Page MenuHomeFreeBSD
Files F102546297

D42147.diff
No OneTemporary
Actions

D42147.diff
View Options

diff --git a/sys/kern/tty_ttydisc.c b/sys/kern/tty_ttydisc.c
--- a/sys/kern/tty_ttydisc.c
+++ b/sys/kern/tty_ttydisc.c
@@ -844,23 +844,25 @@
*/
ttyinq_write(&tp->t_inq, bytes,
UTF8_STACKBUF, 0);
+ ttyinq_unputchar(&tp->t_inq);
} else {
/* Find codepoint and width. */
codepoint =
teken_utf8_bytes_to_codepoint(bytes,
nb);
- if (codepoint !=
- TEKEN_UTF8_INVALID_CODEPOINT) {
- cwidth = teken_wcwidth(
- codepoint);
- } else {
+ if (codepoint ==
+ TEKEN_UTF8_INVALID_CODEPOINT ||
+ (cwidth = teken_wcwidth(
+ codepoint)) == -1) {
/*
* Place all bytes back into the
* inq and fall back to
* default behaviour.
*/
+ cwidth = 1;
ttyinq_write(&tp->t_inq, bytes,
nb, 0);
+ ttyinq_unputchar(&tp->t_inq);
}
}
tp->t_column -= cwidth;
diff --git a/sys/teken/teken_wcwidth.h b/sys/teken/teken_wcwidth.h
--- a/sys/teken/teken_wcwidth.h
+++ b/sys/teken/teken_wcwidth.h
@@ -128,15 +128,32 @@
teken_utf8_bytes_to_codepoint(uint8_t bytes[4], int nbytes)
{
- /* Check for malformed characters. */
- if (__bitcount(bytes[0] & 0xf0) != nbytes)
+ /*
+ * Check for malformed characters by comparing 'nbytes'
+ * to the byte length of the character.
+ *
+ * The table in section 3 of RFC 3629 defines 4 different
+ * values indicating the length of a UTF-8 byte sequence.
+ *
+ * 0xxxxxxx -> 1 byte
+ * 110xxxxx -> 2 bytes
+ * 1110xxxx -> 3 bytes
+ * 11110xxx -> 4 bytes
+ *
+ * The length is determined by the higher-order bits in
+ * the leading octet (except in the first case, where an MSB
+ * of 0 means a byte length of 1). Here we flip the 4 upper
+ * bits and count the leading zeros using __builtin_clz()
+ * to determine the number of bytes.
+ */
+ if (__builtin_clz(~(bytes[0] & 0xf0) << 24) != nbytes)
return (TEKEN_UTF8_INVALID_CODEPOINT);
switch (nbytes) {
case 1:
return (bytes[0] & 0x7f);
case 2:
- return (bytes[0] & 0xf) << 6 | (bytes[1] & 0x3f);
+ return (bytes[0] & 0x1f) << 6 | (bytes[1] & 0x3f);
case 3:
return (bytes[0] & 0xf) << 12 | (bytes[1] & 0x3f) << 6 | (bytes[2] & 0x3f);
case 4:

File Metadata

Mime Type
text/plain
Expires
Thu, Nov 14, 9:30 PM (7 h, 50 m)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
14633981
Default Alt Text
D42147.diff (2 KB)

Event Timeline