By default, spigot shifts chat over to an unbounded thread pool,
on a normal server, this really offers no gains, the creation of a thread
on submitting to the pool on these servers eats more time vs just running it in
the netty pipeline, however, on servers using plugins which do work in here, there
could be some overall benefits to moving this stuff outside of the pipeline.
In general, this patch does two things:
1) Exposes the core size for the pool, this allows for ensuring that a number of threads
sit around in the pool, mitigating the need for creating new threads; This IS however
caveated, the ThreadPoolExecutor will ONLY create core threads as they're needed, it
just won't allow for us to dip back under the # of core threads, this can potentially
be mitigated by calling prestartCoreThread, however, I'm not sure if there is much justification
for this
2) Exposes a max size for the pool, as stated, by default this is unbounded, for most
servers limiting the size of the pool is going to have 0 effects given how fast chat
is actually processed, this is honestly really just exposed for the misnomers or people
who just wanna ensure that this won't grow over a specific size if chat gets stupidly active
While Velocity supports BungeeCord-style IP forwarding, it is not secure. Users
have a lot of problems setting up firewalls or setting up plugins like IPWhitelist.
Further, the BungeeCord IP forwarding protocol still retains essentially its original
form, when there is brand new support for custom login plugin messages in 1.13.
Velocity's modern IP forwarding uses an HMAC-SHA256 code to ensure authenticity
of messages, is packed into a binary format that is smaller than BungeeCord's
forwarding, and is integrated into the Minecraft login process by using the 1.13
login plugin message packet.
Beyond calling the BlockFadeEvent in more places, this patch also aims
to pass the proper replacement state to the event, specifically for
potentially waterlogged block states fading.
Co-authored-by: Lulu13022002 <41980282+Lulu13022002@users.noreply.github.com>
Minecraft's prediction system does not handle block entities, so if we are manually sending block entities during
block breaking we need to set it after the prediction is finished. This fixes block entities not showing when cancelling the BlockBreakEvent.
Paper makes the entity random thread-safe
and constructing an entity off the main thread
should be supported. Some entities (for whatever
reason) use the level's random in some places.
These can be called while an entity is being added to the world,
and if the entity is being added from a chunk load context the
sync load will block indefinitely (because the chunk load context
is for completing the chunk to FULL).
This does raise questions about the current system for these
dynamic registrations, as it looks like there is _zero_ logic
to account for the case where the chunk is _not_ currently loaded
and then later loaded.
The previous solution caused a bunch of bandaid fixes inorder to resolve edge cases where minecraft/the api might spawn items that are air.
Just simply prevent them from being added to the world instead.
Craftbukkit attempts to translate worlds that use the
same generation as the Overworld, The Nether, or The End
to use those dimensions when checking the `changed_dimension`
criteria trigger, or whether to trigger the `NETHER_TRAVEL`
distance trigger. This adds a config option to ignore that
and use the exact dimension key of the worlds involved.
When calling getBedSpawnLocation on an
instance of CraftOfflinePlayer the world was incorrect
due to the logic for reading the NBT not being up-to-date.
- add auto expire setting
- add setter for block data
- add accessors for block state
== AT ==
public net.minecraft.world.entity.item.FallingBlockEntity blockState
Co-authored-by: Lukas Planz <lukas.planz@web.de>
Fixes https://github.com/PaperMC/Paper/issues/7276 and https://github.com/PaperMC/Paper/issues/8118
by using a config option that, when set to false, does not add markers to the entity
tick list at all and ignores them in Spigot's activation range checks. The entity tick
list is only used in the tick and tickPassenger methods, so we can safely not add the
markers to it. When the config option is set to true, markers are ticked as normal.
By checking if the world passed into StructureTemplate's placeInWorld
is not a WorldGenRegion, we can bypass the deadlock entirely.
See https://bugs.mojang.com/browse/MC-246262
Also standardizes how to handle EntityChangeBlockEvent before a removeBlock or destroyBlock
call. Always use 'state.getFluidState().createLegacyBlock()' to get the new state instead of
just using the 'air' state.
Also fixes the new block data for EntityBreakDoorEvent (a sub-event from
EntityChangeBlockEvent)
Co-authored-by: Lulu13022002 <41980282+Lulu13022002@users.noreply.github.com>
Co-authored-by: Jake Potrebic <jake.m.potrebic@gmail.com>
lets debug mode throw a trace in order to potentially see where
such calls are cascading from easier, but, generally, if you see one setBlock
call, you're gonna see more, and this just potentially causes a flood of logs
which can cause issues for slower terminals, etc.
We can limit the flood by just allowing one for a single gen region,
we'll also only gen a trace for the first one, I see no real pressing need
to generate more, given that that would *massively* negate this patch otherwise
The worldgen thread will attempt to get structure references
via the world's getChunkAt method, which is fine if the gen is
not cancelled - but if the chunk was unloaded, the call will block
indefinitely. Instead of using the world state, we use the already
supplied ServerLevelAccessor which will always have the chunk available.
== AT ==
public net.minecraft.world.entity.projectile.FishingHook timeUntilLured
public net.minecraft.world.entity.projectile.FishingHook fishAngle
public net.minecraft.world.entity.projectile.ShulkerBullet targetDeltaX
public net.minecraft.world.entity.projectile.ShulkerBullet targetDeltaY
public net.minecraft.world.entity.projectile.ShulkerBullet targetDeltaZ
public net.minecraft.world.entity.projectile.ShulkerBullet currentMoveDirection
public net.minecraft.world.entity.projectile.ShulkerBullet flightSteps
public net.minecraft.world.entity.projectile.AbstractArrow soundEvent
public net.minecraft.world.entity.projectile.AbstractArrow setPickupItemStack(Lnet/minecraft/world/item/ItemStack;)V
public net.minecraft.world.entity.projectile.ThrownTrident dealtDamage
public net.minecraft.world.entity.projectile.Arrow NO_EFFECT_COLOR
public net.minecraft.world.entity.projectile.Projectile hasBeenShot
public net.minecraft.world.entity.projectile.Projectile leftOwner
public net.minecraft.world.entity.projectile.Projectile ownerUUID
public net.minecraft.world.entity.projectile.Projectile preOnHit(Lnet/minecraft/world/phys/HitResult;)V
public net.minecraft.world.entity.projectile.Projectile canHitEntity(Lnet/minecraft/world/entity/Entity;)Z
public net.minecraft.world.entity.projectile.FireworkRocketEntity getDefaultItem()Lnet/minecraft/world/item/ItemStack;
public net.minecraft.world.item.CrossbowItem FIREWORK_POWER
Co-authored-by: Nassim Jahnke <nassim@njahnke.dev>
Co-authored-by: SoSeDiK <mrsosedik@gmail.com>
Co-authored-by: MelnCat <melncatuwu@gmail.com>
Co-authored-by: Lulu13022002 <41980282+Lulu13022002@users.noreply.github.com>
Piercing arrows search for multiple entities inside a while
loop that is checking the projectile entity's removed state.
If the hit event is cancelled on the first entity, the event will
be called over and over again inside that while loop until the event
is not cancelled. The solution here, is to make use of an
already-existing field on AbstractArrow for tracking entities hit by
piercing arrows to avoid duplicate damage being applied.
== AT ==
protected net.minecraft.world.entity.projectile.Projectile hitCancelled
Restores the API behavior from previous versions of the server
- Do not call API events
- Do not replace the existing block in the world
== AT ==
public net.minecraft.world.entity.item.FallingBlockEntity <init>(Lnet/minecraft/world/level/Level;DDDLnet/minecraft/world/level/block/state/BlockState;)V
Dead players would be stuck in the world loading screen and other players may
miss messages and similar sent in the join event if chunk loading is slow.
Paper already circumvents falling through the world before chunks are loaded,
so we do not need that. The client only needs the chunk it is currently in to
be loaded to close the loading screen, so we just send an empty one.
By default, only LevelStem's that specifically match the ResourceKey for
OVERWORLD will have the 5 (currently) impls of CustomSpawner (for
phantoms, wandering traders, etc.). This adds an option to instead of
just looking at the LevelStem key, look at the DimensionType key which
is one level below that. Defaults to off to keep vanilla behavior.
General purpose patch adding missing getters/setters to BlockData and
its child types.
Co-authored-by: SoSeDiK <mrsosedik@gmail.com>
Co-authored-by: Fabrizio La Rosa <lr.fabrizio@gmail.com>
The 4 missing structure set seed configs are strongholds, mineshafts,
buried treasure, and ancient cities.
Strongholds use a ring placement scheme which isn't random so they
utilize the world seed by default, this adds a config to override it
for just generating the ring positions.
Mineshafts and Buried Treasure structure sets are special cases
where the "salt" that can be defined for them via datapacks has 0
effect because the difference between the spacing and separation is 1
which is used as the upper bound in the random with salt. So the random
always returns the same int (0) so the salt has no effect. This adds
seeds/salts to the frequency reducer which has a similar effect.
Co-authored-by: William Blake Galbreath <blake.galbreath@gmail.com>
There is an explicit check in the handling code for empty pistons that
prevents sticky pistons from firing the event. However when we look back
at the history we see that this check was originally added so that ONLY
sticky pistons would fire the retract event. I'm not sure why.
1092acbddf
Over the course of several updates, the meaning of that field appears to
have changed from "is NOT sticky" to "is sticky". So now its having the
opposite effect. Only normal pistons fire the retraction event. And like
all things in CB, it's just been carried around since.
If we are to believe the history, the correct fix for this issue is to
flip it so it only fires for sticky pistons, but that puts us in a
bind. It's already firing for non-sticky pistons, changing it now would
likely result in breakage. Furthermore, there is little documentation as
to WHY that was ever intended to be the case.
Instead we opt to remove the check entirely so that the event fires for
all piston types.
Co-authored-by: Zach Brown <1254957+zachbr@users.noreply.github.com>
Co-authored-by: Madeline Miller <mnmiller1@me.com>
HashMapPalette uses an instance of CrudeIncrementalIntIdentityHashBiMap
internally. A Palette has a preset maximum size = 1 << bits.
