Age | Commit message (Collapse) | Author | |
---|---|---|---|
2013-04-21 | Don't crop the channel name if it's too long in Channel::Channel() | attilamolnar | |
...and especially don't use the shortened name in one place and the original in another Having different <limits:maxchan> values on the same network is not supported | |||
2013-04-19 | Immediately stop processing whenever we detect and handle a RecvQ overrun | attilamolnar | |
Thanks to @SimosNap for the report and cooperation | |||
2013-04-19 | Fix uuids getting truncated in the nick hash in UserManager::AddUser() if ↵ | attilamolnar | |
nickmax is < 9 | |||
2013-04-19 | m_ident Invalidate cache after changing User::ident | attilamolnar | |
2013-04-19 | m_callerid Allow messaging yourself while +g regardless of the ACCEPT list | attilamolnar | |
2013-04-18 | m_spanningtree Fix IS_LOCAL() check in OnRehash handler | attilamolnar | |
2013-04-18 | m_callerid Ignore duplicate entries when unserializing callerid_data | attilamolnar | |
2013-04-18 | m_callerid Fix bookkeeping error introduced when unserializing callerid_data | attilamolnar | |
This also fixes a memory leak that didn't occur naturally but was triggerable by remote servers Thanks to @SimosNap for the report | |||
2013-04-16 | Fix m_ssl_gnutls and perhaps some other things on Windows by recognizing ↵ | Adam | |
WSAEWOULDBLOCK | |||
2013-04-16 | m_kicknorejoin Store and compare uuids instead pointers | attilamolnar | |
Fixes the off chance scenario where we disallow a join because a previously kicked user has quit and the User who is trying to join happens to be allocated at the exact same memory location | |||
2013-04-16 | m_kicknorejoin Limit time to 30m by default | attilamolnar | |
In the current implementation we only expire entries when someone joins, without a limit it was possible to make us practically never remove entries and consume (a tiny amount of) memory for each entry until the mode was removed/parameter was changed The default limit of 30m is chosen to not surprise people when they upgrade. If you need to prevent rejoins for more than a minute then you should set a (timed)ban instead Config option is available to change the limit (2.0 only) | |||
2013-04-16 | m_kicknorejoin Minor improvements | attilamolnar | |
- Ignore remote users - Remove expired items in one pass | |||
2013-04-16 | Close listening sockets regardless of the return value of shutdown() | attilamolnar | |
2013-04-16 | Do not enable SO_LINGER on our sockets | attilamolnar | |
Using this option allowed close() to block for up to a second Thanks to Shamsdeen and Rix for their assistance that made this fix possible Fixes issue #445 reported by @shaggie76 Fixes issue #494 reported by @Rixcho | |||
2013-04-10 | m_spanningtree Create new TreeServers for incoming connections only when ↵ | attilamolnar | |
they've accepted our credentials, not when they send SERVER | |||
2013-04-09 | Remove some uline checks that ran after an IS_LOCAL() check | attilamolnar | |
2013-04-05 | Don't attempt to unload or reload modules that are waiting to be unloaded | attilamolnar | |
2013-04-05 | Fix LUSERS not working in a PURE_STATIC build | attilamolnar | |
2013-04-01 | cmd_stats List opers without iterating the whole userlist | attilamolnar | |
2013-04-01 | cmd_invite Check if the inviting user is on the channel before potentially ↵ | attilamolnar | |
telling him the target is already on it | |||
2013-03-29 | Release 2.0.11 | attilamolnar | |
2013-03-29 | cmd_kill Send snotice to servers when both the oper and the victim are on ↵ | attilamolnar | |
the same server | |||
2013-03-29 | Update doc and fix the description of m_allowinvite | attilamolnar | |
2013-03-29 | m_ldapoper Make the module actually work | attilamolnar | |
Fixes #439 reported by @rhylan | |||
2013-03-29 | m_ldapoper Fix memory leak | attilamolnar | |
See 1813369adecc1efc9812e90c40c21dc32e4965c9 for details | |||
2013-03-28 | Fix premature overwriting of OLDTIME that resulted in time drifts never ↵ | attilamolnar | |
being detected | |||
2013-03-27 | cmd_lusers Fix miscounting | attilamolnar | |
2013-03-26 | Merge pull request #443 from ShutterQuick/sasl_external | Attila Molnar | |
Added support for SASL EXTERNAL | |||
2013-03-25 | Added support for EXTERNAL | Daniel Vassdal | |
2013-03-25 | m_xline_db Fix pointless excessive stack usage in ReadDatabase() | attilamolnar | |
2013-03-25 | m_xline_db Major enhancements | attilamolnar | |
- Save database on a timer, not on every single change - Use the internal database instead of keeping track of xlines ourselves | |||
2013-03-25 | Accept -c as a command line argument (equivalent to --config) | attilamolnar | |
2013-03-24 | Initialize the FakeClient pointer in InspIRCd::InspIRCd() | attilamolnar | |
2013-03-24 | m_timedbans Notice halfops and above when a timed ban is placed | attilamolnar | |
Fixes #415 reported by @killerrabbit | |||
2013-03-20 | Fix infinite loop when all DNS request slots are in use | attilamolnar | |
This is not the best way to detect this scenario, a better detection mechanism will replace this in the future | |||
2013-03-20 | Add support for mandatory tags in included config files | attilamolnar | |
If the mandatory tag is not found in the included config, the rehash is aborted. This is especially useful for remote includes, as it allows users to have a dummy tag at the end of the included config to indicate that the config has been wholly read. This method does not depend on exit codes so even situations where wget returns an empty or a wrong page that we would otherwise accept can be detected and an error can be generated before we assume that the contents have disappeared (and unload all modules, if the included file is supposed to contain module tags, for example). Usage: <include ... mandatorytag="namehere"> - if the included config doesn't contain a <namehere> tag then the rehash is aborted Fixes #30 reported by @Justasic | |||
2013-03-20 | m_spanningtree Fix incoming channel modes being lost for 0-user permanent ↵ | attilamolnar | |
channels on the losing side of a netmerge This did not affect listmodes | |||
2013-03-20 | m_permchannels Load channels from the config only at module load time and ↵ | attilamolnar | |
only when there are no linked servers to avoid desync | |||
2013-03-19 | m_connflood Ignore e-lined users | attilamolnar | |
2013-03-15 | m_ssl_openssl Make it clear that a CA file is not mandatory | attilamolnar | |
2013-03-15 | Modified m_timedbans to behave more like setting mode +b manually. | ShutterQuick | |
It now accepts extbans. If one enters something not an extban, or valid banmask, it'll add "!*@*", making it a nick-ban instead of borking. | |||
2013-03-11 | Remove dead code from DNS | attilamolnar | |
2013-03-11 | m_connectban Switch to OnSetUserIP hook to handle cgiirc users properly, ↵ | attilamolnar | |
ignore e-lined users | |||
2013-03-11 | Fix m_dnsbl not checking cgiirc users when the cgiirc address is elined | attilamolnar | |
2013-03-11 | Either store the Resolver in DNS::AddResolverClass() or delete it before ↵ | attilamolnar | |
returning | |||
2013-03-11 | Stop user initialization when the user is marked as quitting after OnSetUserIP | attilamolnar | |
2013-03-11 | m_dnsbl Fix connecting users timeouting when a dnsbl result is in the dns cache | attilamolnar | |
Fixes #431 reported by @killerrabbit | |||
2013-03-03 | cmd_lusers Be smart and know how many invisible users are there without ↵ | attilamolnar | |
iterating the userlist | |||
2013-03-03 | cmd_lusers Use of size() instead of counting elements | attilamolnar | |
2013-03-03 | cmd_lusers Refresh max local/global user counters after a user connects | attilamolnar | |
Fixes #423 reported by @RawrDragon |