core
/
testserver
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
testserver/Spigot-Server-Patches/0101-Optimize-Chunk-Unload-...

353 lines
16 KiB
Diff
Raw Normal View History

Clean up title and tab list patch Remember kids, the Ballmer peak is hard to hit https://xkcd.com/323/ Finally fixes #126 and #94 (I tested, it does, please, really, promise)
2016-03-23 06:05:09 +00:00
From e6b943c7e0b18b2ce33fa12fe27c298e1a836c2e Mon Sep 17 00:00:00 2001
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
From: Aikar <aikar@aikar.co>
Date: Fri, 18 Mar 2016 17:57:25 -0400
Subject: [PATCH] Optimize Chunk Unload Queue
Removing chunks from the unload queue when performing chunk lookups is a costly activity.
It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid
to skip removing chunks from the unload queue.
This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark
if the chunk is active, and then insert into a LinkedList queue.
The benefits here is that all chunk unload queue actions are now O(1) constant time.
A LinkedList will never need to resize, and can be removed from in constant time when
used in a queue like method.
We mark the chunk as active in many places that notify it is still being used, so that
when the chunk unload queue reaches that chunk, and sees the chunk became active again,
it will skip it and move to next.
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
Also optimize EAR to use these methods.
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
diff --git a/src/main/java/net/minecraft/server/Chunk.java b/src/main/java/net/minecraft/server/Chunk.java
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
index b6d84d7..7ba45c8 100644
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
--- a/src/main/java/net/minecraft/server/Chunk.java
+++ b/src/main/java/net/minecraft/server/Chunk.java
@@ -55,6 +55,8 @@ public class Chunk {
// Keep this synced with entitySlices.add() and entitySlices.remove()
private final int[] itemCounts = new int[16];
private final int[] inventoryEntityCounts = new int[16];
+ public boolean isChunkActive = true;
+ public boolean isInUnloadQueue = false;
// Paper end
// CraftBukkit start - Neighbor loaded cache for chunk lighting and entity ticking
diff --git a/src/main/java/net/minecraft/server/ChunkProviderServer.java b/src/main/java/net/minecraft/server/ChunkProviderServer.java
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
index 9ef6246..f9375eb 100644
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
--- a/src/main/java/net/minecraft/server/ChunkProviderServer.java
+++ b/src/main/java/net/minecraft/server/ChunkProviderServer.java
@@ -21,7 +21,7 @@ import org.bukkit.event.world.ChunkUnloadEvent;
public class ChunkProviderServer implements IChunkProvider {
private static final Logger a = LogManager.getLogger();
- public final LongHashSet unloadQueue = new LongHashSet(); // CraftBukkit - LongHashSet
+ public final ChunkUnloadQueue unloadQueue = new ChunkUnloadQueue(); // CraftBukkit - LongHashSet // Paper -> ChunkUnloadQueue
public final ChunkGenerator chunkGenerator; // CraftBukkit - public
private final IChunkLoader chunkLoader;
public LongObjectHashMap<Chunk> chunks = new LongObjectHashMap<Chunk>(); // CraftBukkit
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
@@ -83,14 +83,24 @@ public class ChunkProviderServer implements IChunkProvider {
return (chunk == null) ? getChunkAt(x, z) : chunk;
}
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
- public Chunk getChunkIfLoaded(int x, int z) {
- return chunks.get(LongHash.toLong(x, z));
Fix chunk unload issues - Resolves #97
2016-03-19 15:29:46 +00:00
+ // Paper start
+ public Chunk getLoadedChunkAtWithoutMarkingActive(int i, int j) {
+ return chunks.get(LongHash.toLong(i, j));
Improve chunk unload queue to maintain some previous expectations While the previous logic was logically correct, some CB API's before would request a chunk without removing it from the unload queue. While this is logically wrong, some plugins seem to be causing unload issues. This change will make anything using that one API that use to not remove from queue, no longer remove from queue. Hopefully other activities on the server will touch the chunk if it REALLY is in use.
2016-03-20 04:13:20 +00:00
}
+ // getChunkIfLoaded -> getChunkIfActive
+ // this is only used by CraftBukkit now, and plugins shouldnt mark things active
+ public Chunk getChunkIfActive(int x, int z) {
+ Chunk chunk = chunks.get(LongHash.toLong(x, z));
+ return (chunk != null && chunk.isChunkActive) ? chunk : null;
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
+ }
Improve chunk unload queue to maintain some previous expectations While the previous logic was logically correct, some CB API's before would request a chunk without removing it from the unload queue. While this is logically wrong, some plugins seem to be causing unload issues. This change will make anything using that one API that use to not remove from queue, no longer remove from queue. Hopefully other activities on the server will touch the chunk if it REALLY is in use.
2016-03-20 04:13:20 +00:00
+ // Paper end
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
+
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
public Chunk getLoadedChunkAt(int i, int j) {
Chunk chunk = chunks.get(LongHash.toLong(i, j)); // CraftBukkit
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
- this.unloadQueue.remove(i, j); // CraftBukkit
+ //this.unloadQueue.remove(i, j); // CraftBukkit // Paper
+ markChunkActive(chunk); // Paper
return chunk;
}
@@ -118,6 +128,7 @@ public class ChunkProviderServer implements IChunkProvider {
// CraftBukkit end
}
+ markChunkActive(chunk); // Paper
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
return chunk;
}
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
@@ -126,7 +137,7 @@ public class ChunkProviderServer implements IChunkProvider {
}
public Chunk getChunkAt(int i, int j, Runnable runnable) {
- unloadQueue.remove(i, j);
+ //unloadQueue.remove(i, j); // Paper
Chunk chunk = chunks.get(LongHash.toLong(i, j));
ChunkRegionLoader loader = null;
@@ -150,12 +161,13 @@ public class ChunkProviderServer implements IChunkProvider {
if (runnable != null) {
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
runnable.run();
}
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
+ markChunkActive(chunk); // Paper
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
return chunk;
}
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
public Chunk originalGetChunkAt(int i, int j) {
- this.unloadQueue.remove(i, j);
+ //this.unloadQueue.remove(i, j); // Paper
Chunk chunk = this.chunks.get(LongHash.toLong(i, j));
boolean newChunk = false;
// CraftBukkit end
@@ -201,7 +213,7 @@ public class ChunkProviderServer implements IChunkProvider {
Fix chunk unload issues - Resolves #97
2016-03-19 15:29:46 +00:00
continue;
}
- Chunk neighbor = this.getChunkIfLoaded(chunk.locX + x, chunk.locZ + z);
+ Chunk neighbor = this.getLoadedChunkAtWithoutMarkingActive(chunk.locX + x, chunk.locZ + z); // Paper
if (neighbor != null) {
neighbor.setNeighborLoaded(-x, -z);
chunk.setNeighborLoaded(x, z);
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
@@ -212,6 +224,7 @@ public class ChunkProviderServer implements IChunkProvider {
chunk.loadNearby(this, this.chunkGenerator);
world.timings.syncChunkLoadTimer.stopTiming(); // Spigot
}
+ markChunkActive(chunk); // Paper
return chunk;
}
@@ -300,10 +313,17 @@ public class ChunkProviderServer implements IChunkProvider {
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
if (!this.world.savingDisabled) {
// CraftBukkit start
Server server = this.world.getServer();
- for (int i = 0; i < 100 && !this.unloadQueue.isEmpty(); ++i) {
- long chunkcoordinates = this.unloadQueue.popFirst();
- Chunk chunk = this.chunks.get(chunkcoordinates);
- if (chunk == null) continue;
+ // Paper start
+ final Iterator<Chunk> iterator = this.unloadQueue.iterator();
+ for (int i = 0; i < 100 && iterator.hasNext(); ++i) {
+ Chunk chunk = iterator.next();
+ iterator.remove();
+ long chunkcoordinates = LongHash.toLong(chunk.locX, chunk.locZ);
+ chunk.isInUnloadQueue = false;
+ if (chunk.isChunkActive) {
+ continue;
+ }
+ // Paper end
ChunkUnloadEvent event = new ChunkUnloadEvent(chunk.bukkitChunk);
server.getPluginManager().callEvent(event);
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
@@ -325,7 +345,7 @@ public class ChunkProviderServer implements IChunkProvider {
Fix chunk unload issues - Resolves #97
2016-03-19 15:29:46 +00:00
continue;
}
- Chunk neighbor = this.getChunkIfLoaded(chunk.locX + x, chunk.locZ + z);
+ Chunk neighbor = this.getLoadedChunkAtWithoutMarkingActive(chunk.locX + x, chunk.locZ + z); // Paper
if (neighbor != null) {
neighbor.setNeighborUnloaded(-x, -z);
chunk.setNeighborUnloaded(x, z);
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
@@ -367,4 +387,69 @@ public class ChunkProviderServer implements IChunkProvider {
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
public boolean e(int i, int j) {
return this.chunks.containsKey(LongHash.toLong(i, j)); // CraftBukkit
}
+
+ // Paper start
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
+ public static void markChunkActive(Chunk chunk) {
+ if (chunk != null) {
+ chunk.isChunkActive = true;
+ }
+ }
+ public class ChunkUnloadQueue extends LongHashSet { // Provide compat with NMS plugins
+ private java.util.LinkedList<Chunk> unloadQueue = new java.util.LinkedList<Chunk>();
+
+ @Override
+ public boolean isEmpty() {
+ return unloadQueue.isEmpty();
+ }
+
+ @Override
+ public boolean contains(long value) {
+ throw new UnsupportedOperationException("contains on unload queue");
+ }
+
+ @Override
+ public boolean add(long value) {
+ throw new UnsupportedOperationException("add on unload queue");
+ }
+
+ @Override
+ public boolean remove(long value) {
+ throw new UnsupportedOperationException("remove on unload queue");
+ }
+
+ @Override
+ public int size() {
+ return unloadQueue.size();
+ }
+
+ @Override
+ public Iterator iterator() {
+ return unloadQueue.iterator();
+ }
+
+ @Override
+ public boolean contains(int x, int z) {
+ final Chunk chunk = chunks.get(LongHash.toLong(x, z));
+ return (chunk != null && chunk.isInUnloadQueue);
+ }
+
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
+ public void remove(int x, int z) {
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
+ final Chunk chunk = chunks.get(LongHash.toLong(x, z));
+ if (chunk != null) {
+ chunk.isChunkActive = true;
+ }
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
+ }
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
+ public boolean add(int x, int z) {
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
+ final Chunk chunk = chunks.get(LongHash.toLong(x, z));
+ if (chunk != null) {
+ chunk.isChunkActive = false;
+ if (!chunk.isInUnloadQueue) {
+ chunk.isInUnloadQueue = true;
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
+ return unloadQueue.add(chunk);
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
+ }
+ }
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
+ return false;
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
+ }
+ }
+ // Paper end
}
diff --git a/src/main/java/net/minecraft/server/SpawnerCreature.java b/src/main/java/net/minecraft/server/SpawnerCreature.java
index 63e118d..721bcae 100644
--- a/src/main/java/net/minecraft/server/SpawnerCreature.java
+++ b/src/main/java/net/minecraft/server/SpawnerCreature.java
@@ -28,7 +28,7 @@ public final class SpawnerCreature {
Long coord = it.next();
int x = LongHash.msw( coord );
int z = LongHash.lsw( coord );
- if ( !((ChunkProviderServer)server.chunkProvider).unloadQueue.contains( coord ) && server.isChunkLoaded( x, z, true ) )
+ if ( server.isChunkLoaded( x, z, true ) ) // Paper
{
i += server.getChunkAt( x, z ).entityCount.get( oClass );
}
Improve chunk unload queue to maintain some previous expectations While the previous logic was logically correct, some CB API's before would request a chunk without removing it from the unload queue. While this is logically wrong, some plugins seem to be causing unload issues. This change will make anything using that one API that use to not remove from queue, no longer remove from queue. Hopefully other activities on the server will touch the chunk if it REALLY is in use.
2016-03-20 04:13:20 +00:00
diff --git a/src/main/java/net/minecraft/server/World.java b/src/main/java/net/minecraft/server/World.java
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
index a6c8e53..d8bd36c 100644
Improve chunk unload queue to maintain some previous expectations While the previous logic was logically correct, some CB API's before would request a chunk without removing it from the unload queue. While this is logically wrong, some plugins seem to be causing unload issues. This change will make anything using that one API that use to not remove from queue, no longer remove from queue. Hopefully other activities on the server will touch the chunk if it REALLY is in use.
2016-03-20 04:13:20 +00:00
--- a/src/main/java/net/minecraft/server/World.java
+++ b/src/main/java/net/minecraft/server/World.java
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
@@ -163,9 +163,15 @@ public abstract class World implements IBlockAccess {
// Paper end
Improve chunk unload queue to maintain some previous expectations While the previous logic was logically correct, some CB API's before would request a chunk without removing it from the unload queue. While this is logically wrong, some plugins seem to be causing unload issues. This change will make anything using that one API that use to not remove from queue, no longer remove from queue. Hopefully other activities on the server will touch the chunk if it REALLY is in use.
2016-03-20 04:13:20 +00:00
public Chunk getChunkIfLoaded(int x, int z) {
- return ((ChunkProviderServer) this.chunkProvider).getChunkIfLoaded(x, z);
+ return ((ChunkProviderServer) this.chunkProvider).getLoadedChunkAtWithoutMarkingActive(x, z); // Paper - This is added by CB, and will not mark as active. Simply an alias
}
+ // Paper start
+ public Chunk getChunkIfActive(int x, int z) {
+ return ((ChunkProviderServer) this.chunkProvider).getChunkIfActive(x, z);
+ }
+ // Paper end
+
protected World(IDataManager idatamanager, WorldData worlddata, WorldProvider worldprovider, MethodProfiler methodprofiler, boolean flag, ChunkGenerator gen, org.bukkit.World.Environment env) {
this.spigotConfig = new org.spigotmc.SpigotWorldConfig( worlddata.getName() ); // Spigot
this.paperConfig = new com.destroystokyo.paper.PaperWorldConfig(worlddata.getName(), this.spigotConfig); // Paper
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
diff --git a/src/main/java/org/bukkit/craftbukkit/CraftWorld.java b/src/main/java/org/bukkit/craftbukkit/CraftWorld.java
Clean up title and tab list patch Remember kids, the Ballmer peak is hard to hit https://xkcd.com/323/ Finally fixes #126 and #94 (I tested, it does, please, really, promise)
2016-03-23 06:05:09 +00:00
index c51c74b..caa5e62 100644
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
--- a/src/main/java/org/bukkit/craftbukkit/CraftWorld.java
+++ b/src/main/java/org/bukkit/craftbukkit/CraftWorld.java
Improve chunk unload queue to maintain some previous expectations While the previous logic was logically correct, some CB API's before would request a chunk without removing it from the unload queue. While this is logically wrong, some plugins seem to be causing unload issues. This change will make anything using that one API that use to not remove from queue, no longer remove from queue. Hopefully other activities on the server will touch the chunk if it REALLY is in use.
