Introduce a tier-1 JIT compiler based on x86-64 architecture

When the using frequency of a block exceeds a predetermined threshold,
the tier-1 JIT compiler traces the chained block and generate
corresponding low quailty machine code. The resulting target machine
code is stored in the code cache for future utilization.

The primary objective of introducing the tier-1 JIT compiler is to
enhance the execution speed of RISC-V instructions. This implementation
requires two additional components: a tier-1 machine code generator,
and code cache. Furthermore, this tier-1 JIT compiler serves as the
foundational target for future improvements.

In addition, we have developed a Python script that effectively traces
code templates and automatically generates JIT code templates. This
approach eliminates the need for manually writing duplicated code.

As shown in the performance analysis below, the tier-1 JIT compiler's
performance closely parallels that of QEMU in benchmarks with a
constrained dynamic instruction count. However, for benchmarks
featuring a substantial dynamic instruction count or lacking specific
hotspots—examples include pi and STRINGSORT—the tier-1 JIT compiler
demonstrates noticeably slower execution compared to QEMU.

Hence, a robust tier-2 JIT compiler is essential to generate optimized
machine code across diverse execution paths, coupled with a runtime
profiler for detecting hotspots.

* Perfromance
| Metric   | rv32emu-T1C | qemu  |
|----------+-------------+-------|
|aes	   |         0.02|  0.031|
|mandelbrot|	    0.029| 0.0115|
|puzzle	   |       0.0115|  0.009|
|pi        |       0.0413| 0.0177|
|dhrystone |	    0.331|  0.393|
|Nqeueens  |	    0.854|  0.749|
|qsort-O2  |	    2.384|   2.16|
|miniz-O2  |	     1.33|   1.01|
|primes-O2 |	     2.93|  1.069|
|sha512-O2 |	    2.057|  0.939|
|stream	   |       12.747|  10.36|
|STRINGSORT|       89.012| 11.496|

As demonstrated in the memory usage analysis below, the tier-1 JIT
compiler utilizes less memory than QEMU across all benchmarks.

* Memory usage
| Metric   | rv32emu-T1C |   qemu  |
|----------+-------------+---------|
|aes	   |      186,228|1,343,012|
|mandelbrot|	  152,203|  841,841|
|puzzle	   |      153,423|  890,225|
|pi        |      152,923|  879,957|
|dhrystone |	  154,466|  856,404|
|Nqeueens  |	  154,880|  858,618|
|qsort-O2  |	  155,091|  933,506|
|miniz-O2  |	  165,627|1,076,682|
|primes-O2 |	  150,540|  928,446|
|sha512-O2 |	  153,553|  978,177|
|stream	   |      165,911|  957,845|
|STRINGSORT|      167,871|1,104,702|

Related: sysprog21#238

.github/workflows/main.yml

@@ -32,7 +32,9 @@ jobs:
     - name: gdbstub test
       run: |
             make distclean ENABLE_GDBSTUB=1 gdbstub-test
-
+    - name: JIT test
+      run: |
+            make ENABLE_JIT=1 clean check
   coding-style:
     runs-on: ubuntu-22.04
     steps:

.gitignore

@@ -21,3 +21,4 @@ build/path/
 tests/**/*.elf
 tests/arch-test-target/config.ini
 __pycache__/
+src/rv32_jit_template.c

Makefile

@@ -118,6 +118,21 @@ gdbstub-test: $(BIN)
 	$(Q).ci/gdbstub-test.sh && $(call notice, [OK])
 endif
 
+ENABLE_JIT ?= 0
+$(call set-feature, JIT)
+ifeq ($(call has, JIT), 1)
+OBJS_EXT += jit_x64.o 
+ifneq ($(processor), x86_64)
+$(error JIT mode only supports for x64 target currently.)
+endif
+
+src/rv32_jit_template.c:
+	$(Q)tools/gen-jit-template.py $(CFLAGS) > $@
+
+$(OUT)/jit_x64.o: src/jit_x64.c src/rv32_jit_template.c
+	$(VECHO) "  CC\t$@\n"
+	$(Q)$(CC) -o $@ $(CFLAGS) -c -MMD -MF $@.d $<
+endif
 # For tail-call elimination, we need a specific set of build flags applied.
 # FIXME: On macOS + Apple Silicon, -fno-stack-protector might have a negative impact.
 $(OUT)/emulate.o: CFLAGS += -foptimize-sibling-calls -fomit-frame-pointer -fno-stack-check -fno-stack-protector
@@ -214,7 +229,7 @@ endif
 endif
 
 clean:
-	$(RM) $(BIN) $(OBJS) $(HIST_BIN) $(HIST_OBJS) $(deps) $(CACHE_OUT)
+	$(RM) $(BIN) $(OBJS) $(HIST_BIN) $(HIST_OBJS) $(deps) $(CACHE_OUT) src/rv32_jit_template.c
 distclean: clean
 	-$(RM) $(DOOM_DATA) $(QUAKE_DATA)
 	$(RM) -r $(OUT)/id1

mk/tools.mk

@@ -3,6 +3,7 @@ HIST_BIN := $(OUT)/rv_histogram
 # FIXME: riscv.o and map.o are dependencies of 'elf.o', not 'rv_histogram'.
 HIST_OBJS := \
 	riscv.o \
+	utils.o \
 	map.o \
 	elf.o \
 	decode.o \

src/cache.c

@@ -13,10 +13,13 @@
 #include "mpool.h"
 #include "utils.h"
 
