From 1e46ee1009bf679717416f1543b2a95bc671cb49 Mon Sep 17 00:00:00 2001
From: Brooke Vibber <bvibber@pobox.com>
Date: Mon, 29 Dec 2025 13:30:26 -0800
Subject: [PATCH] more compact page tables, just need 8 bits per 4 KiB

---
 src/main.rs | 57 +++++++++++++++++++++++++++++++++--------------------
 1 file changed, 36 insertions(+), 21 deletions(-)

diff --git a/src/main.rs b/src/main.rs
index 07bea61..b58db88 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -1,45 +1,47 @@
 use std::collections::HashMap;
 use std::vec::Vec;
 
-const PAGE_SIZE: usize = 4096;
+const PAGE_BITS: usize = 12;
+const PAGE_SIZE: usize = 2 << PAGE_BITS;
 
 // Actively used bits in our limited implementation:
-const BIT_V: i64 = 1;
-const BIT_R: i64 = 2;
-const BIT_W: i64 = 4;
-const BIT_X: i64 = 8;
+const BIT_V: u8 = 1;
+const BIT_R: u8 = 2;
+const BIT_W: u8 = 4;
+const BIT_X: u8 = 8;
 
 // (Not used) - user mode accessible
-const BIT_U: i64 = 16;
+const BIT_U: u8 = 16;
 
 // (Not used) - global mapping
-const BIT_G: i64 = 32;
+const BIT_G: u8 = 32;
 
 // (Not presently used) accessed -- set if we've touched this since last time A cleared
-const BIT_A: i64 = 64;
+const BIT_A: u8 = 64;
 
 // (Will be used for JIT on writes to executable pags) dirty -- written since last time D cleared
-const BIT_D: i64 = 128;
+const BIT_D: u8 = 128;
 
-// (Not used)
+/*
+// (Not used) reserved for supervisor
 const BITS_RSW_LO: i64 = 256;
 const BITS_RSW_HI: i64 = 512;
 const BITS_RSW: i64 = BITS_RSW_LO | BITS_RSW_HI;
-
+*/
 
 type ExecutorFunc = fn(i64, &mut CoreState, &mut MachineState) -> i64;
 
 #[repr(C)]
 struct MachineState {
     memory: Vec<u8>,
-    pages: Vec<u32>,
+    pages: Vec<u8>,
 }
 
 /**
  * Note that physical memory accessors can traverse page boundaries;
  * we lay out linear memory from 0 to +4 gigabytes and will allocate
  * as many page table entries as are needed to cover RAM. These eat
- * up an extra 4 bytes per 4 kilobytes of address space used, initially
+ * up an extra 1 byte per 4 kilobytes of address space used, initially
  * allocating enough for all physical memory initially allocated.
  *
  * This will be relatively space-inefficient for sparse address spaces
@@ -50,11 +52,11 @@ struct MachineState {
  * page tables, even if running on different threads.
  */
 impl MachineState {
-    fn restore(memory: Vec<u8>, pages: Vec<u32>) -> Self {
-        if ((memory.len() >> 12) << 12) != memory.len() {
+    fn restore(memory: Vec<u8>, pages: Vec<u8>) -> Self {
+        if ((memory.len() >> PAGE_BITS) << PAGE_BITS) != memory.len() {
             panic!("memory size must be a multiple of 4096 bytes");
         }
-        if ((memory.len() >> 12)) != pages.len() {
+        if ((memory.len() >> PAGE_BITS)) != pages.len() {
             panic!("page data is wrong length for memory size");
         }
         return Self {
@@ -65,16 +67,26 @@ impl MachineState {
 
     fn new(memory_size: usize) -> Self {
         let memory = vec![0u8; memory_size];
-        let pages= vec![0u32; memory_size >> 12];
+        let pages = vec![0u8; memory_size >> PAGE_BITS];
         return Self::restore(memory, pages);
     }
 
-    fn get_page_info(&mut self, address: usize) -> u32 {
-        return self.pages[address >> 12];
+    fn get_page_table_entry(&mut self, address: usize) -> u32 {
+        let page = address >> PAGE_BITS;
+        if page < self.pages.len() {
+            return self.pages[page] as u32;
+        } else {
+            return 0;
+        }
     }
 
-    fn set_page_info(&mut self, address: usize, value: u32) {
-        self.pages[address >> 12] = value;
+    fn set_page_table_entry(&mut self, address: usize, bits: u32) {
+        let page = address >> PAGE_BITS;
+        if page < self.pages.len() {
+            self.pages[address >> 12] = bits as u8;
+        } else {
+            panic!("@fixme: handle attempts to expand address space");
+        }
     }
 
     fn lb_physical(&mut self, address: usize) -> i64 {
@@ -157,6 +169,9 @@ impl MachineState {
         self.memory[address + 7] = (value >> 56) as u8;
     }
 
+    fn store_buffer_physical(&mut self, address: usize, bytes: &[u8]) {
+        self.memory[address..address + bytes.len()].copy_from_slice(bytes);
+    }
 }
 
 #[repr(C)]