This patch replaces the old grsecurity kernels with a single NixOS
specific grsecurity kernel. This kernel is intended as a general
purpose kernel, tuned for casual desktop use.
Providing only a single kernel may seem like a regression compared to
offering a multitude of flavors. It is impossible, however, to
effectively test and support that many options. This is amplified by
the reality that very few seem to actually use grsecurity on NixOS,
meaning that bugs go unnoticed for long periods of time, simply because
those code paths end up never being exercised. More generally, it is
hopeless to anticipate imagined needs. It is better to start from a
solid foundation and possibly add more flavours on demand.
While the generic kernel is intended to cover a wide range of use cases,
it cannot cover everything. For some, the configuration will be either
too restrictive or too lenient. In those cases, the recommended
solution is to build a custom kernel --- this is *strongly* recommended
for security sensitive deployments.
Building a custom grsec kernel should be as simple as
```nix
linux_grsec_nixos.override {
extraConfig = ''
GRKERNSEC y
PAX y
# and so on ...
'';
}
```
The generic kernel should be usable both as a KVM guest and host. When
running as a host, the kernel assumes hardware virtualisation support.
Virtualisation systems other than KVM are *unsupported*: users of
non-KVM systems are better served by compiling a custom kernel.
Unlike previous Grsecurity kernels, this configuration disables `/proc`
restrictions in favor of `security.hideProcessInformation`.
Known incompatibilities:
- ZFS: can't load spl and zfs kernel modules; claims incompatibility
with KERNEXEC method `or` and RAP; changing to `bts` does not fix the
problem, which implies we'd have to disable RAP as well for ZFS to
work
- `kexec()`: likely incompatible with KERNEXEC (unverified)
- Xen: likely incompatible with KERNEXEC and UDEREF (unverified)
- Virtualbox: likely incompatible with UDEREF (unverified)
Close #15803. This avoids the error:
while setting up the build environment: executing
‘/nix/store/7sb42axk5lrxqz45nldrb2pchlys14s1-bash-4.3-p42/bin/bash’:
Argument list too long
Note: I wanted to make it optional based on buildCommand length,
but that seems pointless as I'm sure it's less performant.
Amended by vcunat:
https://github.com/NixOS/nixpkgs/pull/15803#issuecomment-224841225
This takes another approach at binding FHS directory structure. We
now bind-mount all the root filesystem to directory "/host" in the target tree.
From that we symlink all the directories into the tree if they do not already
exist in FHS structure.
This probably makes `CHROOTENV_EXTRA_BINDS` unnecessary -- its main usecase was
to add bound directories from the host to the sandbox, and we not just symlink
all of them. I plan to get some feedback on its usage and maybe deprecate it.
This also drops old `buildFHSChrootEnv` infrastructure. The main problem with it
is it's very difficult to unmount a recursive-bound directory when mount is not
sandboxed. This problem is a bug even without these changes -- if
you have for example `/home/alice` mounted to somewhere, you wouldn't see
it in `buildFHSChrootEnv` now. With the new directory structure, it's
impossible to use regular bind at all. After some tackling with this I realized
that the fix would be brittle and dangerous (if you don't unmount everything
clearly and proceed to removing the temporary directory, bye-bye fs!). It also
probably doesn't worth it because I haven't heard that someone actually uses it
for a long time, and `buildFHSUserEnv` should cover most cases while being much
more maintainable and safe for the end-user.
1. When multiple versions of the same package are required
$revs is an array.
2. When cargo fetch is run it usually doesn't need a network
connection. But when it does SSL_CERT_FILE isn't set.
find fails when called with an inexistent search path.
That situation may arise when the output is created after by a postFixup hook.
vcunat amended the PR by clarifying one more `return` to `return 0`.
Checkinstall had two problems:
1. when it was called without a version (e.g. with a derivation created
by fetchFromGitHub) it would use `src` as debian version, which caused
dpkg to fail
2. when dpkg failed, it would invoke the pager with the log, which hangs
the build
So now
1. the default version is the dummy `0.0.0`
2. the used pager is `cat`
Source of this change goes back to 2009 and original version of
fetchgit at 205fb0c87e.
The nondeterminism is really caused by changing .git so leave other
files alone as they might be interesting.
Note: this causes a hash mismatch with Hydra's version of Git Plugin
which we should fix to comply.
Move all the dependencies to their own derivation, so that we don't publish all
of them if the wrapper is installed in a profile.
The previous solution just moved them to a custom directory to avoid conflicts,
this refactors that and completely hides them, while preserving the desired
improvement of adding only one directory to each of the emacs search paths
The chroot caps restriction disallows chroot'ed processes from running
any command that requires `CAP_SYS_ADMIN`, breaking `nixos-rebuild`. See
e.g., https://github.com/NixOS/nixpkgs/issues/15293
This significantly weakens chroot protections, but to break
nixos-rebuild out of the box is too severe.
There are no users of it in main tree and recent merge
of multiple outputs branch makes it obsolete for private trees
too.
At the time hook was created, recently merged multiple output
branch was relying on passing flags to autotools to split
outputs, which obviously wasn't working for other build systems
Scatter output was taking different approach where files were
moved out from a build tree based on known paths, which is more
or less what current multiple-outputs.sh hook is able to do too.
A nix specific set of tools for converting icon files
that are not in a freedesktop ready format.
I plan on using these tools for both `keepass` and
`retroarch` packages. It may benifit many other packages.
For every *.{exe,dll} in $output/bin/ we try to find all (potential)
transitive dependencies and symlink those DLLs into $output/bin
so they are found on invocation.
(DLLs are first searched in the directory of the running exe file.)
The links are relative, so relocating whole /nix/store won't break them.
The hook is activated on cygwin and when cross-compiling to mingw.
- the default --docdir is typically DATAROOTDIR/doc/pkgName
- I saw no other way than to employ some magic to guess this `pkgName`
- user can override it by setting $shareDocName
I supplied meta.outputsToInstall automatically in all
mkDerivation products, but some packages still don't use it.
The reported case: jekyll -> bundlerEnv -> buildEnv -> runCommand.
The /nix path in 4d200538 of the layer tar didn't exist for some
packages, such as cacert. This is because cacert just creates an /etc
directory and doesn't depend on any other /nix paths. If we tried
putting this directory in the tar and using overlayfs with it, we'd get
"Invalid argument" when trying to remove the directory.
We now check whether the closure is non-empty before telling tar to
store the /nix directory.
Fixes #14710.
There were two sources of non-determinisim coming into the images. The
first was tar mtimes, the second was pigz/gzip times.
An example image now passes with the --check flag.
I think the intention of this functionality was to provide a simple
alternative to the "runAsRoot" and "contents" attributes.
The implementation caused very slow builds of Docker images. Almost all
of the build time was spent in IO for tar, due to tarballs being
created, immediately extracted, then recreated. I had 30 minute builds
on some of my images which are now down to less than 2 minutes. A couple
of other users on #nix IRC have observed similar improvements.
The implementation also mutated the produced Docker layers without
changing their hashes. Using non-empty tarballs would produce images
which got cached incorrectly in Docker.
I have a commit which just fixes the performance problem but I opted to
completely remove the tarball feature after I found out that it didn't
correctly implement the Docker Image Specification due to the broken
hashing.
This is to avoid unwanted side effects when installing a wrapped emacs in the environment:
* All executables in the dependencies become available in the user environment
* All site-lisp binaries in the dependencies become accessible to unwrapped emacs
Also, both bin and site-lisp would generate conflicts so installing a wrapped emacs becomes really cumbersome
1. Update bower2nix version and add new/updated dependencies into
node-packages-generated.nix. This was done manually, with npm2nix
generating the initial set of derivations. In future, it would be
nice to have an automatic process (see #10358, #9332).
2. Add an override to nodePackages.bower2nix wrapping the commands so
that git is on the PATH.
3. Update fetchbower to support new command-line options of bower2nix,
and to allow github URL tag versions.
Previously, nix-prefetch-git would report the same JSON whether submodules were being fetched or not; with this change, the --fetch-submodules option will cause the JSON output to include "fetchSubmodules": true, so that fetchgit (builtins.fromJSON (builtins.readFile ./path/to/output.json)) will work.
The following parameters are now available:
* hardeningDisable
To disable specific hardening flags
* hardeningEnable
To enable specific hardening flags
Only the cc-wrapper supports this right now, but these may be reused by
other wrappers, builders or setup hooks.
cc-wrapper supports the following flags:
* fortify
* stackprotector
* pie (disabled by default)
* pic
* strictoverflow
* format
* relro
* bindnow
Some recent perl version introduced "keys" to return the keys
in random order. As some of the packages are solved by "provides" and
based on the order, this randomness affects what packages get into the
closure.
This problem may be in other nix perl scripts.
The importance of glibc makes it worthwhile to provide debug
symbols. However, this revealed an issue with separateDebugInfo: it
was indiscriminately adding --build-id to all ld invocations, while in
fact it should only do that for final links. Glibc also uses non-final
("relocatable") links, leading to subsequent failure to apply a build
ID ("Cannot create .note.gnu.build-id section, --build-id
ignored"). So now ld-wrapper.sh only passes --build-id for final
links.
