Bug#992612: linux-image-5.10.0-8-amd64: Macbook3,1 fails to boot without nomodeset with nouveau
...well that's _fascinating_.
# grep -C3 nouveau /var/log/Xorg.0.log
[ 30.712] (II) modeset(0): Initializing kms color map for depth 24, 8 bpc.
[ 30.712] (==) modeset(0): DPMS enabled
[ 30.713] (II) modeset(0): [DRI2] Setup complete
[ 30.713] (II) modeset(0): [DRI2] DRI driver: nouveau
[ 30.713] (II) modeset(0): [DRI2] VDPAU driver: nouveau
[ 30.713] (II) Initializing extension Generic Event Extension
[ 30.713] (II) Initializing extension SHAPE
[ 30.714] (II) Initializing extension MIT-SHM
--
[ 30.728] (II) Initializing extension SELinux
[ 30.729] (II) SELinux: Disabled on system
[ 30.729] (II) Initializing extension GLX
[ 30.745] (II) AIGLX: Loaded and initialized nouveau
[ 30.745] (II) GLX: Initialized DRI2 GL provider for screen 0
[ 30.745] (II) Initializing extension XFree86-VidModeExtension
[ 30.746] (II) Initializing extension XFree86-DGA
# journalctl -k -b 0 | grep -B10 nouveau | head -n [mumble]
Aug 22 02:54:24 struth kernel: scsi host4: ahci
Aug 22 02:54:24 struth kernel: ata1: SATA max UDMA/133 abar
m2048@0xdb504000 port 0xdb504100 irq 30
Aug 22 02:54:24 struth kernel: ata2: DUMMY
Aug 22 02:54:24 struth kernel: ata3: DUMMY
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: enabling device
(0002 -> 0003)
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: vgaarb:
deactivate vga console
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: NVIDIA G84 (084700a2)
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: pci: MSI enabled
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: trying PRAMIN...
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: ... not enabled
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: trying PROM...
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: 00000000:
ROM signature (0201) unknown
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: image 0 invalid
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: scored 0
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: trying ACPI...
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: trying ACPI...
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: trying PCIROM...
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: Invalid PCI ROM
header signature: expecting 0xaa55, got 0x0001
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: trying PLATFORM...
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: 000000f0:
NPDE signature (30303638) unknown
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: 00000000:
type 00, 53248 bytes
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: scored 4
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: using image
from PLATFORM
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: 000000f0:
NPDE signature (30303638) unknown
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: 000000f0:
NPDE signature (30303638) unknown
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: BIT signature found
Aug 22 02:54:24 struth kernel: nouveau 0000:01:00.0: bios: version
60.84.49.03.00
[...]
What changed? I didn't lob a vbios manually in or anything, if the
output there didn't make that clear...
The bootloader changed. I had the system booting from grubx64.efi
directly, but when trying to reboot into macOS, I found that the
autogenerated "boot macOS" GRUB entries just panicked, so I used alt
on boot to tell it to boot macOS without passing through GRUB or
rEFInd, and reconfigured it to launch rEFInd on boot instead.
Then I directly booted vmlinuz-5.10.0-8-amd64 from rEFInd, without
passing through GRUB, and it didn't steal my nomodeset from the grub
parameters...
...but instead of dying on switching, it happily booted.
Womp, and I say this advisedly, womp.
(GRUB chainloaded from rEFInd breaks the same.)
Curiously, when I changed it from debug=debug to debug=info, this
started appearing on successive boots from rEFInd-only:
[ 66.323195] nouveau 0000:01:00.0: firmware: failed to load
nouveau/nv84_xuc00f (-2)
[ 66.323210] nouveau 0000:01:00.0: Direct firmware load for
nouveau/nv84_xuc00f failed with error -2
[ 66.323218] nouveau 0000:01:00.0: vp: unable to load firmware
nouveau/nv84_xuc00f
[ 66.323223] nouveau 0000:01:00.0: vp: init failed, -2
[ 66.323273] nouveau 0000:01:00.0: firmware: failed to load
nouveau/nv84_xuc103 (-2)
[ 66.323278] nouveau 0000:01:00.0: Direct firmware load for
nouveau/nv84_xuc103 failed with error -2
[ 66.323284] nouveau 0000:01:00.0: bsp: unable to load firmware
nouveau/nv84_xuc103
[ 66.323288] nouveau 0000:01:00.0: bsp: init failed, -2
Video accel still works fine, but it's definitely not in my log for
the first time this worked.
Curious.
Even better, if you disable gfxterm, it works with GRUB, too...
