This makes packages use lapack and blas, which can wrap different
BLAS/LAPACK implementations.
treewide: cleanup from blas/lapack changes
A few issues in the original treewide:
- can’t assume blas64 is a bool
- unused commented code
This is based on previous work for switching between BLAS and LAPACK
implementation in Debian[1] and Gentoo[2]. The goal is to have one way
to depend on the BLAS/LAPACK libraries that all packages must use. The
attrs “blas” and “lapack” are used to represent a wrapped BLAS/LAPACK
provider. Derivations that don’t care how BLAS and LAPACK are
implemented can just use blas and lapack directly. If you do care what
you get (perhaps for some CPP), you should verify that blas and lapack
match what you expect with an assertion.
The “blas” package collides with the old “blas” reference
implementation. This has been renamed to “blas-reference”. In
addition, “lapack-reference” is also included, corresponding to
“liblapack” from Netlib.org.
Currently, there are 3 providers of the BLAS and LAPACK interfaces:
- lapack-reference: the BLAS/LAPACK implementation maintained by netlib.org
- OpenBLAS: an optimized version of BLAS and LAPACK
- MKL: Intel’s unfree but highly optimized BLAS/LAPACK implementation
By default, the above implementations all use the “LP64” BLAS and
LAPACK ABI. This corresponds to “openblasCompat” and is the safest way
to use BLAS/LAPACK. You may received some benefits from “ILP64” or
8-byte integer BLAS at the expense of breaking compatibility with some
packages.
This can be switched at build time with an override like:
import <nixpkgs> {
config.allowUnfree = true;
overlays = [(self: super: {
lapack = super.lapack.override {
lapackProvider = super.lapack-reference;
};
blas = super.blas.override {
blasProvider = super.lapack-reference;
};
})];
}
or, switched at runtime via LD_LIBRARY_PATH like:
$ LD_LIBRARY_PATH=$(nix-build -E '(with import <nixpkgs> {}).lapack.override { lapackProvider = pkgs.mkl; is64bit = true; })')/lib:$(nix-build -E '(with import <nixpkgs> {}).blas.override { blasProvider = pkgs.mkl; is64bit = true; })')/lib ./your-blas-linked-binary
By default, we use OpenBLAS LP64 also known in Nixpkgs as
openblasCompat.
[1]: https://wiki.debian.org/DebianScience/LinearAlgebraLibraries
[2]: https://wiki.gentoo.org/wiki/Blas-lapack-switch
kwallet sets a limit of 1000 for a single characters for environment
variables read from the socket[1]. wrapQtApps gives us a huge value
for QT_PLUGIN_PATH (up to 13000 bytes on my system!) Since this was
overflowing, the Qt plugin loading mechanism was hitting a segfault
when it was trying to parse the truncated QT_PLUGIN_PATH.
So for now, we can just unset QT_PLUGIN_PATH in the pam_kwallet_init
script. kwalletd5 has its own QT_PLUGIN_PATH which it can use.
This problem occured on 20.03, but not 19.09. It’s unclear what
changes were made in that time, but likely that previously we weren’t
getting a QT_PLUGIN_PATH set in the plasma5 startup at all. This means
that in 19.09 our QT_PLUGIN_PATH value must have been small enough to
fit into the 1000 char limit.
Fixes#77290
[1]: https://github.com/KDE/kwallet/blob/bc9713e2725ab1c4311866f751c674a38584bd92/src/runtime/kwalletd/main.cpp#L44
/cc @ttuegel
This should fix the regression found in
https://hydra.nixos.org/eval/1577308#tabs-now-fail. The libretro
scripts use “osx” as the identifier for macOS, and “unix” seems to
mean linux.
If anyone could help out by confirming this works. I don’t currently
have access to a macOS machine.
These files never existed, so best to not leave the reference. If
someone want to step up to maintain this, that would be fine. I don’t
have the hardware to test these out. In addition, someone tried to use
the bootstrap-tools currently built by Hydra and found that they were
broken in some unclear way.
* nixos/nixpkgs.nix: Allow just using config in system
This assertion requires system to work properly. We might not have
this in cases where the user just sets config and wants Nixpkgs to
infer system from that. This adds a default for when this happens,
using doubleFromSystem.
* parens
- Avoid using overrides unless necessary
- Set platform and ARCH by default
- Don’t set dontConfigure unless absolutely necessary
- Use preBuild instead of overriding entire configurePhase
newlib is the default for most tools when no kernel is provided. Other
exist, but this seems like a safe default.
(cherry picked from commit 8009c20711)
Unfortunately, CMake looks in CMAKE_PREFIX_PATH for binaries when
cross-compiling. This means that it will use gettext from
CMAKE_PREFIX_PATH even when we provide a gettext binary in PATH! This
is bad because the on in CMAKE_PREFIX_PATH is for the cross system,
not the native one. The only documented way I can find to change this
behavior is by manually setting the CMAKE_IGNORE_PATH variable.
Cross needs some extra native tools:
- glib
- wayland
- orc
- glib is needed in nativeBuildInputs
Also:
- bash-completion needs to be available for PKG_CONFIG_PATH
libxslt is actually both a target and native build input because some
libxslt binary is needed to generate files and the built binaries also
link against libxslt.
Add a cage module to nixos. This can be used to make kiosk-style
systems that boot directly to a single application. The user (demo by
default) is automatically logged in by this service and the
program (xterm by default) is automatically started.
This is useful for some embedded, single-user systems where we want
automatic booting. To keep the system secure, the user should have
limited privileges.
Based on the service provided in the Cage wiki here:
https://github.com/Hjdskes/cage/wiki/Starting-Cage-on-boot-with-systemd
Co-Authored-By: Florian Klink <flokli@flokli.de>
In https://github.com/NixOS/nixpkgs/issues/76927, we found that
xmemdup is missing when compiling some conftest for gnustep-base. It’s
unclear why this happens but it appears to be from the update to gcc9.
Adding libiberty gives us that symbol and fixes things. What’s less
clear is why this happens to begin with.
Fixes#76927
Matthew Bauer