- The new web-vault version supports fastmail.com anon email, add the
correct api host to support it.
- Removed Firefox Relay, this seems only to be supported on SaaS.
- Added a function to the two-factor api to prevent 404 errors.
Previously FlashMessage was used to provide an error message during login.
This PR changes that flow to not use redirect for this, but renders the HTML and responds using the correct status code where needed. This should solve some issues which were reported in the past.
Thanks to @RealOrangeOne, for initiating this with a PR.
Fixes#2448Fixes#2712Closes#2715
Co-authored-by: Jake Howard <git@theorangeone.net>
This PR adds a the new v2022.8.x revoke feature which allows an
organization owner or admin to revoke access for one or more users.
This PR also fixes several permissions and policy checks which were faulty.
- Modified some functions to use DB Count features instead of iter/count aftwards.
- Rearanged some if statements (faster matching or just one if instead of nested if's)
- Added and fixed several policy checks where needed
- Some small updates on some response models
- Made some functions require an enum instead of an i32
This PR attends to mitigate (not fix) #2644.
There seems to be an issue when uploading files either as attachment or
via send via the mobile (Android) client.
The binary data gets transfered correctly to Vaultwarden (Checked via
Wireshark), but the data is not parsed correctly for some reason.
Since the parsing is not done by Vaultwarden it self, i think we should
at least try to prevent saving the data and letting users think all
fine.
Further investigation is needed to actually fix this issue.
This is just a quick patch.
The previous persistent folder check worked by checking if a file
exists. If you used a bind-mount, then this file is not there. But when
using a docker/podman volume those files are copied, and caused the
container to not start.
This change checks the `/proc/self/mountinfo` for a specific patern to
see if the data folder is persistent or not.
Fixes#2622
When using anything else but the `internal` icon service it would
trigger an CSP block because the redirects were not allowed.
This PR fixes#2623 by dynamically adding the needed CSP strings.
This should also work with custom services.
For Google i needed to add an extra check because that does a redirect
it self to there gstatic.com domain.
A bit inspired by @paolobarbolini from this commit at lettre https://github.com/lettre/lettre/pull/784 .
I added a few more clippy lints here, and fixed the resulted issues.
Overall i think this could help in preventing future issues, and maybe
even peformance problems. It also makes some code a bit more clear.
We could always add more if we want to, i left a few out which i think
arn't that huge of an issue. Some like the `unused_async` are nice,
which resulted in a few `async` removals.
Some others are maybe a bit more estatic, like `string_to_string`, but i
think it looks better to use `clone` in those cases instead of `to_string` while they already are a string.
This is to support scenarios where the attachments and sends folder are to be stored on a separate device from the tmp_folder (i.e. fuse-mounted S3 storage), due to having the tmp_dir on the same device being undesirable.
Example being fuse-mounted S3 storage with the reasoning that because S3 basically requires a copy+delete operations to rename files, it's inefficient to rename files on device, if it's even allowed.
Regarding some recent issues with sending attachments, but previously
also some changes to the API for example which could cause a `400` error
it just returned that there is something wrong, but not to much details
on what exactly.
To help with getting a bit more detailed information, we should set the
log-level for `_` to at least `Warn`.
- Updated dependencies (html5gum for favicon downloading)
* Also openssl, time, jsonwebtoken and r2d2
- Small optimizations on downloading favicons.
It now only emits tokens/tags which needs to be parsed, all others are
being skipped. This prevents unneeded items within the for-loop being
parsed.
This PR fixes#2566
If Organizational syncs returned a FolderId it would cause the web-vault
to hide the cipher because there is a FolderId set. Upstream seems to
not return FolderId and Favorite. When set to null/false it will behave
the same.
In this PR I have added a new CipherSyncType enum to select which type
of sync to execute, and return an empty list for both Folders and Favorites if this is for Orgs.
This also reduces the database load a bit since it will not execute those queries.
- Updated some Rust dependencies
- Fixed an issue with CSP header, this was not configured correctly
- Prevent sending CSP and Frame headers for the MFA connector.html files.
Else some clients will fail to handle these protocols.
- Add `unsafe-inline` for `script-src` only to the CSP for the Admin Interface
- Updated JavaScript and CSS files for the Admin interface
- Changed the layout for showing overridden settings, better visible now.
- Made the version check cachable to prevent hitting the Github API rate limits
- Hide the `database_url` as if it is a password in the Admin Interface
Else for MariaDB/MySQL or PostgreSQL this was plain text.