Seems GRUB screws up the world when it makes the pretty menu.
- Rich
On Sat, Aug 21, 2021 at 1:29 PM Rich <rincebrain@gmail.com> wrote:
>
> https://forums.developer.nvidia.com/t/error-loading-nvidia-kernel-module-with-macbookpro3-1/39313
>
> I found other threads, but apparently this is just a thing people have found happens with nouveau or nvidia.ko if you're booting in EFI mode?
>
> That user said they booted in BIOS emu and smuggled their vBIOS out that way, I'll see what I can do...
>
>
> On Sat, Aug 21, 2021, 6:00 AM Rich <rincebrain@gmail.com> wrote:
>>
>> Okay, so, I did just reboot and verify this...
>>
>> * kernel command line "... ro quiet":
>> displays a blinking cursor in the top-left for a couple seconds and
>> then abruptly cuts to full black, no response to VT switching or any
>> other keyboard input, waited 5 minutes to see if this ever changed
>>
>> * kernel command line "...ro"
>> displays normal output up to the aforementioned EFI VGA lines, then
>> just...stops ever updating again, waited for 5m on this one too, also
>> doesn't respond to VT switching or anything else
>>
>> https://i.imgur.com/jmcHHhm.jpg is a photo
>>
>> * kernel command line "...quiet nomodeset"
>> We get all the way to X and are cooking
>>
>> As far as it's concerned, it keeps going normally after it stops
>> outputting, and isn't sure why I hard power cycled it.
>>
>> Here's the first lines after that divergence on the "...ro" case:
>>
>> Aug 21 04:57:09 struth kernel: checking generic (c0030000 710000) vs
>> hw (c0000000 10000000)
>> Aug 21 04:57:09 struth kernel: fb0: switching to nouveaufb from EFI VGA
>> Aug 21 04:57:09 struth kernel: hub 6-0:1.0: 2 ports detected
>> Aug 21 04:57:09 struth kernel: uhci_hcd 0000:00:1d.2: UHCI Host Controller
>> Aug 21 04:57:09 struth kernel: uhci_hcd 0000:00:1d.2: new USB bus
>> registered, assigned bus number 7
>> Aug 21 04:57:09 struth kernel: uhci_hcd 0000:00:1d.2: detected 2 ports
>> Aug 21 04:57:09 struth kernel: uhci_hcd 0000:00:1d.2: irq 21, io base 0x00006040
>> Aug 21 04:57:09 struth kernel: Console: switching to colour dummy device 80x25
>> Aug 21 04:57:09 struth kernel: nouveau 0000:01:00.0: vgaarb:
>> deactivate vga console
>> Aug 21 04:57:09 struth kernel: usb usb7: New USB device found,
>> idVendor=1d6b, idProduct=0001, bcdDevice= 5.10
>> Aug 21 04:57:09 struth kernel: usb usb7: New USB device strings:
>> Mfr=3, Product=2, SerialNumber=1
>> Aug 21 04:57:09 struth kernel: usb usb7: Product: UHCI Host Controller
>> Aug 21 04:57:09 struth kernel: usb usb7: Manufacturer: Linux
>> 5.10.0-8-amd64 uhci_hcd
>> Aug 21 04:57:09 struth kernel: usb usb7: SerialNumber: 0000:00:1d.2
>> Aug 21 04:57:09 struth kernel: nouveau 0000:01:00.0: NVIDIA G84 (084700a2)
>> Aug 21 04:57:09 struth kernel: hub 7-0:1.0: USB hub found
>> Aug 21 04:57:09 struth kernel: hub 7-0:1.0: 2 ports detected
>> Aug 21 04:57:09 struth kernel: ata4.00: ATAPI: HL-DT-ST DVDRW
>> GSA-S10N, AP09, max UDMA/33
>> Aug 21 04:57:09 struth kernel: nouveau 0000:01:00.0: Invalid PCI ROM
>> header signature: expecting 0xaa55, got 0x0000
>> Aug 21 04:57:09 struth kernel: nouveau 0000:01:00.0: bios: unable to
>> locate usable image
>> Aug 21 04:57:09 struth kernel: nouveau 0000:01:00.0: bios ctor failed, -22
>> Aug 21 04:57:09 struth kernel: nouveau: probe of 0000:01:00.0 failed