Otherwise, when building glibc and other packages, the "strip" from
bootstrapTools is used, which doesn't recognise some tags produced by
the newer "ld" from binutils.
The two lines I removed technically assert the exact same thing, since `!a -> b`
is equivalent to `a || b`. So, I replaced the two lines with the more symmetric
form to make it clearer.
This commit fixes #6651.
Before this change the `nix-prefetch-git` script would use a different store
name than nix's `fetchgit` function. Because of that it was not possible to
use `nix-prefetch-git` as a way to pre-populate the store (for example when
the user it using private git dependencies that needs access to the ssh agent)
In some cases the $sourceRoot is missing. Skip the hook instead
of showing the following cryptic error:
find: cannot search `': No such file or directory
/nix/store/0p1afvl8jcpi6dvsq2n58i90w9c59vz1-set-source-date-epoch-to-latest.sh: line 12: [: : integer expression expected
vcunat removed the warning; the hook will just skip silently in these cases.
Perhaps someone can improve on it some time.
Regression introduced by 4529ed1259.
I've missed this in #5096, not because of a messed up rebase as I have
guessed from a comment on #12635 but missed this in the first place.
The testing I did while working on the pull request weren't exhaustive
enough to cover this, because I haven't tested with packages that use
the propagatedUserEnvPkgs attribute.
In order to make the test a bit more exhaustive this time, let's test it
using:
nix-build -E 'with import ./. {}; buildEnv {
name = "testenv";
paths = [
pkgs.hello pkgs.binutils pkgs.libsoup pkgs.gnome3.yelp
pkgs.gnome3.totem
];
}'
And with this commit the errors no longer show up and the environment is
built correctly.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Fixes: #12635
- Now `pkg.outputUnspecified = true` but this attribute is missing in
every output, so we can recognize whether the user chose or not.
If (s)he didn't choose, we put `pkg.bin or pkg.out or pkg` into
`systemPackages`.
- `outputsToLink` is replaced by `extraOutputsToLink`.
We add extra outputs *regardless* of whether the user chose anything.
It's mainly meant for outputs with docs and debug symbols.
- Note that as a result, some libraries will disappear from system path.
* authorization token is optional
* registry url is taken from X-Docker-Endpoints header
* pull.sh correctly resumes partial layer downloads
* detjson.py does not fail on missing keys
The comment related to the `deepClone` and `no-deepClone` options was
misleading as these options have no relation with submodules, but on the
the depth in `git clone --depth n`.
Login mode can cause hidden problems, e.g. #12406. Generally we don't want
to read user's .bash_profile when we don't start an interactive shell inside
a chroot.
These environment variables allow using fetchgit with git:// URLs using
the SOCKS proxy technique described in 'Using Git with a SOCKS proxy':
http://www.patthoyts.tk/blog/using-git-with-socks-proxy.html
Briefly, GIT_PROXY_COMMAND is set to a script which invokes connect[1],
which reads SOCKS_PROXY, which might be pointing to a local instance of
'ssh -D'.
[1] pkgs/tools/networking/connect
The ld-wrapper.sh script calls `readlink` in some circumstances. We need
to ensure that this is the `readlink` from the `coreutils` package so
that flag support is as expected.
This is accomplished by explicitly setting PATH at the top of each shell
script.
Without doing this, the following happens with a trivial `main.c`:
```
nix-env -f "<nixpkgs>" -iA pkgs.clang
$ clang main.c -L /nix/../nix/store/2ankvagznq062x1gifpxwkk7fp3xwy63-xnu-2422.115.4/Library -o a.out
readlink: illegal option -- f
usage: readlink [-n] [file ...]
```
The key element is the `..` in the path supplied to the linker via a
`-L` flag. With this patch, the above invocation works correctly on
darwin, whose native `/usr/bin/readlink` does not support the `-f` flag.
The explicit path also ensures that the `grep` called by `cc-wrapper.sh`
is the one from Nix.
Fixes #6447
Building packages requires package-build.el from Melpa, but installing
packages only requires package.el. Packages from ELPA are already built,
so there is no need to involve package-build.el.
When building a package from a Melpa recipe file, get the Emacs package
name from the recipe. Nix is more restrictive about packages names than
Emacs, so the Nix name for a package is sometimes different.
Checking file contents is redundant in this case, because we will go
ahead anyway, regardless of whether the content is the same.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Originally wanted to include ignoreCollisions in cups-progs, but I think
it's better if we use ignoreCollisions only if there are _real_
collisions between files with different contents.
Of course, we also check whether the file permissions match, so you get
a collision if contents are the same but the permissions are different.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
