1,979 Commits over 488 Days - 0.17cph!
Update: apped hex prefix when emitting native method address
Tests: none, trivial change
Update: emit address of native function
Trying to chase up why sometimes names missmatch with what VS disassembly window shows
Tests: viewed BasePlayer::BoxTest
Bugfix: use mnemonics instead of bytes when checking if decoding should finish
My initial attempt to inpsect raw opcode didn't work in all cases, leading to false-positive parsing aborts.
Tests: viewed BasePlayer::ServerUpdate and BasePlayer::BoxTest
Update: better detection when to stop decoding stream
- now tracks max referenced jump address and stops on last ret or jmp -offset
Tests: viewed nested call in BasePlayer::BoxTest
Bugfix: ensure instruction pointer gets saved to list only if we stopped parsing teh stream
Previously we could add the ip to the list before we decided to stop decoding, leading to out of sync issues
Tests: viewed BasePlayer::BoxTest
Update: change asm text color if it's a potential address
Tests: viewed BasePlayer::BoxTest
Bugfix: avoid double-draw
Tests: viewed BasePlayer::BoxTest
Update: Add relative jump arrows
This was much more painful than expected, but a bit of hodge-podge and it's gud
Tests: viewed BasePlayer::BoxTest
Update: hook up native symbol resolution
- ServerProfiler.Core built from 0b3b7d86
This gets us various mono_... calls, but still missing some function names - guessing those are mono-related but not part of mono dll itself
Tests: Viewed BasePlayer::BoxTest and stepped through nested calls
Update: separate out JITViewer logic from ServerProfiler
- ServerProfiler.Core built from 1074407d
- also contains native symbol resolving logic, but not plugged in yet
JITViewer functionality is now editor only(dll won't ship with builds).
Tests: viewed BasePlayerr::BoxTest
Update: couple improvements to JIT viewer display
- relative jumps are shown as "signed hex" offsets, to avoid jnz 0xffffffffffffffe cases
- avoid duplication of address when we have a method name for a jitted address
Tests: viewed BasePlayer::BoxTest and BasePlayer::ServerUpdateParallel
Clean: get rid of debug logs
Tests: none, trivial change
Update: remember selecter row index between dissasm states
Tests: viewed BasePlayer::BoxTest
Update: use consecutive int3s to stop decoding
Doesn't always apply, but when it does saves a lot
Tests: viewed BasePlayer::BoxTest
Update: allow inspecting instructions at address-like values
Need to improve stream decoding in those cases, as it can run away to 12k lines of assembly even though the method was over on line 10.
Tests: viewed BasePlayer::BoxTest
Clean: logic refactor in prep for multiple view support
Tests: viewed BasePlayer::BoxTest
Update: resolve addresses to methods/trampolines where possible
- ServerPorfiler built from 18130132
A bit ugly, but don't care for now
Tests: viewed BasePlayer::ServerUpdateParallel
Bugfix: jump addresses are now valid
- Also added ability to show opcode bytes
- ServerProfiler built from 95cd43f7
Was due to invalid use of library API (I should've RTFMd)
Tests: viewed BasePlayer::ServerUpdateParallel
Clean: remove binary dumping debug code
Mystery byte was caused by fopen using "w" mode isntead of "wb", which allows windows to handle special characters. In our case it was injecting an extra byte that was screwing over jumps. But it was all debug code anyway.
Tests: none, trivial change
Bugfix: treat instruction stream as 64bit, not 32bit
- ServerProfiler built from 6c5c94dd
Got mislead by Unity's mono --version reporting it as x86. Last bug left - an unexpected byte appearing mid instruction stream when copying from unmanaged, but missing on managed side
Tests: viewed BasePlayer::ServerUpdateParallel
Bugfix: use ListView+labels intead of multiline TextField
Fixes weird disappearing text and gets rid of the vertex limit for text
Tests: viewed BasePlayer.ServerUpdateParallel
Update: display 32-bit disassembled instruction stream in editor window
- ServerProfiler built from 5bff6c48
Couple weird issues - textfield goes invisible half-way through (but can copy the text out of it and goes back visible before end), binary stream differs between managed and unmanaged sides. Will chase up next
Tests: viewed BasePlayer::ServerUpdateParallel
New: ServerProfiler can now dump raw binary of a managed method after JIT
- using debug binary built from facc83cb
Baby steps. No idea how to read it yet, and will need to explore how to control optimization levels.
Tests: Exported Assembly-CSharp!BasePlayer::ServerUpdateParallel
Merge: from spectate_related_cmd_fixes
- sleeping-/usersinrange commands now run from spectated player's position, if spectating
Tests: used commands when spectating
Update: when running sleeping-/usersinrange, prefer to query from spectating target position
Tests: used the commands while spectating
Merge: from spectate_command_fix
- Bugfix for spectate doing nothing when spectating far-away player
- Bugfix for error on respawn after spectate if previously didn't have a spectate target
Tests: used spectate out of network range 10 times in a row, used keybind to spectate twice in a row - no issues
Bugfix: Spectating - when no valid spectator found when starting to spectate, ensure we still set the right SubStrategy
- If no spectate targets were found, chat message now contains the used filter
Could easily happen for people using keybinds, as keybind sends "spectate True"
Tests: using keybind, spectated other player twice in a row - no more errors
Bugfix: Spectating - ensure Spectating flag always gets sent to the owner
Fallout from the queue skipping optimization
Tests: Went outside of network range, waited for subscription to unsub and and started spectating on a 2p server. Did it 10 times - no failures
Update: Adopt Scientist2FSM to be aware of ghostship navmesh
Need to double check aiming logic (sometiems they aim at the sky), and investigate how initial on-navmesh spawning is done. Also need to convert variant FSMs
Tests: spawned ship navmesh offset and with a rotation. Ran around into different areas of the ship and rooms - they pursued
Update: serverocclusiondebug - clarify what invalid grid means
Now explicitly saying that server occlusion treats path with one or more invalid grid positions as unblocked
Tests: none, trivial change
Bugfix: server occlusion - skip occlusion for players outside of the occlusion grid
Tests: checked occlusion between y:90 and y:110 - previously would stutter(OcclusionV1) or would occlude(OcclusionV2). Tested hiding behind a hill - still works
Update: introduce BaseEntity.ServerNavMeshPos property and use it instead of the methods
Incomplete work(need to expose transform matrices), will return after bugfixing
Tests: none
Update: LimitedTurnNavAgent sets npc's position via helper functions
- ScientistNPC2 overrides it to optionally map on to the detached navmesh if it's on a ghost ship
- LimitedTurnNavAgent uses navmesh positions to calculate steering to waypoint in FaceTarget mode (think it was the original intent anyway)
There's a bunch more code that needs to be converted just to be safe with strong waves
Tests: spawned ghost ship away from craggy, enabled AI, and ran around to different areas - scientist caught up, were able to shoot me
Update: GhostShip spawns navmesh instead of NavMeshSurface
- Disabled NavMeshSurface on GhostSHip prefab (will need to do it for the rest later)
This stops NavMeshSurface constrantly running add-remove navmesh every server frame. Need to figure out how to tie it to NPCs
Tests: visualzied navmesh in editor via gizmos, checked profiler to not see any more NavMeshSurface::UpdateActive calls
Merge: from baseplayer_serverupdateparallel
- Optim for OcclusionV2 - remove extra hash lookup in a busy loop
Tests: with UsePlayerUpdateJobs 2 - toggled invis in close proximity and far away, suicided away from spawn and flew back, teleported around
Optim: OcclusionV2 - skip resetting lastPlayerVisibility every frame
Turns out didn't need to make the other half of changes
Tests: used invis close and far away
Update: UseOcclusionV2 - simplify toggling invis
- when enabling invis, notify occlusion tracking that we destroyed entity on subscriber's end
- when disabling invis, skip sending snapshots - occlusion will handle it
Halfway to remove the constant hashSet.Remove calls in GatherPairs loop
Tests: UsePlayerUpdateJobs 2, toggled invis in front of player. went invis, flew outside network range, flew back, disabled invis - spawned correctly.
Merge: from baseplayer_serverupdateparallel
- Bugfix for UsePlayerUpdateJobs 2 not replicating players for spectators
- Bugfix for leaking players in occlusion groups
Tests: spectated players from far away. Inspected server state after killing a sleeper
Bugfix: ServerOcclusion - clean up players from all of their occlusion groups on server destroy
Turns out we were leaving a lot of "zombie" players once sleepers were killed. This would slow down server occlusion eventually.
Tests: on Craggy, killed a sleeper, moved to a different network group, and checked in debugger if the occlusion group only contains my player
Bugfix: UseOcclusionV2 - handle spectating players
Since we no longer do bidirectional checks when iterating occlusion groups, we ended up missing out on spectators and wouldn't allow replication to them.
Tests: on craggy with UsePlayerUpdateJobs 2, flew far away to get outside of network range then started spectating - could see target moving.
Merge: from spectate_dontfloodsnapshots
- Bugfix for occlusion not destroying all players in UsePlayerUpdateJobs 2 mode
Tests: 2p on craggy - used hill as obstruction, checked entity lists on both clients in editor
Bugfix: UsePlayerTasks - calculate end of batch correctly in SendEntityDestroyMessages_AsyncState
This was causing server occlusion to skip sending some of destroy commands (from last batch), leaving a phantom player
Tests: on Craggy went on behind hill, then checked editor entity list - confirmed that replicated players got removed on both ends
Merge: from spectate_dontfloodsnapshots
- Fixes spectate not working in UsePlayerUpdateJobs 2 case
Tests: tried to spectate occluded player
Bugfix: Ensure initial player snapshot gets sent to spectator in UsePlayerUpdateJobs 2 case
Tests: on craggy with UsePlayerUpdateJobs 2, took 2nd player behind a hill to occlude, then tried to spectate them - confirmed that the first snapshot got sent.
Clean: post-submit feedback
- rename entity_prefabId field to entity_prefab_id
- rename AddShortEntity to AddShortEntityField
More consistent with other examples
Tests: compiles in editor
Merge: from analytics_entity_createdestroy
- minor code clean
Tests: editor compiles
Merge: from analytics_entity_createdestroy
- we now collect entity spawn and destroy events once the server has started up (part of `gameplay_analytics` servervar)
Tests: observed logs of analytics in CLIENT+SERVER editor
