A test added recently checks that when trying to deserialize a NAR with
two files that Unicode-normalize to the same result either succeeds on
Linux, or fails with an "already exists" error on Darwin. However,
failing with an "already exists" error can in fact also happen on Linux,
when using ZFS with the proper utf8 and Unicode normalization options
set.
This commit fixes the issue by not assuming the behavior from the
current system, but just by blindly checking that either one of the two
aforementioned possibilities happen, whether on Darwin or on Linux.
Additionally, we check that the Unicode normalization behaviour of
nix-store is the same as the host file system.
Because of an objc quirk[1], calling curl_global_init for the first time
after fork() will always result in a crash.
Up until now the solution has been to set
OBJC_DISABLE_INITIALIZE_FORK_SAFETY for every nix process to ignore
that error.
This is less than ideal because we were setting it in package.nix,
which meant that running nix tests locally would fail because
that variable was not set.
Instead of working around that error we address it at the core -
by calling curl_global_init inside initLibStore, which should mean
curl will already have been initialized by the time we try to do so in
a forked process.
[1] 01edf1705f/runtime/objc-initialize.mm (L614-L636)
(cherry-picked and adapted from c7d97802e4)
Note: in general, we rely on the OS to tell us if a name is invalid or
if two names normalize in the same way. But for security, we do want
to make sure that we catch '.', '..', slashes and NUL characters. (NUL
characters aren't really a security issue, but since they would be
truncated when we pass them to the OS, it would be canonicity problem.)
Relative path flakes ("subflakes") are basically fundamentally
broken, since they produce lock file entries like
"locked": {
"lastModified": 1,
"narHash": "sha256-/2tW9SKjQbRLzfcJs5SHijli6l3+iPr1235zylGynK8=",
"path": "./flakeC",
"type": "path"
},
that don't specify what "./flakeC" is relative to. They *sometimes*
worked by accident because the `narHash` field allowed
`fetchToStore()` to get the store path of the subflake *if* it
happened to exist in the local store or in a substituter.
Subflakes are properly fixed in #10089 (which adds a "parent" field to
the lock file). Rather than come up with some crazy hack to make them
work in the interim, let's just disable the only test that depends on
the broken behaviour for now.
When `nix fmt` is called without an argument, Nix appends the "." argument before calling the formatter. The comment in the code is:
> Format the current flake out of the box
This also happens when formatting sub-folders.
This means that the formatter is now unable to distinguish, as an interface, whether the "." argument is coming from the flake or the user's intent to format the current folder. This decision should be up to the formatter.
Treefmt, for example, will automatically look up the project's root and format all the files. This is the desired behaviour. But because the "." argument is passed, it cannot function as expected.
Fixes
$ nix flake metadata --store /tmp/nix nixpkgs
error: path '/tmp/nix/nix/store/65xpqkz92d9j7k5ric4z8lzhiigxsfbg-source/flake.nix' is not in the Nix store
This has been broken since 598deb2b23.
On macOS, `mkdir("x/')` behaves differently than `mkdir("x")` if `x` is
a dangling symlink (the formed succeed while the latter fails). So make
sure we always strip the trailing slash.
Fixes
```
umount: /tmp/nix-shell.i3xRwX/nix-test/local-overlay-store/delete-refs/stores/merged-store/nix/store: filesystem was unmounted, but failed to update userspace mount table.
make: *** [mk/lib.mk:93: tests/functional/local-overlay-store/delete-refs.sh.test] Error 16
```
in a dev shell.
Note: this previously worked before we didn't have umount in the dev
shell, so we got /run/wrappers/bin/umount.
This ensures just `nix build`-ing the flake doesn't forget to run all
tests. One can still specifiy specific attributes to just build one
thing.
Co-authored-by: Robert Hensing <roberth@users.noreply.github.com>
* docs: unify documentation on search paths
- put all the information on search path semantics into `builtins.findFile`
- put all the information on determining the value of `builtins.nixPath` into the
`nix-path` setting
maybe `builtins.nixPath` is a better place for this, but those bits
can still be moved around now that it's all next to each other.
- link to the syntax page for lookup paths from all places that are
concerned with it
- add or clarify examples
- add a test verifying a claim from documentation
This also bans various sneaking of negative numbers from the language
into unsuspecting builtins as was exposed while auditing the
consequences of changing the Nix language integer type to a newtype.
It's unlikely that this change comprehensively ensures correctness when
passing integers out of the Nix language and we should probably add a
checked-narrowing function or something similar, but that's out of scope
for the immediate change.
During the development of this I found a few fun facts about the
language:
- You could overflow integers by converting from unsigned JSON values.
- You could overflow unsigned integers by converting negative numbers
into them when going into Nix config, into fetchTree, and into flake
inputs.
The flake inputs and Nix config cannot actually be tested properly
since they both ban thunks, however, we put in checks anyway because
it's possible these could somehow be used to do such shenanigans some
other way.
Note that Lix has banned Nix language integer overflows since the very
first public beta, but threw a SIGILL about them because we run with
-fsanitize=signed-overflow -fsanitize-undefined-trap-on-error in
production builds. Since the Nix language uses signed integers, overflow
was simply undefined behaviour, and since we defined that to trap, it
did.
Trapping on it was a bad UX, but we didn't even entirely notice
that we had done this at all until it was reported as a bug a couple of
months later (which is, to be fair, that flag working as intended), and
it's got enough production time that, aside from code that is IMHO buggy
(and which is, in any case, not in nixpkgs) such as
https://git.lix.systems/lix-project/lix/issues/445, we don't think
anyone doing anything reasonable actually depends on wrapping overflow.
Even for weird use cases such as doing funny bit crimes, it doesn't make
sense IMO to have wrapping behaviour, since two's complement arithmetic
overflow behaviour is so *aggressively* not what you want for *any* kind
of mathematics/algorithms. The Nix language exists for package
management, a domain where bit crimes are already only dubiously in
scope to begin with, and it makes a lot more sense for that domain for
the integers to never lose precision, either by throwing errors if they
would, or by being arbitrary-precision.
Fixes: https://github.com/NixOS/nix/issues/10968
Original-CL: https://gerrit.lix.systems/c/lix/+/1596
Change-Id: I51f253840c4af2ea5422b8a420aa5fafbf8fae75
