Bring up Linux kernel

The 8250 UART and PLIC peripherals, defined in src/devices/uart.[ch]
and src/devices/plic.[ch], have been integrated. Modifications to
src/system.c enable MMIO read and write operations for these
peripherals. The minimal device tree source file is also integrated in
src/devices/minimal.dts.

Default termios control flags are set to ICANON | ECHO | ISIG, allowing
all keyboard input to be captured by the guestOS.

The CLI parser has been updated to support custom Linux Kernel Images,
rootfs, and Device Tree Blobs.

The ELF_LOADER has been introduced to support test suites involving
single ELF files that use the rv32emu built-in syscall. It helps to
choose the right executable loader and requires differentiation from
guestOS syscalls.

The csrrw instruction is now branchable since it may overwrite the SATP
CSR during guestOS process scheduling.

last_pc and is_branch_taken are now updated only when not in a trapped
state, otherwise might lost the original state of them.

A system make target has been added to simplify running system
emulation.

Misc changes:
- Rename 'ARRAYS_SIZE' to 'ARRAY_SIZE' for better readability
- Move JIT related variables to proper conditional build flag

Author: ChinYikMing

Makefile

@@ -9,8 +9,16 @@ CONFIG_FILE := $(OUT)/.config
 
 CFLAGS = -std=gnu99 -O2 -Wall -Wextra
 CFLAGS += -Wno-unused-label
-CFLAGS += -include src/common.h
+CFLAGS += -include src/common.h -Isrc/
 
+# In the system test suite, the executable is an ELF file (e.g., MMU).
+# However, the Linux kernel emulation includes the Image, DT, and
+# root filesystem (rootfs). Therefore, the test suite needs this
+# flag to load the ELF and differentiate it from the kernel emulation.
+ENABLE_ELF_LOADER ?= 0
+$(call set-feature, ELF_LOADER)
+
+# Enable system emulation
 ENABLE_SYSTEM ?= 0
 $(call set-feature, SYSTEM)
 
@@ -45,10 +53,6 @@ CFLAGS += $(CFLAGS_NO_CET)
 
 OBJS_EXT :=
 
-ifeq ($(call has, SYSTEM), 1)
-OBJS_EXT += system.o
-endif
-
 # Integer Multiplication and Division instructions
 ENABLE_EXT_M ?= 1
 $(call set-feature, EXT_M)
@@ -71,7 +75,14 @@ src/softfloat/build/Linux-RISCV-GCC/Makefile:
 SOFTFLOAT_LIB := $(SOFTFLOAT_OUT)/softfloat.a
 $(SOFTFLOAT_LIB): src/softfloat/build/Linux-RISCV-GCC/Makefile
 	$(MAKE) -C $(dir $<) BUILD_DIR=$(SOFTFLOAT_OUT) CC=$(CC) AR=$(AR)
-$(OUT)/decode.o $(OUT)/riscv.o: $(SOFTFLOAT_LIB)
+ifeq ($(call has, SYSTEM), 1)
+DEV_OUT := $(OUT)/devices
+endif
+OBJS_NEED_SOFTFLOAT := $(OUT)/decode.o \
+                       $(OUT)/riscv.o \
+                       $(DEV_OUT)/uart.o \
+                       $(DEV_OUT)/plic.o
+$(OBJS_NEED_SOFTFLOAT): $(SOFTFLOAT_LIB)
 LDFLAGS += $(SOFTFLOAT_LIB)
 LDFLAGS += -lm
 endif
@@ -204,8 +215,9 @@ $(OUT)/emulate.o: CFLAGS += -foptimize-sibling-calls -fomit-frame-pointer -fno-s
 include mk/external.mk
 include mk/artifact.mk
 include mk/wasm.mk
+include mk/system.mk
 
-all: config $(BIN)
+all: config $(BUILD_DTB) $(BIN)
 