2016-03-20 04:13:20 +00:00
@@ -206,7 +206,7 @@ public class CraftWorld implements World {
}
// Paper start - Don't create a chunk just to unload it
- net.minecraft.server.Chunk chunk = world.getChunkProviderServer().getChunkIfLoaded(x, z);
+ net.minecraft.server.Chunk chunk = world.getChunkProviderServer().getChunkIfActive(x, z); // Paper
if (chunk == null) {
return false;
}
Fix chunk unload issues - Resolves #97
2016-03-19 15:29:46 +00:00
@@ -232,7 +232,7 @@ public class CraftWorld implements World {
continue;
}
- net.minecraft.server.Chunk neighbor = world.getChunkProviderServer().getChunkIfLoaded(chunk.locX + x, chunk.locZ + z);
+ net.minecraft.server.Chunk neighbor = world.getChunkProviderServer().getLoadedChunkAtWithoutMarkingActive(chunk.locX + x, chunk.locZ + z); // Paper
if (neighbor != null) {
neighbor.setNeighborUnloaded(-xx, -zz);
chunk.setNeighborUnloaded(xx, zz);
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
@@ -294,6 +294,7 @@ public class CraftWorld implements World {
// Use the default variant of loadChunk when generate == true.
return world.getChunkProviderServer().getChunkAt(x, z) != null;
}
+ if (true) { return world.getChunkProviderServer().getOrLoadChunkAt(x, z) != null; } // Paper
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
world.getChunkProviderServer().unloadQueue.remove(x, z);
net.minecraft.server.Chunk chunk = world.getChunkProviderServer().chunks.get(LongHash.toLong(x, z));
@@ -319,7 +320,7 @@ public class CraftWorld implements World {
Fix chunk unload issues - Resolves #97
2016-03-19 15:29:46 +00:00
continue;
}
- net.minecraft.server.Chunk neighbor = world.getChunkProviderServer().getChunkIfLoaded(chunk.locX + x, chunk.locZ + z);
+ net.minecraft.server.Chunk neighbor = world.getChunkProviderServer().getLoadedChunkAtWithoutMarkingActive(chunk.locX + x, chunk.locZ + z); // Paper
if (neighbor != null) {
neighbor.setNeighborLoaded(-x, -z);
chunk.setNeighborLoaded(x, z);
Clean up title and tab list patch Remember kids, the Ballmer peak is hard to hit https://xkcd.com/323/ Finally fixes #126 and #94 (I tested, it does, please, really, promise)
2016-03-23 06:05:09 +00:00
@@ -1543,7 +1544,7 @@ public class CraftWorld implements World {
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
}
// Already unloading?
- if (cps.unloadQueue.contains(chunk.locX, chunk.locZ)) {
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
+ if (chunk.isInUnloadQueue) { // Paper
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
continue;
}
Fix chunk unload issues - Resolves #97
2016-03-19 15:29:46 +00:00
diff --git a/src/main/java/org/bukkit/craftbukkit/chunkio/ChunkIOProvider.java b/src/main/java/org/bukkit/craftbukkit/chunkio/ChunkIOProvider.java
index 482af17..a1a6d5a 100644
--- a/src/main/java/org/bukkit/craftbukkit/chunkio/ChunkIOProvider.java
+++ b/src/main/java/org/bukkit/craftbukkit/chunkio/ChunkIOProvider.java
@@ -62,7 +62,7 @@ class ChunkIOProvider implements AsynchronousExecutor.CallBackProvider<QueuedChu
continue;
}
- Chunk neighbor = queuedChunk.provider.getChunkIfLoaded(chunk.locX + x, chunk.locZ + z);
+ Chunk neighbor = queuedChunk.provider.getLoadedChunkAtWithoutMarkingActive(chunk.locX + x, chunk.locZ + z); // Paper
if (neighbor != null) {
neighbor.setNeighborLoaded(-x, -z);
chunk.setNeighborLoaded(x, z);
Improve chunk unload queue to maintain some previous expectations While the previous logic was logically correct, some CB API's before would request a chunk without removing it from the unload queue. While this is logically wrong, some plugins seem to be causing unload issues. This change will make anything using that one API that use to not remove from queue, no longer remove from queue. Hopefully other activities on the server will touch the chunk if it REALLY is in use.
2016-03-20 04:13:20 +00:00
diff --git a/src/main/java/org/spigotmc/ActivationRange.java b/src/main/java/org/spigotmc/ActivationRange.java
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
index daed1db..e4af40c 100644
Improve chunk unload queue to maintain some previous expectations While the previous logic was logically correct, some CB API's before would request a chunk without removing it from the unload queue. While this is logically wrong, some plugins seem to be causing unload issues. This change will make anything using that one API that use to not remove from queue, no longer remove from queue. Hopefully other activities on the server will touch the chunk if it REALLY is in use.
2016-03-20 04:13:20 +00:00
--- a/src/main/java/org/spigotmc/ActivationRange.java
+++ b/src/main/java/org/spigotmc/ActivationRange.java
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
@@ -100,9 +100,9 @@ public class ActivationRange
{
for ( int j1 = k; j1 <= l; ++j1 )
{
- if ( world.getWorld().isChunkLoaded( i1, j1 ) )
- {
- activateChunkEntities( world.getChunkAt( i1, j1 ) );
+ Chunk chunk = world.getChunkIfActive(i1, j1); // Paper
+ if (chunk != null) { // Paper
+ activateChunkEntities( chunk ); // Paper
}
}
}
Improve chunk unload queue to maintain some previous expectations While the previous logic was logically correct, some CB API's before would request a chunk without removing it from the unload queue. While this is logically wrong, some plugins seem to be causing unload issues. This change will make anything using that one API that use to not remove from queue, no longer remove from queue. Hopefully other activities on the server will touch the chunk if it REALLY is in use.
2016-03-20 04:13:20 +00:00
@@ -250,7 +250,7 @@ public class ActivationRange
int x = MathHelper.floor( entity.locX );
int z = MathHelper.floor( entity.locZ );
// Make sure not on edge of unloaded chunk
- Chunk chunk = entity.world.getChunkIfLoaded( x >> 4, z >> 4 );
Another attempt at unload queue, including EAR improvements. should be fully working now as I pretty much fell back to existing methods so anything touching the unloadQueue set should behave correctly. And maintained NMS Reflection safe change too
2016-03-22 02:51:14 +00:00
+ Chunk chunk = isActive ? entity.world.getChunkIfActive( x >> 4, z >> 4 ) : null; // Paper
Improve chunk unload queue to maintain some previous expectations While the previous logic was logically correct, some CB API's before would request a chunk without removing it from the unload queue. While this is logically wrong, some plugins seem to be causing unload issues. This change will make anything using that one API that use to not remove from queue, no longer remove from queue. Hopefully other activities on the server will touch the chunk if it REALLY is in use.
2016-03-20 04:13:20 +00:00
if ( isActive && !( chunk != null && chunk.areNeighborsLoaded( 1 ) ) )
{
isActive = false;
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00
--
Clean up title and tab list patch Remember kids, the Ballmer peak is hard to hit https://xkcd.com/323/ Finally fixes #126 and #94 (I tested, it does, please, really, promise)
2016-03-23 06:05:09 +00:00
2.7.4.windows.1
Optimize Chunk Unload Queue Removing chunks from the unload queue when performing chunk lookups is a costly activity. It drastically slows down server performance as many methods call getChunkAt, resulting in a bandaid to skip removing chunks from the unload queue. This patch optimizes the unload queue to instead use a boolean on the Chunk object itself to mark if the chunk is active, and then insert into a LinkedList queue. The benefits here is that all chunk unload queue actions are now O(1) constant time. A LinkedList will never need to resize, and can be removed from in constant time when used in a queue like method. We mark the chunk as active in many places that notify it is still being used, so that when the chunk unload queue reaches that chunk, and sees the chunk became active again, it will skip it and move to next.
2016-03-18 22:03:44 +00:00