Age | Commit message (Collapse) | Author |
|
(cherry picked from commit 6c6fe821de20eaf18640285587e960726c1ea39b)
|
|
(cherry picked from commit c910d5843ac4f7f96d9cb2d9c564f7fe8587c50f)
|
|
(cherry picked from commit 13a2c6ad63478ee575e33dc389e50e9d409db948)
|
|
(cherry picked from commit eb1a10558717239f2025737ee4e56dd01cba4355)
|
|
Add kernel modules for Hyper-V
Based on user reviews the following settings must be made:
```
CONFIG_CONNECTOR=y
CONFIG_HYPERV_UTILS=m
```
(cherry picked from commit 6f6e826977bd59947a660c9131ffeaf41d01d4bb)
|
|
(cherry picked from commit cc04a75932345ba37e4e18f82ae3f85330cedf68)
|
|
Mellanox/NVIDIA NICs require Infiniband support for proper communication with
user space, which is used by tools like DPDK.
This commit enables Infiniband with user access support and adds it to
`mlx4`/`mlx5`.
(cherry picked from commit 0a5b69b8472ce1b39f5a6f8af575aeecd459c34f)
|
|
(cherry picked from commit c237941c3c95de16d2581b198b679288898758f3)
|
|
This reverts commit 88be901bc103d1c47adbbc874d02e8ec5cde3397.
(cherry picked from commit ad214dba3ec08a2fa8c9fb37f22db367ff663e8a)
|
|
TL;DR: systemd does not require the performance-sensitive bits of Linux control
groups enabled in the kernel. However, it does require some
non-performance-sensitive bits of the control group logic.
http://0pointer.de/blog/projects/cgroups-vs-cgroups.html
The only controllers required for VyOS to function are the memory and PID
controller required by the container feature. All other controlles can be
disabled.
(cherry picked from commit 86f80f99c57240bda43b99d110de736aef4dea76)
|
|
With all the minor fixes applied to the Kernel, also options change - this
just syncs the previous config to the latest version with the new defaults
applied that slipped in during all those bugfixes.
(cherry picked from commit 33e78bde1a9d9da98c37269819b8b8e8e73ee39c)
|
|
(cherry picked from commit 97ddcaec1f488bef18c0d8dfa37cb272115e548a)
|
|
|
|
Enable virtiofs to support lxd-agent.
See https://github.com/canonical/lxd/blob/005bd8d473002a1f72b19b3e4f9e05557bece639/lxd/instance/drivers/driver_qemu.go#L2510-L2549
|
|
I added also additional expansion card with 2 Intel i211 NIC’s.
I want to utilize all 4 NIC’s, but I have an issue because only NIC on expansion cards are properly detected.
|
|
Jun 16 20:39:24 systemd[1]: Starting hostapd@wlan0.service - Access point and authentication server for Wi-Fi and Ethernet (wlan0)…
Jun 16 20:39:25 hostapd[7198]: rfkill: Cannot open RFKILL control device
Jun 16 20:39:25 hostapd[7198]: wlan0: interface state UNINITIALIZED->COUNTRY_UPDATE
Jun 16 20:39:25 systemd[1]: Started hostapd@wlan0.service - Access point and authentication server for Wi-Fi and Ethernet (wlan0).
Seems RFKILL is missing, or could there be someting else?
|
|
Without this option the 'vhost' modules are not included
|
|
The kernel 'vhost-net' options required for creating TAP devices
vpp# create tap
create tap: open '/dev/vhost-net': No such file or directory
vpp#
vpp# create tap id 1 host-if-name tap1
create tap: open '/dev/vhost-net': No such file or directory
vpp#
|
|
Add Mellanox Technologies firmware flash module mlxfw to kernel
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This enables Device Mapper support for either RAID targets or encrypted
partitions.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Enabled build of the SG module to make possible communication with RAID
adapters from Areca (https://www.areca.com.tw/).
|
|
Required for IPSec XFRM interfaces.
|
|
|
|
|
|
|
|
The latest Intel QAT drivers make the 5.10 series Kernel crash and there was
no real and good response from Intel so far.
Today the build failed one more time by a source file moving around in the
Intel webservers. It's time to try to move the QAT functionality to the Kernel
buildin drivers.
|
|
|
|
cards"
Newer cards must use QMI it seems. It also feels more "robust".
|
|
|
|
This allows us to use kexec in the future to support fast reboots. This is more
important on physical hardware which can take 10+ minutes to POST.
Signed-Off-By: Rob Thomas <xrobau@gmail.com>
|
|
All throughout the history of vyos-build, NUMA has been disabled.
It WAS enabled back in Vyatta days, so something somewhere has
accidentally turned it off - and it's ALSO mentioned explicitly
in various pieces of documentation, so it's MEANT to be on!
Note for future researchers: CONFIG_INOTIFY_STACKFS=y is provided
by a patch, so patch the kernel BEFORE running make menuconfig, or
manually put it back. If not, it'll vanish and is a wanted part of
the build.
Signed-Off-By: Rob Thomas <xrobau@gmail.com>
|