add duplicate function merging to nucode backend

totalspectrum · Jan 7, 2024 · 5f2b1a4 · 5f2b1a4
1 parent e7e24c2
commit 5f2b1a4
Show file tree

Hide file tree

Showing 6 changed files with 99 additions and 16 deletions.
diff --git a/backends/asm/optimize_ir.c b/backends/asm/optimize_ir.c
@@ -5809,14 +5809,6 @@ HashIR(SHA256_CTX *ctx, IR *ir, Function *f)
 
 static FunctionList *funchash[256];
 
-static FunctionList *
-NewFunctionList(Function *f)
-{
-    FunctionList *entry = (FunctionList *)calloc(1, sizeof(FunctionList));
-    entry->func = f;
-    return entry;
-}
-
 void
 HashFuncIRL(Function *f)
 {

diff --git a/backends/asm/outasm.h b/backends/asm/outasm.h
@@ -8,6 +8,7 @@
 #define OUTASM_H
 
 #include "instr.h"
+#include "becommon.h"
 #include "util/sha256.h"
 
 // functions for producing local identifiers

diff --git a/backends/becommon.c b/backends/becommon.c
@@ -1,7 +1,7 @@
 /*
  * Spin to C/C++ translator
- * Copyright 2011-2023 Total Spectrum Software Inc.
- * Copyright 2021-2022 Ada Gottensträter
+ * Copyright 2011-2024 Total Spectrum Software Inc.
+ * Copyright 2021-2024 Ada Gottensträter
  * +--------------------------------------------------------------------
  * ¦  TERMS OF USE: MIT License
  * +--------------------------------------------------------------------
@@ -690,3 +690,12 @@ int DecomposeBits(unsigned val, int *shifts)
     }
     return r;
 }
+
+FunctionList *
+NewFunctionList(Function *f)
+{
+    FunctionList *entry = (FunctionList *)calloc(1, sizeof(FunctionList));
+    entry->func = f;
+    return entry;
+}
+
diff --git a/backends/becommon.h b/backends/becommon.h
@@ -178,4 +178,7 @@ static inline bool isPowerOf2(uint32_t x)
 // shifts[2] to initial shift
 int DecomposeBits(unsigned val, int *shifts);
 
+// allocate a FunctionList entry
+FunctionList *NewFunctionList(Function *f);
+
 #endif /* BACKEND_COMMON_H */
diff --git a/backends/nucode/outnu.c b/backends/nucode/outnu.c
@@ -2353,7 +2353,7 @@ static int NuConvertFunctions(void *vptr, Module *P) {
 }
 
 static int NuOptimizeFunction(Function *pf, NuIrList *irl) {
-    // for now, do nothing
+    // for now, do very little
     int change;
     int all_changes = 0;
     unsigned flags = pf->optimize_flags;
@@ -2383,15 +2383,25 @@ static int NuRevisitFunctions(void *vptr, Module *P) {
     Function *pf;
     NuIrList **global_ptr = (NuIrList **)vptr;
     NuIrList *globalList = *global_ptr;
+    bool isDup;
+
     current = P;
 
     for (pf = P->functions; pf; pf = pf->next) {
         // optimize function
         irl = &FunData(pf)->irl;
         NuOptimizeFunction(pf, irl);
-        // thread into global list
-        irl->nextList = globalList;
-        globalList = irl;
+        if (pf->optimize_flags & OPT_MERGE_DUPLICATES) {
+            // see if it matches any other function
+            isDup = NuHashFunc(pf);
+        } else {
+            isDup = false;
+        }
+        if (!isDup) {
+            // thread into global list
+            irl->nextList = globalList;
+            globalList = irl;
+        }
     }
 
     current = save;
@@ -2426,6 +2436,13 @@ static int NuCompileObject(void *vptr, Module *P) {
         flexbuf_delete(&datBuf);
     }
     for (pf = P->functions; pf; pf = pf->next) {
+        FunctionList *fl;
+        if (FunData(pf)->isRemoved) continue;
+        for (fl = FunData(pf)->funcdups; fl; fl = fl->next) {
+            Function *f2 = fl->func;
+            flexbuf_printf(fb, "'--- Function: %s\n", f2->name);
+            NuOutputLabelNL(fb, FunData(f2)->entryLabel);
+        }
         flexbuf_printf(fb, "'--- Function: %s\n", pf->name);
         NuOutputIrList(fb, &FunData(pf)->irl);
         if (FunData(pf)->dataLabel) {
@@ -2436,6 +2453,64 @@ static int NuCompileObject(void *vptr, Module *P) {
     return 1;
 }
 