 OBJS := \
 	map.o \
@@ -222,7 +234,7 @@ OBJS := \
 	main.o
 
 OBJS := $(addprefix $(OUT)/, $(OBJS))
-deps := $(OBJS:%.o=%.o.d)
+deps += $(OBJS:%.o=%.o.d) # mk/system.mk includes prior this line, so declare deps at there
 
 ifeq ($(call has, EXT_F), 1)
 $(OBJS): $(SOFTFLOAT_LIB)
@@ -236,7 +248,7 @@ $(OUT)/%.o: src/%.c $(deps_emcc)
 	$(VECHO) "  CC\t$@\n"
 	$(Q)$(CC) -o $@ $(CFLAGS) $(CFLAGS_emcc) -c -MMD -MF $@.d $<
 
-$(BIN): $(OBJS)
+$(BIN): $(OBJS) $(DEV_OBJS)
 	$(VECHO) "  LD\t$@\n"
 	$(Q)$(CC) -o $@ $(CFLAGS_emcc) $^ $(LDFLAGS)
 
@@ -333,9 +345,10 @@ endif
 endif
 
 clean:
-	$(RM) $(BIN) $(OBJS) $(HIST_BIN) $(HIST_OBJS) $(deps) $(WEB_FILES) $(CACHE_OUT) src/rv32_jit.c
+	$(RM) $(BIN) $(OBJS) $(DEV_OBJS) $(BUILD_DTB) $(HIST_BIN) $(HIST_OBJS) $(deps) $(WEB_FILES) $(CACHE_OUT) src/rv32_jit.c
 distclean: clean
-	-$(RM) $(DOOM_DATA) $(QUAKE_DATA)
+	-$(RM) $(DOOM_DATA) $(QUAKE_DATA) $(BUILDROOT_DATA) $(LINUX_DATA)
+	$(RM) -r $(OUT)/linux-image
 	$(RM) -r $(TIMIDITY_DATA)
 	$(RM) -r $(OUT)/id1
 	$(RM) -r $(DEMO_DIR)

mk/artifact.mk

@@ -34,10 +34,14 @@ TEST_BENCHES += \
 SCIMARK2_URL := https://math.nist.gov/scimark2/scimark2_1c.zip
 SCIMARK2_SHA1 := de278c5b8cef84ab6dda41855052c7bfef919e36
 
-SHELL_HACK := $(shell mkdir -p $(BIN_DIR)/linux-x86-softfp $(BIN_DIR)/riscv32)
+SHELL_HACK := $(shell mkdir -p $(BIN_DIR)/linux-x86-softfp $(BIN_DIR)/riscv32 $(BIN_DIR)/linux-image)
 
 ifeq ($(call has, PREBUILT), 1)
-  LATEST_RELEASE := $(shell wget -q https://api.github.com/repos/sysprog21/rv32emu-prebuilt/releases/latest -O- | grep '"tag_name"' | sed -E 's/.*"tag_name": "([^"]+)".*/\1/')
+ifeq ($(call has, SYSTEM), 1)
+  LATEST_RELEASE := $(shell wget -q https://api.github.com/repos/sysprog21/rv32emu-prebuilt/releases -O- | grep '"tag_name"' | grep "Linux-Image" | head -n 1 | sed -E 's/.*"tag_name": "([^"]+)".*/\1/')
+else
+  LATEST_RELEASE := $(shell wget -q https://api.github.com/repos/sysprog21/rv32emu-prebuilt/releases -O- | grep '"tag_name"' | grep "ELF" | head -n 1 | sed -E 's/.*"tag_name": "([^"]+)".*/\1/')
+endif
 else
   # Since rv32emu only supports the dynamic binary translation of integer instruction in tiered compilation currently,
   # we disable the hardware floating-point and the related SIMD operation of x86.
@@ -53,24 +57,40 @@ endif
 artifact: fetch-checksum ieeelib scimark2
 ifeq ($(call has, PREBUILT), 1)
 	$(Q)$(PRINTF) "Checking SHA-1 of prebuilt binaries ... "
+	$(Q)$(eval RES := 0)
+
+ifeq ($(call has, SYSTEM), 1)
+	$(Q)$(eval PREBUILT_LINUX_IMAGE_FILENAME := $(shell cat $(BIN_DIR)/sha1sum-linux-image | awk '{  print $$2 };'))
 
+	$(Q)$(eval $(foreach FILE,$(PREBUILT_LINUX_IMAGE_FILENAME), \
+	    $(call verify,$(shell grep -w $(FILE) $(BIN_DIR)/sha1sum-linux-image | awk '{ print $$1 };'),$(BIN_DIR)/linux-image/$(FILE),RES) \
+	))
+
+	$(Q)$(eval RV32EMU_PREBUILT_TARBALL := rv32emu-linux-image-prebuilt.tar.gz)
+else
 	$(Q)$(eval PREBUILT_X86_FILENAME := $(shell cat $(BIN_DIR)/sha1sum-linux-x86-softfp | awk '{  print $$2 };'))
 	$(Q)$(eval PREBUILT_RV32_FILENAME := $(shell cat $(BIN_DIR)/sha1sum-riscv32 | awk '{ print $$2 };'))
 
-	$(Q)$(eval RES := 0)
 	$(Q)$(eval $(foreach FILE,$(PREBUILT_X86_FILENAME), \
 	    $(call verify,$(shell grep -w $(FILE) $(BIN_DIR)/sha1sum-linux-x86-softfp | awk '{ print $$1 };'),$(BIN_DIR)/linux-x86-softfp/$(FILE),RES) \
 	))
 	$(Q)$(eval $(foreach FILE,$(PREBUILT_RV32_FILENAME), \
 	    $(call verify,$(shell grep -w $(FILE) $(BIN_DIR)/sha1sum-riscv32 | awk '{ print $$1 };'),$(BIN_DIR)/riscv32/$(FILE),RES) \
 	))
 
+	$(Q)$(eval RV32EMU_PREBUILT_TARBALL := rv32emu-prebuilt.tar.gz)
+endif
+
 	$(Q)if [ "$(RES)" = "1" ]; then \
 	    $(PRINTF) "\n$(YELLOW)SHA-1 verification fails! Re-fetching prebuilt binaries from \"rv32emu-prebuilt\" ...\n$(NO_COLOR)"; \
-	    wget -q --show-progress https://github.com/sysprog21/rv32emu-prebuilt/releases/download/$(LATEST_RELEASE)/rv32emu-prebuilt.tar.gz -O- | tar -C build --strip-components=1 -xz; \
+	    wget -q --show-progress https://github.com/sysprog21/rv32emu-prebuilt/releases/download/$(LATEST_RELEASE)/$(RV32EMU_PREBUILT_TARBALL) -O- | tar -C build --strip-components=1 -xz; \
 	else \
 	    $(call notice, [OK]); \
 	fi
+else
+ifeq ($(call has, SYSTEM), 1)
+	$(Q)(cd $(BIN_DIR) && $(SHA1SUM) Image >> sha1sum-linux-image)
+	$(Q)(cd $(BIN_DIR) && $(SHA1SUM) rootfs.cpio >> sha1sum-linux-image)
 else
 	git submodule update --init $(addprefix ./tests/,$(foreach tb,$(TEST_SUITES),$(tb)))
 	$(Q)for tb in $(TEST_SUITES); do \
@@ -103,17 +123,24 @@ else
 	$(Q)(cd $(BIN_DIR)/linux-x86-softfp; for fd in *; do $(SHA1SUM) "$$fd"; done) >> $(BIN_DIR)/sha1sum-linux-x86-softfp
 	$(Q)(cd $(BIN_DIR)/riscv32; for fd in *; do $(SHA1SUM) "$$fd"; done) >> $(BIN_DIR)/sha1sum-riscv32
 endif
+endif
 
 fetch-checksum:
 ifeq ($(call has, PREBUILT), 1)
 	$(Q)$(PRINTF) "Fetching SHA-1 of prebuilt binaries ... "
+ifeq ($(call has, SYSTEM), 1)
+	$(Q)wget -q -O $(BIN_DIR)/sha1sum-linux-image https://github.com/sysprog21/rv32emu-prebuilt/releases/download/$(LATEST_RELEASE)/sha1sum-linux-image
+	$(Q)$(call notice, [OK])
+else
 	$(Q)wget -q -O $(BIN_DIR)/sha1sum-linux-x86-softfp https://github.com/sysprog21/rv32emu-prebuilt/releases/download/$(LATEST_RELEASE)/sha1sum-linux-x86-softfp
 	$(Q)wget -q -O $(BIN_DIR)/sha1sum-riscv32 https://github.com/sysprog21/rv32emu-prebuilt/releases/download/$(LATEST_RELEASE)/sha1sum-riscv32
 	$(Q)$(call notice, [OK])
 endif
+endif
 
 scimark2:
 ifeq ($(call has, PREBUILT), 0)
+ifeq ($(call has, SYSTEM), 0)
 	$(Q)$(call prologue,"scimark2")
 	$(Q)$(call download,$(SCIMARK2_URL))
 	$(Q)$(call verify,$(SCIMARK2_SHA1),$(notdir $(SCIMARK2_URL)))
@@ -126,6 +153,7 @@ ifeq ($(call has, PREBUILT), 0)
 	$(Q)$(MAKE) -C ./tests/scimark2 CC=$(CROSS_COMPILE)gcc CFLAGS="-march=rv32imf -mabi=ilp32 -O2"
 	$(Q)cp ./tests/scimark2/scimark2 $(BIN_DIR)/riscv32/scimark2
 endif
+endif
 
 ieeelib:
 ifeq ($(call has, PREBUILT), 0)

mk/system.mk

@@ -0,0 +1,28 @@
+# Peripherals for system emulation
+ifeq ($(call has, SYSTEM), 1)
+
+DEV_SRC := src/devices
+
+DTC ?= dtc
+$(OUT)/minimal.dtb: $(DEV_SRC)/minimal.dts
+	$(VECHO) " DTC\t$@\n"
+	$(Q)$(DTC) $^ -o $@
+BUILD_DTB := $(OUT)/minimal.dtb
+
+$(DEV_OUT)/%.o: $(DEV_SRC)/%.c $(deps_emcc)
+	$(Q)mkdir -p $(DEV_OUT)
+	$(VECHO) "  CC\t$@\n"
+	$(Q)$(CC) -o $@ $(CFLAGS) $(CFLAGS_emcc) -c -MMD -MF $@.d $<
+DEV_OBJS := $(patsubst $(DEV_SRC)/%.c, $(DEV_OUT)/%.o, $(wildcard $(DEV_SRC)/*.c))
+deps := $(DEV_OBJS:%.o=%.o.d)
+
+OBJS_EXT += system.o
+
+# system target execution by using default dependencies
+LINUX_IMAGE_DIR := linux-image
+system_action := ($(BIN) -k $(OUT)/$(LINUX_IMAGE_DIR)/Image -i $(OUT)/$(LINUX_IMAGE_DIR)/rootfs.cpio -b $(OUT)/minimal.dtb)
+system_deps += artifact $(BUILD_DTB) $(BIN)
+system: $(system_deps)
+	$(system_action)
+
+endif

src/common.h

@@ -5,6 +5,9 @@
 
 #pragma once
 
+#include <assert.h>
+#include <stdint.h>
+
 #include "feature.h"
 
 #if defined(__GNUC__) || defined(__clang__)
@@ -23,10 +26,64 @@
 #endif
 #endif
 
-#define ARRAYS_SIZE(arr) (sizeof(arr) / sizeof(arr[0]))
+#define ARRAY_SIZE(arr) (sizeof(arr) / sizeof(arr[0]))
 
 #define MASK(n) (~((~0U << (n))))
 
+#if defined(_MSC_VER)
+#include <intrin.h>
+static inline int rv_clz(uint32_t v)
+{
+    /* 0 is considered as undefined behavior */
+    assert(v);
+
+    uint32_t leading_zero = 0;
+    _BitScanReverse(&leading_zero, v);
+    return 31 - leading_zero;
+}
+#elif defined(__GNUC__) || defined(__clang__)
+static inline int rv_clz(uint32_t v)
+{
+    /* https://gcc.gnu.org/onlinedocs/gcc/Other-Builtins.html */
+    /* 0 is considered as undefined behavior */
+    assert(v);
+
+    return __builtin_clz(v);
+}
+#else /* generic implementation */
+static inline int rv_clz(uint32_t v)
+{
+    /* 0 is considered as undefined behavior */
+    assert(v);
+
+    /* http://graphics.stanford.edu/~seander/bithacks.html#IntegerLogDeBruijn */
+    static const uint8_t mul_debruijn[] = {
+        0, 9,  1,  10, 13, 21, 2,  29, 11, 14, 16, 18, 22, 25, 3, 30,
+        8, 12, 20, 28, 15, 17, 24, 7,  19, 27, 23, 6,  26, 5,  4, 31,
+    };
+
+    v |= v >> 1;
+    v |= v >> 2;
+    v |= v >> 4;
+    v |= v >> 8;
+    v |= v >> 16;
+
+    return mul_debruijn[(uint32_t) (v * 0x07C4ACDDU) >> 27];
+}
+#endif
+
+/*
+ * Integer log base 2
+ *
+ * The input x must not be zero.
+ * Otherwise, the result is undefined on some platform.
+ *
+ */
+static inline uint8_t ilog2(uint32_t x)
+{
+    return 31 - rv_clz(x);
+}
+
 /* Alignment macro */
 #if defined(__GNUC__) || defined(__clang__)
 #define __ALIGNED(x) __attribute__((aligned(x)))

src/decode.h

@@ -90,7 +90,7 @@ enum op_field {
     )                                                  \
     /* RV32 Zicsr Standard Extension */                \
     IIF(RV32_HAS(Zicsr))(                              \
-        _(csrrw, 0, 4, 0, ENC(rs1, rd))                \
+        _(csrrw, 1, 4, 0, ENC(rs1, rd))                \
         _(csrrs, 0, 4, 0, ENC(rs1, rd))                \
         _(csrrc, 0, 4, 0, ENC(rs1, rd))                \
         _(csrrwi, 0, 4, 0, ENC(rs1, rd))               \

src/devices/minimal.dts

@@ -0,0 +1,69 @@
+/dts-v1/;
+
+/ {
+    #address-cells = <1>;
+    #size-cells = <1>;
+    model = "rv32emu";
+
+    aliases {
+        serial0 = "/soc@F0000000/serial@4000000";
+    };
+
+    chosen {
+        bootargs = "earlycon console=ttyS0";
+        stdout-path = "serial0";
+        linux,initrd-start = <0x1f700000>; /* @403 MiB (503 * 1024 * 1024)     */
+        linux,initrd-end = <0x1fefffff>;   /* @511 MiB (511 * 1024 * 1024 - 1) */
+    };
+
+    cpus {
+        #address-cells = <1>;
+        #size-cells = <0>;
+        timebase-frequency = <65000000>;
+        cpu0: cpu@0 {
+            device_type = "cpu";
+            compatible = "riscv";
+            reg = <0>;
+            riscv,isa = "rv32ima";
+            mmu-type = "riscv,sv32";
+            cpu0_intc: interrupt-controller {
+                #interrupt-cells = <1>;
+                #address-cells = <0>;
+                interrupt-controller;
+                compatible = "riscv,cpu-intc";
+            };
+        };
+    };
+
+    sram: memory@0 {
+        device_type = "memory";
+        reg = <0x00000000 0x20000000>;
+        reg-names = "sram0";
+    };
+
+    soc: soc@F0000000 {
+        #address-cells = <1>;
+        #size-cells = <1>;
+        compatible = "simple-bus";
+        ranges = <0x0 0xF0000000 0x10000000>;
+        interrupt-parent = <&plic0>;
+
+        plic0: interrupt-controller@0 {
+            #interrupt-cells = <1>;
+            #address-cells = <0>;
+            compatible = "sifive,plic-1.0.0";
+            reg = <0x0000000 0x4000000>;
+            interrupt-controller;
+            interrupts-extended = <&cpu0_intc 9>;
+            riscv,ndev = <31>;
+        };
+
+        serial@4000000 {
+            compatible = "ns16550";
+            reg = <0x4000000 0x100000>;
+            interrupts = <1>;
+            no-loopback-test;
+            clock-frequency = <5000000>; /* the baudrate divisor is ignored */
+        };
+    };
+};