You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hey Pike - I appreciate all this impressive work, particularly the clean and well-commented code. Many thanks to you and all the others involved! I'm hoping that once I get things up and going I will be able to offer some contributions of my own.
I've been able to figure out most of the details, and have been working through the code bit by bit to resolve the various problems, but I've reached something of an impasse.
So far as I can tell, everything in the boot process succeeds and hands off control to the kernel. I have DEBUG_BOOT and DEBUG_BOOT_GRAPHICS enabled as well as -v in com.apple.Boot.plist. I've verified that the plist is found and -v is recognized. I'm not seeing any output from -v that I do not see without it. When the Apple logo appears is verbose mode complete? Not clear on this point.
Based on this issue #3 I hypothesize that this is a graphics initialization issue due to EFI? My GPU is a Radeon Vega, no iGPU (C610, 2xx5-2683v4). I attempted to follow advice here https://github.com/Piker-Alpha/RevoBoot/wiki/About-EFI-Device-Properties and here https://www.insanelymac.com/forum/topic/259090-revoboot/ (while booted via Clover), resulting in a hex string that I pasted into EFI/One.h. Specifically, I used Lizard to produce device-properties.hex, which I then processed per create.sh to output the hex string that I used in EFI/One.h (I used Lizard because output from gfxutil -fsvn devprop.plist produced errors with create.sh). Static EFI string did not seem to make a difference.
I am using an SSD formatted GPT with the first partition as unencrypted HFS+. I installed boot0 and boot1 as fdisk440 -f sym/i386/boot0 -u -y /dev/disk2 and dd if=sym/i386/boot1h of=/dev/rdisk2s2. I let the Makefile copy boot to the root of SSD. My model is specified as IMAC_161.
I had to modify the Makefile MODEL naming schema because the #defines with the same names in platform.h were causing the compiler to think that the TARGET_MODEL was something like (IMAC | (13 << 15)). I'm not sure how this was working at all previously? My changes are available here https://github.com/Asher-/RevoBoot. I believe that the result is clean separation between MODEL and TARGET_MODEL, but it's possible that I missed something.
Any insight as to what is going on, how to approach the problem, or suggestions as to how to begin researching / diagnosing the matter would be very much appreciated.
I realized I didn't have LOAD_DSDT_TABLE_FROM_EXTRA_ACPI enabled, but enabling it doesn't seem to have changed anything. My DSDT has GPU configuration added (verified working when used with Clover). I also enabled INJECT_EFI_DEVICE_PROPERTIES based on the IM post above, but that did not change anything (which is what I expected from inspecting the code paths).
Hey Pike - I appreciate all this impressive work, particularly the clean and well-commented code. Many thanks to you and all the others involved! I'm hoping that once I get things up and going I will be able to offer some contributions of my own.
I've been able to figure out most of the details, and have been working through the code bit by bit to resolve the various problems, but I've reached something of an impasse.
So far as I can tell, everything in the boot process succeeds and hands off control to the kernel. I have DEBUG_BOOT and DEBUG_BOOT_GRAPHICS enabled as well as -v in com.apple.Boot.plist. I've verified that the plist is found and -v is recognized. I'm not seeing any output from -v that I do not see without it. When the Apple logo appears is verbose mode complete? Not clear on this point.
Based on this issue #3 I hypothesize that this is a graphics initialization issue due to EFI? My GPU is a Radeon Vega, no iGPU (C610, 2xx5-2683v4). I attempted to follow advice here https://github.com/Piker-Alpha/RevoBoot/wiki/About-EFI-Device-Properties and here https://www.insanelymac.com/forum/topic/259090-revoboot/ (while booted via Clover), resulting in a hex string that I pasted into EFI/One.h. Specifically, I used Lizard to produce device-properties.hex, which I then processed per create.sh to output the hex string that I used in EFI/One.h (I used Lizard because output from gfxutil -fsvn devprop.plist produced errors with create.sh). Static EFI string did not seem to make a difference.
I am using an SSD formatted GPT with the first partition as unencrypted HFS+. I installed boot0 and boot1 as fdisk440 -f sym/i386/boot0 -u -y /dev/disk2 and dd if=sym/i386/boot1h of=/dev/rdisk2s2. I let the Makefile copy boot to the root of SSD. My model is specified as IMAC_161.
I had to modify the Makefile MODEL naming schema because the #defines with the same names in platform.h were causing the compiler to think that the TARGET_MODEL was something like (IMAC | (13 << 15)). I'm not sure how this was working at all previously? My changes are available here https://github.com/Asher-/RevoBoot. I believe that the result is clean separation between MODEL and TARGET_MODEL, but it's possible that I missed something.
Any insight as to what is going on, how to approach the problem, or suggestions as to how to begin researching / diagnosing the matter would be very much appreciated.
Thanks!
Asher
SETTINGS/One.h
One.h.txt
EFI/One.h
One.h.txt
The text was updated successfully, but these errors were encountered: