Hello Marin,
Will there be a new install ISO image with the new kernel backed in?
cheers Conny
Kernel 5.11
- Thread starter martin
- Start date
Will there be a new install ISO image with the new kernel backed in?
cheers Conny
Hi guys, if someone want's to build the Kernel yourself, i made a small 08/15 Helper Script.
You can get it from here: (kernel-update.sh)
https://cloud.golima.de/s/E7KiWLFgzSp4MNP
Or here as Code:
It's mostly self explanatory, i will try to update it with time!
Note: Please take care, that the "build" dependencies are mostly Debian related, but Ubuntus package names should be the same.
This script won't work with RHEL or Arch based Distros! Only with apt based distros like debian/ubuntu/mint/etc.
Cheers xD
EDIT:
v1.1: Added Build Flags, PVE_BUILD_CC and PVE_BUILD_CFLAGS are now functional, sorry xD
You can get it from here: (kernel-update.sh)
https://cloud.golima.de/s/E7KiWLFgzSp4MNP
Or here as Code:
Bash:
#!/bin/bash
#
# Config:
#
# Check Building dependencies? 1=yes / 0=no
# - This is a check only, nothing will get installed by this script!
KUPD_CHECK_DEPENDENCIES=1
KUPD_BUILD_DEPENDENCIES=(gcc devscripts asciidoc-base automake bc bison cpio dh-python flex git kmod libdw-dev libelf-dev libiberty-dev libnuma-dev libpve-common-perl libslang2-dev libssl-dev libtool lintian lz4 perl-modules python2-minimal rsync sed sphinx-common tar xmlto zlib1g-dev dwarves systemtap-sdt-dev libunwind-dev libzstd-dev libcap-dev libbabeltrace-dev libpython3.9-dev libperl-dev binutils-dev)
#
# Update to another Kernel as PVE provides? 1=yes / 0=no
# - If you don't need a Branch, leave empty: KUPD_KERNEL_BRANCH=""
KUPD_REPLACE_KREPO=1
KUPD_KERNEL_REPO="git://kernel.ubuntu.com/ubuntu/ubuntu-hirsute.git"
KUPD_KERNEL_BRANCH="master-next"
#
KUPD_PVE_REPO="git://git.proxmox.com/git/pve-kernel.git"
#
# Do you want to define Cflags & Compiler? PVE_BUILD_CFLAGS & PVE_BUILD_CC
# - If you don't enable this, making with PVE_BUILD_CFLAGS & PVE_BUILD_CC wont make any effect.
KUPD_ENABLE_BUILD_FLAGS=1
#
# Changelog:
# Version 1.1: Added Build Flags, forgot about those before.
#
# --- Script Section, do not modify ----
C_RED='\e[0;31m'
C_GREEN='\e[0;32m'
C_NC='\e[0m'
git clone $KUPD_PVE_REPO
cd pve-kernel
if [ "$KUPD_REPLACE_KREPO" -eq "1" ]; then
git submodule set-url submodules/ubuntu-hirsute $KUPD_KERNEL_REPO
if [ ! -z "$KUPD_KERNEL_BRANCH" ]; then
git submodule set-branch --branch $KUPD_KERNEL_BRANCH submodules/ubuntu-hirsute
fi
echo -e "${C_GREEN}INFO:${C_NC} Using Custom Kernel Repository!"
fi
make submodule
if [ "$KUPD_REPLACE_KREPO" -eq "1" ]; then
git submodule update --recursive --remote
KUPD_PLVL_AKRN=$(grep ^SUBLEVEL submodules/ubuntu-hirsute/Makefile)
KUPD_PLVL_AREP=$(grep ^KERNEL_PATCHLEVEL Makefile)
KUPD_PLVL_VKRN=$(echo $KUPD_PLVL_AKRN | cut -d = -f 2 | xargs)
KUPD_PLVL_VREP=$(echo $KUPD_PLVL_AREP | cut -d = -f 2 | xargs)
KUPD_PLVL_NKRN=$(echo $KUPD_PLVL_AKRN | cut -d = -f 1 | xargs)
KUPD_PLVL_NREP=$(echo $KUPD_PLVL_AREP | cut -d = -f 1 | xargs)
KUPD_PLVL_RKRN="$KUPD_PLVL_NREP=$KUPD_PLVL_VKRN"
if [ "$KUPD_PLVL_VREP" != "$KUPD_PLVL_VKRN" ]; then
sed -i "s/^$KUPD_PLVL_AREP/$KUPD_PLVL_RKRN/" Makefile
echo -e "${C_GREEN}INFO:${C_NC} Makefile updated to 5.11.${C_GREEN}$KUPD_PLVL_VKRN${C_NC}"
fi
fi
KUPD_B_CHECK() {
return $(dpkg-query -W -f '${Status}\n' "${1}" 2>&1|awk '/ok installed/{print 0;exit}{print 1}')
}
if [ "$KUPD_CHECK_DEPENDENCIES" -eq "1" ]; then
KUPD_BUILD_M=""
for KUPD_PKG in ${KUPD_BUILD_DEPENDENCIES[@]}; do
if ! $(KUPD_B_CHECK $KUPD_PKG) ; then
KUPD_BUILD_M+=" $KUPD_PKG"
fi
done
if [[ ${KUPD_BUILD_M} != *"dwarves"* ]]; then
KUPD_DRWAVES_C="$(dpkg-query -W -f="\${Version}\n" dwarves 2>/dev/null)"
if [ "$(printf '%s\n' "1.16" "$KUPD_DRWAVES_C" | sort -V | head -n1)" = "1.16" ]; then
echo -e "${C_GREEN}INFO:${C_NC} dwarves >= 1.16.0, good!"
else
echo -e "${C_RED}WARNING:${C_NC} Please update dwarves >= 1.16.0"
fi
fi
if [ ! -z "$KUPD_BUILD_M" ]; then
echo -e "${C_RED}WARNING:${C_NC} Missing$KUPD_BUILD_M"
echo -e "${C_RED}WARNING:${C_NC} Build will probably fail!"
fi
fi
if [ "$KUPD_ENABLE_BUILD_FLAGS" -eq "1" ]; then
# Add default Compiler if nothing is set:
sed -i '/^PVE_CONFIG_OPTS=.*/i PVE_BUILD_CC ?= \${CC}\n' debian/rules
# Add Flags to .config_mark:
sed -i "s/\${MAKE} -C \${KERNEL_SRC} oldconfig/\${MAKE} -C \${KERNEL_SRC} CC=\${PVE_BUILD_CC} oldconfig/" debian/rules
# Add Flags to .compile_mark:
sed -i "s/\${MAKE} -C \${KERNEL_SRC} KBUILD_BUILD_VERSION_TIMESTAMP=\"PVE \${DEB_VERSION} (\${CHANGELOG_DATE})\"/\${MAKE} -C \${KERNEL_SRC} CC=\${PVE_BUILD_CC} KCFLAGS=\"\${PVE_BUILD_CFLAGS}\" KBUILD_BUILD_VERSION_TIMESTAMP=\"PVE \${DEB_VERSION} (\${CHANGELOG_DATE})\"/" debian/rules
echo -e "${C_GREEN}INFO:${C_NC} Added build Flag options to debian/rules"
fi
echo ""
echo -e "${C_GREEN}DONE:${C_NC} You can build the Kernel with:"
echo -e "${C_GREEN}NOTE:${C_NC} make PVE_BUILD_CC=gcc-11 PVE_BUILD_CFLAGS=\"-O3 -march=znver3\""
echo -e "${C_GREEN}NOTE:${C_NC} PVE_BUILD_CC and PVE_BUILD_CFLAGS are optional."
echo -e "${C_GREEN}NOTE:${C_NC} For -march please read: https://gcc.gnu.org/onlinedocs/gcc/x86-Options.html"It's mostly self explanatory, i will try to update it with time!
Note: Please take care, that the "build" dependencies are mostly Debian related, but Ubuntus package names should be the same.
This script won't work with RHEL or Arch based Distros! Only with apt based distros like debian/ubuntu/mint/etc.
Cheers xD
EDIT:
v1.1: Added Build Flags, PVE_BUILD_CC and PVE_BUILD_CFLAGS are now functional, sorry xD
Last edited:
Thank you that worked.
For anyone who is interested:
1) add
Code:
deb http://deb.debian.org/debian buster-backports main
Code:
/etc/apt/sources.list
Code:
apt update
Code:
apt install -t buster-backports btrfs-progsEnabling pvetest repo is not needed anymore. I am using the default deb https://enterprise.proxmox.com/debian/pve buster pve-enterprise and just installed it via apt-get install pve-kernel-5.11
Awesome: Ryzen 5000 CPU temperature monitoring is working now with Proxmox!
Bash:
~# uname -r
5.11.7-1-pve
~# sensors
...
k10temp-pci-00c3
Adapter: PCI adapter
Tctl: +63.0°C
Tdie: +63.0°C
Unfortunately, the kernel-headers for 5.11-pve appear to not be available in the subscription repository, which is a problem for vendor-reset and dkms.
