Age | Commit message (Collapse) | Author | Files | Lines |
|
- NVME_AUTH must be module if NVME_CORE is module on 6.7+
- NVME_TARGET_PASSTHRU is spelled differently, and does not exist before 5.9
- PCI_P2PDMA does not exist before 4.20
|
|
TLS.
|
|
|
|
|
|
Necessary to support the backlight on the GPD Pocket, GPD Win and several other devices.
|
|
These modules are very useful for power management via GPIO pins on
embedded platforms such as the Raspberry Pi. The POWER_RESET_GPIO module
has been tested on a Raspberry Pi 3B+.
|
|
DAMON_DBGFS has been renamed to DAMON_DBGFS_DEPRECATED. Let's just
drop it now.
|
|
Arch and Fedora have been shipping it for a while now, we should
probably join them in the happy non-fbdev future. Legacy fbdev
drivers are still enabled for now, so non-simpledrm-friendly
devices don't break.
|
|
Disabled by default in 5.10, needed by MEDIA_CEC_RC
|
|
Was disabled in dbdcb3fad435fc531e0bfc71e1b93cc0ff5b0882
for "powertop" kernel variants, which was then merged into default
kernels in 77a8ea6fbd474bca969371506ee663e1517. Not really sure why.
|
|
Previously, when a kernel configuration option was defined multiple times (in
different sections), only the first value was honored and all subsequent ones
were silently ignored. Remove this footgun by throwing in this situation
instead.
In addition, fix all instances where an option was in fact defined multiple
times. In two cases (`RCU_TORTURE_TEST` and `CRASH_DUMP`), the value was
effectively the same both times. In the other two cases (`SCHEDSTATS` and
`DRM_SIMPLEDRM`) the value was in fact different; the one that was actually
applied was kept here to prevent a rebuild.
|
|
This is just useful things I've found while diffing our kernel configs
with Arch and Fedora. Some of these have a small performance overhead,
most are completely free.
Switch default performance governor to schedutil, because performance
is almost always a bad default.
|
|
As per https://docs.kernel.org/sound/designs/powersave.html
We update our defaults for better power saving on laptops
which does not perform powertop --auto-tune, TLP or manual tweaking, e.g.
power-profiles-daemon.
|
|
|
|
Upstream disabled NFSD_V2 entirely, so disable this also.
|
|
This enables support for ZAC/ZBC/ZNS host-managed and host-aware zoned
block devices. It also enables zone features in various filesystems
including BTRFS and F2FS, and other relative subsystems like nvme and
virtio-blk.
|
|
This enables support for 802.11s mesh networking in drivers that
support it.
|
|
|
|
|
|
|
|
|
|
This reverts commit 096639c548e2b0256309c92c53542e052eaa3761.
The kernel bug that broke this has now been fixed, and we now set
CONFIG_X86_GENERIC, which would have stopped us being affected by it anyway.
|
|
In-kernel Rust on Linux (Optional) (2nd Attempt)
|
|
With the current patching situation around enabling Linux to build
with whatever rustc/bindgen version we have in nixpkgs, it feels like
supporting a Rust-by-default build at this time is not fully
justified.
Co-authored-by: Julian Stecklina <julian.stecklina@cyberus-technology.de>
|
|
linux/kernel/common-config: Reenable the rfkill-input module
|
|
|
|
> This is really intended for distributors who need more generic
> optimizations.
Apparently it's expected that as a generic distribution, we set this,
and if we had we wouldn't have been affected by the recent 32-bit x86
Xen breakage[1][2]:
> In normal distro kernels, CONFIG_X86_GENERIC is enabled
[1]: https://lore.kernel.org/lkml/20231204084722.3789473-1-arnd@kernel.org/
[2]: 096639c548e2 ("linux: drop XEN on 32-bit")
|
|
The default value of the RFKILL_INPUT Kconfig option depends on the
EXPERT option which was changed in 3b07356d2d55c5eacc7a11eb08c3a8de97884b2f.
However, disabling the rfkill-input module was unintentional and causes
some airplane mode buttons on laptops to not function [1].
[1]: https://github.com/NixOS/nixpkgs/issues/261880
|
|
linux: enable UDMABUF
|
|
|
|
> A driver to let userspace turn memfd regions into dma-bufs.
> Qemu can use this to create host dmabufs for guest framebuffers.
Enabled by Fedora and Arch.
|
|
It doesn't build anymore and I doubt anyone wants to maintain it:
https://hydra.nixos.org/build/243596962/nixlog/1
|
|
|
|
This adds two kernel config settings, one to make MediaTek network
modules visible (so the modules get built) and one to support the
SoC-integrated MT7915E variant in the module.
This gets the MT798X series of SoCs working with a stock NixOS kernel.
|
|
|
|
ZSMALLOC is now the default allocator for ZSWAP, so if the allocator
isn't changed, ZSMALLOC must be y on 6.7 (and will default to that).
The script will set all options that can be modules to m anyway, so
there was no need for us to be explicitly setting it to module before
anyway.
|
|
|
|
|
|
kernel/common-config: arm: configure alignment traps
|
|
These options were already enabled in previous versions, but Kconfig
changes in 6.6 made NixOS kernel disable them. Therefore, we enable
unconditionally, to be explicit that they're needed.
Without them, the fbcon/console on systems with DRM devices freeze/blank
on early boot, even though the system boots normally.
|
|
|
|
|
|
...into staging
|
|
Previous (4.9): https://github.com/NixOS/nixpkgs/pull/201845
|
|
|
|
under armv7l, config.gz already had CONFIG_ALIGNMENT_TRAP=y, but
explicitly enable it and provide some context for the option.
under aarch64, on kernels that are new enough to support it, specify
CONFIG_COMPAT_ALIGNMENT_FIXUPS=y to attempt to resolve alignment
faults identically to how they are handled under 32-bit kernels. This
minimizes the potential for aarch32 userspace to behave differently
under an aarch64 kernel.
|
|
Exposes /proc/fs/fscache/stats that shows some handy stats about fs caching. Can
be useful for debugging, so why not.
|
|
rc1: https://lwn.net/Articles/944122/
|
|
|
|
`KUNIT` knob was added around 5.5 release:
https://github.com/torvalds/linux/commit/914cc63eea6fbe11ed46dba5e4438d81b0cd42d2
|