+//
+// hashing function
+// returns true if function is found to be a duplicate of another one,
+// otherwise false
+//
+static FunctionList *buckets[256];
+
+bool NuHashFunc(Function *f) {
+    static int unique_func_count = 0;
+    SHA256_CTX ctx;
+    FunctionList *fl;
+
+    if (!(f->optimize_flags & OPT_MERGE_DUPLICATES)) {
+        ++unique_func_count;
+        memcpy(FunData(f)->hash, &unique_func_count, sizeof(unique_func_count));
+        return false;
+    }
+    NuIrList *irl = &FunData(f)->irl;
+    NuIr *ir;
+
+    sha256_init(&ctx);
+    for (ir = irl->head; ir; ir = ir->next) {
+        sha256_update(&ctx, (unsigned char *)&ir->op, sizeof(ir->op));
+    }
+    sha256_final(&ctx, FunData(f)->hash);
+
+    fl = buckets[FunData(f)->hash[0]];
+    if (!fl) {
+        buckets[FunData(f)->hash[0]] = NewFunctionList(f);
+        return false;
+    }
+    for(;;) {
+        Function *f2 = fl->func;
+        if (FunData(f2)) {
+            if (memcmp(FunData(f2)->hash, FunData(f)->hash, SHA256_BLOCK_SIZE) == 0) {
+                FunData(f)->isRemoved = true;
+                // make sure we keep f2
+                f2->used_as_ptr |= f->used_as_ptr;
+                f2->cog_task |= f->cog_task;
+                f2->callSites += f->callSites;
+
+                FunctionList *entry = NewFunctionList(f);
+                entry->next = FunData(f2)->funcdups;
+                FunData(f2)->funcdups = entry;
+                FunData(f)->isRemoved = true;
+                return true;
+            }
+        }
+        if (!fl->next) {
+            fl->next = NewFunctionList(f);
+            break;
+        }
+        fl = fl->next;
+    }
+
+    return false;
+}
+
 #ifdef NEVER
 static void NuAddHeap(ByteOutputBuffer *bob, Module*P) {
     Symbol *sym = LookupSymbolInTable(&systemModule->objsyms, "__heap_base");
@@ -2548,3 +2623,4 @@ void OutputNuCode(const char *asmFileName, Module *P)
     current = saveCurrent;
     curfunc = saveFunc;
 }
+
diff --git a/backends/nucode/outnu.h b/backends/nucode/outnu.h
@@ -19,7 +19,9 @@ typedef struct {
     NuIrList irl;
     NuIrLabel *entryLabel;
     NuIrLabel *dataLabel;
-    Flexbuf dataBuf;  // inline data for the function
+    Flexbuf dataBuf;         // inline data for the function
+    FunctionList *funcdups;  // functions that are aliases to this one
+    bool isRemoved;          // flag, if set function is a duplicate of something else
     unsigned char hash[SHA256_BLOCK_SIZE];
 } NuFunData;
 
@@ -35,7 +37,7 @@ NuIrOpcode NuInvertCondition(NuIrOpcode op);
 NuIrOpcode NuReverseDirCondition(NuIrOpcode op);
 void NuDeleteIr(NuIrList *irl, NuIr *ir);
 void NuIrInsertBefore(NuIrList *irl, NuIr *anchor, NuIr *newitem);
-void NuFuncHash(Function *f);
+bool NuHashFunc(Function *f);
 
 /* ultimate output function */
 void OutputNuCode(const char *fname, Module *P);