Changes in / [fb75979:a65ccc4] in mainline

Files:

• common/include/str.h (modified) (2 diffs)
• common/str.c (modified) (3 diffs)
• kernel/generic/src/console/console.c (modified) (1 diff)

common/include/str.h

@@ -149 +149 @@
 extern char *str_tok(char *, const char *, char **);
 
+extern size_t str_sanitize(char *, size_t, uint8_t);
+
 extern errno_t str_uint8_t(const char *, const char **, unsigned int, bool,
     uint8_t *);
@@ -165 +167 @@
 extern void bin_order_suffix(const uint64_t, uint64_t *, const char **, bool);
 
-extern size_t str_sanitize(char *str, size_t n, uint8_t replacement);
-
 /*
  * TODO: Get rid of this.

common/str.c

@@ -547 +547 @@
 
         for (; n > 0 && b[0]; b++, n--) {
+                if (b[0] < ' ') {
+                        /* C0 control codes */
+                        b[0] = replacement;
+                        count++;
+                        continue;
+                }
+
                 int cont = _continuation_bytes(b[0]);
                 if (__builtin_expect(cont, 0) == 0)
@@ -584 +591 @@
                 }
 
+                bool c1_control = (b[0] == 0b11000010 && b[1] < 0b10100000);
+                if (cont == 1 && c1_control) {
+                        b[0] = replacement;
+                        count++;
+                        continue;
+                }
+
                 /* 0b1110!!!! 0b10!xxxxx 0b10xxxxxx */
                 if (cont == 2 && !(b[0] & 0b00001111) && !(b[1] & 0b00100000)) {
@@ -619 +633 @@
 }
 
+/** Replaces any byte that's not part of a complete valid UTF-8 character
+ * encoding with a replacement byte.
+ * Also replaces C0 and C1 control codes.
+ */
 size_t str_sanitize(char *str, size_t n, uint8_t replacement)
 {

kernel/generic/src/console/console.c

@@ -384 +384 @@
                 data[size] = 0;
 
+                uint8_t substitute = '\x1a';
+                str_sanitize(data, size, substitute);
+
                 switch (cmd) {
                 case KIO_WRITE: