summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorThomas Lively <tlively@google.com>2024-06-18 20:08:25 -0700
committerGitHub <noreply@github.com>2024-06-19 03:08:25 +0000
commit765c61445550c6e4ecfd250e1893d776d570b4fd (patch)
treeab60cbac0ba59fa649d93e9b62a2c0b0b0915a69
parent3acacac34c9ebe949fbc7d9eb3649266760104fe (diff)
downloadbinaryen-765c61445550c6e4ecfd250e1893d776d570b4fd.tar.gz
binaryen-765c61445550c6e4ecfd250e1893d776d570b4fd.tar.bz2
binaryen-765c61445550c6e4ecfd250e1893d776d570b4fd.zip
Validate that names are valid UTF-8 (#6682)
Add an `isUTF8` utility and use it in both the text and binary parsers. Add missing checks for overlong encodings and overlarge code points in our WTF8 reader, which the new utility uses. Re-enable the spec tests that test UTF-8 validation.
-rw-r--r--scripts/test/shared.py4
-rw-r--r--src/parser/lexer.h9
-rw-r--r--src/support/string.cpp22
-rw-r--r--src/support/string.h3
-rw-r--r--src/wasm-binary.h2
-rw-r--r--src/wasm/wasm-binary.cpp8
6 files changed, 36 insertions, 12 deletions
diff --git a/scripts/test/shared.py b/scripts/test/shared.py
index 4034f95ef..909b20c4b 100644
--- a/scripts/test/shared.py
+++ b/scripts/test/shared.py
@@ -400,10 +400,6 @@ os.chdir(options.out_dir)
# expected-output/ if any.
SPEC_TESTS_TO_SKIP = [
# Malformed module accepted
- 'utf8-custom-section-id.wast',
- 'utf8-import-field.wast',
- 'utf8-import-module.wast',
- 'utf8-invalid-encoding.wast',
'const.wast',
'address.wast',
diff --git a/src/parser/lexer.h b/src/parser/lexer.h
index 83cbcfc53..37c3fe04a 100644
--- a/src/parser/lexer.h
+++ b/src/parser/lexer.h
@@ -25,6 +25,7 @@
#include "support/name.h"
#include "support/result.h"
+#include "support/string.h"
#ifndef parser_lexer_h
#define parser_lexer_h
@@ -124,11 +125,11 @@ public:
std::optional<std::string> takeString();
std::optional<Name> takeName() {
- // TODO: Validate UTF.
- if (auto str = takeString()) {
- return Name(*str);
+ auto str = takeString();
+ if (!str || !String::isUTF8(*str)) {
+ return std::nullopt;
}
- return std::nullopt;
+ return Name(*str);
}
bool takeSExprStart(std::string_view expected) {
diff --git a/src/support/string.cpp b/src/support/string.cpp
index 31d0e9170..01fe4e522 100644
--- a/src/support/string.cpp
+++ b/src/support/string.cpp
@@ -195,9 +195,21 @@ std::optional<uint32_t> takeWTF8CodePoint(std::string_view& str) {
}
str = str.substr(1 + trailingBytes);
+
if (!valid) {
return std::nullopt;
}
+
+ size_t expectedTrailing = u < 0x80 ? 0
+ : u < 0x800 ? 1
+ : u < 0x10000 ? 2
+ : u < 0x110000 ? 3
+ : -1;
+ if (trailingBytes != expectedTrailing) {
+ // Overlong encoding or overlarge code point.
+ return std::nullopt;
+ }
+
return u;
}
@@ -404,4 +416,14 @@ std::ostream& printEscapedJSON(std::ostream& os, std::string_view str) {
return os << '"';
}
+bool isUTF8(std::string_view str) {
+ while (str.size()) {
+ auto u = takeWTF8CodePoint(str);
+ if (!u || (0xD800 <= *u && *u < 0xE000)) {
+ return false;
+ }
+ }
+ return true;
+}
+
} // namespace wasm::String
diff --git a/src/support/string.h b/src/support/string.h
index af120ab4e..24eb570c2 100644
--- a/src/support/string.h
+++ b/src/support/string.h
@@ -99,6 +99,9 @@ bool convertWTF16ToWTF8(std::ostream& os, std::string_view str);
// unit. Returns `true` if the input was valid UTF-16.
bool convertUTF16ToUTF8(std::ostream& os, std::string_view str);
+// Whether the string is valid UTF-8.
+bool isUTF8(std::string_view str);
+
} // namespace wasm::String
#endif // wasm_support_string_h
diff --git a/src/wasm-binary.h b/src/wasm-binary.h
index 46d81064e..9ce9bf181 100644
--- a/src/wasm-binary.h
+++ b/src/wasm-binary.h
@@ -1505,7 +1505,7 @@ public:
HeapType getIndexedHeapType();
Type getConcreteType();
- Name getInlineString();
+ Name getInlineString(bool requireValid = true);
void verifyInt8(int8_t x);
void verifyInt16(int16_t x);
void verifyInt32(int32_t x);
diff --git a/src/wasm/wasm-binary.cpp b/src/wasm/wasm-binary.cpp
index cd1ca7dfd..dc97dbbc3 100644
--- a/src/wasm/wasm-binary.cpp
+++ b/src/wasm/wasm-binary.cpp
@@ -2201,11 +2201,13 @@ Type WasmBinaryReader::getConcreteType() {
return type;
}
-Name WasmBinaryReader::getInlineString() {
+Name WasmBinaryReader::getInlineString(bool requireValid) {
BYN_TRACE("<==\n");
auto len = getU32LEB();
auto data = getByteView(len);
-
+ if (requireValid && !String::isUTF8(data)) {
+ throwError("invalid UTF-8 string");
+ }
BYN_TRACE("getInlineString: " << data << " ==>\n");
return Name(data);
}
@@ -3027,7 +3029,7 @@ void WasmBinaryReader::readStrings() {
}
size_t num = getU32LEB();
for (size_t i = 0; i < num; i++) {
- auto string = getInlineString();
+ auto string = getInlineString(false);
// Re-encode from WTF-8 to WTF-16.
std::stringstream wtf16;
if (!String::convertWTF8ToWTF16(wtf16, string.str)) {