Commands such as "cp -p" also use fsetxattr() in addition to fchown(),
so we need to make sure these syscalls always return successful as well
in order to avoid nasty "Invalid value" errors.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
What we basically want is a seccomp mode 2 BPF program like this but for
every architecture:
BPF_STMT(BPF_LD+BPF_W+BPF_ABS, offsetof(struct seccomp_data, nr)),
BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, __NR_chown, 4, 0),
BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, __NR_fchown, 3, 0),
BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, __NR_fchownat, 2, 0),
BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, __NR_lchown, 1, 0),
BPF_STMT(BPF_RET+BPF_K, SECCOMP_RET_ALLOW),
BPF_STMT(BPF_RET+BPF_K, SECCOMP_RET_ERRNO)
However, on 32 bit architectures we do have chown32, lchown32 and
fchown32, so we'd need to add all the architecture blurb which
libseccomp handles for us.
So we only need to make sure that we add the 32bit seccomp arch while
we're on x86_64 and otherwise we just stay at the native architecture
which was set during seccomp_init(), which more or less replicates
setting 32bit personality during runChild().
The FORCE_SUCCESS() macro here could be a bit less ugly but I think
repeating the seccomp_rule_add() all over the place is way uglier.
Another way would have been to create a vector of syscalls to iterate
over, but that would make error messages uglier because we can either
only print the (libseccomp-internal) syscall number or use
seccomp_syscall_resolve_num_arch() to get the name or even make the
vector a pair number/name, essentially duplicating everything again.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
We're going to use libseccomp instead of creating the raw BPF program,
because we have different syscall numbers on different architectures.
Although our initial seccomp rules will be quite small it really doesn't
make sense to generate the raw BPF program because we need to duplicate
it and/or make branches on every single architecture we want to suuport.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
This reverts commit ff0c0b645c.
We're going to use seccomp to allow "cp -p" and force chown-related
syscalls to always return 0.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
This solves a problem whereby if /gnu/store/.links had enough entries,
ext4's directory index would be full, leading to link(2) returning
ENOSPC.
* nix/libstore/optimise-store.cc (LocalStore::optimisePath_): Upon
ENOSPC from link(2), print a message and return instead of throwing a
'SysError'.
The SSHStore PR adds this functionality to the daemon, but we have to
handle the case where the Nix daemon is 1.11.
Also, don't require signatures for trusted users. This restores 1.11
behaviour.
Fixes https://github.com/NixOS/hydra/issues/398.
For example, you can now set
build-sandbox-paths = /dev/nvidiactl?
to specify that /dev/nvidiactl should only be mounted in the sandbox
if it exists in the host filesystem. This is useful e.g. for EC2
images that should support both CUDA and non-CUDA instances.
The removal of CachedFailure caused the value of TimedOut to change,
which broke timed-out handling in Hydra (so timed-out builds would
show up as "aborted" and would be retried, e.g. at
http://hydra.nixos.org/build/42537427).
The store parameter "write-nar-listing=1" will cause BinaryCacheStore
to write a file ‘<store-hash>.ls.xz’ for each ‘<store-hash>.narinfo’
added to the binary cache. This file contains an XZ-compressed JSON
file describing the contents of the NAR, excluding the contents of
regular files.
E.g.
{
"version": 1,
"root": {
"type": "directory",
"entries": {
"lib": {
"type": "directory",
"entries": {
"Mcrt1.o": {
"type": "regular",
"size": 1288
},
"Scrt1.o": {
"type": "regular",
"size": 3920
},
}
}
}
...
}
}
(The actual file has no indentation.)
This is intended to speed up the NixOS channels programs index
generator [1], since fetching gazillions of large NARs from
cache.nixos.org is currently a bottleneck for updating the regular
(non-small) channel.
[1] https://github.com/NixOS/nixos-channel-scripts/blob/master/generate-programs-index.cc
We can now write
throw Error("file '%s' not found", path);
instead of
throw Error(format("file '%s' not found") % path);
and similarly
printError("file '%s' not found", path);
instead of
printMsg(lvlError, format("file '%s' not found") % path);
The fact that queryPathInfo() is synchronous meant that we needed a
thread for every concurrent binary cache lookup, even though they end
up being handled by the same download thread. Requiring hundreds of
threads is not a good idea. So now there is an asynchronous version of
queryPathInfo() that takes a callback function to process the
result. Similarly, enqueueDownload() now takes a callback rather than
returning a future.
Thus, a command like
nix path-info --store https://cache.nixos.org/ -r /nix/store/slljrzwmpygy1daay14kjszsr9xix063-nixos-16.09beta231.dccf8c5
that returns 4941 paths now takes 1.87s using only 2 threads (the main
thread and the downloader thread). (This is with a prewarmed
CloudFront.)
It's a slight misnomer now because it actually limits *all* downloads,
not just binary cache lookups.
Also add a "enable-http2" option to allow disabling use of HTTP/2
(enabled by default).
The binary cache store can now use HTTP/2 to do lookups. This is much
more efficient than HTTP/1.1 due to multiplexing: we can issue many
requests in parallel over a single TCP connection. Thus it's no longer
necessary to use a bunch of concurrent TCP connections (25 by
default).
For example, downloading 802 .narinfo files from
https://cache.nixos.org/, using a single TCP connection, takes 11.8s
with HTTP/1.1, but only 0.61s with HTTP/2.
This did require a fairly substantial rewrite of the Downloader class
to use the curl multi interface, because otherwise curl wouldn't be
able to do multiplexing for us. As a bonus, we get connection reuse
even with HTTP/1.1. All downloads are now handled by a single worker
thread. Clients call Downloader::enqueueDownload() to tell the worker
thread to start the download, getting a std::future to the result.
This largely reverts c68e5913c7. Running
builds as root breaks "cp -p", since when running as root, "cp -p"
assumes that it can succesfully chown() files. But that's not actually
the case since the user namespace doesn't provide a complete uid
mapping. So it barfs with a fatal error message ("cp: failed to
preserve ownership for 'foo': Invalid argument").
This fixes an assertion failure in "assert(goal);" in
Worker::waitForInput() after a substitution goal is cancelled by the
termination of another goal. The problem was the line
//worker.childTerminated(shared_from_this()); // FIXME
in the SubstitutionGoal destructor. This was disabled because
shared_from_this() obviously doesn't work from a destructor. So we now
use a real pointer for object identity.
The inner lambda was returning a SQLite-internal char * rather than a
std::string, leading to Hydra errors liks
Caught exception in Hydra::Controller::Root->narinfo "path âø£â is not in the Nix store at /nix/store/6mvvyb8fgwj23miyal5mdr8ik4ixk15w-hydra-0.1.1234.abcdef/libexec/hydra/lib/Hydra/Controller/Root.pm line 352."
For example, you can now say:
configureFlags = "--prefix=${placeholder "out"} --includedir=${placeholder "dev"}";
The strings returned by the ‘placeholder’ builtin are replaced at
build time by the actual store paths corresponding to the specified
outputs.
Previously, you had to work around the inability to self-reference by doing stuff like:
preConfigure = ''
configureFlags+=" --prefix $out --includedir=$dev"
'';
or rely on ad-hoc variable interpolation semantics in Autoconf or Make
(e.g. --prefix=\$(out)), which doesn't always work.
This makes it easier to create a diverted store, i.e.
NIX_REMOTE="local?root=/tmp/root"
instead of
NIX_REMOTE="local?real=/tmp/root/nix/store&state=/tmp/root/nix/var/nix" NIX_LOG_DIR=/tmp/root/nix/var/log
For one particular NixOS configuration, this cut the runtime of
"nix-store -r --dry-run" from 6m51s to 3.4s. It also fixes a bug in
the size calculation that was causing certain paths to be counted
twice, e.g. before:
these paths will be fetched (1249.98 MiB download, 2995.74 MiB unpacked):
and after:
these paths will be fetched (1219.56 MiB download, 2862.17 MiB unpacked):
This way, all builds appear to have a uid/gid of 0 inside the
chroot. In the future, this may allow using programs like
systemd-nspawn inside builds, but that will require assigning a larger
UID/GID map to the build.
Issue #625.
This allows an unprivileged user to perform builds on a diverted store
(i.e. where the physical store location differs from the logical
location).
Example:
$ NIX_LOG_DIR=/tmp/log NIX_REMOTE="local?real=/tmp/store&state=/tmp/var" nix-build -E \
'with import <nixpkgs> {}; runCommand "foo" { buildInputs = [procps nettools]; } "id; ps; ifconfig; echo $out > $out"'
will do a build in the Nix store physically in /tmp/store but
logically in /nix/store (and thus using substituters for the latter).
This is a convenience command to allow users who are not privileged to
create /nix/store to use Nix with regular binary caches. For example,
$ NIX_REMOTE="local?state=$HOME/nix/var&real=/$HOME/nix/store" nix run firefox bashInteractive
will download Firefox and bash from cache.nixos.org, then start a
shell in which $HOME/nix/store is mounted on /nix/store.
This is primarily to subsume the functionality of the
copy-from-other-stores substituter. For example, in the NixOS
installer, we can now do (assuming we're in the target chroot, and the
Nix store of the installation CD is bind-mounted on /tmp/nix):
$ nix-build ... --option substituters 'local?state=/tmp/nix/var&real=/tmp/nix/store'
However, unlike copy-from-other-stores, this also allows write access
to such a store. One application might be fetching substitutes for
/nix/store in a situation where the user doesn't have sufficient
privileges to create /nix, e.g.:
$ NIX_REMOTE="local?state=/home/alice/nix/var&real=/home/alice/nix/store" nix-build ...
The function builtins.fetchgit fetches Git repositories at evaluation
time, similar to builtins.fetchTarball. (Perhaps the name should be
changed, being confusing with respect to Nixpkgs's fetchgit function,
with works at build time.)
Example:
(import (builtins.fetchgit git://github.com/NixOS/nixpkgs) {}).hello
or
(import (builtins.fetchgit {
url = git://github.com/NixOS/nixpkgs-channels;
rev = "nixos-16.03";
}) {}).hello
Note that the result does not contain a .git directory.
If --no-build-output is given (which will become the default for the
"nix" command at least), show the last 10 lines of the build output if
the build fails.
This allows commands like "nix verify --all" or "nix path-info --all"
to work on S3 caches.
Unfortunately, this requires some ugly hackery: when querying the
contents of the bucket, we don't want to have to read every .narinfo
file. But the S3 bucket keys only include the hash part of each store
path, not the name part. So as a special exception
queryAllValidPaths() can now return store paths *without* the name
part, and queryPathInfo() accepts such store paths (returning a
ValidPathInfo object containing the full name).
Caching path info is generally useful. For instance, it speeds up "nix
path-info -rS /run/current-system" (i.e. showing the closure sizes of
all paths in the closure of the current system) from 5.6s to 0.15s.
This also eliminates some APIs like Store::queryDeriver() and
Store::queryReferences().
For convenience, you can now say
$ nix-env -f channel:nixos-16.03 -iA hello
instead of
$ nix-env -f https://nixos.org/channels/nixos-16.03/nixexprs.tar.xz -iA hello
Similarly,
$ nix-shell -I channel:nixpkgs-unstable -p hello
$ nix-build channel:nixos-15.09 -A hello
Abstracting over the NixOS/Nixpkgs channels location also allows us to
use a more efficient transport (e.g. Git) in the future.
This specifies the number of distinct signatures required to consider
each path "trusted".
Also renamed ‘--no-sigs’ to ‘--no-trust’ for the flag that disables
verifying whether a path is trusted (since a path can also be trusted
if it has no signatures, but was built locally).
