| .. |
|
coremods
|
watch: notify WATCH/MONITOR via nick->watchers reverse indexes (watch_by/monitor_by) instead of scanning every user on each online/offline/nick-change flip — O(watchers) not O(users); maintained through centralized watch_index_*/monitor_index_* helpers + quit cleanup
|
2026-08-19 02:34:22 +00:00 |
|
modules
|
channels: replace Member's six parallel prefix bools (oprefix/owner/admin/op/halfop/voice) with a single u8 bitfield (PFX_*) + inline bool accessors/mutators — same semantics, one byte instead of six, no more risk of the flags drifting out of sync; all call sites go through op()/set_op()-style methods
|
2026-08-19 02:47:04 +00:00 |
|
accounts.rs
|
operprefix + ojoin: server oper prefix (!/mode y, above owner) auto-granted to opers + OJOIN command
|
2026-08-10 13:01:07 +00:00 |
|
bcrypt.rs
|
bcrypt: native $2b$ hashing (blowfish + eksblowfish; pi constants via exact fixed-point machin) wired into password_hash + MKPASSWD
|
2026-08-11 11:32:35 +00:00 |
|
channels.rs
|
channels: replace Member's six parallel prefix bools (oprefix/owner/admin/op/halfop/voice) with a single u8 bitfield (PFX_*) + inline bool accessors/mutators — same semantics, one byte instead of six, no more risk of the flags drifting out of sync; all call sites go through op()/set_op()-style methods
|
2026-08-19 02:47:04 +00:00 |
|
command.rs
|
operprefix + ojoin: server oper prefix (!/mode y, above owner) auto-granted to opers + OJOIN command
|
2026-08-10 13:01:07 +00:00 |
|
config.rs
|
refactor: WEBIRC gateways and +G censor rules are named structs (WebircGateway/CensorRule) instead of positional tuples — self-documenting field access, no (_, g, _) index guessing; extends the OperBlock pattern
|
2026-08-18 23:31:21 +00:00 |
|
extensible.rs
|
perf: mimalloc global allocator + aHash maps + memchr line framer + LTO/codegen-units=1 — ~29% faster channel fanout; and drop the bogus openssl+mio dependency whitelist from the guard (any perf crate is welcome now)
|
2026-08-18 19:45:33 +00:00 |
|
http.rs
|
http: verify TLS certs by default + cap the response body — set_verify(NONE) unconditionally let a MITM read/forge the account-registration & captcha traffic (usernames, emails, plaintext passwords) POSTed to operator-configured https URLs; now the connector's cert+hostname check stays on (opt out with http_tls_verify=no), and read_to_end is bounded to 4 MiB so a hostile endpoint can't OOM the worker
|
2026-08-19 00:37:50 +00:00 |
|
ircd.rs
|
metrics: optional OpenMetrics/Prometheus endpoint (metrics_bind, off by default) — commands/messages/connects counters bumped inline via shared atomics, users/channels/servers/links gauges republished each tick; no event round-trip on the hot path
|
2026-08-18 22:34:36 +00:00 |
|
lib.rs
|
test: deterministic two-node S2S convergence simulator — real handshake + join/part/rejoin/FJOIN-TS-arbitration driven through actual code paths, proptest-randomised churn asserts both sides always converge; extract Server::part so the sim and the PART command share one path
|
2026-08-18 22:24:19 +00:00 |
|
link.rs
|
s2s: extend source_behind to FTOPIC/RENAME/UID/OPERTYPE/REDACT — the guard was on plain TOPIC but not its S2S twin FTOPIC (a peer could forge a network-wide topic overwrite), nor channel RENAME, UID (SID announce unchecked), OPERTYPE, or REDACT; all now validate the source lives behind the arriving link
|
2026-08-19 04:45:06 +00:00 |
|
main.rs
|
metrics: optional OpenMetrics/Prometheus endpoint (metrics_bind, off by default) — commands/messages/connects counters bumped inline via shared atomics, users/channels/servers/links gauges republished each tick; no event round-trip on the hot path
|
2026-08-18 22:34:36 +00:00 |
|
map.rs
|
perf: mimalloc global allocator + aHash maps + memchr line framer + LTO/codegen-units=1 — ~29% faster channel fanout; and drop the bogus openssl+mio dependency whitelist from the guard (any perf crate is welcome now)
|
2026-08-18 19:45:33 +00:00 |
|
message.rs
|
message: parse the IRCv3 tag block in a single pass instead of four separate split(';') scans plus an intermediate Vec for ctags
|
2026-08-19 02:54:24 +00:00 |
|
mode.rs
|
channels/whois: validate +k/+l, cap and case-fold ban lists, honour -n for externals, hide +s/+p/+i from whois/who/names, 301 in whois, conf-key the join/nickflood lockout
|
2026-08-16 17:56:24 +00:00 |
|
module.rs
|
registration: add a Hold verdict so captcha/challenge modules pend the client for the challenge instead of tearing the link down
|
2026-08-17 19:06:17 +00:00 |
|
numeric.rs
|
add HELP command with a built-in index and per-topic help (704/705/706)
|
2026-08-16 18:20:57 +00:00 |
|
proxy.rs
|
test: property-based fuzzing (proptest) for every untrusted-input parser — message line, PROXY header, WebSocket frame, regex engine, ban-mask, duration; asserts no-panic + round-trip/idempotence/bounds invariants
|
2026-08-18 22:17:17 +00:00 |
|
regex.rs
|
test: property-based fuzzing (proptest) for every untrusted-input parser — message line, PROXY header, WebSocket frame, regex engine, ban-mask, duration; asserts no-panic + round-trip/idempotence/bounds invariants
|
2026-08-18 22:17:17 +00:00 |
|
resolver.rs
|
resolver: close an off-path DNS spoof — connect() the UDP socket so the kernel drops replies from any IP but the nameserver, and use a CSPRNG per-query transaction id instead of the hardcoded 0x4543/0x4544 (which were in the public source, so rDNS/DNSBL answers could be forged with no guessing)
|
2026-08-19 00:33:18 +00:00 |
|
s2s_sim.rs
|
server: scrub a departed user's pending invites via a User.invited reverse index instead of scanning every channel on the network per quit — the old O(channels)-per-quit path was O(channels*quits) on a netsplit; the index is maintained at the 4 invite add/remove sites (INVITE cmd, S2S INVITE, join-consume, UNINVITE)
|
2026-08-19 02:09:36 +00:00 |
|
server.rs
|
channels: replace Member's six parallel prefix bools (oprefix/owner/admin/op/halfop/voice) with a single u8 bitfield (PFX_*) + inline bool accessors/mutators — same semantics, one byte instead of six, no more risk of the flags drifting out of sync; all call sites go through op()/set_op()-style methods
|
2026-08-19 02:47:04 +00:00 |
|
socketengine.rs
|
socketengine: cap bytes drained from one socket per readable event (MAX_READ_PER_TURN=64KiB), then re-arm epoll and yield — bounds the per-turn line buffer and stops one flooding client from monopolising the reactor; the leftover waits in the kernel buffer and is re-delivered next turn (verified: a 133KB single-write burst gets every reply back)
|
2026-08-19 03:28:51 +00:00 |
|
tls.rs
|
perf: mimalloc global allocator + aHash maps + memchr line framer + LTO/codegen-units=1 — ~29% faster channel fanout; and drop the bogus openssl+mio dependency whitelist from the guard (any perf crate is welcome now)
|
2026-08-18 19:45:33 +00:00 |
|
tls_rustls.rs
|
tls: add opt-in pure-Rust rustls backend behind the TlsBackend trait (tls_backend = rustls; default stays openssl) — TLS1.3, CertFP via handshake-sig verify, SNI, REHASH reload; bump integration timeouts for loaded-host robustness
|
2026-08-18 20:20:21 +00:00 |
|
users.rs
|
server: carry accepted_nicks count across a nick change — set_nick rewrites ACCEPT entries old->new in place (bypassing accept_add/remove) so the reverse count desynced; the count-gated quit scrub could then skip a reused nick, letting it inherit +g acceptance. move the count old->new (also fixes a slow map leak)
|
2026-08-19 04:45:07 +00:00 |
|
watch.rs
|
watch: notify WATCH/MONITOR via nick->watchers reverse indexes (watch_by/monitor_by) instead of scanning every user on each online/offline/nick-change flip — O(watchers) not O(users); maintained through centralized watch_index_*/monitor_index_* helpers + quit cleanup
|
2026-08-19 02:34:22 +00:00 |
|
websocket.rs
|
test: property-based fuzzing (proptest) for every untrusted-input parser — message line, PROXY header, WebSocket frame, regex engine, ban-mask, duration; asserts no-panic + round-trip/idempotence/bounds invariants
|
2026-08-18 22:17:17 +00:00 |
|
xline.rs
|
s2s/xline: reject a malformed ADDLINE duration instead of coercing it to 0 (= a silent permanent ban), and make add_xline expiry saturating (n.saturating_add) so a peer sending duration=u64::MAX can't overflow-panic the debug build or wrap in release
|
2026-08-19 00:39:40 +00:00 |