EDIT1: Nevermind, it is just not automatically installed (by pve-kernel-5.11 and/or pve-headers) but it is available as pve-headers-5.11.7-1-pve.
EDIT2: I seem to run into issue 26 of vendor-reset with 5.11. Are the following kernel options still set to
EDIT1: Nevermind, it is just not automatically installed (by pve-kernel-5.11 and/or pve-headers) but it is available as pve-headers-5.11.7-1-pve.
EDIT2: I seem to run into issue 26 of vendor-reset with 5.11. Are the following kernel options still set to
y in 5.11.7-1-pve:CONFIG_KALLSYMS CONFIG_KALLSYMS_ALL CONFIG_KALLSYMS_ABSOLUTE_PERCPU CONFIG_KALLSYMS_BASE_RELATIVE CONFIG_KPROBES CONFIG_PCI_QUIRKS CONFIG_FTRACE?
Last edited:
Headers were never automatically installed, the
pve-headers packages still points at pve-headers-5.4 as that is still the default kernel.There's as always also the respective meta package to get the latest headers of a major release automatically:
apt install pve-headers-5.11 - available on all repositories.You can check the used config in
/boot/config-5.11.7-1-pve.
Bash:
$ grep -P 'CONFIG_(KALLSYMS|KPROBES|PCI_QUIRKS|FTRACE)' /boot/config-5.11.7-1-pve
CONFIG_KALLSYMS=y
CONFIG_KALLSYMS_ALL=y
CONFIG_KALLSYMS_ABSOLUTE_PERCPU=y
CONFIG_KALLSYMS_BASE_RELATIVE=y
CONFIG_KPROBES=y
CONFIG_KPROBES_ON_FTRACE=y
CONFIG_PCI_QUIRKS=y
CONFIG_FTRACE=y
CONFIG_FTRACE_SYSCALLS=y
CONFIG_FTRACE_MCOUNT_RECORD=y
# CONFIG_FTRACE_RECORD_RECURSION is not set
# CONFIG_FTRACE_STARTUP_TEST is not set
# CONFIG_KPROBES_SANITY_TEST is not setThank you for showing me how to install and check all this. The Proxmox kernel 5.11.7-1-pve does support all required kernel config flags, but the master-branch of vendor-reset that should support 5.11 still fails with
The dkms.conf for this module includes a BUILD_EXCLUSIVE directive which does not match this kernel/arch. . This is unfortunate but probably not the fault of the Proxmox kernel.
Hmm weird, their
Can you check the following commands:
My dkms knowledge is a bit rusty at the moment, but maybe it gets pinned locally, and you can just drop that pinning (the line with the
dkms.conf in the repo does not have a BUILD_EXCLUSIVE entry, and git log mentions fixing build for 5.11 already.Can you check the following commands:
Bash:
dkms status
grep -r BUILD_EXCLUSIVE_KERNEL /usr/src/My dkms knowledge is a bit rusty at the moment, but maybe it gets pinned locally, and you can just drop that pinning (the line with the
BUILD_EXCLUSIVE) from the respective local dkms.conf file.
vendor-reset, 0.1.1, 5.4.106-1-pve, x86_64: installed (WARNING! Diff between built and installed module!)
wireguard, 1.0.20210219, 4.19.0-16-amd64, x86_64: installed (WARNING! Diff between built and installed module!)
wireguard, 1.0.20210219, 5.4.106-1-pve, x86_64: installed/usr/src/wireguard-1.0.20210219/dkms.conf: BUILD_EXCLUSIVE_KERNEL="^((5\.[0-4]($|[.-]))|(4\.)|(3\.1[0-9]))"
/usr/src/wireguard-1.0.20210219/dkms.conf: BUILD_EXCLUSIVE_KERNEL="^((5\.[0-5]($|[.-]))|(4\.)|(3\.1[0-9]))"grep -R 'BUILD_EXCLUSIVE' does not show anything in the local vendor-reset directory.
Thanks for your help but after removing wireguard-dkms, I could not get vendor-reset to build and had some problems switching back to 5.4. I'll try all this in a VM with a fresh Proxmox 6, install 5.11 and then try vendor-reset to see if it is a problem with my setup or some kind of incompatibility.
EDIT: vendor-reset does not build with a freshly installed and dist-upgraded Proxmox 6.3 and kernel+headers 5.11.7-1-pve.
