monitor: support init mem with a gz file (OpenXiangShan#7)

This commit adds support for initializing memory with a compressed
gzip file. When CONFIG_MEM_COMPRESS is enabled, files with names
ending with ".gz" will be seen as compressed gzip files and be
decompressed before loading into memory.

Only non-zero bytes are copied into pmem, to avoid using too much
memory (by copy-on-write).

Makefile

@@ -47,6 +47,10 @@ LDFLAGS += $(CFLAGS_BUILD)
 
 NAME  = nemu-$(ENGINE)
 
+ifdef CONFIG_MEM_COMPRESS
+LDFLAGS += -lz
+endif
+
 ifndef CONFIG_SHARE
 LDFLAGS += -lreadline -ldl -pie
 else

src/memory/Kconfig

@@ -25,4 +25,11 @@ config MEM_RANDOM
   help
     This may help to find undefined behaviors.
 
+config MEM_COMPRESS
+  depends on MODE_SYSTEM && !DIFFTEST
+  bool "Initialize the memory with a compressed gz file"
+  default n
+  help
+    Must have zlib installed.
+
 endmenu #MEMORY

src/monitor/monitor.c

@@ -31,12 +31,68 @@ static inline void welcome() {
 }
 
 #ifndef CONFIG_MODE_USER
+
+#ifdef CONFIG_MEM_COMPRESS
+#include <zlib.h>
+
+static long load_gz_img(const char *filename) {
+  gzFile compressed_mem = gzopen(filename, "rb");
+  Assert(compressed_mem, "Can not open '%s'", filename);
+
+  const uint32_t chunk_size = 16384;
+  uint8_t *temp_page = (uint8_t *)calloc(chunk_size, sizeof(long));
+  uint8_t *pmem_start = (uint8_t *)guest_to_host(RESET_VECTOR);
+  uint8_t *pmem_current;
+
+  // load file byte by byte to pmem
+  uint64_t curr_size = 0;
+  while (curr_size < CONFIG_MSIZE) {
+    uint32_t bytes_read = gzread(compressed_mem, temp_page, chunk_size);
+    if (bytes_read == 0) {
+      break;
+    }
+    for (uint32_t x = 0; x < bytes_read; x++) {
+      pmem_current = pmem_start + curr_size + x;
+      uint8_t read_data = *(temp_page + x);
+      if (read_data != 0 || *pmem_current != 0) {
+        *pmem_current = read_data;
+      }
+    }
+    curr_size += bytes_read;
+  }
+
+  // check again to ensure the bin has been fully loaded
+  uint32_t left_bytes = gzread(compressed_mem, temp_page, chunk_size);
+  Assert(left_bytes == 0, "File size is larger than buf_size!\n");
+
+  free(temp_page);
+  Assert(!gzclose(compressed_mem), "Error closing '%s'\n", filename);
+  return curr_size;
+}
+
+// Return whether a file is a gz file, determined by its name.
+// If the filename ends with ".gz", we treat it as a gz file.
+bool is_gz_file(const char *filename) {
+  if (filename == NULL || strlen(filename) < 3) {
+    return false;
+  }
+  return !strcmp(filename + (strlen(filename) - 3), ".gz");
+}
+#endif // CONFIG_MEM_COMPRESS
+
 static inline long load_img() {
   if (img_file == NULL) {
     Log("No image is given. Use the default build-in image.");
     return 4096; // built-in image size
   }
 
+#ifdef CONFIG_MEM_COMPRESS
+  if (is_gz_file(img_file)) {
+    Log("The image is %s", img_file);
+    return load_gz_img(img_file);
+  }
+#endif
+
   FILE *fp = fopen(img_file, "rb");
   Assert(fp, "Can not open '%s'", img_file);
 
@@ -52,7 +108,7 @@ static inline long load_img() {
   fclose(fp);
   return size;
 }
-#endif
+#endif // CONFIG_MODE_USER
 
 static inline int parse_args(int argc, char *argv[]) {
   const struct option table[] = {