Because downstream code expects to use `==` on platform attrsets, we
are unfortunately not able to throw a useful error message when the
`sharedLibrary` attribute is accessed.
When users do a comparison like:
stdenv.hostPlatform == pkgsStatic.stdenv.hostPlatform
... in a situation where `stdenv.hostPlatform.hasSharedLibraries`,
they expect this to return `false`. Unfortunately Nix does a deep
equality comparison here, and ends up forcing the
`pkgsStatic.stdenv.hostPlatform.extensions.sharedLibrary` attribute,
which throws the error.
Rather than returning `null`, this commit instead simply omits the
`extensions.sharedLibrary` attribute. This provides the user with a
more-useful error message: instead of waiting until the `null` is
used (and hoping that produces an error), the user will get an error
about the `extensions.sharedLibrary` attribute being missing, at the
position where it was referenced.
Big thanks to @trofi for his PR to add
`NIX_VALIDATE_EVAL_NONDETERMINISM` to Nix, which I am now using. It
made tracking this down really easy!
Fixes#244045
This commit adds `hasSharedLibraries` to `lib.systems`.
We need `plat.hasSharedLibraries` in order to know whether or not to
expect `gcc` (and many other tools) to emit shared libraries (like
`libgcc_s.so`). Many of the GNU build scripts are smart enough that
if you configure them with `--enable-shared` on a platform (such as
`arm-none-eabi`) that doesn't support dynamic linking, they will
simply skip the shared libraries instead of aborting the
`configurePhase`. Unfortunately the missing shared libraries in the
final build product cause very hard-to-troubleshoot problems later
on.
The alternative to introducing `hasSharedLibraries` would be to set
`isStatic` in these situations. However doing so causes
`make-derivation.nix` to insert `-static` between the `pname` and
`hostPlatform` suffix, which is undesirable.
If at some point in the future we eliminate the `-static` suffix,
then `hasSharedLibraries` can be made equal to `!isStatic`.
toLosslessStringMaybe is not used by anything other than lib/tests,
so it can be private to that file.
I don't think this function was terribly well thought-through. If
people start using it, we will become permanently dependent on the
ability to test platforms for equality. It also makes the
elaboration process more fragile, because it encourages code outside
of nixpkgs to become sensitive to the minute details of how
elaboration happens.
The eminent Donald E. Knuth should be recognized as having equal
standing with such entities as IBM, Apple, and the Personal
Computer. We should acknowledge this by including him as a "vendor".
Also, `gnu-config` recognizes `mmix-knuth-*` triples (and in fact
requires `vendor="knuth"` when `cpu="mmix"`) -- so we sort of have
to. But we should do it anyways.
We already have examples for these, but since we didn't actually
recognise the doubles, it wasn't possible to build any packages for
them without setting allowUnsupportedSystem.
I imagine this was supposed to be rustc = args.rustc, like the other
two lines. This meant that we accepted both rust and rustc
attributes, with the same effect. I doubt anybody was using the
undocumented, probably-accidental "rust" spelling, but we should
remove it before somebody starts.
In fact, we don't need to set rustc here at all, because no value
platforms.select could return will ever include a rustc key (unlike
the other two), so then rustc will be filled in later, when args is
merged into final.
It makes sense to allow platform definitions to opt out of having libc
at all. One use case would be targetting some obscure new Linux
target that doesn't have a libc implementation yet, and another is
UEFI, which is basically libc-less Windows.
Not having libc is not commonly specified in (GNU) triples (even
Linux's build system will just target either -gnu or -musl depending
on the platform), so instead, we use a separate attribute for it.
- Christmas is over!
- Upstream has changed the name of the target triplet used for the JS
backend from js-unknown-ghcjs to javascript-unknown-ghcjs, since Cabal
calls the architecture "javascript":
6636b67023
Since the triplet is made up anyways, i.e. autoconf does not support
it and Rust uses different triplets for its emscripten backends, we'll
just change it as well.
- Upstream fixed the problem with ar(1) being invoked incorrectly by stage0:
e987e345c8
There is no "aarch64" CPU family — it counts as "arm", as can be seen
from the definition of isAarch64 above.
Checked that stdenv.hostPlatform.isEfi is still true on aarch64-linux.
In the past, most (if not all) armv8 CPUs could also execute armv7. However,
with the advent of Apple Silicon, aarch64 CPUs without any aarch32 capabilities
are now wide-spread among users.