Bash:
make -C /lib/modules/5.11.7-1-pve/build M=/var/lib/dkms/vendor-reset/0.1.1/build modules
...
LD [M] /var/lib/dkms/vendor-reset/0.1.1/build/vendor-reset.o
MODPOST /var/lib/dkms/vendor-reset/0.1.1/build/Module.symvers
make[3]: *** No rule to make target 'scripts/module.lds', needed by '/var/lib/dkms/vendor-reset/0.1.1/build/vendor-reset.ko'. Stop.
make[3]: *** Waiting for unfinished jobs....
CC [M] /var/lib/dkms/vendor-reset/0.1.1/build/vendor-reset.mod.o
make[2]: *** [scripts/Makefile.modpost:117: __modpost] Error 2
make[1]: *** [Makefile:1738: modules] Error 2
make[1]: Leaving directory '/usr/src/linux-headers-5.11.7-1-pve'
make: *** [Makefile:8: build] Error 2
Last edited:
Code:
root@p1:~/realtek2.5g/r8152-2.14.0# make
make -C /lib/modules/5.11.7-1-pve/build M=/root/realtek2.5g/r8152-2.14.0 modules
make[1] : on entre dans le répertoire « /usr/src/linux-headers-5.11.7-1-pve »
CC [M] /root/realtek2.5g/r8152-2.14.0/r8152.o
MODPOST /root/realtek2.5g/r8152-2.14.0/Module.symvers
CC [M] /root/realtek2.5g/r8152-2.14.0/r8152.mod.o
make[3]: *** Aucune règle pour fabriquer la cible « scripts/module.lds », nécessaire pour « /root/realtek2.5g/r8152-2.14.0/r8152.ko ». Arrêt.
make[2]: *** [scripts/Makefile.modpost:117: __modpost] Error 2
make[1]: *** [Makefile:1738: modules] Error 2
make[1] : on quitte le répertoire « /usr/src/linux-headers-5.11.7-1-pve »
make: *** [Makefile:24: modules] Error 2
root@p1:~/realtek2.5g/r8152-2.14.0#I have problem to compile the realtek driver r8152
Same problem trying to compile the last version ixgbe driver
Code:
root@pxzen:/usr/src/ixgbe-5.11.3# cat /var/lib/dkms/ixgbe/5.11.3/build/make.log
DKMS make.log for ixgbe-5.11.3 for kernel 5.11.7-1-pve (x86_64)
Thu 22 Apr 2021 09:26:39 AM -04
make: Entering directory '/var/lib/dkms/ixgbe/5.11.3/build/src'
make[1]: Entering directory '/usr/src/linux-headers-5.11.7-1-pve'
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_main.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_api.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_common.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_dcb.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_dcb_82598.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_dcb_82599.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_ethtool.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_lib.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_mbx.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_sriov.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_param.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_phy.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_procfs.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_xsk.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_82598.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_82599.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_x540.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_x550.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_dcb_nl.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_debugfs.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_fcoe.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_ptp.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe_sysfs.o
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/kcompat.o
LD [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe.o
MODPOST /var/lib/dkms/ixgbe/5.11.3/build/src/Module.symvers
CC [M] /var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe.mod.o
make[3]: *** No rule to make target 'scripts/module.lds', needed by '/var/lib/dkms/ixgbe/5.11.3/build/src/ixgbe.ko'. Stop.
make[2]: *** [scripts/Makefile.modpost:117: __modpost] Error 2
make[1]: *** [Makefile:1738: modules] Error 2
make[1]: Leaving directory '/usr/src/linux-headers-5.11.7-1-pve'
make: *** [Makefile:99: default] Error 2
make: Leaving directory '/var/lib/dkms/ixgbe/5.11.3/build/src'1. Don't use ixgbe from intel on 5.11 kernel.
The module inside the kernel is probably newer, works better and has more features as intels corresponding driver that you can download.
2. https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1906131
This whole module linking issue with dkms, comes from a change since the 5.10 kernel.
They mention there, that it's a wanted change, and all the dkms packages needs to get updates and fixed.
It's probably super stupid, but it's how it is.
https://www.phoronix.com/scan.php?page=news_item&px=Realtek-RTL8153-RTL8156-Linux
maybe soon , no need to compile the driver... wait and see
maybe soon , no need to compile the driver... wait and see
I also got the "No rule to make target 'scripts/module.lds'" error while installing the pve-kernel-5.11 from the 6.4 stable release with the nvidia-driver from buster-backports (460.67-1~bpo10+1) already installed.
