Re: nouveau unhappy with EFI
On Wed, Jan 15, 2020 at 01:22:00PM +0100, Lars Reichardt wrote: > On Mon, 29 Jul 2019 10:16:00 +0100 > Patrick Welche wrote: > > > With biosboot set on the root partition of a GPT disk, I can now > > biosboot or EFI boot the same computer. > > > > Thanks to mlelstv@'s hint, I can get a serial console on the EFI side > > with consdev com,0x3f8,115200 > > as well as the biosbooting side. > > > > In both cases, booting with the serial console, X works on this > > nouveau GeForce GTX 680. > > > > However, if I boot without the serial console. I see nothing on > > screen in the EFI case, but all is OK in the biosboot case. (Unsure: > > typing blindly I did not get X to start in the EFI case, but fat > > fingering possible.) > > > > Hi, > > I see about the same behavior with a GT710 booted with efi, either the > machine stops the moment the frame buffer should get activated or the > screen is all black afterwards with the machine running otherwise. > The last lines displayed are: > nouveau0: error: priv: HUB0: 086014 (1a70828c) > nouveau0: info: fb 2048 MiB GDDR5 > > The Situation with a radeon card (either R5 230 or R7 240) is like it > but not entirely. > With bios boot both cards initialize properly during drmkms startup. > With efi boot both hang about 11sec during initialization with the > following messages: > *ERROR* UVD not responding, trying to reset the VCPU!!! > *ERROR* UVD not responding, trying to reset the VCPU!!! > *ERROR* UVD not responding, trying to reset the VCPU!!! > *ERROR* UVD not responding, trying to reset the VCPU!!! > *ERROR* UVD not responding, trying to reset the VCPU!!! > *ERROR* UVD not responding, trying to reset the VCPU!!! > *ERROR* UVD not responding, trying to reset the VCPU!!! > *ERROR* UVD not responding, trying to reset the VCPU!!! > *ERROR* UVD not responding, trying to reset the VCPU!!! > *ERROR* UVD not responding, trying to reset the VCPU!!! > *ERROR* UVD not responding, trying to reset the VCPU!!! > *ERROR* UVD not responding, trying to reset the VCPU!!! > *ERROR* UVD not responding, trying to reset the VCPU!!! > *ERROR* UVD not responding, giving up!!! > failed initializing UVD (-1). > > the code printing the error messages has some delays and retries that > add up to the delay... > but the framebuffer comes to live afterwards. > In case of the R7 240 no glamor acceleration but that's a different > story. > > I'm not very deep into the drm code but my guess is, something is > missing about taking over/releasing the efi console and thats blocking > proper initialization as said just a guess. > > > Lars > > > - > You will continue to suffer > if you have an emotional reaction to everything that is said to you. > True power is sitting back and observing everything with logic. > If words control you that means everyone else can control you. > Breathe and allow things to pass. > > --- Bruce Lee > Just FYI: there's something wrong about how nouveau sets up interrupts. (We had to force disable MSI, and I think it enabled interrupts once in its own code and once in generic drm code, or something.) It's possible EFI just happens to run into it.
Re: nouveau unhappy with EFI
On Mon, 29 Jul 2019 10:16:00 +0100 Patrick Welche wrote: > With biosboot set on the root partition of a GPT disk, I can now > biosboot or EFI boot the same computer. > > Thanks to mlelstv@'s hint, I can get a serial console on the EFI side > with consdev com,0x3f8,115200 > as well as the biosbooting side. > > In both cases, booting with the serial console, X works on this > nouveau GeForce GTX 680. > > However, if I boot without the serial console. I see nothing on > screen in the EFI case, but all is OK in the biosboot case. (Unsure: > typing blindly I did not get X to start in the EFI case, but fat > fingering possible.) > Hi, I see about the same behavior with a GT710 booted with efi, either the machine stops the moment the frame buffer should get activated or the screen is all black afterwards with the machine running otherwise. The last lines displayed are: nouveau0: error: priv: HUB0: 086014 (1a70828c) nouveau0: info: fb 2048 MiB GDDR5 The Situation with a radeon card (either R5 230 or R7 240) is like it but not entirely. With bios boot both cards initialize properly during drmkms startup. With efi boot both hang about 11sec during initialization with the following messages: *ERROR* UVD not responding, trying to reset the VCPU!!! *ERROR* UVD not responding, trying to reset the VCPU!!! *ERROR* UVD not responding, trying to reset the VCPU!!! *ERROR* UVD not responding, trying to reset the VCPU!!! *ERROR* UVD not responding, trying to reset the VCPU!!! *ERROR* UVD not responding, trying to reset the VCPU!!! *ERROR* UVD not responding, trying to reset the VCPU!!! *ERROR* UVD not responding, trying to reset the VCPU!!! *ERROR* UVD not responding, trying to reset the VCPU!!! *ERROR* UVD not responding, trying to reset the VCPU!!! *ERROR* UVD not responding, trying to reset the VCPU!!! *ERROR* UVD not responding, trying to reset the VCPU!!! *ERROR* UVD not responding, trying to reset the VCPU!!! *ERROR* UVD not responding, giving up!!! failed initializing UVD (-1). the code printing the error messages has some delays and retries that add up to the delay... but the framebuffer comes to live afterwards. In case of the R7 240 no glamor acceleration but that's a different story. I'm not very deep into the drm code but my guess is, something is missing about taking over/releasing the efi console and thats blocking proper initialization as said just a guess. Lars - You will continue to suffer if you have an emotional reaction to everything that is said to you. True power is sitting back and observing everything with logic. If words control you that means everyone else can control you. Breathe and allow things to pass. --- Bruce Lee
nouveau unhappy with EFI
With biosboot set on the root partition of a GPT disk, I can now biosboot or EFI boot the same computer. Thanks to mlelstv@'s hint, I can get a serial console on the EFI side with consdev com,0x3f8,115200 as well as the biosbooting side. In both cases, booting with the serial console, X works on this nouveau GeForce GTX 680. However, if I boot without the serial console. I see nothing on screen in the EFI case, but all is OK in the biosboot case. (Unsure: typing blindly I did not get X to start in the EFI case, but fat fingering possible.) Apart from hard disk probe ordering differences, this is all I see when comparing the dmesgs: $ diff -u bios_with_nv_com.dmesg efi_with_nv_com.dmesg --- bios_with_nv_com.dmesg 2019-07-29 09:58:10.390398272 +0100 +++ efi_with_nv_com.dmesg 2019-07-29 08:59:02.186251104 +0100 @@ -13,29 +13,31 @@ Kernelized RAIDframe activated running cgd selftest aes-xts-256 aes-xts-512 done timecounter: Timecounter "i8254" frequency 1193182 Hz quality 100 +efi: systbl at pa db67f018 System manufacturer System Product Name (System Version) mainbus0 (root) -ACPI: RSDP 0x000F05B0 24 (v02 ALASKA) -ACPI: XSDT 0xD143E098 AC (v01 ALASKA A M I01072009 AMI 00010013) -ACPI: FACP 0xD144A788 000114 (v06 ALASKA A M I01072009 AMI 00010013) -ACPI: DSDT 0xD143E1E0 00C5A3 (v02 ALASKA A M I01072009 INTL 20120913) +ACPI: RSDP 0xCA802000 24 (v02 ALASKA) +ACPI: XSDT 0xCA8020A0 B4 (v01 ALASKA A M I01072009 AMI 00010013) +ACPI: FACP 0xCA80E790 000114 (v06 ALASKA A M I01072009 AMI 00010013) +ACPI: DSDT 0xCA8021E8 00C5A3 (v02 ALASKA A M I01072009 INTL 20120913) ACPI: FACS 0xDA931D80 40 -ACPI: APIC 0xD144A8A0 00015E (v03 ALASKA A M I01072009 AMI 00010013) -ACPI: FPDT 0xD144AA00 44 (v01 ALASKA A M I01072009 AMI 00010013) -ACPI: FIDT 0xD144AA48 9C (v01 ALASKA A M I01072009 AMI 00010013) -ACPI: SSDT 0xD144AAE8 FC (v02 ALASKA CPUSSDT 01072009 AMI 01072009) -ACPI: SSDT 0xD145BCA8 BF (v01 AMDAMD PT 1000 INTL 20120913) -ACPI: SSDT 0xD144AC40 008C98 (v02 AMDAMD ALIB 0002 MSFT 0400) -ACPI: SSDT 0xD14538D8 00322C (v01 AMDAMD AOD 0001 INTL 20120913) -ACPI: MCFG 0xD1456B08 3C (v01 ALASKA A M I01072009 MSFT 00010013) -ACPI: HPET 0xD1456B48 38 (v01 ALASKA A M I01072009 AMI 0005) -ACPI: SSDT 0xD1456B80 24 (v01 AMDFCH FCHZP1000 INTL 20120913) -ACPI: UEFI 0xD1456BA8 42 (v01 ) -ACPI: IVRS 0xD1456BF0 D0 (v02 AMDAMD IVRS 0001 AMD ) -ACPI: SSDT 0xD1456CC0 002314 (v01 AMDAMD CPU 0001 AMD 0001) -ACPI: CRAT 0xD1458FD8 000F50 (v01 AMDAMD CRAT 0001 AMD 0001) -ACPI: CDIT 0xD1459F28 29 (v01 AMDAMD CDIT 0001 AMD 0001) -ACPI: SSDT 0xD1459F58 001D4A (v01 AMDAmdTable 0001 INTL 20120913) +ACPI: APIC 0xCA80E8A8 00015E (v03 ALASKA A M I01072009 AMI 00010013) +ACPI: FPDT 0xCA80EA08 44 (v01 ALASKA A M I01072009 AMI 00010013) +ACPI: FIDT 0xCA80EA50 9C (v01 ALASKA A M I01072009 AMI 00010013) +ACPI: SSDT 0xCA80EAF0 FC (v02 ALASKA CPUSSDT 01072009 AMI 01072009) +ACPI: SSDT 0xCA81FCE8 BF (v01 AMDAMD PT 1000 INTL 20120913) +ACPI: SSDT 0xCA80EC48 008C98 (v02 AMDAMD ALIB 0002 MSFT 0400) +ACPI: SSDT 0xCA8178E0 00322C (v01 AMDAMD AOD 0001 INTL 20120913) +ACPI: MCFG 0xCA81AB10 3C (v01 ALASKA A M I01072009 MSFT 00010013) +ACPI: HPET 0xCA81AB50 38 (v01 ALASKA A M I01072009 AMI 0005) +ACPI: SSDT 0xCA81AB88 24 (v01 AMDFCH FCHZP1000 INTL 20120913) +ACPI: UEFI 0xCA81ABB0 42 (v01 ) +ACPI: BGRT 0xCA81ABF8 38 (v01 ALASKA A M I01072009 AMI 00010013) +ACPI: IVRS 0xCA81AC30 D0 (v02 AMDAMD IVRS 0001 AMD ) +ACPI: SSDT 0xCA81AD00 002314 (v01 AMDAMD CPU 0001 AMD 0001) +ACPI: CRAT 0xCA81D018 000F50 (v01 AMDAMD CRAT 0001 AMD 0001) +ACPI: CDIT 0xCA81DF68 29 (v01 AMDAMD CDIT 0001 AMD 0001) +ACPI: SSDT 0xCA81DF98 001D4A (v01 AMDAmdTable 0001 INTL 20120913) ACPI: 8 ACPI AML tables successfully acquired and loaded @@ -475,7 +477,7 @@ nouveau0: info: NVIDIA GK104 (0e4000a2) nouveau0: info: bios: version 80.04.09.00.01 nouveau0: info: fb: 2048 MiB GDDR5 -Zone kernel: Available graphics memory: 11312638 kiB +Zone kernel: Available graphics memory: 11312672 kiB Zone dma32: Available graphics memory: 2097152 kiB nouveau0: info: DRM: VRAM: 2048 MiB nouveau0: info: DRM: GART: 1048576 MiB @@ -495,10 +497,65 @@ kern inf
