I set up my new machine with EFI on a gigabyte X470 aorus gaming wifi board with ryzen 2600X.

I read that, the EFI support fie Nvidia drivers is poor, but nevertheless I tried.
I get the following message in the dmesg output:

[ 24.404715] NVRM: RmInitAdapter failed! (0x26:0x65:1125)
[ 24.404739] NVRM: rm_init_adapter failed for device bearing minor number 0

Itried using the grub bootloader and the kernel stub, with the same outcome, the X server does not find a usable screen.

can anyone explain the message above?
Is there a combination of kernel and nvidia driver which is known to run with EFI?

Nvidia, when is there proper support for EFI booting with linux? It also works in windows, so it cannot be for technical reasons?

This the rest of the log. I appreciate any input to help make it work.

[ 6.135591] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 410.73 Sat Oct 20 22:12:33 CDT 2018 (using threaded interrupts)

[ 6.475541] resource sanity check: requesting [mem 0x000c0000-0x000fffff], which spans more than PCI Bus 0000:00 [mem 0x000c0000-0x000dffff window]
[ 6.475620] caller _nv001126rm+0xe3/0x1d0 [nvidia] mapping multiple BARs
[ 6.720895] input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci0000:00/0000:00:03.1/0000:0a:00.1/sound/card0/input21
[ 6.720931] input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci0000:00/0000:00:03.1/0000:0a:00.1/sound/card0/input22
[ 6.720962] input: HDA NVidia HDMI/DP,pcm=8 as /devices/pci0000:00/0000:00:03.1/0000:0a:00.1/sound/card0/input23
[ 6.721173] input: HDA NVidia HDMI/DP,pcm=9 as /devices/pci0000:00/0000:00:03.1/0000:0a:00.1/sound/card0/input24
[ 10.684493] NVRM: RmInitAdapter failed! (0x26:0x65:1125)
[ 10.684515] NVRM: rm_init_adapter failed for device bearing minor number 0
[ 10.784727] nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms 410.73 Sat Oct 20 22:08:51 CDT 2018
[ 10.785636] [drm] [nvidia-drm] [GPU ID 0x00000a00] Loading driver
[ 10.785637] [drm] Initialized nvidia-drm 0.0.0 20160202 for 0000:0a:00.0 on minor 0

[ 20.323195] resource sanity check: requesting [mem 0x000c0000-0x000fffff], which spans more than PCI Bus 0000:00 [mem 0x000c0000-0x000dffff window]
[ 20.323267] caller _nv001126rm+0xe3/0x1d0 [nvidia] mapping multiple BARs
[ 24.404715] NVRM: RmInitAdapter failed! (0x26:0x65:1125)
[ 24.404739] NVRM: rm_init_adapter failed for device bearing minor number 0

Thomas,

Can you run

nvidia-bug-report.sh

on your machine and email the output file to linux-bugs [at] nvidia.com ? In the rare chance that this issue causes your machine to lock up while collecting this bug report please run

nvidia-bug-report.sh --safe-mode

.

thanks, did this. Hope for the best.

Did you check for a bios update?

I run the latest, version F5

The driver seems to be timing out while waiting for the GPU to initialize, which is a weird symptom to start happening just because EFI is enabled. GPUs in other systems work fine in EFI mode, so I’m tempted to agree with generix that this sounds like a system-level problem.

When you enabled the grub bootloader, did it initialize the GPU and display the boot menu on it, or did it still boot completely headless? If you do use grub, do you still get exactly the same RmInitAdapter error, or do the numbers change? Finally, can you please give the 415.13 beta driver a try?

thanks for the hint.
When I run with grub, The text menu is displayed, after that the screen is blank, because I disabled all the framebuffer support.

in dmesg, I get the same error:

[ 10.651190] NVRM: RmInitAdapter failed! (0x26:0x65:1125)
[ 10.651215] NVRM: rm_init_adapter failed for device bearing minor number 0

actually twice. Going to try the beta driver, unfortunately there is no ebuild yet…

with the 415.13 module I get a slightly different output, but still the same effect.

[ 10.746509] nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms 415.13 Wed Oct 31 18:49:37 CDT 2018

[ 24.003372] NVRM: RmInitAdapter failed! (0x26:0x65:1098)
[ 24.003399] NVRM: rm_init_adapter failed for device bearing minor number 0

also with and without using grub.