>> with error -22
>> Aug 21 04:57:09 struth kernel: scsi 1:0:0:0: CD-ROM
>> HL-DT-ST DVDRW GSA-S10N AP09 PQ: 0 ANSI: 5
>> Aug 21 04:57:09 struth kernel: random: fast init done
>> Aug 21 04:57:09 struth kernel: ata1: SATA link up 1.5 Gbps (SStatus
>> 113 SControl 300)
>> Aug 21 04:57:09 struth kernel: ata1.00: ACPI cmd ef/10:03:00:00:00:a0
>> (SET FEATURES) filtered out
>> Aug 21 04:57:09 struth kernel: ata1.00: ATA-8: WDC WD2500LPVT-00G33T0,
>> 01.01A01, max UDMA/133
>> Aug 21 04:57:09 struth kernel: ata1.00: 488397168 sectors, multi 0:
>> LBA48 NCQ (depth 32), AA
>> Aug 21 04:57:09 struth kernel: ata1.00: ACPI cmd ef/10:03:00:00:00:a0
>> (SET FEATURES) filtered out
>> Aug 21 04:57:09 struth kernel: ata1.00: configured for UDMA/133
>> Aug 21 04:57:09 struth kernel: scsi 0:0:0:0: Direct-Access ATA
>> WDC WD2500LPVT-0 1A01 PQ: 0 ANSI: 5
>> Aug 21 04:57:09 struth kernel: sr 1:0:0:0: [sr0] scsi3-mmc drive:
>> 24x/24x writer cd/rw xa/form2 cdda caddy
>> Aug 21 04:57:09 struth kernel: cdrom: Uniform CD-ROM driver Revision: 3.20
>> Aug 21 04:57:09 struth kernel: clocksource: tsc: mask:
>> 0xffffffffffffffff max_cycles: 0x1fa284fbd4e, max_idle_ns:
>> 440795223138 ns
>> Aug 21 04:57:09 struth kernel: clocksource: Switched to clocksource tsc
>> Aug 21 04:57:09 struth kernel: sr 1:0:0:0: Attached scsi CD-ROM sr0
>> Aug 21 04:57:09 struth kernel: sd 0:0:0:0: [sda] 488397168 512-byte
>> logical blocks: (250 GB/233 GiB)
>>
>>
>> When we check out why Xorg doesn't ever spawn if the system keeps
>> running when this happens, the Xorg.0.log diff says...
>> (EE) open /dev/fb0: No such file or directory
>> vesa: Refusing to run on UEFI
>> (EE) Screen 0 deleted because of no matching config section.
>> (II) UnloadModule: "modesetting"
>> (EE) Screen 0 deleted because of no matching config section.
>> (II) UnloadModule: "fbdev"
>> (II) UnloadSubModule: "fbdevhw"
>> (EE) Device(s) detected, but none match those in the config file.
>> (EE)
>> error:
>> (EE) no screens found(EE)
>> (EE)
>>
>> (Full logs attached, .old is the one from "...ro" boot that didn't
>> display, the other is from a boot with nomodeset)
>>
>> Can provide other information as useful.
>>
>> - Rich
>>
>>
>> On Sat, Aug 21, 2021 at 4:28 AM Rich <rincebrain@gmail.com> wrote:
>> >
>> > Also (I've never used this before, so I might be holding it wrong, but...):
>> > $ isenkram-lookup
>> > cheese
>> > ethtool
>> > pcscd
>> > scdaemon
>> > $
>> >
>> > On Sat, Aug 21, 2021 at 4:15 AM Rich <rincebrain@gmail.com> wrote:
>> > >
>> > > Hi Salvatore,
>> > > I used firmware-11.0.0-amd64-netinst.iso. And yes, I did not manually
>> > > install firmware-misc-nonfree since this initially failed, and I still
>> > > got a black screen which did not ever resolve or respond to VT
>> > > switching unless I used nomodeset and deleted quiet. (With quiet
>> > > removed, it doesn't black out, but it does stop ever responding after
>> > > it prints the "switching from EFI VGA" line.)
>> > >
>> > > I'll reboot in a bit and wait longer and see what happens...
>> > >
>> > > grep -C3 nouveau on syslog for, I believe, one of the failed boots says:
>> > > Aug 21 02:28:06 struth kernel: [ 2.209185] TSC found unstable after
>> > > boot, most likely due to broken BIOS. Use 'tsc=unstable'.