Could solve it by copying
As I am a linux newbie my question is when and by whom I/we can expect a fix for it?
For what I read and understand there already exists a patch from/for ubuntu.
But @Ramalama also states:
So the real question: Can we expect a fix(/temporary workaround?) with the next pve-kernel/headers update or do we need to wait for a updated driver-package (in my case from nvidia)?
If the latter one, does someone know if nvidia already fixed it and in which version and I/we only need to wait to get the specific version to the buster-backports?
PS.: As this is my first post, I also want to thank all of you for this awesome product very much!
Could solve it by copying
/usr/src/linux-headers-5.11.7-1-pve/scripts/module.lds.S to /usr/src/linux-headers-5.11.7-1-pve/scripts/module.lds and removing the last line #include <asm/module.lds.h> from it.As I am a linux newbie my question is when and by whom I/we can expect a fix for it?
For what I read and understand there already exists a patch from/for ubuntu.
But @Ramalama also states:
So the real question: Can we expect a fix(/temporary workaround?) with the next pve-kernel/headers update or do we need to wait for a updated driver-package (in my case from nvidia)?
If the latter one, does someone know if nvidia already fixed it and in which version and I/we only need to wait to get the specific version to the buster-backports?
PS.: As this is my first post, I also want to thank all of you for this awesome product very much!
I looked shortly into it last week, hoping that there's something in upstream dkms addressing that which could be backported or re-used, but I did not find anything specific then.
We can do the copy+adapt of the linker script, while it seems a bit hacky it is definitely better than not working.
I'll see if I can find some time to further investigate this..
I tried the new kernel yesterday but for me it breaks USB-passthrough functionality. With Proxmox 6.4 and kernel 5.4 it's working, with kernel 5.11 after a while one VM stops working and this comes up in the syslog:
After this, that VM is not working anymore and new VMs with USB-passthrough can't be started, too. Only VMs without USB devices can be started then. And only a reboot helps until it happens again after a few minutes. Switches back to kernel 5.4 and there are no problems anymore.[ 846.440540] INFO: task kvm:9359 blocked for more than 241 seconds.
[ 846.441201] Tainted: P O 5.11.7-1-pve #1
[ 846.441707] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[ 846.442150] task:kvm statestack: 0 pid: 9359 ppid: 1 flags:0x00000000
[ 846.442613] Call Trace:
[ 846.443077] __schedule+0x3ae/0x830
[ 846.443563] schedule+0x40/0xb0
[ 846.444020] schedule_timeout+0x211/0x2a0
[ 846.444517] ? ttwu_do_activate+0x69/0xd0
[ 846.444961] wait_for_completion+0xa6/0x110
[ 846.445380] kthread_stop+0x6c/0x160
[ 846.445793] __w1_remove_master_device+0x88/0x190 [wire]
[ 846.446234] w1_remove_master_device+0x57/0x60 [wire]
[ 846.446673] ds_disconnect+0x5f/0x90 [ds2490]
[ 846.447078] usb_unbind_interface+0x77/0x250
[ 846.447488] ? kernfs_remove_by_name_ns+0x75/0xa0
[ 846.447875] device_release_driver_internal+0xfe/0x1d0
[ 846.448261] device_release_driver+0x12/0x20
[ 846.448676] usb_driver_release_interface+0x46/0x80
[ 846.449083] proc_ioctl+0x1ea/0x230
[ 846.449485] usbdev_ioctl+0x33f/0xfc0
[ 846.449900] ? __fput+0x11c/0x240
[ 846.450310] ? _cond_resched+0x19/0x30
[ 846.450706] __x64_sys_ioctl+0x92/0xd0
[ 846.451112] do_syscall_64+0x37/0x80
[ 846.451564] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 846.452012] RIP: 0033:0x7ff8706a1427
[ 846.452466] RSP: 002b:00007fffff662148 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
[ 846.452925] RAX: ffffffffffffffda RBX: 00007fffff662160 RCX: 00007ff8706a1427
[ 846.453346] RDX: 00007fffff662150 RSI: 00000000c0105512 RDI: 000000000000002e
[ 846.453770] RBP: 0000559e4435ae00 R08: 00007fffff662190 R09: 0000559e43d27380
[ 846.454196] R10: 0000000000000006 R11: 0000000000000246 R12: 000000000000002e
[ 846.454638] R13: 0000000000000000 R14: 0000000000000060 R15: 0000000000000000
That trace isn't super helpfull. Probably you could provide more details, like your hardware/chipset/usb controller device, dmesg -loglevel warn,error, etc...
Cheers