Commit graph

271 commits

Author SHA1 Message Date
Olivia Lee
5f9e85cb49
return an error when attempting to delete a nonexistent local alias
The spec doesn't spell out explicitly that the server should check this,
but it does list M_NOT_FOUND "room alias ... not found" as an example
error response.
2025-05-18 14:13:42 -07:00
Olivia Lee
a34bca3986
transfer local canonical aliases to new room on upgrade
When upgrading rooms, we reassign any local aliases from the old room to
the new room. This commit updates the m.room.canonical_alias events in
the old and new rooms to reflect which aliases were moved. The spec is
unclear on whether the server should do this[1], but it's consistent
with synapse's behavior.

I went with putting the canonical alias update logic inline, rather than
something like add_canonical_alias and remove_canonical_alias helper
functions to the alias service, because it's desirable to have the alias
updates be sent as a single event than a separate event for each change.

[1]: https://github.com/matrix-org/matrix-spec/issues/2142
2025-05-18 14:13:41 -07:00
Lambda
6bcc4e310e
Immediately trigger EDU sending after client read receipt
Previously, read receipts would only be forwarded via federation
incidentally when some PDU was later sent to the destination server.
Trigger a send without any event to collect EDUs and get read receipts
out directly.
2025-03-30 16:48:06 -07:00
Lambda
175a62007d Update MSRV to 1.84.0
And appease clippy (`__CARGO_FIX_YOLO=1 cargo clippy --fix` plus some
manual type shuffling).
2025-02-18 17:06:08 +00:00
Lambda
5616510727 Use UInt instead of u32 for max request size
Sometimes you just really want to upload a full disk image as media.
2025-02-18 16:48:47 +00:00
Olivia Lee
472f51c350
allow adding canonical aliases from remote servers
Like is mentioned in the comment, this isn't explicitly required by the
spec, but it's reasonable and what synapse does.
2025-01-23 09:20:10 -08:00
Olivia Lee
29d8fbaefa
only validate canonical aliases that are new
Previously we required every alias in a canonical alias event sent by a
client to be valid, and would only validate local aliases. This
prevented clients from adding/removing canonical aliases if there were
existing remote or invalid aliases.
2025-01-23 09:20:10 -08:00
Olivia Lee
50c1e77cd6
factor canonical alias event validation into a helper function
Not all the semantics from the spec quote in the doc comment are
implemented yet.
2025-01-23 09:20:10 -08:00
Olivia Lee
051221c0ab
validate schema of new canonical alias events sent by clients
Previously, we would only attempt to validate the aliases in the event
content if we were able to parse the event, and would silently allow it
otherwise.
2025-01-23 09:20:10 -08:00
Olivia Lee
c748c7c188
return M_BAD_ALIAS when trying to set non-existent canonical aliases
This is the error code named in the spec.
2025-01-20 16:59:01 -08:00
Olivia Lee
ba72616672
do not backoff remote device key queries when a request fails due to backoff
The previous logic would increment the backoff counter both when a
request actually fails and when we do not make a request because the
server was already in backoff. This lead to a positive feedback loop
where every request made while a server is in backoff increases the
backoff delay, making it impossible to recover from backoff unless the
entire backoff delay elapses with zero requests.
2024-12-11 13:02:11 -08:00
Olivia Lee
4ee8312068
reset device key query backoff after a successful request
Failing to reset the backoff state resulted in a monotonically
increasing backoff delay. If a remote server was temporarily
unavailable, we would have a persistently increased rate of key query
failures until the backoff state was reset by a server restart. If
enough key queries were attempted while the remote was unavailable, it
can accumulate an arbitrarily long backoff delay and effectively block
all future key queries to this server.
2024-12-11 13:02:10 -08:00
Olivia Lee
79cedccdb6
factor remote device key request logic into helper functions
This is pure code-motion, with no behavior changes. The new structure
will make it easier to fix the backoff behavior, and makes the code
somewhat less of a nightmare to follow.
2024-12-11 13:02:10 -08:00
Lambda
daceadb310 sync/v3: factor out into separate functions
This is both easier to read and produces much better tracing spans.
2024-12-11 17:38:49 +00:00
Lambda
8a7f87e9b4 sync/v3: move readonly data to context struct
This makes it a lot easier to factor out parts of the big
sync_events_route().
2024-12-11 17:38:49 +00:00
Lambda
55a04f77c6 sync/v3: record relevant span fields 2024-12-11 17:38:49 +00:00
Lambda
79783ebe56 sync: split into separate files 2024-12-11 17:38:49 +00:00
Charles Hall
51b30d9ba3
largely stop using RoomCreateEventContent
This became a problem because  #foundation-office:matrix.org has a
malformed create event with its `predecessor` set to a string instead of
a map.

The solution to this is, unfortunately, to do more shotgun parsing to
extract only the desired fields rather than trying to parse the entire
content every time. To prevent this kind of problem from happening
again, `RoomCreateEventContent` must only be used for creating new PDUs,
existing PDUs must be shotgun-parsed.
2024-11-08 20:36:46 -08:00
Charles Hall
a4e1522875
generalize get_room_version
There are other fields of `m.room.create` events that are useful to
individually extract without caring about the values of other fields.
2024-11-08 18:38:16 -08:00
Charles Hall
c9c30fba30
upgrade to latest ruma 2024-11-08 17:26:44 -08:00
Andreas Fuchs
76a633cb66
Log failed remote resident server join requests 2024-10-13 19:26:35 -07:00
Andreas Fuchs
e001356653
Return local join error if all remote joins fail
If all join requests to resident servers fail or if the joining server
is the only resident server (i.e. the room is local-only), we would
previously send a 500 error, even if the more correct response would be
M_UNAUTHORIZED (e.g. if the user tries to join an invite-only room).

To fix this, we now return the error generated by attempting the join
locally, which correctly informs the client about why their request
failed.
2024-10-13 19:10:58 -07:00
mikoto
287f6b9163
refactor calculate_invite_state
That was terribly named and terribly implemented.

Co-authored-by: Charles Hall <charles@computer.surgery>
2024-10-03 10:52:07 -07:00
Benjamin Lee
9add9a1e96
fix room version comparisons
Fixes a set of bugs introduced by 00b77144c1,
where we replaced explicit `RoomVersionId` matches with `version < V11`
comparisons. The `Ord` impl on `RoomVersionId` does not work like that,
and is in fact a lexicographic string comparison[1]. The most visible
effect of these bugs is that incoming redaction events would sometimes
be ignored.

Instead of reverting to the explicit matches, which were quite verbose,
I implemented a `RoomVersion` struct that has flags for each property
that we care about. This is similar to the approach used by ruma[2] and
synapse[3].

[1]: 7cfa3be0c6/crates/ruma-common/src/identifiers/room_version_id.rs (L136)
[2]: 7cfa3be0c6/crates/ruma-state-res/src/room_version.rs
[3]: c856ae4724/synapse/api/room_versions.py
2024-09-26 13:01:25 -07:00
Charles Hall
c24f79b79b
update rust deps except rocksdb and otel clownery
* OTel v0.25.0 requires downgrading Tokio to 1.38 [0]
* They have a fix for this but aren't cutting a release just for release
  schedule reasons [1]
* Prometheus support (at least for server-pull) was dropped at OTel
  v0.23 and isn't planned to be picked up again until OTel v1 [2]
* No real reasoning was provided for this decision AFAICT [3] [4]
* So many compiler errors
* Unhelpful changelogs

The last two points are what made me give up on trying to upgrade to
OTel v0.24 too.

RocksDB isn't updated because we'd need to update our nixpkgs too but
that causes other problems, such as an upstream bug in liburing when
building for musl.

[0]: https://github.com/open-telemetry/opentelemetry-rust/issues/2094
[1]: https://github.com/open-telemetry/opentelemetry-rust/issues/2094#issuecomment-2346834030
[2]: https://docs.rs/opentelemetry-prometheus/0.17.0/opentelemetry_prometheus/index.html
[3]: https://github.com/open-telemetry/opentelemetry-rust/pull/1792
[4]: https://github.com/open-telemetry/opentelemetry-rust/pull/1792#issuecomment-2121514344
2024-09-23 14:22:55 -07:00
Benjamin Lee
7672cc8473
use OwnedMxcUri in media service
Not using `MxcData` because it borrows it's fields, and so we wouldn't
be able to return an owned `MxcData` from functions that read the db.
2024-09-15 00:32:17 -07:00
Lambda
3a55684623
Update MSRV to 1.81.0
Plus a "__CARGO_FIX_YOLO=1 cargo clippy --fix"
2024-09-08 09:54:08 -07:00
Lambda
cce83beedb Properly type stored EDUs 2024-09-06 17:20:14 +00:00
Lambda
26322d5a95 Add PduId wrapper struct
Death to Vec<u8>
2024-09-06 17:20:14 +00:00
Lambda
b4fecbc517 client/media: work around matrix-media-repo bug
It rejects Authenticated Media requests if X-Matrix values aren't
unnecessarily quoted:
https://github.com/t2bot/matrix-media-repo/issues/609
2024-09-01 09:42:45 +00:00
Lambda
3fe0110649 media: convert allow_remote to enum 2024-09-01 09:42:45 +00:00
Lambda
a4b7df1b3a media: use FileMeta instead of tuples 2024-09-01 09:42:45 +00:00
Lambda
84850a163d Factor content out of FileMeta
That's not what *meta*data means
2024-09-01 09:42:45 +00:00
Lambda
a06c8db996 Fetch and thumbnail original media if fetching thumbnail fails
E.g. because the remote server's thumbnail endpoint is broken in
hilarious ways: https://github.com/element-hq/synapse/issues/17518
2024-09-01 09:42:45 +00:00
Lambda
79053ad052 client_server: use and provide authenticated media API 2024-09-01 09:42:45 +00:00
Lambda
7f6ab63752 client_server: factor out width/height conversion 2024-09-01 09:42:45 +00:00
Lambda
94204415ee Add MxcData helper 2024-09-01 09:42:45 +00:00
Lambda
64b3c357dd media: put old API behind legacy_media import 2024-09-01 09:42:45 +00:00
Lambda
5a5bea3217 cargo update ruma httparse
This adds authenticated media APIs.
2024-09-01 09:42:45 +00:00
avdb13
00b77144c1 chore: deprecate support for unstable room versions 2024-08-12 06:18:28 +02:00
Benjamin Lee
91739899e7
changelog entry for remote device key query logging 2024-08-11 16:39:08 -07:00
Benjamin Lee
141c60e4a3
log more detailed backoff state for remote device key requests 2024-08-11 16:39:08 -07:00
Benjamin Lee
1551833501
log failed remote device key requests 2024-08-11 16:39:08 -07:00
Lambda
4893c54f4f Use TokenSet for roomid_mutex_insert 2024-07-22 18:07:03 +00:00
Lambda
34ccb2cd06 Use TokenSet for roomid_mutex_state 2024-07-22 18:07:03 +00:00
Lambda
aea6019c0a upgrade_room_route: fix state lock
This was using the lock for the replacement room to send events to the
original room, which may or may not cause problems.
2024-07-22 18:06:11 +00:00
Lambda
ef1b6fe111 upgrade_room_route: give more descriptive names to locks
This makes the next commit, which is an actual fix, much easier to read.
2024-07-22 18:06:11 +00:00
tranquillity-codes
c70cfd3d25
Return 504 when a file is missing to be by-spec
The spec defines that the media endpoints should return 504 when a file
is not-yet-uploaded, which has been interpreted to include when a file
was deleted. Modifies the /media/v3/download/ and /media/r0/thumbnail
endpoints.
2024-07-21 20:42:07 -07:00
Charles Hall
ee95b6be3c
fix api/client_server/sync events
Some duplicated-ish code from src/database/key_value/rooms/timeline.rs
about handling errors from `pdus_since`/`pdus_until`, it seems like the
error message was actually directly copy-pasted from there because it
referred to the wrong function lol
2024-07-16 11:16:08 -07:00
Charles Hall
dacb2ae7dd
fix api/client_server/context events 2024-07-16 11:12:08 -07:00