Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Installation problems with Qubes 4.0.2 final #5553

Open
linuxfan1000 opened this issue Jan 3, 2020 · 12 comments
Open

Installation problems with Qubes 4.0.2 final #5553

linuxfan1000 opened this issue Jan 3, 2020 · 12 comments

Comments

@linuxfan1000
Copy link

@linuxfan1000 linuxfan1000 commented Jan 3, 2020

Qubes OS version
I am using Qubes 4.0.2 final and downloaded the ISO file immediately after release, yesterday.

Affected component(s) or functionality
Installer.

Brief summary
I used dd to write the image to an USB stick and verified the installation media. Qubes 4.0.2 installs fine until the last step when one has to choose the installation setup and check the boxes (create default vault, personal, etc; use Tor for updates) or "quit". After configuration of each template it reboots, one has to enter the pre-boot password in non-graphical mode with the error message "Failed to start Load Kernel Modules" and then it brings one back to that checkbox form configuration page. Then it will install the next template and reboot and the same process follows. The final time when I enter the password it boots and then I get something that looks like a kernel panic.

The laptop (Acer Aspire V) is fine. I have the latest BIOS firmware. I checked the memory and SSD. I have been using it with Qubes 3.1 until 4.0.1 without problems. I installed in BIOS Legacy mode.

To Reproduce
Install Qubes 4.0.2 final from USB Stick.

Expected behavior
Start install then log in normally after finished installation.

Actual behavior
Start install, few reboots with pre-boot password authorization and multiple post-config screens, then finally kernel panic(?).

Screenshots
At the very end I get a very similar screen of what GaijinSystems posted below.

Solutions you've tried
Downloaded ISO again. Same problem.
Used a different USB stick. Same problem.
Upgraded BIOS to latest version. Same problem.
Tried to install in UEFI mode. Same problem.

I also tried adding the options mapbs=1 and noexitboot=1 to the xen.cfg without success:
https://www.qubes-os.org/doc/uefi-troubleshooting/#installation-completes-successfully-but-then-boot-loops-or-hangs-on-black-screen

Relevant documentation you've consulted
https://www.qubes-os.org/doc/installation-guide/

@GaijinSystems

This comment has been minimized.

Copy link

@GaijinSystems GaijinSystems commented Jan 3, 2020

I'm experiencing the same problem with a thinkpad T450s. I did not have this issue when installing 4.0.1.

My apologies for this being an image, but when switched to tty1 to see kernel/console output, it looks like a kernel panic with blk-core.c

IMG_20200102_144757

@pure007

This comment has been minimized.

Copy link

@pure007 pure007 commented Jan 3, 2020

I can confirm this bug on my Inspiron 5379 after installing 4.0.2

@marmarek marmarek added this to the Release 4.0 updates milestone Jan 3, 2020
@linuxfan1000

This comment has been minimized.

Copy link
Author

@linuxfan1000 linuxfan1000 commented Jan 4, 2020

I have updated the original post with more information.

I have another question. Is it recommended to install in Legacy BIOS mode or UEFI?

@linuxfan1000 linuxfan1000 changed the title Installation problems with 4.0.2 Installation problems with Qubes 4.0.2 final Jan 4, 2020
@mfp20

This comment has been minimized.

Copy link

@mfp20 mfp20 commented Jan 7, 2020

Similar problem here with threadripper on x399 mobo. I installed in UEFI mode, had to comment out mapbs and noexitboot in the USB installer boot cfg but I didn't in the installed system. I tried to do the post-install finalization procedure 2 times, and two times I got a sudden reboot. Third time I just ticked "do not configure anything" and continued to the installed system, but without the default VMs is hard for me to figure out what to do to manually configure them.

EDIT1: I repeated the install process from the beginning (wipefs and reboot to the installer), and on reboot I got a random reboot during the x11 initialization (before the post-install finalization prompt).

EDIT2: I repeated (again) the install process from the beginning. This time I added "efi=attr=uc" option in xen.cfg as stated in the UEFI troubleshooting page and don't know if it worked because the debian templatevm install bar stopped moving, so after a couple of minutes I tried to switch to a console (alt+ctrl+Fx) and as soon as I arrived to F7... I had been able to read the first line only (something about "kernel BUG"), it looked like a kernel panic and xen triggered the watchdog to reboot dom0.

EDIT3: After a new fresh install, I tried to disable SMT in BIOS (because I noticed it is disable in xen.cfg) and reduce mem_max to 2048M (because it did the trick for xcp-next install). But no luck. Same behaviour.

EDIT4: I tried adding the mapbs and noexitboot options, one at a time, with no luck. Mapbs simply hangs the computer with black screen (same of the installer, I had to remove those in order to see the installer welcome screen). Noexitboot makes no difference at all. At this point I'm short of options. The numa arch of this machine might be the issue, but I've no idea how to troubleshoot it.

EDIT5: After adding noreboot boot option I've been able to read the kernel panic message. The first line asserts some kind of kernel bug in blk-core.c:1779. The second line complaints about invalid opcode 0000 SMP NOPTI. I also tried to disable memory interleaving (NUMA) in the mobo BIOS, in order to have a more traditional UMA topology, but didn't change anything. Still: it may be a NUMA related issue, as the mobo connects the network devices to one numa node and the nvme block devices to the other node; but I don't know if I can disable numa for I/O as well, as I did with memory. It looks like it fails on "blk put" function... block device write issue? The first stage of the install process is massive on block device and have no bug! I also run badblocks to check for block device problems and ... the block device is fully functional.

@marmarek

This comment has been minimized.

Copy link
Member

@marmarek marmarek commented Jan 7, 2020

Yes, it's a known issue in the dom0 kernel, that's also why 4.0.2 is taken down from the download page. For now you can use 4.0.2-rc3.

@mfp20

This comment has been minimized.

Copy link

@mfp20 mfp20 commented Jan 7, 2020

Yes, it's a known issue in the dom0 kernel, that's also why 4.0.2 is taken down from the download page. For now you can use 4.0.2-rc3.

I can confirm 4.0.2-rc3 works and ... is awesome!!! I've been testing QubesOS about once per year since the first public release and ... this is the first time I've the feeling I can replace my current distro with QubesOS. All the operations I tried went smooth, flawless. Even glxgears works! Probably steam games (proton) won't be so easy to run but ... it's a start.

Congratulation guys, it's really a good job. PLEASE, keep going.

@linuxfan1000

This comment has been minimized.

Copy link
Author

@linuxfan1000 linuxfan1000 commented Jan 8, 2020

@marmarek: Thx a lot for bringing back back the 4.0.2 RC version. I will try to install the RC version on the weekend.

Maybe the news section of the homepage should also be updated ASAP, so users are not confused and look for the final version 0f 4.0.2.

Thx for the great work. I love Qubes. It is the best OS ever!

andrewdavidwong added a commit to QubesOS/qubes-posts that referenced this issue Jan 8, 2020
@andrewdavidwong

This comment has been minimized.

Copy link
Member

@andrewdavidwong andrewdavidwong commented Jan 8, 2020

Maybe the news section of the homepage should also be updated ASAP, so users are not confused and look for the final version 0f 4.0.2.

Thank you for the suggestion. I've just added an update to the 4.0.2 announcement post.

@andrewdavidwong andrewdavidwong pinned this issue Jan 8, 2020
@linuxfan1000

This comment has been minimized.

Copy link
Author

@linuxfan1000 linuxfan1000 commented Jan 12, 2020

Maybe the news section of the homepage should also be updated ASAP, so users are not confused and look for the final version 0f 4.0.2.

Thank you for the suggestion. I've just added an update to the 4.0.2 announcement post.

Hi David

I just wanted to note that in the news section of the web site it still says that the new version 4.0.2 is out. I think it would be better to also remove that so not to confuse anybody.

https://www.qubes-os.org/news/

Thx for the great work. I love Qubes OS.

@andrewdavidwong

This comment has been minimized.

Copy link
Member

@andrewdavidwong andrewdavidwong commented Jan 12, 2020

Maybe the news section of the homepage should also be updated ASAP, so users are not confused and look for the final version 0f 4.0.2.

Thank you for the suggestion. I've just added an update to the 4.0.2 announcement post.

Hi David

I just wanted to note that in the news section of the web site it still says that the new version 4.0.2 is out. I think it would be better to also remove that so not to confuse anybody.

https://www.qubes-os.org/news/

Thx for the great work. I love Qubes OS.

Yes, I'm aware. That's because the News page contains a list of all the News posts, and the original 4.0.2 announcement post has not been taken down. We never take down a post after publishing it (we only add updates within the body of the post), because links to the post are immediately shared on mailing lists, social media websites, and elsewhere across the Web immediately after publication. Taking down the post would break all of those links, which would prevent people from seeing the update in the body of the post to learn what is going on. How about this: I'll add [Delayed] in front of the title, so that it's easier to see something's up before viewing the body of the post.

andrewdavidwong added a commit to QubesOS/qubes-posts that referenced this issue Jan 12, 2020
marmarek added a commit to QubesOS/qubesos.github.io that referenced this issue Jan 12, 2020
_posts:
    gpg: Good signature from "Andrew David Wong (Qubes Documentation Signing Key)" [ultimate]
    object 018cea3924c0b77de4a4a067a8c9ff2555b3a702
    type commit
    tag adw_018cea39
    tagger Andrew David Wong <adw@andrewdavidwong.com> 1578871693 -0600

    Tag for commit 018cea3924c0b77de4a4a067a8c9ff2555b3a702

    018cea3 Add "[Delayed]" to 4.0.2 post title (QubesOS/qubes-issues#5553)
@marmarek

This comment has been minimized.

Copy link
Member

@marmarek marmarek commented Jan 14, 2020

Updated installation image, soon to be published as R4.0.3-rc1:
https://openqa.qubes-os.org/tests/5506/asset/iso/Qubes-R4.0.3-rc1pre1-x86_64.iso

@GWeck

This comment has been minimized.

Copy link

@GWeck GWeck commented Jan 19, 2020

The installation process for this iso still has the problem that the second phase of the installation is aborted if manual partitioning is selected and the partition sizes are given in GB instead of in GiB.

See #5003 for a detailed description of this situation.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
7 participants
You can’t perform that action at this time.