Replies: 29 comments 10 replies
Just to be clear, do you mean RELEASE works but DEBUG (same otherwise) does not? |
No, I just don't have a backup of DEBUG version to provide the log. This comment only reflects the OC hang during chainloading. RP not booting is for both RELEASE and DEBUG versions (latest few commits I don't remember where it started) |
|
Bug is said to be that RP is not booting and not logging since the last few commits:
I have just tested again and it works for me. It is that you don't see the boot screen when running RP? |
|
it is not booting and no logging occurs. It is similar to what I had with the MBP with the memory leak except there is no logging at all. If I use the MVC card I stay on the grey screen forever. With a regular card there is a black screen forever. I don't get to the boot selector at all. |
|
I see. Try the attached current build. |
|
I just tried the rEFInd 0.13.01 and while it boots there were no partitions to select just the tools. Here is the config file which is just a copy of the RP file renamed |
|
OK. Please test the previously attached file. |
|
Same thing with the attached build. I have no previous undeleted logs. |
|
OK. Thanks. Might be worth opening a ticket upstream since it seems RP appears to be behaving consistently with rEFInd. |
One odd thing though ... the attached should not be able get to showing tools without outputting a log of progress up to that point. |
I only see tools in refind nothing in RP. |
OK. Either way though, there is an issue there. Maybe test rEFInd with the RP specific tokens removed from the config to make sure there is no conflict. If there is still an issue, then best to open the ticket: https://sourceforge.net/p/refind/tickets/ (Seems tickets might be disabled. Might need to go into the forum) |
|
@dakanji just tested X110 as requested. RP initiated fine and loaded bootpicker on my cMP5,1. See RP logs attached. |
|
Thanks @CornCat78 @startergo Initially looks like it might be something specific to you. |
Can you please try the touch refind config file?just rename it config.conf. |
Good idea. @CornCat78 ... he means to try using the config file here: #24 (comment) Since you are using a MyBootMgr setup, best way will be to temporarily replace your overrides.conf file with this and reboot. |
|
@dakanji - just to be clear, test with the replacement |
|
@dakanji |
|
This is the rEFInd log: |
|
Dayo I found what is "wrong". I will let you find it too through the log. Let me know if you find it (hint: I changed something :) ) |
|
Noted the EFI Revision change you made. |
|
I still consider this a bug even though it is in the upstream rEFInd. This was the only way to initialize an NVIDIA GOP without chain-loading OC through RP on systems with EFI revision 1.1. There is a change in the code between 0.12.0-0.13.1 of rEFInd, which breaks this workaround and leaves the system in unbootable state. @joevt what do you think? |
|
@startergo So you enabled the hack that allows new GOPs that want EFI 2.3 to work. Somehow that hack causes a hang. Is the hack being applied by RefindPlus or by my FakeUEFI2.efi and ReloadPCIRom.efi driver? I believe the RefindPlus implementations are slightly different but I didn't do much comparing or testing of either. |
This was addressed in the RP code earlier so it is not needed when chain-loading OC through RP, but if one uses OC alone without this driver early in the booting process NVIDIA's GOP does not initialize. It broke in the rEFInd code between versions 0.12 and 0.13.1 and Dayo merged the upstream rEFInd code and now both bootloaders hang with this driver loaded. |
A broken hack is not a bug. You basically have the following options:
|
This is the best approach. |
Correct. Note that this can be done by loading the driver in OC instead of injecting it in firmware. |
|
"Probably be done" |
|
I tested it before and it did not work as it needed to be loaded early in the booting process or reloaded on demand like in RP. |
|
There are challenges there indeed. Not straightforward to implement by embedding in their code. Have to consider some of the other options listed instead. |
|
Theoretically though, the actions can still be done in a driver. |
|
At present the best option is chain loading OC through RP. Days may I request for your consideration a self blessing feature? Like R |
Not sure how that would work or if even possible. |
|
OC is pretty consistent in finding itself after NVRAM reset now. |
|
That's not OC doing anything special. This is just how the UEFI specs work. If you install RP as a fallback efi (BOOTx64 file in EFI/BOOT), as per MyBootMgr recommended setup for instance, it will be loaded after an NVRAM reset. Specific "self blessing" is something else beyond this. Implies finding itself even when in a location and/or with a filename inconsistent with the specs. |
In multi disk configuration? |
The first fallback efi found (OC, RP, rEFInd, Clover, whatever) will always be loaded. |
|
Synced with an upstream change that impacts on EFI driver loading. |
|
Very strange. The debug does not boot to RP bootpicker, the release boots but does not chainload OC. Otherwise it boots a regular Mac OS/Windows. |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Describe the bug
No booting occurs through latest few commits and no log file is generated
To Reproduce
Steps to reproduce the behaviour:
Try to boot any OS
Expected behaviour
Expected to see the RP bootpicker or the login screen
Environment:
Log and Config Files
No Log file is generated.
Additional context
Restored working RP (sorry don't know which revision), but chain loading with the MVC GT-640 OC Hangs. Attached please find the OC log file hang:
Archive.zip
After soft blessing OC directly it boots to the login screen. Attached the log file:
opencore-2021-02-24-231046.txt.zip
Sorry the working RP version is the RELEASE version so no RP log file is generated. Here it is attached. Maybe you can find out what version it is:
Refind_x64.efi.zip
All reactions