- Fixed an issue that pressing enter on the SMTP Test would save the config.
resolves#2542
- Prevent user names larger then 50 characters
resolves#2419
This will add a persistent volume check to make sure when running
containers someone is using a volume for persistent storage.
This check can be bypassed if someone configures
`I_REALLY_WANT_VOLATILE_STORAGE=true` as an environment variable.
This should prevent issues like #2493 .
The limits for uploading files were to small in regards to the allowed
maximum filesize of the Bitwarden clients including the web-vault.
Changed both `data-form` (used for Send) and `file` (used for
attachments) to be 525MB, this is the same size we already check our selfs.
Also changed the `json` limit to be 20MB, this should allow very large
imports with 4000/5000+ items depending on if there are large notes or not.
And, also disabled Rocket from outputting colors, these colors were also
send to the log files and syslog. I think this changed in Rocket 0.5rc
somewhere, need to look a bit further into that maybe.
Improved sync speed by resolving the N+1 query issues.
Solves #1402 and Solves #1453
With this change there is just one query done to retreive all the
important data, and matching is done in-code/memory.
With a very large database the sync time went down about 3 times.
Also updated misc crates and Github Actions versions.
`PRAGMA busy_timeout = 5000` tells SQLite to keep trying for up to 5000 ms
when there is lock contention, rather than aborting immediately. This should
hopefully prevent the vast majority of "database is locked" panics observed
since the async transition.
`PRAGMA synchronous = NORMAL` trades better performance for a small potential
loss in durability (the default is `FULL`). The SQLite docs recommend `NORMAL`
as "a good choice for most applications running in WAL mode".
The added endpoints work the same as in their upstream implementations.
Upstream also implements `/api/ip`. This seems to include the server's public
IP address (the one that should be hidden behind Cloudflare), which doesn't
seem like a great idea.
For a while now WebAuthn has replaced u2f.
And since web-vault v2.27.0 the connector files for u2f have been removed.
Also, on the official bitwarden server the endpoint to `/two-factor/get-u2f` results in a 404.
- Removed all u2f code except the migration code from u2f to WebAuthn
- Removed all `thread::sleep` and use `tokio::time::sleep` now.
This solves an issue with updating to Bullseye ( Resolves#1998 )
- Updated all Debian images to Bullseye
- Added MiMalloc feature and enabled it by default for Alpine based images
This increases performance for the Alpine images because the default
memory allocator for MUSL based binaries isn't that fast
- Updated `dotenv` to `dotenvy` a maintained and updated fork
- Fixed an issue with a newer jslib (not fully released yet)
That version uses a different endpoint for `prelogin` Resolves#2378 )
- Updated jsonwebtoken to latest version
- Trim `username` received from the login form ( Fixes#2348 )
- Make uuid and user_uuid a combined primary key for the devices table ( Fixes#2295 )
- Updated crates including regex which contains a CVE ( https://blog.rust-lang.org/2022/03/08/cve-2022-24713.html )
This is a rather large PR which updates the async branch to have all the
database methods as an async fn.
Some iter/map logic needed to be changed to a stream::iter().then(), but
besides that most changes were just adding async/await where needed.
Favicon:
- Replaced HTML tokenizer, much faster now.
- Caching the domain blacklist function.
- Almost all functions are async now.
- Fixed bug on minimizing data to parse
- Changed maximum icon download size to 5MB to match Bitwarden
- Added `apple-touch-icon.png` as a second fallback besides `favicon.ico`
SMTP:
- Deprecated SMTP_SSL and SMTP_EXPLICIT_TLS, replaced with SMTP_SECURITY
Misc:
- Fixed issue when `resolv.conf` contains errors and trust-dns panics (Fixes#2283)
- Updated Javscript and CSS files for admin interface
- Fixed an issue with the /admin interface which did not cleared the login cookie correctly
- Prevent websocket notifications during org import, this caused a lot of traffic, and slowed down the import.
This is also the same as Bitwarden which does not trigger this refresh via websockets.
Rust:
- Updated to use v1.59
- Use the new `strip` option and enabled to strip `debuginfo`
- Enabled `lto` with `thin`
- Removed the strip RUN from the alpine armv7, this is now done automatically
- Changed nightly to stable in Dockerfile and Workflow
- Updated Dockerfile to use stable and updated ENV's
- Removed 0.0.0.0 as default addr it now uses ROCKET_ADDRESS or the default
- Updated Github Workflow actions to the latest versions
- Updated Hadolint version
- Re-orderd the Cargo.toml file a bit and put libs together which are linked
- Updated some libs
- Updated .dockerignore file
This is a rather large PR which updates the async branch to have all the
database methods as an async fn.
Some iter/map logic needed to be changed to a stream::iter().then(), but
besides that most changes were just adding async/await where needed.
The current limit of 19 is an artifact of the implementation, which can be
easily rewritten in terms of a more general string generation function.
The new limit is 255 (max value of a `u8`); using a larger type would
probably be overkill.
Squashed commit of the following:
commit 1bdf1c7954e0731c95703d10118f3874ab5155d3
Merge: 8ba6e617257251
Author: Daniel García <dani-garcia@users.noreply.github.com>
Date: Sun Jan 23 23:40:17 2022 +0100
Merge branch 'remove-bwrs' of https://github.com/RealOrangeOne/vaultwarden into RealOrangeOne-remove-bwrs
commit 7257251ecf
Author: Jake Howard <git@theorangeone.net>
Date: Thu Jan 6 17:48:18 2022 +0000
Use `or_else` to save potentially unnecessary function call
commit 40ae81dd3c
Author: Jake Howard <git@theorangeone.net>
Date: Wed Jan 5 21:18:24 2022 +0000
Move $BWRS_VERSION fallback into build.rs
commit 743ef74b30
Author: Jake Howard <git@theorangeone.net>
Date: Sat Jan 1 23:08:27 2022 +0000
Revert "Add feature to enable use of `Option::or` in const context"
This reverts commit fe8e043b8a.
We want to run on stable soon, where these features are not supported
commit a1f0da638c
Author: Jake Howard <git@theorangeone.net>
Date: Sat Jan 1 13:04:47 2022 +0000
Rename web vault version file
https://github.com/dani-garcia/bw_web_builds/pull/58
commit fe8e043b8a
Author: Jake Howard <git@theorangeone.net>
Date: Sat Jan 1 12:56:44 2022 +0000
Add feature to enable use of `Option::or` in const context
commit 687435c8b2
Author: Jake Howard <git@theorangeone.net>
Date: Sat Jan 1 12:27:28 2022 +0000
Continue to allow using `$BWRS_VERSION`
commit 8e2f708e50
Author: Jake Howard <git@theorangeone.net>
Date: Fri Dec 31 11:41:34 2021 +0000
Remove references to "bwrs"
The only remaining one is getting the version of the web vault, which requires coordinating with the web vault patching.
The CLI seems to send a String instead of an Integer for the maximum access count.
It now accepts both types and converts it to an i32 in all cases.
Fixes#2196
The default code is 307 (temporary) to make it easier to test different icon
services, but once a service has been decided on, users should ideally switch
to using permanent redirects for cacheability.
Browsers are rather smart, but also dumb. This uses the `Expires` header
alongside `cache-control` to better prompt the browser to actually
cache.
Unfortunately, firefox still tries to "race" its own cache, in an
attempt to respond to requests faster, so still ends up making a bunch
of requests which could have been cached. Doesn't appear there's any way
around this.
- Using my own rust-musl build containers we now support all database
types for both Debian and Alpine.
- Added new Alpine containers for armv6 and arm64/aarch64
- The Debian builds can also be done wihout dpkg magic stuff, probably
some fixes in Rust regarding linking (Or maybe OpenSSL or Diesel), in
any case, it works now without hacking dpkg and apt.
- Updated toolchain and crates
If an external icon service is configured, icon requests return an HTTP
redirect to the corresponding icon at the external service.
An external service may be useful for various reasons, such as if:
* The Vaultwarden instance has no external network connectivity.
* The Vaultwarden instance has trouble handling large bursts of icon requests.
* There are concerns that an attacker may probe the instance to try to detect
whether icons for certain sites have been cached, which would suggest that
the instance contains entries for those sites.
* The external icon service does a better job of providing icons than the
built-in fetcher.
- Enabled trust-dns feature which seems to help a bit when DNS is
causing long timeouts. Though in the blocking version it is less visible
then on the async branch.
- Updated crates
- Removed some redundant code
- Updated javascript/css libraries
Resolves#2118Resolves#2119
- Decreased `recursion_limit` from 512 to 87
Mainly done by optimizing the config macro's.