CrudeIncrementalIntIdentityHashBiMap has an initial size but is
automatically resized. The CrudeIncrementalIntIdentityHashBiMap is created
with the maximum size in the constructor of HashMapPalette, with the aim
that it doesn't need to be resized anymore. However, there are two things
that I think Mojang hasn't considered here:
1) The CrudeIncrementalIntIdentityHashBiMap is resized, when its initial
size is reached and not the next time, when a further object is added.
2) HashMapPalette adds objects (unnecessarily) before checking if the
initial size of CrudeIncrementalIntIdentityHashBiMap is reached.
This means to actually avoid resize operations in
CrudeIncrementalIntIdentityHashBiMap, one has to add 2 to the initial size
or add 1 and check the size before adding objects. This commit implements
the second approach. Note that this isn't only an optimization but also
makes async reads of Palettes fail-safe. An async read while the
CrudeIncrementalIntIdentityHashBiMap is resized is fatal and can even lead
to corrupted data. This is also something that Anti-Xray is currently
relying on.
Fixes kelp modifier changing growth for other crops
Also add growth modifiers for glow berries, mangrove propagules,
torchflower crops and pitcher plant crops
Also fix above-mentioned modifiers from having the reverse effect
Co-authored-by: Jake Potrebic <jake.m.potrebic@gmail.com>
Co-authored-by: Noah van der Aa <ndvdaa@gmail.com>
Co-authored-by: Lulu13022002 <41980282+Lulu13022002@users.noreply.github.com>
This patch detects whether or not the server is currently executing as a privileged user and spits out a warning.
The warning serves as a sort-of PSA for newer server admins who don't understand the risks of running as root.
We've seen plenty of bad/malicious plugins hit markets, and there's been a few close-calls with exploits in the past.
Hopefully this helps mitigate some potential damage to servers, even if it is just a warning.
Co-authored-by: Noah van der Aa <ndvdaa@gmail.com>
Sometimes, blocks are changed and then logic is called before the associated
tile entity is removed. When this happens, the factories were relying on the
block at the position, not the tile entity. This change prioritizes using the
tile entity type to determine the block state factory and falls back on
the material type of the block at that location.
== AT ==
public net.minecraft.world.level.block.entity.BlockEntityType validBlocks
In certain areas the player's head rotation could be desynced when teleported/moved.
This is because bukkit uses a separate head rotation field for yaw.
This issue only applies to players.
Preserves overstacked items in loot tables, such as shulker box drops, to prevent the items
from being deleted (as they'd overflow past the bounds of the container)-- or worse, causing
chunk bans via the large amount of NBT created by unstacking the items.
Fixes GH-5140 and GH-4748.
Prior to this commit, the tropical fish bucket meta would set the body
color to the previous metas pattern colour when updating the pattern
colour.
This commit hence simply fixes this by using the proper body colour
value when updating the pattern color.
Example config:
packet-limiter:
kick-message: '&cSent too many packets'
limits:
all:
interval: 7.0
max-packet-rate: 500.0
ServerboundPlaceRecipePacket:
interval: 4.0
max-packet-rate: 5.0
action: DROP
all section refers to all incoming packets, the action for all is
hard coded to KICK.
For specific limits, the section name is the class's name,
and an action can be defined: DROP or KICK
If interval or rate are less-than 0, the limit is ignored
Child nodes are handled by CommandDispatcher#parse checking
requirements.
Vanilla clients only send ServerboundCommandSuggestionPacket when
encountering a command node with ASK_SERVER suggestions, however a
modified client can send this packet whenever it wants.
Just use the iblockdata already retrieved, removes a getType call.
Also save approx. 5% for the raytrace call, as most (expensive)
raytracing tends to go through air and returning early is an
easy win. The remaining problems with this function
are mostly with the block getting itself.
Tux did some profiling some time ago and showed that the
previous getChunkAt method which had inlined logic for loaded
chunks did get inlined, but the standard CPS.getChunkAt
method was not inlined.
Add commands to get the mobcaps for a world, as well as the mobcaps for
each player when per-player mob spawning is enabled.
Also has a hover text on each mob category listing what entity types are
in said category
Log when the async catcher is tripped
The chunk system can swallow the exception given it's all
built with completablefuture, so ensure it is at least printed.
Add/move several async catchers
Async catch modifications to critical entity state
These used to be here from Spigot, but were dropped with 1.17.
Now in 1.17, this state is _even more_ critical than it was before,
so these must exist to catch stupid plugins.
Co-authored-by: Jake Potrebic <jake.m.potrebic@gmail.com>
Still call the event and change the active container though. We
want to avoid close logic because it's possible to load the
chunk through it. This should also be OK from a leak prevention/
state desync POV because the TE is getting unloaded anyways.
Fixes permission checks for vanilla commands which don't have a
requirement, as well as for namespaced vanilla commands.
== AT ==
public-f com.mojang.brigadier.tree.CommandNode requirement
