Arch Linux

JimNeedsHelp

Member

Registered: 2020-12-30

Posts: 2

[SOLVED] Error shown at login promt: nvidia, ucsi_ccg and more

When I get my login prompt after the boot process instead of just "arch login: " it says

"arch login: [    3.520017] nvidia-gpu 0000:26:00.3: i2c timeout error e0000000
[    3.520026] ucsi_ccg 0-0008: i2c_transfer failed -110
[    3.520032] ucsi_ccg 0-0008: ucsi_ccg_init failed - -110
[    3.544793] sd 8:0:0:0: [sdb] No Caching mode page found
[    3.544798] sd 8:0:0:0: [sdb] Assuming drive cache: write through 

Although there is this message I can log in and use Xorg.
I have installed arch two times. Both times and when booting the USB flash installation medium these errors happend but there are some differences and some additional errors. The first time i used EFISTUB directly with efibootmgr, the nouveau driver and the mainboard firmware that was already installed. The second time I used systemd-boot, the nvidia driver and updated the firmware for the mainboard (current install). Both times i used /boot as ESP-mountpoint and implemented amd microcode updates. More information later.

My PC components

• Mainboard: B450 TOMAHAWK MAX (MS-7C02)

• CPU: AMD Ryzen 5 3600

• GPU: Nvidia RTX 2070 Super

Some Error Output:

systemctl --failed
  UNIT LOAD ACTIVE SUB DESCRIPTION
0 loaded units listed.

journalctl -p 3 -xb
Dec 30 23:43:39 arch kernel: __common_interrupt: 1.55 No irq handler for vector
Dec 30 23:43:39 arch kernel: __common_interrupt: 2.55 No irq handler for vector
Dec 30 23:43:39 arch kernel: __common_interrupt: 3.55 No irq handler for vector
Dec 30 23:43:39 arch kernel: __common_interrupt: 4.55 No irq handler for vector
Dec 30 23:43:39 arch kernel: __common_interrupt: 5.55 No irq handler for vector
Dec 30 23:43:39 arch kernel: __common_interrupt: 6.55 No irq handler for vector
Dec 30 23:43:39 arch kernel: __common_interrupt: 7.55 No irq handler for vector
Dec 30 23:43:39 arch kernel: __common_interrupt: 8.55 No irq handler for vector
Dec 30 23:43:39 arch kernel: __common_interrupt: 9.55 No irq handler for vector
Dec 30 23:43:39 arch kernel: __common_interrupt: 10.55 No irq handler for vector
Dec 30 23:43:39 arch kernel: sp5100-tco sp5100-tco: Watchdog hardware is disabled
Dec 30 23:43:40 arch kernel: nvidia-gpu 0000:26:00.3: i2c timeout error e0000000
Dec 30 23:43:40 arch kernel: ucsi_ccg 3-0008: i2c_transfer failed -110
Dec 30 23:43:40 arch kernel: ucsi_ccg 3-0008: ucsi_ccg_init failed - -110
Dec 30 23:43:40 arch kernel: sd 8:0:0:0: [sdb] No Caching mode page found
Dec 30 23:43:40 arch kernel: sd 8:0:0:0: [sdb] Assuming drive cache: write through 

dmesg
(of course these are just some interesting snippets)
[    0.156500] ACPI: [Firmware Bug]: BIOS _OSI(Linux) query ignored
[    0.168738] acpi PNP0A08:00: [Firmware Info]: MMCONFIG for domain 0000 [bus 00-7f] only partially covers this bridge
[    2.206173] acpi_cpufreq: overriding BIOS provided _PSD data
[    2.286362] ccp 0000:28:00.1: ccp: unable to access the device: you might be running a broken BIOS.
[    3.374946] ucsi_ccg: probe of 3-0008 failed with error -110 

With my first install sometimes the system randomly froze during the startup of Arch. Also, the messages after the login prompt weren't there sometimes. There were no freezes with this current install (yet) but since there still were these messages I thought there still is a or even the same problem.
At first I thought it has something to do with the GPU drivers. There were actually some additional errors with nouveau that I could see with dmesg and I also got an error like "nouveau ... pmu:firmware unavailable" when booting the USB flash installation medium. I switched to nvidia drivers when It still was my first install but there were some differences to the current install. I think these "__common_interrupt: No irq handler for vector" were not there (really not sure) and I think the last two error messages that are after the login prompt "sd 8:0...: were also not there yet. Something else that is different is that with my current install at the end of the boot process is this: first there is nothing behind "Arch login:" , but right after that is shown there is a black screen for half a second and then the login prompt is back with the error messages. So that black screen is new. Also with systemd-boot there are two boot entries in my UEFI boot menu now while there was only one with efibootmgr, but that's probably the firmware's fault. (And no it's not that one of them is from the first install.) Also when I enable the timeout option in the systemd-boot config the resolution is lower between selecting Arch and the login prompt for some reason. But those two things don't bother me as much.
I read that there can be some problems with some hardware using EFISTUB so I might try GRUB.
The only other thing I could think of is adding some kernel parameters or maybe switching to linux-lts and nvidia-lts?
Currently the kernel parameters are default + amd microcode.

(rant)But usually the solutions to something nvidia related bring some disadvantages with them. I think most or all of my problems are Nvidia related and they are not fun. I also have some screen tearing and audio problems... I really want to switch to an AMD GPU but that's not easy/cheap because of the GPU situation. Would the problems be the same on other Distros because they use the same driver and bootloader and kernel or should I maybe try another one until I have another GPU if there is no good fix to my problem?

Thanks

Last edited by JimNeedsHelp (2020-12-31 02:14:22)

V1del

Forum Moderator

Registered: 2012-10-16

Posts: 23,134

Re: [SOLVED] Error shown at login promt: nvidia, ucsi_ccg and more

Do you have an actual issue or are you just wondering about the messages? Watchdog HW and the interrupt flood are AMD CPU related, the i2c  timeout seems somewhat normal but I don't see any ill-effects on my setup. the sdb messages aren't errors. The flicker is normal during the modesetting switch, can happen on every GPU and driver, might be alleviated with the nvidia-drm.modeset=1 kernel parameter. The ACPI errors are normal on nearly every BIOS, you seem to get off lucky that only the linux query is ignored (...they are generally unlikely to be relevant, but this one is almost guaranteed), ccp is a AMD CPU/BIOS feature.

Most of those can only be "fixed" with a UEFI update, but they are also completely harmless to normal system usage.

FWIW a variety of errors can happen for all sorts of reasons (racy behaviour due to multi-threaded startup leading to a few things being initalized "too soon" e.g.) and they will be logged regardless of your HW setup they might recover immediately after the error messages or be not actually relevant to how you are intending to use your system. If this annoys you set a lower log level (... with the caveat that you might miss actual issues) or learn to filter and ignore issues that aren't relevant in your head.

Last edited by V1del (2020-12-31 01:33:54)

JimNeedsHelp

Member

Registered: 2020-12-30

Posts: 2

Re: [SOLVED] Error shown at login promt: nvidia, ucsi_ccg and more

Thanks V1del, I thought that these log messages are abnormal. I forgot to mention an important reason though that made me think there is a bigger problem with these messages: With my first install sometimes the system randomly froze during the startup of Arch. I will edit that into my post. Also, the messages after the login prompt weren't there sometimes. There were no freezes with this install (yet) but since there still were these messages I thought there still is a or even the same problem. I will try the kernel parameter tomorrow and ignore these messages after the login prompt now.

Last edited by JimNeedsHelp (2020-12-31 01:49:10)