>> > > Aug 21 02:28:06 struth kernel: [ 2.209187] sched_clock: Marking
>> > > unstable (2208942206, 241638)<-(2215089173, -5905242)
>> > > Aug 21 02:28:06 struth kernel: [ 2.221262] clocksource: Switched to
>> > > clocksource hpet
>> > > Aug 21 02:28:06 struth kernel: [ 2.305362] nouveau 0000:01:00.0:
>> > > enabling device (0002 -> 0003)
>> > > Aug 21 02:28:06 struth kernel: [ 2.305517] checking generic
>> > > (c0030000 710000) vs hw (d2000000 1000000)
>> > > Aug 21 02:28:06 struth kernel: [ 2.305519] checking generic
>> > > (c0030000 710000) vs hw (c0000000 10000000)
>> > > Aug 21 02:28:06 struth kernel: [ 2.305521] fb0: switching to
>> > > nouveaufb from EFI VGA
>> > > Aug 21 02:28:06 struth kernel: [ 2.305686] Console: switching to
>> > > colour dummy device 80x25
>> > > Aug 21 02:28:06 struth kernel: [ 2.305783] nouveau 0000:01:00.0:
>> > > vgaarb: deactivate vga console
>> > > Aug 21 02:28:06 struth kernel: [ 2.305843] nouveau 0000:01:00.0:
>> > > NVIDIA G84 (084700a2)
>> > > Aug 21 02:28:06 struth kernel: [ 2.306955] nouveau 0000:01:00.0:
>> > > Invalid PCI ROM header signature: expecting 0xaa55, got 0x0001
>> > > Aug 21 02:28:06 struth kernel: [ 2.306966] nouveau 0000:01:00.0:
>> > > bios: unable to locate usable image
>> > > Aug 21 02:28:06 struth kernel: [ 2.306975] nouveau 0000:01:00.0:
>> > > bios ctor failed, -22
>> > > Aug 21 02:28:06 struth kernel: [ 2.306984] nouveau: probe of
>> > > 0000:01:00.0 failed with error -22
>> > > Aug 21 02:28:06 struth kernel: [ 2.324639] random: fast init done
>> > > Aug 21 02:28:06 struth kernel: [ 2.385518] ata1.00: ATAPI: HL-DT-ST
>> > > DVDRW GSA-S10N, AP09, max UDMA/33
>> > > Aug 21 02:28:06 struth kernel: [ 2.457249] usb 4-1: new high-speed
>> > > USB device number 2 using ehci-pci
>> > > --
>> > >
>> > > On Sat, Aug 21, 2021 at 4:03 AM Salvatore Bonaccorso <carnil@debian.org> wrote:
>> > > >
>> > > > Control: tags -1 + moreinfo
>> > > >
>> > > > Hi Rich,
>> > > >
>> > > > On Sat, Aug 21, 2021 at 02:45:26AM -0400, Rich Ercolani wrote:
>> > > > > Package: src:linux
>> > > > > Version: 5.10.46-4
>> > > > > Severity: normal
>> > > > > X-Debbugs-Cc: rincebrain@gmail.com
>> > > > >
>> > > > > Dear Maintainer,
>> > > > >
>> > > > > (I have no idea how many people might run into this, but...)
>> > > > >
>> > > > > I just had occasion to set up bullseye on an old Macbook Pro. After
>> > > > > convincing it to boot the installer
>> > > > > in the first place (which was "exciting"), the installer worked
>> > > > > fine...
>> > > > >
>> > > > > ...then on first boot, I selected the top GRUB option, it booted,
>> > > > > and very shortly the display blanked and the system stopped
>> > > > > responding to any input that I could see.
>> > > > >
>> > > > > I left it for a minute or two to see if it would come back, but no,
>> > > > > nothing visibly occurred. So I rebooted.
>> > > > >
>> > > > > Drop the quiet from the kernel parameters, it goes up to something
>> > > > > mentioning nouveau and switching from EFI VGA to modesetting, and
>> > > > > then...stops, entirely, no response to VT switching or any other
>> > > > > commands, for at least the few minutes I waited. (I'll reboot and
>> > > > > take a photo shortly.)
>> > > > >
>> > > > > So I try nomodeset, and we get all the way to X happily (in much
>> > > > > less time than I let it sit to see if it would come back, so I don't
>> > > > > think it would have come back...)
>> > > > >
>> > > > > Slightly inconvenient. Fortunate that I didn't try the graphical
>> > > > > installer, I suppose.
>> > > > >
>> > > > > I'll try fiddling with kdump and non-journald logging so I can get a
>> > > > > better idea what might be happening when it stops outputting...
>> > > >
>> > > > #991941 and https://www.debian.org/releases/bullseye/amd64/ch06s04#completing-installed-system
>> > > > Do you have needed firmware installed, which installer image did you
>> > > > use?
>> > > >
>> > > > Regards,
>> > > > Salvatore
Reply to: