Bug 1913240 - kernel-5.10.4 installation errors; external displays stay black
Summary: kernel-5.10.4 installation errors; external displays stay black
Keywords:
Status: CLOSED WORKSFORME
Alias: None
Product: Fedora
Classification: Fedora
Component: kernel
Version: 33
Hardware: x86_64
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Kernel Maintainer List
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2021-01-06 10:51 UTC by Freddy Willemsen
Modified: 2021-01-13 06:40 UTC (History)
20 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2021-01-12 11:00:47 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)
journalctl --no-hostname -k > dmesg.txt (95.37 KB, text/plain)
2021-01-06 10:51 UTC, Freddy Willemsen
no flags Details

Description Freddy Willemsen 2021-01-06 10:51:40 UTC
Created attachment 1744863 [details]
journalctl --no-hostname -k > dmesg.txt

1. Please describe the problem:
For the kernel 5.10 test days I enabled the kernel stabilization repo. The provided kernel 5.10.4 installs, with errors however:

2021-01-06T05:06:45+0100 INFO sort: fflush failed: 'standard output': Broken pipe
sort: write error

gzip: stdout: Broken pipe

gzip: stdout: Broken pipe
sort: write failed: 'standard output': Broken pipe
sort: write error

INFO Not setting net/ipv4/conf/all/rp_filter (explicit setting exists).
Not setting net/ipv4/conf/default/rp_filter (explicit setting exists).
Not setting net/ipv4/conf/all/accept_source_route (explicit setting exists).
Not setting net/ipv4/conf/default/accept_source_route (explicit setting exists).
Not setting net/ipv4/conf/all/promote_secondaries (explicit setting exists).
Not setting net/ipv4/conf/default/promote_secondaries (explicit setting exists).
Couldn't write '2' to 'net/ipv4/tcp_frto_response', ignoring: No such file or directory

Then after a reboot, I get to the gdm login page but after that everything just stays blank. This only happens when connected to a docking station using dual QHD monitors. Using only the internal (FHD) display everything works. As soon as I dock agian, all goes blank and drm:drm_dp_mst_dpcd_read errors appear in the journal. For the record, I am using xorg (not wayland).

I was able to complete the kernel test days scripts for kernel-5.10.4 using a VirtualBox image.

2. What is the Version-Release number of the kernel:
kernel-5.10.4-200.fc33.x86_64

3. Did it work previously in Fedora? If so, what kernel version did the issue
   *first* appear?  Old kernels are available for download at
   https://koji.fedoraproject.org/koji/packageinfo?packageID=8 :
Up to kernel-5.10.3-200.fc33.x86_64 it works fine.

4. Can you reproduce this issue? If so, please provide the steps to reproduce
   the issue below:
Install kernel-5.10.4 and reboot. 

5. Does this problem occur with the latest Rawhide kernel? To install the
   Rawhide kernel, run ``sudo dnf install fedora-repos-rawhide`` followed by
   ``sudo dnf update --enablerepo=rawhide kernel``:

6. Are you running any modules that not shipped with directly Fedora's kernel?:
Sophos-av

7. Please attach the kernel logs. You can get the complete kernel log
   for a boot with ``journalctl --no-hostname -k > dmesg.txt``. If the
   issue occurred on a previous boot, use the journalctl ``-b`` flag.

Comment 1 Freddy Willemsen 2021-01-07 18:23:31 UTC
Just tried kernel-5.10.5-200.fc33.x86_64, it exhibits the exact same behaviour. "fflush failed: 'standard output': Broken pipe" during installation, blank screens after a reboot. Back to kernel-5.10.3-200 for now.

Comment 2 Freddy Willemsen 2021-01-11 23:35:25 UTC
Some good news, I tried the kernel-5.10.6-200.fc33 and although the "fflush failed: 'standard output': Broken pipe" message remains during installation, the screens no longer stay black but seem to be working properly now. If things continue to work out I will close this ticket some time tomorrow.

Comment 3 Freddy Willemsen 2021-01-12 11:00:47 UTC
Intrigued by what's going on with the postinstall scriptlet error I did some digging. The error originates from the /usr/lib/kernel/install.d/50-dracut.install script. More specifically, the 'dracut -f' line. So I added the debug flag to the script and reinstalled kernel-5.10.6-200.fc33.  


/bin/dracut@1984(): CPIO_REPRODUCIBLE=1
/bin/dracut@1990(): [[ 0 != 0 ]]
/bin/dracut@1992(): [[ yes = yes ]]
/bin/dracut@1993(): echo 1
/bin/dracut@1995(): [[ -n 1 ]]
/bin/dracut@1996(): find /var/tmp/dracut.IT9NAg/earlycpio/d -newer /usr/lib/dracut/dracut-functions.sh -print0
/bin/dracut@1997(): xargs -r -0 touch -h -m -c -r /usr/lib/dracut/dracut-functions.sh
/bin/dracut@2002(): umask 077
/bin/dracut@2002(): cd /var/tmp/dracut.IT9NAg/earlycpio/d
/bin/dracut@2003(): find . -print0
/bin/dracut@2003(): sort -z
/bin/dracut@2004(): cpio --reproducible --null -H newc -o --quiet
/bin/dracut@2012(): umask 077
/bin/dracut@2012(): cd /var/tmp/dracut.IT9NAg/initramfs
/bin/dracut@2013(): find . -print0
/bin/dracut@2013(): sort -z
/bin/dracut@2014(): cpio --reproducible --null -H newc -o --quiet
/bin/dracut@2015(): pigz -9 -n -T -R
/bin/dracut@2021(): ((  maxloglvl >= 5  ))
/bin/dracut@2021(): ((  verbosity_mod_l >= 0  ))
/bin/dracut@2022(): [[ -n '' ]]
/bin/dracut@2025(): lsinitrd /var/tmp/dracut.IT9NAg/initramfs.img
/bin/dracut@2025(): ddebug
/lib/kdump/kdump-logger.sh@305(ddebug): set +x
sort: fflush failed: 'standard output': Broken pipe
sort: write error

gzip: stdout: Broken pipe

gzip: stdout: Broken pipe
sort: write failed: 'standard output': Broken pipe
sort: write error
/bin/dracut@2029(): umask 077
/bin/dracut@2031(): [[ '' = yes ]]
/bin/dracut@2080(): cp --reflink=auto /var/tmp/dracut.IT9NAg/initramfs.img /boot/initramfs-5.10.6-200.fc33.x86_64.img
/bin/dracut@2081(): dinfo '*** Creating initramfs image file '\''/boot/initramfs-5.10.6-200.fc33.x86_64.img'\'' done ***'
/lib/kdump/kdump-logger.sh@315(dinfo): set +x

Looking at the part where the error occurs in dracut:

if (( maxloglvl >= 5 )) && (( verbosity_mod_l >= 0 )); then
    if [[ $allowlocal ]]; then
        "$dracutbasedir/lsinitrd.sh" "${DRACUT_TMPDIR}/initramfs.img"| ddebug
    else
        lsinitrd "${DRACUT_TMPDIR}/initramfs.img"| ddebug
    fi
fi

Not sure where the ddebug part comes from but that one seems to be the culprit. As you can see however, the initramfs is succesfully built, so the generated error actually seems to be irrelevant. Will close this ticket now as the kernel works well.

Comment 4 Thorsten Leemhuis 2021-01-13 06:40:26 UTC
(In reply to Freddy Willemsen from comment #3)
> Intrigued by what's going on with the postinstall scriptlet error I did some
> digging. 

FWIW, see also Bug 1911038


Note You need to log in before you can comment on or make changes to this bug.