These are content-addressed paths or outputs of locally performed
builds. They are trusted even if they don't have signatures, so "nix
verify-paths" won't complain about them.
Unlike "nix-store --verify-path", this command verifies signatures in
addition to store path contents, is multi-threaded (especially useful
when verifying binary caches), and has a progress indicator.
Example use:
$ nix verify-paths --store https://cache.nixos.org -r $(type -p thunderbird)
...
[17/132 checked] checking ‘/nix/store/rawakphadqrqxr6zri2rmnxh03gqkrl3-autogen-5.18.6’
Doing a chdir() is a bad idea in multi-threaded programs, leading to
failures such as
error: cannot connect to daemon at ‘/nix/var/nix/daemon-socket/socket’: No such file or directory
Since Linux doesn't have a connectat() syscall like FreeBSD, there is
no way we can support this in a race-free way.
This enables an optimisation in hydra-queue-runner, preventing a
download of a NAR it just uploaded to the cache when reading files
like hydra-build-products.
This enables an optimisation in hydra-queue-runner, preventing a
download of a NAR it just uploaded to the cache when reading files
like hydra-build-products.
This allows a RemoteStore object to be used safely from multiple
threads concurrently. It will make multiple daemon connections if
necessary.
Note: pool.hh and sync.hh have been copied from the Hydra source tree.
This is currently only used by the Hydra queue runner rework, but like
eff5021eaa it presumably will be useful
for the C++ rewrite of nix-push and
download-from-binary-cache. (@shlevy)
Also, move a few free-standing functions into StoreAPI and Derivation.
Also, introduce a non-nullable smart pointer, ref<T>, which is just a
wrapper around std::shared_ptr ensuring that the pointer is never
null. (For reference-counted values, this is better than passing a
"T&", because the latter doesn't maintain the refcount. Usually, the
caller will have a shared_ptr keeping the value alive, but that's not
always the case, e.g., when passing a reference to a std::thread via
std::bind.)
For example,
$ nix-build --hash -A nix-repl.src
will build the fixed-output derivation nix-repl.src (a fetchFromGitHub
call), but instead of *verifying* the hash given in the Nix
expression, it prints out the resulting hash, and then moves the
result to its content-addressed location in the Nix store. E.g
build produced path ‘/nix/store/504a4k6zi69dq0yjc0bm12pa65bccxam-nix-repl-8a2f5f0607540ffe56b56d52db544373e1efb980-src’ with sha256 hash ‘0cjablz01i0g9smnavhf86imwx1f9mnh5flax75i615ml71gsr88’
The goal of this is to make all nix-prefetch-* scripts unnecessary: we
can just let Nix run the real thing (i.e., the corresponding fetch*
derivation).
Another example:
$ nix-build --hash -E 'with import <nixpkgs> {}; fetchgit { url = "https://github.com/NixOS/nix.git"; sha256 = "ffffffffffffffffffffffffffffffffffffffffffffffffffff"; }'
...
git revision is 9e7c1a4bbd
...
build produced path ‘/nix/store/gmsnh9i7x4mb7pyd2ns7n3c9l90jfsi1-nix’ with sha256 hash ‘1188xb621diw89n25rifqg9lxnzpz7nj5bfh4i1y3dnis0dmc0zp’
(Having to specify a fake sha256 hash is a bit annoying...)
Previously files in the Nix store were owned by root or by nixbld,
depending on whether they were created by a substituter or by a
builder. This doesn't matter much, but causes spurious diffoscope
differences. So use root everywhere.
E.g.
$ nix-build pkgs/stdenv/linux/ -A stage1.pkgs.perl --check
nix-store: src/libstore/build.cc:1323: void nix::DerivationGoal::tryToBuild(): Assertion `buildMode != bmCheck || validPaths.size() == drv->outputs.size()' failed.
when perl.out exists but perl.man doesn't. The fix is to only check
the outputs that exist. Note that "nix-build -A stage1.pkgs.all
--check" will still give a (proper) error in this case.
This was observed in the deb_debian7x86_64 build:
http://hydra.nixos.org/build/29973215
Calling c_str() on a temporary should be fine because the temporary
shouldn't be destroyed until after the execl() call, but who knows...
If repair found a corrupted/missing path that depended on a
multiple-output derivation, and some of the outputs of the latter were
not present, it failed with a message like
error: path ‘/nix/store/cnfn9d5fjys1y93cz9shld2xwaibd7nn-bash-4.3-p42-doc’ is not valid
This makes Darwin consistent with Linux: Nix expressions can't break
out of the sandbox unless relaxed sandbox mode is enabled.
For the normal sandbox mode this will require fixing #759 however.
Otherwise, since the call to write a "d" character to the lock file
can fail with ENOSPC, we can get an unhandled exception resulting in a
call to terminate().
Caused by 8063fc497a. If tmpDir !=
tmpDirInSandbox (typically when there are multiple concurrent builds
with the same name), the *Path attribute would not point to an
existing file. This caused Nixpkgs' writeTextFile to write an empty
file. In particular this showed up as hanging VM builds (because it
would run an empty run-nixos-vm script and then wait for it to finish
booting).
This is arguably nitpicky, but I think this new formulation is even
clearer. My thinking is that it's easier to comprehend when the
calculated hash value is displayed close to the output path. (I think it
is somewhat similar to eliminating double negatives in logic
statements.)
The formulation is inspired / copied from the OpenEmbedded build tool,
bitbake.
Rather than using $<host-TMPDIR>/nix-build-<drvname>-<number>, the
temporary directory is now always /tmp/nix-build-<drvname>-0. This
improves bitwise-exact reproducibility for builds that store $TMPDIR
in their build output. (Of course, those should still be fixed...)
Temporarily allow derivations to describe their full sandbox profile.
This will be eventually scaled back to a more secure setup, see the
discussion at #695
Nix reports a hash mismatch saying:
output path ‘foo’ should have sha256 hash ‘abc’, instead has ‘xyz’
That message is slightly ambiguous and some people read that statement
to mean the exact opposite of what it is supposed to mean. After this
patch, the message will be:
Nix expects output path ‘foo’ to have sha256 hash ‘abc’, instead it has ‘xyz’
- rename options but leav old names as lower-priority aliases,
also "-dirs" -> "-paths" to get closer to the meaning
- update docs to reflect the new names (old aliases are not documented),
including a new file with release notes
- tests need an update after corresponding changes to nixpkgs
- __noChroot is left as it is (after discussion on the PR)
Passing "--option build-repeat <N>" will cause every build to be
repeated N times. If the build output differs between any round, the
build is rejected, and the output paths are not registered as
valid. This is primarily useful to verify build determinism. (We
already had a --check option to repeat a previously succeeded
build. However, with --check, non-deterministic builds are registered
in the DB. Preventing that is useful for Hydra to ensure that
non-deterministic builds don't end up getting published at all.)
This reverts commit 79ca503332. Ouch,
never noticed this. We definitely don't want to allow builds to have
arbitrary access to /bin and /usr/bin, because then they can (for
instance) bring in a bunch of setuid programs. Also, we shouldn't be
encouraging the use of impurities in the default configuration.
If automatic store optimisation is enabled, and a hard-linked file in
the store gets corrupted, then the corresponding .links entry will
also be corrupted. In that case, trying to repair with --repair or
--repair-path won't work, because the new "good" file will be replaced
by a hard link to the corrupted file. We can catch most of these cases
by doing a sanity-check on the file sizes.
This removes the need to have multiple downloads in the stdenv
bootstrap process (like a separate busybox binary for Linux, or
curl/mkdir/sh/bzip2 for Darwin). Now all those files can be combined
into a single NAR.
This makes it consistent with the Nixpkgs fetchurl and makes it work
in chroots. We don't need verification because the hash of the result
is checked anyway.
