. That's why I've kept a couple older Macs on El Capitan. The functionalities are no longer available today. I bought a 2017 5k iMac, and by the time I had the same functionality my 9,1 has, it cost nearly $3,000. Welcome to adapter and dongle hell.[/QUOTE]
FCP is 10.4.6.
it's something wrong with codesignature
when it says - faield to create metal device, it means AMDMTLBronzeDriver was not loaded after all
I have used ldid -S AMDMTLBronzeDriver
to "sign" it and make system ok with it.
you may want to try this kext
https://github.com/mologie/macos-disable-library-validation
when it says - faield to create metal device, it means AMDMTLBronzeDriver was not loaded after all
I have used ldid -S AMDMTLBronzeDriver
to "sign" it and make system ok with it.
you may want to try this kext
https://github.com/mologie/macos-disable-library-validation
yeah i had the problems too with hackintosh since the distance was a bit far and conflicted with wifi. I attached an antenna extension on the bluetooth port and it works fine now (from -70dB to -40 dB)
As for safari lag, what if you hover over links ? that is where the cursor slowness seems to be most apparent for me
EDIT: looks like it also happens in chrome when the iGPU is active - buttons links etc all make the mouse cursor lag. but staying within the blank/white space is fine
Last edited:
A tip - "Processes" panel doesn't work, one can turn it off in "Sections." All other features intact, turn off Update Check and set S.M.A.R.T. to off unless your Mac has an elderly drive inside.
Enjoy.
Nope, no problem like that, not in Safari, not in Chrome or even Firefox. Something specific of your installation
or the hardware since it's a clean install. Reset the SMC/PRAM too. I havent even signed into iCloud - just run the post patcher, reboot, create a new account, open safari and it's happening when the iGPU is active. A google search shows this issue going back to 2011 with no solutions too
Hardware wise, there's no optical drive and i replaced the Bluetooth 2.1 with 4.0
I am continuing to have issues trying to get system updates to work properly. I used the Mojave Patcher (Most likely the 1.2.e or older) on my MacBook Pro 5,5 (Mid-2009). It is currently running OSX14.3. I do the standard system update to get it to OSX14.5 and I get a boot loop. I did do a new install of Mojave in the APFS on a SSD.
I have tried to create an install USB drive using the the patch 1.3.0 and 1.3.3 but when I do the boot management option the only drive that pops up the EFI. I have an older version of the Installer USB that still works.
So question does the name of the Installer Drive matter. (ie does it have to be named macOSX when you format it? as per the instruction on Dosdude1) I found a old 32gb usb drive does the type of USB drive matter?
I have tried to create an install USB drive using the the patch 1.3.0 and 1.3.3 but when I do the boot management option the only drive that pops up the EFI. I have an older version of the Installer USB that still works.
So question does the name of the Installer Drive matter. (ie does it have to be named macOSX when you format it? as per the instruction on Dosdude1) I found a old 32gb usb drive does the type of USB drive matter?
I posted a patch for Xcode 11 on the Catalina thread. It works on Mojave as well. Link.
AMD RX 580 works in the 3,1 now thanks to netkas's work, just took a bit more effort to get working than we had realized.
I had to use the kext you mentioned to disable library validation. I tried using 'codesign' to sign the binary, which it did sign apparently correctly, however that only got me to the next fail point where it was failing to create the metal device.
I don't have 'ldid' on any of my Mac installations, therefore I attempted using codesign instead as suggested by dosdude.
The moment I loaded the kext to disable the display turned on and WindowServer stopped failing to create metal device.
Added Lilu and WhateverGreen along with appropriate nvram args for AMD hardware acceleration, VideoProc apparently can't determine the GPU hardware when it is in the 3,1...
"jellyfish-400-mbps-4k-uhd-hevc-10bit.mkv" plays properly with zero frame drops and nominal CPU usage (this is a 1.5Gb video clip that is 30 seconds long). Note that VideoProc claims no HEVC hardware decode (seems bug in the app) but it does work as without it this video isn't able to be played on a Mac Pro.
AirPlay to an AppleTV does show up now and I can get a black screen on the AppleTV but the display doesn't work and it never shows in display prefs.
I already have Xcode 11 beta "non-metal patched" from @ASentientBot on an external SSD , I noticed that in Catalina PrivateFrameworks there are some new SwiftUI.framework and maybe others related to Xcode11, so don't know, maybe editing the Xcode.app's Info.plist could be an attempt, I'll try on HighSierra 10.13.6 .
Funny: I'm editing from Xcode 10.1 the Xcode 11 Info.plist.
Attachments
I got it to launch, clicked accept and entered my password for changes. But it doesn't seem to be doing anything. Is this normal?
Launching from HighSierra Terminal I got this message:
Code:
./Xcode
dyld: Library not loaded: /usr/lib/libAccessibility.dylib
Referenced from: /Volumes/ExternalUSB/Xcode-beta.app/Contents/Developer/Platforms/MacOSX.platform/Developer/Library/PrivateFrameworks/XCTAutomationSupport.framework/Versions/A/XCTAutomationSupport
Reason: image not found
Abort trap: 6But I guess some other Mojave 10.14.4/5 libraries are needed.