feat: Add snprintf and refine printf family funcs

This patch adds snprintf and refines printf family funcs by changing
function signature to return an integer to match their original
definitions defined by specifications.

The second parameter "size" for snprintf is expected to
accept only zero or positive integer, otherwise as C11 specification
states, it would be undefined behavior.

Co-authored-by: Jim Hsu <[email protected]>

Author: Jim Hsu

lib/c.c

@@ -288,60 +288,63 @@ int __format(char *buffer,
     return bi;
 }
 
-void printf(char *str, ...)
+int __format_to_buf(char *buffer, char *format, int *var_args, int size)
 {
-    int *var_args = &str + 4;
-    char buffer[200];
     int si = 0, bi = 0, pi = 0;
 
-    while (str[si]) {
-        if (str[si] != '%') {
-            buffer[bi] = str[si];
+    if (size == 0)
+        return 0;
+
+    while (format[si]) {
+        if (bi >= size - 1) {
+            break;
+        } else if (format[si] != '%') {
+            buffer[bi] = format[si];
             bi++;
             si++;
         } else {
             int w = 0, zp = 0, pp = 0;
 
             si++;
-            if (str[si] == '#') {
+            if (format[si] == '#') {
                 pp = 1;
                 si++;
             }
-            if (str[si] == '0') {
+            if (format[si] == '0') {
                 zp = 1;
                 si++;
             }
-            if (str[si] >= '1' && str[si] <= '9') {
-                w = str[si] - '0';
+            if (format[si] >= '1' && format[si] <= '9') {
+                w = format[si] - '0';
                 si++;
-                while (str[si] >= '0' && str[si] <= '9') {
+                while (format[si] >= '0' && format[si] <= '9') {
                     w *= 10;
-                    w += str[si] - '0';
+                    w += format[si] - '0';
                     si++;
                 }
             }
-            if (str[si] == 's') {
+            if (format[si] == 's') {
                 /* append param pi as string */
                 int l = strlen(var_args[pi]);
                 strcpy(buffer + bi, var_args[pi]);
                 bi += l;
-            } else if (str[si] == 'c') {
+            } else if (format[si] == 'c') {
                 /* append param pi as char */
                 buffer[bi] = var_args[pi];
                 bi += 1;
-            } else if (str[si] == 'o') {
+            } else if (format[si] == 'o') {
                 /* append param as octal */
                 int v = var_args[pi];
                 bi += __format(buffer + bi, v, w, zp, 8, pp);
-            } else if (str[si] == 'd') {
+            } else if (format[si] == 'd') {
                 /* append param as decimal */
                 int v = var_args[pi];
                 bi += __format(buffer + bi, v, w, zp, 10, 0);
-            } else if (str[si] == 'x') {
+            } else if (format[si] == 'x') {
                 /* append param as hex */
                 int v = var_args[pi];
                 bi += __format(buffer + bi, v, w, zp, 16, pp);
-            } else if (str[si] == '%') {
+            } else if (format[si] == '%') {
                 /* append literal '%' character */
                 buffer[bi] = '%';
                 bi++;
@@ -352,71 +355,27 @@ void printf(char *str, ...)
             si++;
         }
     }
-    buffer[bi] = 0;
-    __syscall(__syscall_write, 1, buffer, bi);
+
+    int len = size - 1 > bi ? bi : size - 1;
+    buffer[len] = 0;
+    return len;
 }
 
-void sprintf(char *buffer, char *str, ...)
+int printf(char *str, ...)
 {
-    int *var_args = &str + 4;
-    int si = 0, bi = 0, pi = 0;
+    char buffer[200];
+    int len = __format_to_buf(buffer, str, &str + 4, 0x7fffffff);
+    return __syscall(__syscall_write, 1, buffer, len);
+}
 
-    while (str[si]) {
-        if (str[si] != '%') {
-            buffer[bi] = str[si];
-            bi++;
-            si++;
-        } else {
-            int w = 0, zp = 0, pp = 0;
+int sprintf(char *buffer, char *str, ...)
+{
+    return __format_to_buf(buffer, str, &str + 4, 0x7fffffff);
+}
 