The stack allocated for the builder was way too small (32 KB). This is
sufficient for normal derivations, because they just do some setup and
then exec() the actual builder. But for the fetchurl builtin
derivation it's not enough. Also, allocating the stack on the caller's
stack was fishy business.
Previously, pkg-config was already queried for libsqlite3's and
libcurl's link flags. However they were not used, but hardcoded
instead. This commit replaces the hardcoded LDFLAGS by the ones
provided by pkg-config in a similar pattern as already used for
libsodium.
Fixes https://github.com/NixOS/nixpkgs/issues/9504.
Note that this means we may have a non-functional /bin/sh in the
chroot while rebuilding Bash or one of its dependencies. Ideally those
packages don't rely on /bin/sh though.
This ensures that 1) the derivation doesn't change when Nix changes;
2) the derivation closure doesn't contain Nix and its dependencies; 3)
we don't have to rely on ugly chroot hacks.
In particular, hydra-queue-runner can now distinguish between remote
build / substitution / already-valid. For instance, if a path already
existed on the remote side, we don't want to store a log file.
Previously, to build a derivation remotely, we had to copy the entire
closure of the .drv file to the remote machine, even though we only
need the top-level derivation. This is very wasteful: the closure can
contain thousands of store paths, and in some Hydra use cases, include
source paths that are very large (e.g. Git/Mercurial checkouts).
So now there is a new operation, StoreAPI::buildDerivation(), that
performs a build from an in-memory representation of a derivation
(BasicDerivation) rather than from a on-disk .drv file. The only files
that need to be in the Nix store are the sources of the derivation
(drv.inputSrcs), and the needed output paths of the dependencies (as
described by drv.inputDrvs). "nix-store --serve" exposes this
interface.
Note that this is a privileged operation, because you can construct a
derivation that builds any store path whatsoever. Fixing this will
require changing the hashing scheme (i.e., the output paths should be
computed from the other fields in BasicDerivation, allowing them to be
verified without access to other derivations). However, this would be
quite nice because it would allow .drv-free building (e.g. "nix-env
-i" wouldn't have to write any .drv files to disk).
Fixes#173.
The following patch is an attempt to address this bug (see
<http://bugs.gnu.org/18994>) by preserving the supplementary groups of
build users in the build environment.
In practice, I would expect that supplementary groups would contain only
one or two groups: the build users group, and possibly the “kvm” group.
[Changed &at(0) to data() and removed tabs - Eelco]
Not substituting builds with "preferLocalBuild = true" was a bad idea,
because it didn't take the cost of dependencies into account. For
instance, if we can't substitute a fetchgit call, then we have to
download/build git and all its dependencies.
Partially reverts 5558652709 and adds a
new derivation attribute "allowSubstitutes" to specify whether a
derivation may be substituted.
Nixpkgs' writeTextAsFile does this:
mv "$textPath" "$n"
Since $textPath was owned by root, if $textPath is on the same
filesystem as $n, $n will be owned as root. As a result, the build
result was rejected as having suspicious ownership.
http://hydra.nixos.org/build/22836807
Hello!
The patch below adds a ‘verifyStore’ RPC with the same signature as the
current LocalStore::verifyStore method.
Thanks,
Ludo’.
>From aef46c03ca77eb6344f4892672eb6d9d06432041 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Ludovic=20Court=C3=A8s?= <ludo@gnu.org>
Date: Mon, 1 Jun 2015 23:17:10 +0200
Subject: [PATCH] Add a 'verifyStore' remote procedure call.
This hook can be used to set system-specific per-derivation build
settings that don't fit into the derivation model and are too complex or
volatile to be hard-coded into nix. Currently, the pre-build hook can
only add chroot dirs/files through the interface, but it also has full
access to the chroot root.
The specific use case for this is systems where the operating system ABI
is more complex than just the kernel-support system calls. For example,
on OS X there is a set of system-provided frameworks that can reliably
be accessed by any program linked to them, no matter the version the
program is running on. Unfortunately, those frameworks do not
necessarily live in the same locations on each version of OS X, nor do
their dependencies, and thus nix needs to know the specific version of
OS X currently running in order to make those frameworks available. The
pre-build hook is a perfect mechanism for doing just that.
This hook can be used to set system specific per-derivation build
settings that don't fit into the derivation model and are too complex or
volatile to be hard-coded into nix. Currently, the pre-build hook can
only add chroot dirs/files.
The specific use case for this is systems where the operating system ABI
is more complex than just the kernel-supported system calls. For
example, on OS X there is a set of system-provided frameworks that can
reliably be accessed by any program linked to them, no matter the
version the program is running on. Unfortunately, those frameworks do
not necessarily live in the same locations on each version of OS X, nor
do their dependencies, and thus nix needs to know the specific version
of OS X currently running in order to make those frameworks available.
The pre-build hook is a perfect mechanism for doing just that.
This is because we don't want to do HTTP requests on every evaluation,
even though we can prevent a full redownload via the cached ETag. The
default is one hour.
This was causing NixOS VM tests to fail mysteriously since
5ce50cd99e. Nscd could (sometimes) no
longer read /etc/hosts:
open("/etc/hosts", O_RDONLY|O_CLOEXEC) = -1 EACCES (Permission denied)
Probably there was some wacky interaction between the guest kernel and
the 9pfs implementation in QEMU.
Thus, for example, to get /bin/sh in a chroot, you only need to
specify /bin/sh=${pkgs.bash}/bin/sh in build-chroot-dirs. The
dependencies of sh will be added automatically.
I'm seeing hangs in Glibc's setxid_mark_thread() again. This is
probably because the use of an intermediate process to make clone()
safe from a multi-threaded program (see
524f89f139) is defeated by the use of
vfork(), since the intermediate process will have a copy of Glibc's
threading data structures due to the vfork(). So use a regular fork()
again.
If ‘build-use-chroot’ is set to ‘true’, fixed-output derivations are
now also chrooted. However, unlike normal derivations, they don't get
a private network namespace, so they can still access the
network. Also, the use of the ‘__noChroot’ derivation attribute is
no longer allowed.
Setting ‘build-use-chroot’ to ‘relaxed’ gives the old behaviour.
chroot only changes the process root directory, not the mount namespace root
directory, and it is well-known that any process with chroot capability can
break out of a chroot "jail". By using pivot_root as well, and unmounting the
original mount namespace root directory, breaking out becomes impossible.
Non-root processes typically have no ability to use chroot() anyway, but they
can gain that capability through the use of clone() or unshare(). For security
reasons, these syscalls are limited in functionality when used inside a normal
chroot environment. Using pivot_root() this way does allow those syscalls to be
put to their full use.
I.e., not readable to the nixbld group. This improves purity a bit for
non-chroot builds, because it prevents a builder from enumerating
store paths (i.e. it can only access paths it knows about).
Especially in WAL mode on a highly loaded machine, this is not a good
idea because it results in a WAL file of approximately the same size
ad the database, which apparently cannot be deleted while anybody is
accessing it.
For the "stdenv accidentally referring to bootstrap-tools", it seems
easier to specify the path that we don't want to depend on, e.g.
disallowedRequisites = [ bootstrapTools ];
It turns out that using clone() to start a child process is unsafe in
a multithreaded program. It can cause the initialisation of a build
child process to hang in setgroups(), as seen several times in the
build farm:
The reason is that Glibc thinks that the other threads of the parent
exist in the child, so in setxid_mark_thread() it tries to get a futex
that has been acquired by another thread just before the clone(). With
fork(), Glibc runs pthread_atfork() handlers that take care of this
(in particular, __reclaim_stacks()). But clone() doesn't do that.
Fortunately, we can use fork()+unshare() instead of clone() to set up
private namespaces.
See also https://www.mail-archive.com/lxc-devel@lists.linuxcontainers.org/msg03434.html.
