There was a bug in the pam_mount module that crypt mount options were
not passed to the mount.crypt command. This is now fixed and
additionally, a cryptMountOptions NixOS option is added to define mount
options that should apply to all crypt mounts.
Fixes#230920
According to Ted Unangst, since doas evaluates rules in a last
matched manner, it is prudent to have the "permit root to do everything
without a password at the end of the file.
Source: https://flak.tedunangst.com/post/doas-mastery
This reverts commit 2265160fc0 and
e56db577a1.
Ideally, we shouldn't cause friction for users that bump `stateVersion`,
and I'd consider having to switch and/or manually hardcode a UID/GID
to supress the warning friction. I think it'd be more beneficial to, in
this rare case of an ID being missed, just let it be until more
discussion happens surrounding this overall issue.
See https://github.com/NixOS/nixpkgs/pull/217785 for more context.
this converts meta.doc into an md pointer, not an xml pointer. since we
no longer need xml for manual chapters we can also remove support for
manual chapters from md-to-db.sh
since pandoc converts smart quotes to docbook quote elements and our
nixos-render-docs does not we lose this distinction in the rendered
output. that's probably not that bad, our stylesheet didn't make use of
this anyway (and pre-23.05 versions of the chapters didn't use quote
elements either).
also updates the nixpkgs manual to clarify that option docs support all
extensions (although it doesn't support headings at all, so heading
anchors don't work by extension).
MD can only do the latter, so change them all over now to keeps diffs reviewable.
this also includes <literal><xref> -> <xref> where options are referenced since
the reference will implicitly add an inner literal tag.
markdown cannot represent those links. remove them all now instead of in
each chapter conversion to keep the diff for each chapter small and more
understandable.
fscrypt can automatically unlock directories with the user's login
password. To do this it ships a PAM module which reads the user's
password and loads the respective keys into the user's kernel keyring.
Significant inspiration was taken from the ecryptfs implementation.
With Go 1.19 calls to setrlimit are required for lego to run.
While we could allow setrlimit alone, I think it is not unreasonable to
allow @resources in general.
Closes: #197513
Lego has a built-in mechanism for sleeping for a random amount
of time before renewing a certificate. In our environment this
is not only unnecessary (as our systemd timer takes care of it)
but also unwanted since it slows down the execution of the
systemd service encompassing it, thus also slowing down the
start up of any services its depending on.
Also added FixedRandomDelay to the timer for more predictability.
Fixes#190493
Check if an actual key file exists. This does not
completely cover the work accountHash does to ensure
that a new account is registered when account
related options are changed.
Fixes#191794
Lego threw a permission denied error binding to port 80.
AmbientCapabilities with CAP_NET_BIND_SERVICE was required.
Also added a test for this.
Before this change, the description for
security.wrappers.<name>.capabilities made it seem like you could just
string together the names of capabilities like this:
capabilities = "CAP_SETUID,CAP_SETGID";
In reality, each item in the list must be a full-on capability clause:
capabilities = "CAP_SETUID=ep,CAP_SETGID+i";
Summary: fix errors with example code in the manual that shows how to set up DNS-01 verification via the acme protocol, e.g. for those who want to get wildcard certificates from Let's Encrypt.
Fix syntax error in nix arrays (there should not be commas.)
Fix permissions on /var/lib/secrets so it can be read by bind daemon. Without this fix bind won't start.
Add the missing feature: put the generated secret into certs.secret
most of these are hidden because they're either part of a submodule that
doesn't have its type rendered (eg because the submodule type is used in
an either type) or because they are explicitly hidden. some of them are
merely hidden from nix-doc-munge by how their option is put together.
conversions were done using https://github.com/pennae/nix-doc-munge
using (probably) rev f34e145 running
nix-doc-munge nixos/**/*.nix
nix-doc-munge --import nixos/**/*.nix
the tool ensures that only changes that could affect the generated
manual *but don't* are committed, other changes require manual review
and are discarded.
there are sufficiently few variable list around, and they are
sufficiently simple, that it doesn't seem helpful to add another
markdown extension for them. rendering differences are small, except in
the tor module: admonitions inside other blocks cannot be made to work
well with mistune (and likely most other markdown processors), so those
had to be shuffled a bit. we also lose paragraph breaks in the list
items due to how we have to render from markdown to docbook, but once we
remove docbook from the pipeline those paragraph breaks will be restored.
mostly no rendering changes. some lists (like simplelist) don't have an
exact translation to markdown, so we use a comma-separated list of
literals instead.
using regular strings works well for docbook because docbook is not as
whitespace-sensitive as markdown. markdown would render all of these as
code blocks when given the chance.
this renders the same in the manpage and a little more clearly in the
html manual. in the manpage there continues to be no distinction from
regular text, the html manual gets code-type markup (which was probably
the intention for most of these uses anyway).
now nix-doc-munge will not introduce whitespace changes when it replaces
manpage references with the MD equivalent.
no change to the manpage, changes to the HTML manual are whitespace only.
make (almost) all links appear on only a single line, with no
unnecessary whitespace, using double quotes for attributes. this lets us
automatically convert them to markdown easily.
the few remaining links are extremely long link in a gnome module, we'll
come back to those at a later date.
we can't embed syntactic annotations of this kind in markdown code
blocks without yet another extension. replaceable is rare enough to make
this not much worth it, so we'll go with «thing» instead. the module
system already uses this format for its placeholder names in attrsOf
paths.
markdown can't represent the difference without another extension and
both the html manual and the manpage render them the same, so keeping the
distinction is not very useful on its own. with the distinction removed
we can automatically convert many options that use <code> tags to markdown.
the manpage remains unchanged, html manual does not render
differently (but class names on code tags do change from "code" to "literal").
Instead of enabling the PAM modules based on config.krb5.enable,
introduce a new option to control the PAM modules specifically.
Users may want to turn on config.krb5.enable, to get a working Kerberos
client config with tools like kinit, while letting pam_sss or something
else handle Kerberos password lookups.
the conversion procedure is simple:
- find all things that look like options, ie calls to either `mkOption`
or `lib.mkOption` that take an attrset. remember the attrset as the
option
- for all options, find a `description` attribute who's value is not a
call to `mdDoc` or `lib.mdDoc`
- textually convert the entire value of the attribute to MD with a few
simple regexes (the set from mdize-module.sh)
- if the change produced a change in the manual output, discard
- if the change kept the manual unchanged, add some text to the
description to make sure we've actually found an option. if the
manual changes this time, keep the converted description
this procedure converts 80% of nixos options to markdown. around 2000
options remain to be inspected, but most of those fail the "does not
change the manual output check": currently the MD conversion process
does not faithfully convert docbook tags like <code> and <package>, so
any option using such tags will not be converted at all.
Fix bug where pam_u2f options would be partially included in other pam.d
files if the module was enable for specific services, resulting in
broken configuration.
Previously, `pam_unix.so` was `required` to set PAM_AUTHTOK so that
dependent pam modules (such as gnome keyering) could use the password
(for example to unlock a keyring) upon login of the user. This however
broke any additional auth providers (such as AD or LDAP): for any
non-local user `pam_unix.so` will not yield success, thus eventually the
auth would fail (even the following auth providers were actually
executed, they could not overrule the already failed auth).
This change replaces `required` by `optional`. Therefore, the
`pam_unix.so` is executed and can set the PAM_AUTHTOK for the following
optional modules, _even_ if the user is not a local user. Therefore, the
gnome keyring for example is unlocked both for local and additional
users upon login, and login is working for non-local users via
LDAP/AD.
activating the configuration...
setting up /etc...
chown: warning: '.' should be ':': ‘root.root’
chown: warning: '.' should be ':': ‘root.messagebus’
chown: warning: '.' should be ':': ‘root.root’
chown: warning: '.' should be ':': ‘root.root’
chown: warning: '.' should be ':': ‘root.root’
chown: warning: '.' should be ':': ‘root.root’
chown: warning: '.' should be ':': ‘root.root’
chown: warning: '.' should be ':': ‘root.root’
chown: warning: '.' should be ':': ‘root.root’
chown: warning: '.' should be ':': ‘root.root’
chown: warning: '.' should be ':': ‘root.root’
chown: warning: '.' should be ':': ‘root.root’
chown: warning: '.' should be ':': ‘root.root’
chown: warning: '.' should be ':': ‘root.root’
chown: warning: '.' should be ':': ‘root.root’
chown: warning: '.' should be ':': ‘root.root’
reloading user units for root...
pam-ussh allows authorizing using an SSH certificate stored in your
SSH agent, in a similar manner to pam-ssh-agent-auth, but for
certificates rather than raw public keys.
In issue #157787 @martined wrote:
Trying to use confinement on packages providing their systemd units
with systemd.packages, for example mpd, fails with the following
error:
system-units> ln: failed to create symbolic link
'/nix/store/...-system-units/mpd.service': File exists
This is because systemd-confinement and mpd both provide a mpd.service
file through systemd.packages. (mpd got updated that way recently to
use upstream's service file)
To address this, we now place the unit file containing the bind-mounted
paths of the Nix closure into a drop-in directory instead of using the
name of a unit file directly.
This does come with the implication that the options set in the drop-in
directory won't apply if the main unit file is missing. In practice
however this should not happen for two reasons:
* The systemd-confinement module already sets additional options via
systemd.services and thus we should get a main unit file
* In the unlikely event that we don't get a main unit file regardless
of the previous point, the unit would be a no-op even if the options
of the drop-in directory would apply
Another thing to consider is the order in which those options are
merged, since systemd loads the files from the drop-in directory in
alphabetical order. So given that we have confinement.conf and
overrides.conf, the confinement options are loaded before the NixOS
overrides.
Since we're only setting the BindReadOnlyPaths option, the order isn't
that important since all those paths are merged anyway and we still
don't lose the ability to reset the option since overrides.conf comes
afterwards.
Fixes: https://github.com/NixOS/nixpkgs/issues/157787
Signed-off-by: aszlig <aszlig@nix.build>
C's assert macro only works when NDEBUG is undefined. Previously
NDEBUG was undefined incorrectly which meant that the assert
macros in wrapper.c did not work.
The `nix.*` options, apart from options for setting up the
daemon itself, currently provide a lot of setting mappings
for the Nix daemon configuration. The scope of the mapping yields
convience, but the line where an option is considered essential
is blurry. For instance, the `extra-sandbox-paths` mapping is
provided without its primary consumer, and the corresponding
`sandbox-paths` option is also not mapped.
The current system increases the maintenance burden as maintainers have to
closely follow upstream changes. In this case, there are two state versions
of Nix which have to be maintained collectively, with different options
avaliable.
This commit aims to following the standard outlined in RFC 42[1] to
implement a structural setting pattern. The Nix configuration is encoded
at its core as key-value pairs which maps nicely to attribute sets, making
it feasible to express in the Nix language itself. Some existing options are
kept such as `buildMachines` and `registry` which present a simplified interface
to managing the respective settings. The interface is exposed as `nix.settings`.
Legacy configurations are mapped to their corresponding options under `nix.settings`
for backwards compatibility.
Various options settings in other nixos modules and relevant tests have been
updated to use structural setting for consistency.
The generation and validation of the configration file has been modified to
use `writeTextFile` instead of `runCommand` for clarity. Note that validation
is now mandatory as strict checking of options has been pushed down to the
derivation level due to freeformType consuming unmatched options. Furthermore,
validation can not occur when cross-compiling due to current limitations.
A new option `publicHostKey` was added to the `buildMachines`
submodule corresponding to the base64 encoded public host key settings
exposed in the builder syntax. The build machine generation was subsequently
rewritten to use `concatStringsSep` for better performance by grouping
concatenations.
[1] - https://github.com/NixOS/rfcs/blob/master/rfcs/0042-config-option.md
SUID wrappers really shouldn't be enabled by default, unless a consumer
relies on them. So in my opinion this falls upon the desktop
environments if needed or a user to explicltly enable this if wanted.
Most desktop environments and services like CUPS already enable polkit
by default, that should really be sufficient.
When running e.g. `aa-genprof` get error:
> ERROR: Syntax Error: Unknown line found in file /etc/apparmor.d/abstractions/pam line 26:
> r /nix/store/XXXXX.pam,mr /nix/store/XXXXX-linux-pam-1.5.1/lib/security/pam_filter/*,
So add an explicit newline as concatMapStringsSep only adds them
between.
In the process I also found that the CapabilityBoundingSet
was restricting the service from listening on port 80, and
the AmbientCapabilities was ineffective. Fixed appropriately.
- Added defaultText for all inheritable options.
- Add docs on using new defaults option to configure
DNS validation for all domains.
- Update DNS docs to show using a service to configure
rfc2136 instead of manual steps.
Allows configuring many default settings for certificates,
all of which can still be overridden on a per-cert basis.
Some options have been moved into .defaults from security.acme,
namely email, server, validMinDays and renewInterval. These
changes will not break existing configurations thanks to
mkChangedOptionModule.
With this, it is also now possible to configure DNS-01 with
web servers whose virtualHosts utilise enableACME. The only
requirement is you set `acmeRoot = null` for each vhost.
