aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* mod_smacks: Fix typosKim Alvefur2021-12-201-2/+2
|
* mod_smacks: Fix error on shutdown if loaded on ComponentKim Alvefur2021-12-201-0/+4
|
* mod_smacks: Fix typo (thanks codespell)Kim Alvefur2021-12-191-1/+1
|
* mod_smacks: Avoid log noise when a non-smacks session is closedKim Alvefur2021-12-191-0/+1
|
* mod_smacks: Check if session was really sleeping before logging messageKim Alvefur2021-12-181-1/+1
| | | | | | It could have been resumed without going into hibernation first, i.e. when the client notices the disconnect before the server, or if it switches networks etc.
* mod_smacks: Fix traceback when bouncing unacked stanzasKim Alvefur2021-12-181-0/+1
| | | | | Errors sent from handle_unacked_stanzas() should usually not be sent to the session itself, but if one is, it should not be queued.
* mod_smacks: Fix duplicate sleep behavior when replacing a sessionKim Alvefur2021-12-181-0/+1
| | | | | Fixes that an extra watchdog was set, leaking the previous one, which went on to do behave as if the session times out.
* mod_c2s: Fire pre-closing event regardless of connected stateKim Alvefur2021-12-181-3/+3
| | | | | | | | With mod_smacks, a session can outlive the connection, so whether the event is fired should not hinge on whether the session is connected or not. Helps mod_smacks remove some state.
* mod_smacks: Remove references to connection on hibernationKim Alvefur2021-12-181-1/+4
| | | | | | Still having the connection on the session may cause unintentional behavior, such as the session being treated as if connected, even tho the connection has been closed.
* mod_smacks: Add more loggingKim Alvefur2021-12-181-0/+9
|
* mod_smacks: Remove useless delay in requesting ack on resumptionKim Alvefur2021-12-171-23/+1
| | | | Was this the last place using the delay? Nice!
* mod_smacks: Fix to not wait for acks from before a resumptionKim Alvefur2021-12-171-0/+1
| | | | | Otherwise it can get stuck waiting indefinitely for an ack that made it notice the connection was stale.
* mod_smacks: Split log messages for when an old session exists in storageKim Alvefur2021-12-171-1/+2
| | | | Non-existent did not seem entirely accurate for this case
* mod_smacks: Bounce unacked stanzas on shutdownKim Alvefur2021-12-171-1/+11
|
* mod_smacks: Preserve counter values on shutdownKim Alvefur2021-12-171-0/+22
|
* mod_smacks: Simplify access to local user sessionsKim Alvefur2021-12-171-2/+2
| | | | | Less to type if per chance the next commit also wants to access sessions.
* mod_smacks: Cancel hibernation when session is closedKim Alvefur2021-12-161-0/+6
| | | | | | To ensure that if a session is replaced after it has gone into hibernation, it does not come back and cause trouble for the new session (see previous commit).
* mod_smacks: Skip hibernation logic if session was closed or replacedKim Alvefur2021-12-161-0/+4
| | | | | | | | | | | | The resumption_token is removed when the session is closed via the pre-session-close event, signaling that it cannot be resumed, and therefore no hibernation timeout logic should be invoked. Fixes that if a session somehow is replaced by a new one using the same resource (which is the common behavior), the old session would still be around until it times out at which point it sends `<presence type="unavailable"/>` which would look as if it came from the new session, ie appearing offline to everyone including MUCs.
* mod_smacks: Fix indentationKim Alvefur2021-12-161-4/+4
| | | | Meld keeps messing up indentation when I merge and rebase...
* mod_smacks: Stop stanzas from being queued on hibernation timeoutKim Alvefur2021-12-161-0/+1
| | | | | | | | | | This overloads that flag a bit, but it has the intended effect of stopping outgoing_stanza_filter() from queueing stanzas. Fixes a traceback because of the queue having been removed somewhere around here, since it is no longer needed. Thanks Martin for reporting
* util.smqueue: Simplify compat table, fix dependent modules (thanks Martin)Kim Alvefur2021-12-163-24/+36
| | | | | | There was an off-by-one in the modulo calculation. Switching to a plain old array-table makes the apparent size of the queue wrong, but since some of the queue may not be available this is likely for the best.
* util.smqueue: Add missing method to Teal record specKim Alvefur2021-12-161-1/+2
|
* mod_smacks: Limit queue memory consumption using new utilKim Alvefur2021-12-142-52/+73
| | | | | | | | | This brings back the queue size limit that was once added, then removed because destroying the session when reaching the limit was not great. Instead, the queue wraps and overwrites the oldest unacked stanza on the assumption that it will probably be acked anyway and thus does not need to be delivered. If those discarded stanzas turn out to be needed on resumption then the resumption fails.
* util.smqueue: Abstract queue with acknowledgements and overflowKim Alvefur2021-12-143-0/+223
| | | | | | | | Meant to be used in mod_smacks for XEP-0198 Meant to have a larger virtual size than actual number of items stored, on the theory that in most cases, the excess will be acked before needed for a resumption event.
* mod_smacks: Set a watchdog to watch sleeping sessionsKim Alvefur2021-12-021-59/+27
| | | | | Extending the timeout by poking the watchdog, and letting it go on resumption, should be much better than the previous method.
* mod_smacks: Switch storage for tracking resumption tokensKim Alvefur2021-12-021-40/+5
| | | | | | | | All that was a complicated way to limit the number of resumable sessions. Let's control resource usage some other way. This leaves the essence of mapping resumption tokens to live sessions. This keeps resumption state across reloads.
* mod_smacks: Clear resumption token from persistent storage on resumptionKim Alvefur2021-12-011-0/+1
|
* mod_smacks: Persist old counter values to storageKim Alvefur2021-12-011-7/+5
| | | | | | | | | This allows clients that try to resume a session after a server restart to at least know which of their pending outgoing stanzas were received and which need to be re-sent. This removes the limit on how many of those counters are kept, which should be fixed eventually.
* mod_smacks: Remove now unused variable [luacheck]Kim Alvefur2021-12-151-1/+1
|
* mod_smacks: Remove unused eventKim Alvefur2021-12-151-1/+0
| | | | | Since nothing uses it. Some equivalent functionality may return in the future.
* mod_smacks: Fix typoKim Alvefur2021-12-151-1/+1
|
* mod_smacks: Flag both sessions to prevent duplicating queue on resumptionKim Alvefur2021-12-141-0/+3
| | | | | | | I have no idea what is going on in this code, which session is which? Something has one of the sessions as an upvalue which is where the filter checks for it.
* mod_smacks: Increment a counter for each resumptionKim Alvefur2021-12-061-0/+1
| | | | | Encountered what looks like a bug where after *many* resumptions, part of the queue was not acked.
* mod_smacks: Back out 48f8fa6cea7b - back in a86ae74da96c againKim Alvefur2021-12-131-2/+4
| | | | Maybe it works now with the session patching in the previous commit
* mod_smacks: Patch various self-references on the resumed sessionKim Alvefur2021-12-131-0/+5
| | | | | The number of places where the session is an upvalue to its own methods is too high!
* mod_smacks: Back out a86ae74da96c: Fixed one problem, caused anotherKim Alvefur2021-12-131-4/+2
| | | | | | | | As per a86ae74da96c the 'session' object here is the wrong session, so the attempt to block stanzas from being added to the queue twice did not work causing something of a leak. Instead we have a leak of the previous session.
* mod_smacks: Back out def20683c161Kim Alvefur2021-12-131-1/+0
| | | | The attempted fix completely broke everything after resumption.
* mod_smacks: Update reference to session from send()Kim Alvefur2021-12-131-0/+1
| | | | | | a86ae74da96c caused the stanza queue to double on resumption because session.send() keeps a reference to the session which is what gets passed to filters, so the added flag was not seen in the filter.
* util.format: Optimize most common integer formatKim Alvefur2021-12-131-0/+2
| | | | | | A search for log formats in use points to %s being the most common, followed by %d, so worth having a fast path for that. %g works well with most numbers and is what Lua 5.1 and 5.2 used
* util.format: Ensure metatable __tostring results are also sanitizedKim Alvefur2021-12-133-2/+20
|
* mod_smacks: Include reason for destroying session on hibernation timeoutKim Alvefur2021-12-121-1/+1
| | | | To aid in debugging
* tools/generate_format_spec: Apply lua-format to silence luacheckKim Alvefur2021-12-111-7/+9
|
* util.format: Fix some formats expecting positive numbers in Lua 5.2Kim Alvefur2021-12-112-4/+9
| | | | Amazing how string.format behaves differently under each Lua version
* util.format: Fix Lua 5.1 quirks thanks to ALL THE TESTSKim Alvefur2021-12-111-1/+6
|
* util.format: ALL THE TESTS!!!Kim Alvefur2021-12-112-0/+891
| | | | | | The more tests I made, the more Lua 5.1 quirks I discovered. Tests generated using a tool plus some touch-up.
* util.format: Also handle the %p format added in Lua 5.4Kim Alvefur2021-12-112-3/+27
|
* util.format: Ensure sanitation of strings passed to wrong formatKim Alvefur2021-12-112-21/+38
| | | | | | | | | Ie. log("debug", "%d", "\1\2\3") should not result in garbage. Also optimizing for the common case of ASCII string passed to %s and early returns everywhere. Returning nil from a gsub callback keeps the original substring.
* util.format: Escape invalid UTF-8 by passing trough serializationKim Alvefur2021-12-102-1/+14
| | | | | | Should prevent invalid UTF-8 from making it into the logs, which can cause trouble with terminals or log viewers or other tools, such as when grep determines that log files are binary.
* mod_mam: Avoid storing bounces for messages from the bare account (thanks Ge0rG)Kim Alvefur2021-12-101-2/+12
| | | | | | This should rule out error replies to PEP notifications, which come from the account bare JID, and would be reflected back to it if they can't be delivered for some reason, e.g. s2s problems.
* MUC: Remove <{muc}x> tags in some errorsKim Alvefur2021-12-086-10/+6
| | | | | Including the payload of the stanza that caused the error is optional and we're generally not doing it anywhere else.