Cross-compilation of anything downstream of gtk3 requires qemu (due to
gobject-introspection) with --target-list=*-linux-user. Without this commit,
those qemu builds will fail on a powerpc64le host due to qemu being configured
with --cpu=powerpc64le instead of --cpu=ppc64le. Unfortunately the build
failure message from qemu in this situation is extremely cryptic.
The root cause turns out not to be the qemu expression, but rather the fact that
on powerpc64le hostPlatform.uname.processor returns the gnu-name (powerpc64le)
for the cpu instead of the linux-name (ppc64le) for the cpu.
uname.processor on mips64el also needs adjustment -- the Linux-name is "mips64"
for both big and little endian (unlike powerpc64, where the Linux-name includes
a "le" suffix):
```
nix@oak:/tmp$ uname -m; lscpu | head -n2
mips64
Architecture: mips64
Byte Order: Little Endian
```
uname.processor on powerpc32 has also been adjusted.
The main purpose of this PR is to make the basis for
`mkSkeletonFromList`'s decision between `cpu-kernel-libcabi` vs
`cpu-vendor-os` clear, without changing its behavior. The existing
code obscures this decision behind a sequence of prioritized matches
(i.e. `if-then`) which jump around between different coordinates.
Two side benefits of this PR:
1. It makes the root cause of #165836 obvious: we are missing a case
for `cpu-vendor-libcabi`. This is why nixpkgs stumbles over
`*-none-*`.
2. It illuminates some very weird corner cases in the existing
logic, like `*-${vendor}-ghcjs` overriding the `vendor` field,
and `mingw32` being transformed into `windows` in some cases.
Co-authored-by: John Ericson <git@JohnEricson.me>
A tricky thing about FreeBSD is that there is no stable ABI across
versions. That means that putting in the version as part of the config
string is paramount.
We have a parsed represenation that separates name versus version to
accomplish this. We include FreeBSD versions 12 and 13 to demonstrate
how it works.
Move already implemented functionality to the upper level so
it could be used in a more generic way.
Signed-off-by: Ivan Nikolaenko <ivan.nikolaenko@unikie.com>
```
nix-repl> pkgsCross.arm-embedded.stdenv.hostPlatform.emulatorAvailable pkgsCross.arm-embedded.buildPackages
false
nix-repl> pkgsCross.aarch64-multiplatform.stdenv.hostPlatform.emulatorAvailable pkgsCross.aarch64-multiplatform.buildPackages
true
```
will be useful for stuff like handling https://github.com/NixOS/nixpkgs/issues/187109
The comment in lib/systems/default.nix for uname.processor indicates that it
should match `uname -p`. I tried that command and found that it reports
`unknown` on all of these machines:
- `x86_64-linux`
- `aarch64-linux`
- `mips64el-linux`
- `powerpc64le-linux`
The command `uname -m` reports the expected value on all of the above.
I think the comment is wrong. So I fixed it.
In Nixpkgs, we assume that the "config" field is a canonicalized GNU
triple. I noticed that non-canonical values were being used here,
because the pkgsCross.mips64el-linux-gnu triples did not contain the
vendor field, but the pkgsCross.mips64el-linux-gnu.pkgsStatic did.
Here, I've run all the MIPS triples in lib.systems.examples through
config.sub to canonicalize them. I think this will avoid nasty
surprises in future.
Tested by building Nix and the bootstrap files for
pkgsCross.mips64el-linux-gnu.
This has been deprecated for a long time, and it's doubtful it had any
users to start with. And having an undisablable warning when
enumarating platforms is not good.
These servers apparently no longer exist, since September 2, 2021[1].
If somebody needs this for non-Scaleway machines, they should suggest
its reintroduction with a different name.
[1]: https://news.ycombinator.com/item?id=27192757
Very confusingly, the `isPowerPC` predicate in
`lib/systems/inspect.nix` does *not* match `powerpc64le`!