-/* THRESHOLD is set to identify hot spots. Once the frequency of use for a block
- * exceeds the THRESHOLD, the JIT compiler flow is triggered.
+/* Currently, THRESHOLD is set to identify hot spots. Once the using frequency
+ * for a block exceeds the THRESHOLD, the tier-1 JIT compiler process is
+ * triggered.
+ * FIXME: Implement effective  profiler to detect hot spots, instead of simply
+ * relying on THRESHOLD.
  */
-#define THRESHOLD 1000
+#define THRESHOLD 4096
 
 static uint32_t cache_size, cache_size_bits;
 static struct mpool *cache_mp;
@@ -530,3 +533,62 @@ void cache_free(cache_t *cache)
     free(cache->map);
     free(cache);
 }
+
+#if !RV32_HAS(ARC)
+uint32_t cache_freq(struct cache *cache, uint32_t key)
+{
+    if (!cache->capacity ||
+        hlist_empty(&cache->map->ht_list_head[cache_hash(key)]))
+        return 0;
+    lfu_entry_t *entry = NULL;
+#ifdef __HAVE_TYPEOF
+    hlist_for_each_entry (entry, &cache->map->ht_list_head[cache_hash(key)],
+                          ht_list)
+#else
+    hlist_for_each_entry (entry, &cache->map->ht_list_head[cache_hash(key)],
+                          ht_list, lfu_entry_t)
+#endif
+    {
+        if (entry->key == key)
+            return entry->frequency;
+    }
+    return 0;
+}
+#endif
+
+#if RV32_HAS(JIT)
+bool cache_hot(struct cache *cache, uint32_t key)
+{
+    if (!cache->capacity ||
+        hlist_empty(&cache->map->ht_list_head[cache_hash(key)]))
+        return false;
+#if RV32_HAS(ARC)
+    arc_entry_t *entry = NULL;
+#ifdef __HAVE_TYPEOF
+    hlist_for_each_entry (entry, &cache->map->ht_list_head[cache_hash(key)],
+                          ht_list)
+#else
+    hlist_for_each_entry (entry, &cache->map->ht_list_head[cache_hash(key)],
+                          ht_list, arc_entry_t)
+#endif
+    {
+        if (entry->key == key && entry->frequency == THRESHOLD)
+            return true;
+    }
+#else
+    lfu_entry_t *entry = NULL;
+#ifdef __HAVE_TYPEOF
+    hlist_for_each_entry (entry, &cache->map->ht_list_head[cache_hash(key)],
+                          ht_list)
+#else
+    hlist_for_each_entry (entry, &cache->map->ht_list_head[cache_hash(key)],
+                          ht_list, lfu_entry_t)
+#endif
+    {
+        if (entry->key == key && entry->frequency == THRESHOLD)
+            return true;
+    }
+#endif
+    return false;
+}
+#endif

src/cache.h

@@ -5,6 +5,7 @@
 
 #pragma once
 
+#include <stdbool.h>
 #include <stdint.h>
 
 struct cache;
@@ -38,3 +39,17 @@ void *cache_put(struct cache *cache, uint32_t key, void *value);
  * @callback: a function for freeing cache entry completely
  */
 void cache_free(struct cache *cache);
+
+#if RV32_HAS(JIT)
+/**
+ * cache_hot - check whether the frequency of the cache entry exceeds the
+ * threshold or not
+ * @cache: a pointer points to target cache
+ * @key: the key of the specified entry
+ */
+bool cache_hot(struct cache *cache, uint32_t key);
+#endif
+
+#if !RV32_HAS(ARC)
+uint32_t cache_freq(struct cache *cache, uint32_t key);
+#endif

src/decode.h

@@ -179,13 +179,31 @@ enum op_field {
     )
 /* clang-format on */
 
+/* Macro operation fusion */
+
+/* macro operation fusion: convert specific RISC-V instruction patterns
+ * into faster and equivalent code
+ */
+#define FUSE_INSN_LIST \
+    _(fuse1)           \
+    _(fuse2)           \
+    _(fuse3)           \
+    _(fuse4)           \
+    _(fuse5)           \
+    _(fuse6)           \
+    _(fuse7)
+
 /* clang-format off */
 /* IR list */
 enum {
 #define _(inst, can_branch, insn_len, reg_mask) rv_insn_##inst,
     RV_INSN_LIST
 #undef _
-    N_RV_INSNS
+    N_RV_INSNS,
+#define _(inst) rv_insn_##inst,
+    FUSE_INSN_LIST
+#undef _
+    N_TOTAL_INSNS,
 };
 /* clang-format on */