From 1e39ea08122ef6db924f2ae622e03d4bedde8cd5 Mon Sep 17 00:00:00 2001 From: Ian Lance Taylor Date: Sat, 22 Sep 2012 06:51:59 +0000 Subject: [PATCH] compiler, runtime: Reject surrogate pair converting int to string. From-SVN: r191636 --- gcc/go/gofrontend/lex.cc | 6 ++++++ libgo/runtime/go-int-to-string.c | 9 +++++++++ 2 files changed, 15 insertions(+) diff --git a/gcc/go/gofrontend/lex.cc b/gcc/go/gofrontend/lex.cc index 25aaeb794a55..6add84ed1f7d 100644 --- a/gcc/go/gofrontend/lex.cc +++ b/gcc/go/gofrontend/lex.cc @@ -1312,6 +1312,12 @@ Lex::append_char(unsigned int v, bool is_character, std::string* str, // Turn it into the "replacement character". v = 0xfffd; } + if (v >= 0xd800 && v < 0xe000) + { + warning_at(location, 0, + "unicode code point 0x%x is invalid surrogate pair", v); + v = 0xfffd; + } if (v <= 0xffff) { buf[0] = 0xe0 + (v >> 12); diff --git a/libgo/runtime/go-int-to-string.c b/libgo/runtime/go-int-to-string.c index e9645bf98feb..17a5fcb04c01 100644 --- a/libgo/runtime/go-int-to-string.c +++ b/libgo/runtime/go-int-to-string.c @@ -17,6 +17,11 @@ __go_int_to_string (int v) unsigned char *retdata; struct __go_string ret; + /* A negative value is not valid UTF-8; turn it into the replacement + character. */ + if (v < 0) + v = 0xfffd; + if (v <= 0x7f) { buf[0] = v; @@ -34,6 +39,10 @@ __go_int_to_string (int v) "replacement character". */ if (v > 0x10ffff) v = 0xfffd; + /* If the value is a surrogate pair, which is invalid in UTF-8, + turn it into the replacement character. */ + if (v >= 0xd800 && v < 0xe000) + v = 0xfffd; if (v <= 0xffff) {