The test suite has been revamped to cover these additions
and also to generally make it easier to maintain. Test config
for apache and nginx has been fully standardised, and it
is now much easier to add a new web server if it follows
the same configuration patterns as those two. I have also
optimised the use of switch-to-configuration which should
speed up testing.
Closes#129838
It is possible for the CA to revoke a cert that has not yet
expired. We must run lego to validate this before expiration,
but we must still ignore failures on unexpired certs to retain
compatibility with #85794
Also changed domainHash logic such that a renewal will only
be attempted at all if domains are unchanged, and do a full
run otherwises. Resolves#147540 but will be partially
reverted when go-acme/lego#1532 is resolved + available.
ClosesNixOS/nixpkgs#108237
When a user first adds an ACME cert to their configuration,
it's likely to fail to renew due to DNS misconfig. This is
non-fatal for other services since selfsigned certs are
(usually) put in place to let dependant services start.
Tell the user about this in the logs, and exit 2 for
differentiation purposes.
selfsignedDeps is already appended to the after and wants
of a cert's renewal service, making these redundant.
You can see this if you run the following command:
systemctl list-dependencies --all --reverse acme-selfsigned-mydomain.com.service
This is horrible if you want to debug failures that happened during
system switches but your 30-ish acme clients spam the log with the same
messages over and over again.
ClosesNixOS/nixpkgs#147348
I was able to reproduce this intermittently in the
test suite during the tests for HTTPd. Adding
StartLimitIntervalSec=0 to disable rate limiting
for these services works fine. I added it anywhere
there was a ConditionPathExists.
This should mirror the behavior we implement for sudo: The TERMINFO and
TERMINFO_DIRS variables are inherited from the normal user's
environment, so terminfo files installed in the user's profile can be
found by ncurses applications running as root.
pam_mkhomedir should create homedirs with the same umask as the rest
of the system. Currently it creates homedirs with go+rx which makes
it readable for other non-privileged users.
nixos-rebuild test causes pam_mount to prompt for a password when running with
an encrypted home:
building '/nix/store/p6bflh7n5zy2dql8l45mix9qnzq65hbk-nixos-system-mildred-18.09.git.98592c5da79M.drv'...
activating the configuration...
setting up /etc...
reenter password for pam_mount:
(mount.c:68): Messages from underlying mount program:
(mount.c:72): crypt_activate_by_passphrase: File exists
(pam_mount.c:522): mount of /dev/mapper/vg0-lv_home_peter failed
kbuildsycoca5 running...
This change makes pam_mount not prompt. It still tries to remount (and fails in
the process) but that message can be ignored.
Fixes: #44586
The cacert package can now generate p11-kit-compatible output itself,
as well as generating the correct set of outputs for fully-joined
and unbundled "traditional" outputs (in standard PEM and
OpenSSL-compatible formats).
Since 7a10478ea7, all /var except
/var/lib/acme gets mounted in a read-only fashion. This behavior
breaks the existing acme deployments having a webroot set outside of
/var/lib/acme.
Collecting the webroots and adding them to the paths read/write
mounted to the systemd service runtime tree.
Fixes#139310
Moving the service before multi-user.target (so the `hardened` test
continue to work the way it did before) can result in locking the kernel
too early. It's better to lock it a bit later and changing the test to
wait specifically for the disable-kernel-module-loading.service.
Add a shell script that checks if the paths of all wrapped programs
actually exist to catch mistakes. This only checks for Nix store paths,
which are always expected to exist at build time.
To keep backward compatibility and have a typing would require making
all options null by default, adding a defaultText containing the actual
value, write the default value logic based on `!= null` and replacing
the nulls laters. This pretty much defeats the point of having used
a submodule type.
The security.wrappers option is morally a set of submodules but it's
actually (un)typed as a generic attribute set. This is bad for several
reasons:
1. Some of the "submodule" option are not document;
2. the default values are not documented and are chosen based on
somewhat bizarre rules (issue #23217);
3. It's not possible to override an existing wrapper due to the
dumb types.attrs.merge strategy;
4. It's easy to make mistakes that will go unnoticed, which is
really bad given the sensitivity of this module (issue #47839).
This makes the option a proper set of submodule and add strict types and
descriptions to every sub-option. Considering it's not yet clear if the
way the default values are picked is intended, this reproduces the current
behavior, but it's now documented explicitly.
bind mounting directories into the nix-store breaks nix commands.
In particular it introduces character devices that are not supported
by nix-store as valid files in the nix store. Use `/var/empty` instead
which is designated for these kind of use cases. We won't create any
files beause of the tmpfs mounted.