This fixes an issue with the rust-analyzer which doesn't go beyond 128
- Removed Regex for masking sensitive values and replaced it with a map()
This is much faster then using a Regex.
- Refactored the get_support_json macro's
- All items above also lowered the binary size and possibly compile-time
- Removed `_conn: DbConn` from several functions, these caused unnecessary database connections for functions who didn't used that at all
- Decreased json response for `/plans`
- Updated libraries and where needed some code changes
This also fixes some rare issues with SMTP https://github.com/lettre/lettre/issues/678
- Using Rust 2021 instead of 2018
- Updated rust nightly
Rejecting an emergency access request should transition the grantor/grantee
relationship back into the `Confirmed` state, and the grantor's encrypted key
should remain in escrow rather than being cleared, or else future emergency
access requsts from that grantee will fail.
For one of these flags to be in effect for a cipher, upstream requires all of
(rather than any of) the collections the cipher is in to have that flag set.
Also, some of the logic for loading access restrictions was wrong. I think
that only malicious clients that also had knowledge of the UUIDs of ciphers
they didn't have access to would have been able to take advantage of that.
An incomplete 2FA login is one where the correct master password was provided,
but the 2FA token or action required to complete the login was not provided
within the configured time limit. This potentially indicates that the user's
master password has been compromised, but the login was blocked by 2FA.
Be aware that the 2FA step can usually still be completed after the email
notification has already been sent out, which could be confusing. Therefore,
the incomplete 2FA time limit should be long enough that this situation would
be unlikely. This feature can also be disabled entirely if desired.
During a small discusson on Matrix it seems logical to have the /alive
endpoint also check if the database connection still works.
The reason for this was regarding a certificate which failed/expired
while vaultwarden and the database were still up-and-running, but
suddenly vaultwarden couldn't connect anymore.
With this `DbConn` added to `/alive`, it will be more accurate, because
of vaultwarden can't reach the database, it isn't alive.
- Fixed a bug in JavaScript which caused no messages to be shown to the
user in-case of an error send by the server.
- Changed mail error handling for better error messages
- Changed user/org actions from a to buttons, this should prevent
strange issues in-case of javascript issues and the page does re-load.
- Added Alpine and Debian info for the running docker image
During the mail error testing i encountered a bug which caused lettre to
panic. This panic only happens on debug builds and not release builds,
so no need to update anything on that part. This bug is also already
fixed. See https://github.com/lettre/lettre/issues/678 and https://github.com/lettre/lettre/pull/679Resolves#2021
Could also fix the issue reported here #2022, or at least no hash `#` in
the url.
- Unicode websites could break (www.post.japanpost.jp for example).
regex would fail because it was missing the unicode-perl feature.
- Be less verbose in logging with icon downloads
- Removed duplicate info/error messages
- Added err_silent! macro to help with the less verbose error/info messages.
- Changed the date of the migration folders to be from this date.
- Removed a lot is_email_domain_allowed checks.
This check only needs to be done during the invite it self, else
everything else will fail even if a user has an account created via the
/admin interface which bypasses that specific check! Also, the check was
at the wrong place anyway's, since it would only not send out an e-mail,
but would still have allowed an not allowed domain to be used when
e-mail would have been disabled. While that check always works, even if
sending e-mails is disasbled.
- Added an extra allowed route during password/key-rotation change which
updates/checks the public-key afterwards.
- A small change with some `Some` and `None` orders.
- Change the new invite object to only generate the UTC time once, since
it could be possible that there will be a second difference, and we only
need to call it just once.
by black.dex@gmail.com
Signed-off-by: thelittlefireman <thelittlefireman@users.noreply.github.com>
- Updated JS Libraries
- Downgraded bootstrap.css to v5.0.2 which works with Bootstrap-Native.
- Fixed issue with settings being able to open/collapse on some systems.
- Added .js and .css to the exclude list for the end-of-file-fixer pre-commit
For user management within the organization view you are able to select
multiple users to re-invite, confirm or delete them.
These actions were not working which this PR fixes by adding support for
these endpoints. This will make it easier to confirm and delete multiple
users at once instead of having to do this one-by-one.
In the case when SMTP is disabled and.
when inviting new users either via the admin interface or into an
organization and using uppercase letters, this would fail for those
users to be able to register since the checks which were done are
case-sensitive and never matched.
This PR fixes that issue by ensuring everything is lowercase.
Fixes#1963