Also with the EFI framebuffer enabled I get the same error.
But then I at least have a useable console.

Sounds like you’re using Gentoo with a stripped-down kernel. Any chance you can do small ubuntu install so you can evaluate if you’re missing kernel options?

Did this. Messed around with a lot of PCI sounding features, now I got it loaded.
Next Problem is I am facing a black sddm, but this is a different story.
Will provide a diff of the configs…

ok select the correct opengl and the graphics are back :-)

< # CONFIG_CGROUP_BPF is not set
155c154
< CONFIG_EXPERT=y

CONFIG_EXPERT is not set

160d158
< # CONFIG_SYSCTL_SYSCALL is not set
183,184c181,182
< CONFIG_BPF_SYSCALL=y
< CONFIG_USERFAULTFD=y

CONFIG_BPF_SYSCALL is not set

CONFIG_USERFAULTFD is not set

187d184
< # CONFIG_DEBUG_RSEQ is not set
190d186
< # CONFIG_PC104 is not set
201d196
< # CONFIG_SLOB is not set
253c248
< CONFIG_X86_X2APIC=y

CONFIG_X86_X2APIC is not set

257c252
< CONFIG_INTEL_RDT=y

CONFIG_INTEL_RDT is not set

259d253
< # CONFIG_X86_NUMACHIP is not set
261d254
< # CONFIG_X86_UV is not set
285d277
< # CONFIG_PROCESSOR_SELECT is not set
305c297
< CONFIG_X86_REROUTE_FOR_BROKEN_BOOT_IRQS=y

CONFIG_X86_REROUTE_FOR_BROKEN_BOOT_IRQS is not set

334c326
< # CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT is not set

CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y
441d432
< # CONFIG_ACPI_REDUCED_HARDWARE_ONLY is not set
506c497
< CONFIG_PCI_MMCONFIG=y

CONFIG_PCI_MMCONFIG is not set

508,509d498
< CONFIG_MMCONF_FAM10H=y
< # CONFIG_PCI_CNB20LE_QUIRK is not set
565d553
< # CONFIG_ISA_BUS is not set
841d828
< # CONFIG_XDP_SOCKETS is not set
913d899
< # CONFIG_BPF_STREAM_PARSER is not set
1323d1308
< # CONFIG_PATA_PLATFORM is not set
1645d1629
< # CONFIG_TTY_PRINTK is not set
2608d2591
< # CONFIG_USB_OTG_BLACKLIST_HUB is not set
3696c3679
< # CONFIG_DEBUG_MEMORY_INIT is not set

CONFIG_DEBUG_MEMORY_INIT=y
3801d3783
< CONFIG_BPF_EVENTS=y
3803d3784
< # CONFIG_BPF_KPROBE_OVERRIDE is not set
3884d3864
< # CONFIG_UNWINDER_GUESS is not set

seems one of these must be relevant in this case. Hope it helps somebody.

CONFIG_PCI_MMCONFIG is essential for most kind of hardware.

few kernels later, I think CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT is the relevant one. Should be disabled.

So it is running with EFI boot and EFI framebuffer.
Grub and kernel stub both work ok.

Hi Thomas,

I had reproduce the issue internally few weeks back but today when I tried with the same configuration, I don’t see any problem while running nvidia-smi.
Can you please verify if problem still persists with you.
If yes, please share VBIOS for GPU.

Hi Thomas,

Additionally, can you please use below parameter at grub menu & boot OS and share results/observation wth us.

mem_encrypt=off

Hi Thomas, Do you have any updates for the requested information?

1. In earlier repro scenario we observed NVRM: RmInitAdapter failed! in the log in text mode but now we are not able to repro NVRM: RmInitAdapter failed! issue.
2. Now X is not starting with the error [ 650.314] (EE) NVIDIA(GPU-0): Failed to initialize DMA.
   [ 650.315] (EE) NVIDIA(0): Failed to allocate push buffer
   [ 651.544] (EE) AddScreen/ScreenInit failed for driver 0

We will continue the investigation. Additionally, It would be nice to know testing feedback with kernel parameter mem_encrypt=off