Currently, we hardcode the use of --http.webroot, even if no webroot is
configured. This has the effect of disabling the built-in server.
Co-authored-by: Chris Forno <jekor@jekor.com>
Reusing the same private/public key on renewal has two issues:
- some providers don't accept to sign the same public key
again (Buypass Go SSL)
- keeping the same private key forever partly defeats the purpose of
renewing the certificate often
Therefore, let's remove this option. People wanting to keep the same
key can set extraLegoRenewFlags to `[ --reuse-key ]` to keep the
previous behavior. Alternatively, we could put this as an option whose
default value is true.
According to the ABNF grammar for PEM files described in [RFC
7468][1], an eol character (i.e. a newline) is not mandatory after the
posteb line (i.e. "-----END CERTIFICATE-----" in the case of
certificates).
This commit makes our CA certificate bundler expression account for
the possibility that files in config.security.pki.certificateFiles
might not have final newlines, by using `awk` instead of `cat` to
concatenate them. (`awk` prints a final newline from each input file
even if the file doesn't end with a newline.)
[1]: https://datatracker.ietf.org/doc/html/rfc7468#section-3
As per #121293, I ensured the UMask is set correctly
and removed any unnecessary chmod/chown/chgrp commands.
The test suite already partially covered permissions
checking but I added an extra check for the selfsigned
cert permissions.
Currently if fprintd is enabled, pam will ask for fingerprint
regardless of other configured authentication modules (e.g. yubikey).
This change make fingerprint the last resort of authentication before asking for password.
First because IFD (import-from-derivation) is not allowed on hydra.nixos.org,
and second because without https://github.com/NixOS/hydra/pull/825
hydra-eval-jobs crashes instead of skipping aggregated jobs which fail
(here because they required an IFD).
With the UMask set to 0023, the
mkdir -p command which creates the webroot
could end up unreadable if the web server
changes, as surfaced by the test suite in #114751
On top of this, the following commands
to chown the webroot + subdirectories was
mostly unnecessary. I stripped it back to
only fix the deepest part of the directory,
resolving #115976, and reintroduced a
human readable error message.
With libcap 2.41 the output of cap_to_text changed, also the original
author of code hoped that this would never happen.
To counter this now the security-wrapper only relies on the syscall
ABI, which is more stable and robust than string parsing. If new
breakages occur this will be more obvious because version numbers will
be incremented.
Furthermore all errors no make execution explicitly fail instead of
hiding errors behind debug environment variables and the code style was
more consistent with no goto fail; goto fail; vulnerabilities (https://gotofail.com/)
I found a logical error in the bash script, but during
debugging I enabled command echoing and realised it
would be a good idea to have it enabled all the time for
ease of bug reporting.
For in NixOS it is beneficial if both plasma5 and pam use the same Qt5
version. Because the plasma5 desktop may use a different version as the
default Qt5 version, we introduce plasma5Packages.
- Added an ExecPostStart to acme-$cert.service when webroot is defined to create the acme-challenge
directory and fix required permissions. Lego always tries to create .well-known and acme-challenge,
thus if any permissions in that tree are wrong it will crash and break cert renewal.
- acme-fixperms now configured with acme User and Group, however the script still runs as root. This
ensures the StateDirectories are owned by the acme user.
- Switched to list syntax for systemd options where multiple values are specified.
Closes#106603
Some webservers (lighttpd) require that the
files they are serving are world readable. We
do our own chmods in the scripts anyway, and
lego has sensible permissions on its output
files, so this change is safe enough.
systemd-tmpfiles is no longer required for
most of the critical paths in the module. The
only one that remains is the webroot
acme-challenge directory since there's no
other good place for this to live and forcing
users to do the right thing alone will only
create more issues.
Closes#106565
When generating multiple certificates which all
share the same server + email, lego will attempt
to create an account multiple times. By adding an
account creation target certificates which share
an account will wait for one service (chosen at
config build time) to complete first.
This means that all systems running from master will trigger
new certificate creation on next rebuild. Race conditions around
multiple account creation are fixed in #106857, not this commit.
See https://www.redhat.com/sysadmin/fedora-31-control-group-v2 for
details on why this is desirable, and how it impacts containers.
Users that need to keep using the old cgroup hierarchy can re-enable it
by setting `systemd.unifiedCgroupHierarchy` to `false`.
Well-known candidates not supporting that hierarchy, like docker and
hidepid=… will disable it automatically.
Fixes#73800
