From 8b1f0ea23c92ad1ea91b9e7cbbc1fa2391ff8141 Mon Sep 17 00:00:00 2001
From: Jussi Saurio <jussi.saurio@gmail.com>
Date: Mon, 3 Feb 2025 10:28:51 +0200
Subject: [PATCH] Use vec for label resolution, not hashmap

---
 core/vdbe/builder.rs | 36 +++++++++++++++++++++++-------------
 core/vdbe/mod.rs     |  6 +++---
 2 files changed, 26 insertions(+), 16 deletions(-)

diff --git a/core/vdbe/builder.rs b/core/vdbe/builder.rs
index 1752e9dd3..a029a0161 100644
--- a/core/vdbe/builder.rs
+++ b/core/vdbe/builder.rs
@@ -15,7 +15,7 @@ use super::{BranchOffset, CursorID, Insn, InsnReference, Program};
 #[allow(dead_code)]
 pub struct ProgramBuilder {
     next_free_register: usize,
-    next_free_label: i32,
+    next_free_label: u32,
     next_free_cursor_id: usize,
     insns: Vec<Insn>,
     // for temporarily storing instructions that will be put after Transaction opcode
@@ -24,7 +24,7 @@ pub struct ProgramBuilder {
     // Cursors that are referenced by the program. Indexed by CursorID.
     pub cursor_ref: Vec<(Option<String>, CursorType)>,
     // Hashmap of label to insn reference. Resolved in build().
-    label_to_resolved_offset: HashMap<i32, u32>,
+    label_to_resolved_offset: Vec<Option<InsnReference>>,
     // Bitmask of cursors that have emitted a SeekRowid instruction.
     seekrowid_emitted_bitmask: u64,
     // map of instruction index to manual comment (used in EXPLAIN)
@@ -57,7 +57,7 @@ impl ProgramBuilder {
             next_insn_label: None,
             cursor_ref: Vec::new(),
             constant_insns: Vec::new(),
-            label_to_resolved_offset: HashMap::new(),
+            label_to_resolved_offset: Vec::with_capacity(4), // 4 is arbitrary, we guess to assign at least this much
             seekrowid_emitted_bitmask: 0,
             comments: HashMap::new(),
             parameters: Parameters::new(),
@@ -91,8 +91,10 @@ impl ProgramBuilder {
 
     pub fn emit_insn(&mut self, insn: Insn) {
         if let Some(label) = self.next_insn_label {
-            self.label_to_resolved_offset
-                .insert(label.to_label_value(), self.insns.len() as InsnReference);
+            self.label_to_resolved_offset.insert(
+                label.to_label_value() as usize,
+                Some(self.insns.len() as InsnReference),
+            );
             self.next_insn_label = None;
         }
         self.insns.push(insn);
@@ -183,8 +185,10 @@ impl ProgramBuilder {
     }
 
     pub fn allocate_label(&mut self) -> BranchOffset {
-        self.next_free_label -= 1;
-        BranchOffset::Label(self.next_free_label)
+        let label_n = self.next_free_label;
+        self.next_free_label += 1;
+        self.label_to_resolved_offset.push(None);
+        BranchOffset::Label(label_n)
     }
 
     // Effectively a GOTO <next insn> without the need to emit an explicit GOTO instruction.
@@ -197,8 +201,8 @@ impl ProgramBuilder {
     pub fn resolve_label(&mut self, label: BranchOffset, to_offset: BranchOffset) {
         assert!(matches!(label, BranchOffset::Label(_)));
         assert!(matches!(to_offset, BranchOffset::Offset(_)));
-        self.label_to_resolved_offset
-            .insert(label.to_label_value(), to_offset.to_offset_int());
+        self.label_to_resolved_offset[label.to_label_value() as usize] =
+            Some(to_offset.to_offset_int());
     }
 
     /// Resolve unresolved labels to a specific offset in the instruction list.
@@ -209,10 +213,16 @@ impl ProgramBuilder {
     pub fn resolve_labels(&mut self) {
         let resolve = |pc: &mut BranchOffset, insn_name: &str| {
             if let BranchOffset::Label(label) = pc {
-                let to_offset = *self.label_to_resolved_offset.get(label).unwrap_or_else(|| {
-                    panic!("Reference to undefined label in {}: {}", insn_name, label)
-                });
-                *pc = BranchOffset::Offset(to_offset);
+                let to_offset = self
+                    .label_to_resolved_offset
+                    .get(*label as usize)
+                    .unwrap_or_else(|| {
+                        panic!("Reference to undefined label in {}: {}", insn_name, label)
+                    });
+                *pc = BranchOffset::Offset(
+                    to_offset
+                        .unwrap_or_else(|| panic!("Unresolved label in {}: {}", insn_name, label)),
+                );
             }
         };
         for insn in self.insns.iter_mut() {
diff --git a/core/vdbe/mod.rs b/core/vdbe/mod.rs
index 16156347a..3f70dab45 100644
--- a/core/vdbe/mod.rs
+++ b/core/vdbe/mod.rs
@@ -74,7 +74,7 @@ pub enum BranchOffset {
     /// A label is a named location in the program.
     /// If there are references to it, it must always be resolved to an Offset
     /// via program.resolve_label().
-    Label(i32),
+    Label(u32),
     /// An offset is a direct index into the instruction list.
     Offset(InsnReference),
     /// A placeholder is a temporary value to satisfy the compiler.
@@ -103,7 +103,7 @@ impl BranchOffset {
     }
 
     /// Returns the label value. Panics if the branch offset is an offset or placeholder.
-    pub fn to_label_value(&self) -> i32 {
+    pub fn to_label_value(&self) -> u32 {
         match self {
             BranchOffset::Label(v) => *v,
             BranchOffset::Offset(_) => unreachable!("Offset cannot be converted to label value"),
@@ -116,7 +116,7 @@ impl BranchOffset {
     /// label or placeholder.
     pub fn to_debug_int(&self) -> i32 {
         match self {
-            BranchOffset::Label(v) => *v,
+            BranchOffset::Label(v) => *v as i32,
             BranchOffset::Offset(v) => *v as i32,
             BranchOffset::Placeholder => i32::MAX,
         }