This is because `isPowerPC` is defined as
isPowerPC = { cpu = cpuTypes.powerpc; };
Where `cpuTypes.powerpc` is:
{ bits = 32; significantByte = bigEndian; family = "power"; };
This means that the `isPowerPC` predicate actually only matches the
subset of machines marketed under this name which happen to be 32-bit
and running in big-endian mode which is equivalent to:
with stdenv.hostPlatform; isPower && isBigEndian && is32bit
This seems like a sharp edge that people could easily cut themselves
on. In fact, that has already happened: in
`linux/kernel/common-config.nix` there is a test which will always
fail:
(stdenv.hostPlatform.isPowerPC && stdenv.hostPlatform.is64bit)
A more subtle case of the strict isPowerPC being used instead of the
moreg general isPower accidentally are the GHC expressions:
Update pkgs/development/compilers/ghc/8.10.7.nix
Update pkgs/development/compilers/ghc/8.8.4.nix
Update pkgs/development/compilers/ghc/9.2.2.nix
Update pkgs/development/compilers/ghc/9.0.2.nix
Update pkgs/development/compilers/ghc/head.nix
Since the remaining legitimate use sites of isPowerPC are so few, remove
the isPowerPC predicate completely. The alternative expression above is
noted in the release notes as an alternative.
Co-authored-by: sternenseemann <sternenseemann@systemli.org>
canExecute is like isCompatible, but also checks that the Kernels are
_equal_, i.e. that both platforms use the same syscall interface. This
is crucial in order to actually be able to execute binaries for the
other platform.
isCompatible is dropped, since it has changed semantically and there's
no use case left in nixpkgs.
Since we (exclusively) use isCompatible to gauge whether platform a can
execute binaries built for platform b, mode switching CPUs are not to be
considered compatible for our purposes: Switching the mode of a CPU
usually requires a reset. At the very least we can't execute a mix of
executables for the two modes which would usually be the case in nixpkgs
where we may want to execute buildInputs for the hostPlatform in
addition to nativeBuildInputs for the buildPlatform.
Since the list only gates the platforms the nixpkgs flake exposes
packages to build on, the `hydra` label made little sense. It was also
only used for this purpose, so the `tier*` attributes were largely
unnecessary.
To reflect the intention more accurately, we expose
`lib.systems.flakeExposed` and use it to gate flake.nix's system list.
This patch causes the autodetection code in lib/systems/platforms.nix
to return {} if it cannot detect the platform and one of the
platform.nix-detection-provided attributes (linux-kernel, gcc, and
rustc) are accessed, rather than silently assuming the "pc" platform
as was previously done.
It is definitely safe to assume that code using these attributes is
prepared to deal with `gcc` and `rustc` not being defined, because
many of the working entries in this file don't define it.
Regarding `linux-kernel` the situation is less certain, but some code
(`lib/systems/default.nix` for example) is already designed to deal
with that attribute being missing. At worst it would result in an
"attribute not found" error.
While adding mips64el bootstrap support to nixpkgs, the silent
assumption that mips64el routers are actually Intel PCs caused
significant frustration. This commit removes that assumption in order
to save people who port nixpkgs to new platforms in the future from
this frustration.
For other platforms like Intel and ARM, we can do
e.g. lib.platforms.aarch64 to get only the 64-bit ARM platorms, but
until now there were no equivalents for RISC-V.
Prior to this commit, nixpkgs would assume that every little-endian
mips32 system was a "fuloong2f_n32".
Not only are there plenty of mips32 chips other than the fuloong, but
the fuloong is actually a mips64 chip! Note that the "n32" ABI is
(confusingly) an ABI for 64-bit mips chips (like the "x32" ABI for
amd64 chips -- both are ABIs which use 32-bit pointers on an
otherwise-64-bit system).
This error causes far-ranging problems. One of them was particularly
difficult to track down: it caused GCC to select 128-bit `long double`
types, which is invalid for the mips32 ABI. This isn't noticed until
you try to build musl-libc, which is careful to check for these things.
Prior to this commit,
nix-build . -A pkgsCross.mipsel-linux-gnu.pkgsStatic.hello
would fail. With this commit and #170736, it succeeds.
There is only one ABI for 32-bit MIPS chips. Before mips64, it didn't
really have a name.
The 64-bit MIPS ABI comes in two flavors, "n64" and "n32". It is
commonplace to refer to the old 32-bit ABI as "o32" (MIPS and SGI
documents do this).
However, when configuring gcc, one must use --with-abi=32, not
--with-abi=o32.
Let's keep GCC happy with this commit.
This commit adds an `isPower64` predicate to the two existing
predicates for this architecture (`isPower` and `isPowerPC`).
Note that `isPowerPC` matches only 32-bit machines, whereas `isPower`
matches both 64-bit and 32-bit machines. Prior to this commit there
was no single `isXXX` predicate for `powerpc64le`.