-            si++;
-            if (str[si] == '#') {
-                pp = 1;
-                si++;
-            }
-            if (str[si] == '0') {
-                zp = 1;
-                si++;
-            }
-            if (str[si] >= '1' && str[si] <= '9') {
-                w = str[si] - '0';
-                si++;
-                if (str[si] >= '0' && str[si] <= '9') {
-                    w *= 10;
-                    w += str[si] - '0';
-                    si++;
-                }
-            }
-            switch (str[si]) {
-            case 37: /* % */
-                buffer[bi++] = '%';
-                si++;
-                continue;
-            case 99: /* c */
-                buffer[bi++] = var_args[pi];
-                break;
-            case 115: /* s */
-                strcpy(buffer + bi, var_args[pi]);
-                bi += strlen(var_args[pi]);
-                break;
-            case 111: /* o */
-                bi += __format(buffer + bi, var_args[pi], w, zp, 8, pp);
-                break;
-            case 100: /* d */
-                bi += __format(buffer + bi, var_args[pi], w, zp, 10, 0);
-                break;
-            case 120: /* x */
-                bi += __format(buffer + bi, var_args[pi], w, zp, 16, pp);
-                break;
-            default:
-                abort();
-                break;
-            }
-            pi++;
-            si++;
-        }
-    }
-    buffer[bi] = 0;
+int snprintf(char *buffer, int n, char *str, ...)
+{
+    return __format_to_buf(buffer, str, &str + 4, n);
 }
 
 int __free_all();

src/defs.h

@@ -17,17 +17,17 @@
 #define MAX_VAR_LEN 32
 #define MAX_TYPE_LEN 32
 #define MAX_PARAMS 8
-#define MAX_LOCALS 1500
+#define MAX_LOCALS 1600
 #define MAX_FIELDS 64
 #define MAX_FUNCS 512
 #define MAX_TYPES 64
-#define MAX_IR_INSTR 50000
+#define MAX_IR_INSTR 60000
 #define MAX_BB_PRED 128
 #define MAX_BB_DOM_SUCC 64
 #define MAX_BB_RDOM_SUCC 256
 #define MAX_GLOBAL_IR 256
 #define MAX_LABEL 4096
-#define MAX_SOURCE 327680
+#define MAX_SOURCE 524288
 #define MAX_CODE 262144
 #define MAX_DATA 262144
 #define MAX_SYMTAB 65536
@@ -39,7 +39,7 @@
 #define MAX_CASES 128
 #define MAX_NESTING 128
 #define MAX_OPERAND_STACK_SIZE 32
-#define MAX_ANALYSIS_STACK_SIZE 750
+#define MAX_ANALYSIS_STACK_SIZE 800
 
 /* Default capacities for common data structures */
 /* Default arena size is initialized with 256 KiB */

src/parser.c

@@ -3401,17 +3401,15 @@ void load_source_file(char *file)
         }
         if (!strncmp(buffer, "#include ", 9) && (buffer[9] == '"')) {
             char path[MAX_LINE_LEN];
-            int c = strlen(file) - 1;
+            int c = strlen(file) - 1, inclusion_path_len = strlen(buffer) - 11;
             while (c > 0 && file[c] != '/')
                 c--;
             if (c) {
                 /* prepend directory name */
-                strncpy(path, file, c + 1);
-                c++;
+                snprintf(path, c + 2, "%s", file);
             }
-            path[c] = 0;
-            buffer[strlen(buffer) - 2] = 0;
-            strcpy(path + c, buffer + 10);
+
+            snprintf(path + c + 1, inclusion_path_len, "%s", buffer + 10);
             load_source_file(path);
         } else {
             strcpy(SOURCE + source_idx, buffer);

tests/driver.sh

@@ -1611,4 +1611,66 @@ int main()
 }
 EOF
 
+# printf family, including truncation and zero size input
+try_output 11 "Hello World" << EOF
+int main() {
+    int written = printf("Hello World");
+    return written;
+}
+EOF
+
+try_output 247 "" << EOF
+int main()
+{
+    __syscall(__syscall_close, 1);
+    int written = printf("Hello\n");
+    return written;
+}
+EOF
+
+try_output 11 "Hello World" << EOF
+int main() {
+    char buffer[50];
+    int written = sprintf(buffer, "Hello World");
+    printf("%s", buffer);
+    return written;
+}
+EOF
+
+try_output 16 "Hello World 1123" << EOF
+int main() {
+    char buffer[50];
+    int written = sprintf(buffer, "Hello %s %d", "World", 1123);
+    printf("%s", buffer);
+    return written;
+}
+EOF
+
+try_output 16 "Hello World 1123" << EOF
+int main() {
+    char buffer[50];
+    int written = snprintf(buffer, 50, "Hello %s %d", "World", 1123);
+    printf("%s", buffer);
+    return written;
+}
+EOF
+
+try_output 0 "" << EOF
+int main() {
+    char buffer[20];
+    int written = snprintf(buffer, 0, "Number: %d", -37);
+    printf("%s", buffer);
+    return written;
+}
+EOF
+
+try_output 9 "Number: -" << EOF
+int main() {
+    char buffer[10];
+    int written = snprintf(buffer, 10, "Number: %d", -37);
+    printf("%s", buffer);
+    return written;
+}
+EOF
+
 echo OK