mirror of
https://github.com/LadybirdBrowser/ladybird.git
synced 2024-11-21 23:20:20 +00:00
AK+Everywhere: Fix data corruption due to code-point-to-char conversion
In particular, StringView::contains(char) is often used with a u32 code point. When this is done, the compiler will for some reason allow data corruption to occur silently. In fact, this is one of two reasons for the following OSS Fuzz issue: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=49184 This is probably a very old bug. In the particular case of URLParser, AK::is_url_code_point got confused: return /* ... */ || "!$&'()*+,-./:;=?@_~"sv.contains(code_point); If code_point is a large code point that happens to have the correct lower bytes, AK::is_url_code_point is then convinced that the given code point is okay, even if it is actually problematic. This commit fixes *only* the silent data corruption due to the erroneous conversion, and does not fully resolve OSS-Fuzz#49184.
This commit is contained in:
parent
f07e0180d6
commit
3aeb57ed09
Notes:
sideshowbarker
2024-07-17 06:09:03 +09:00
Author: https://github.com/BenWiederhake Commit: https://github.com/SerenityOS/serenity/commit/3aeb57ed09 Pull-request: https://github.com/SerenityOS/serenity/pull/15211 Reviewed-by: https://github.com/ADKaster Reviewed-by: https://github.com/IdanHo Reviewed-by: https://github.com/bgianfo Reviewed-by: https://github.com/trflynn89
7 changed files with 29 additions and 13 deletions
|
@ -9,6 +9,7 @@
|
|||
#include <AK/Find.h>
|
||||
#include <AK/Function.h>
|
||||
#include <AK/Memory.h>
|
||||
#include <AK/StringBuilder.h>
|
||||
#include <AK/StringView.h>
|
||||
#include <AK/Vector.h>
|
||||
|
||||
|
@ -137,6 +138,20 @@ bool StringView::contains(char needle) const
|
|||
return false;
|
||||
}
|
||||
|
||||
bool StringView::contains(u32 needle) const
|
||||
{
|
||||
// A code point should be at most four UTF-8 bytes, which easily fits into StringBuilder's inline-buffer.
|
||||
// Therefore, this will not allocate.
|
||||
StringBuilder needle_builder;
|
||||
auto result = needle_builder.try_append_code_point(needle);
|
||||
if (result.is_error()) {
|
||||
// The needle is invalid, therefore the string does not contain it.
|
||||
return false;
|
||||
}
|
||||
|
||||
return contains(needle_builder.string_view());
|
||||
}
|
||||
|
||||
bool StringView::contains(StringView needle, CaseSensitivity case_sensitivity) const
|
||||
{
|
||||
return StringUtils::contains(*this, needle, case_sensitivity);
|
||||
|
|
|
@ -88,6 +88,7 @@ public:
|
|||
[[nodiscard]] bool matches(StringView mask, CaseSensitivity = CaseSensitivity::CaseInsensitive) const;
|
||||
[[nodiscard]] bool matches(StringView mask, Vector<MaskSpan>&, CaseSensitivity = CaseSensitivity::CaseInsensitive) const;
|
||||
[[nodiscard]] bool contains(char) const;
|
||||
[[nodiscard]] bool contains(u32) const;
|
||||
[[nodiscard]] bool contains(StringView, CaseSensitivity = CaseSensitivity::CaseSensitive) const;
|
||||
[[nodiscard]] bool equals_ignoring_case(StringView other) const;
|
||||
|
||||
|
|
|
@ -406,7 +406,7 @@ bool URL::code_point_is_in_percent_encode_set(u32 code_point, URL::PercentEncode
|
|||
case URL::PercentEncodeSet::EncodeURI:
|
||||
// NOTE: This is the same percent encode set that JS encodeURI() uses.
|
||||
// https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/encodeURI
|
||||
return code_point >= 0x7E || (!is_ascii_alphanumeric(code_point) && !";,/?:@&=+$-_.!~*'()#"sv.contains(code_point));
|
||||
return code_point >= 0x7E || (!is_ascii_alphanumeric(code_point) && !";,/?:@&=+$-_.!~*'()#"sv.contains(static_cast<char>(code_point)));
|
||||
default:
|
||||
VERIFY_NOT_REACHED();
|
||||
}
|
||||
|
|
|
@ -32,9 +32,9 @@ static void report_validation_error(SourceLocation const& location = SourceLocat
|
|||
|
||||
static Optional<String> parse_opaque_host(StringView input)
|
||||
{
|
||||
auto forbidden_host_code_points_excluding_percent = "\0\t\n\r #/:<>?@[\\]^|"sv;
|
||||
for (auto code_point : forbidden_host_code_points_excluding_percent) {
|
||||
if (input.contains(code_point)) {
|
||||
auto forbidden_host_characters_excluding_percent = "\0\t\n\r #/:<>?@[\\]^|"sv;
|
||||
for (auto character : forbidden_host_characters_excluding_percent) {
|
||||
if (input.contains(character)) {
|
||||
report_validation_error();
|
||||
return {};
|
||||
}
|
||||
|
@ -72,9 +72,9 @@ static Optional<String> parse_host(StringView input, bool is_not_special = false
|
|||
// FIXME: Let asciiDomain be the result of running domain to ASCII on domain.
|
||||
auto& ascii_domain = domain;
|
||||
|
||||
auto forbidden_host_code_points = "\0\t\n\r #%/:<>?@[\\]^|"sv;
|
||||
for (auto code_point : forbidden_host_code_points) {
|
||||
if (ascii_domain.view().contains(code_point)) {
|
||||
auto forbidden_host_characters = "\0\t\n\r #%/:<>?@[\\]^|"sv;
|
||||
for (auto character : forbidden_host_characters) {
|
||||
if (ascii_domain.view().contains(character)) {
|
||||
report_validation_error();
|
||||
return {};
|
||||
}
|
||||
|
|
|
@ -350,7 +350,7 @@ static ThrowCompletionOr<String> encode(VM& vm, String const& string, StringView
|
|||
auto code_unit = utf16_string.code_unit_at(k);
|
||||
// c. If C is in unescapedSet, then
|
||||
// NOTE: We assume the unescaped set only contains ascii characters as unescaped_set is a StringView.
|
||||
if (code_unit < 0x80 && unescaped_set.contains(code_unit)) {
|
||||
if (code_unit < 0x80 && unescaped_set.contains(static_cast<char>(code_unit))) {
|
||||
// i. Set k to k + 1.
|
||||
k++;
|
||||
|
||||
|
@ -420,8 +420,8 @@ static ThrowCompletionOr<String> decode(VM& vm, String const& string, StringView
|
|||
continue;
|
||||
}
|
||||
|
||||
if ((decoded_code_unit & 0x80) == 0) {
|
||||
if (reserved_set.contains(decoded_code_unit))
|
||||
if (decoded_code_unit < 0x80) {
|
||||
if (reserved_set.contains(static_cast<char>(decoded_code_unit)))
|
||||
decoded_builder.append(string.substring_view(k - 2, 3));
|
||||
else
|
||||
decoded_builder.append(decoded_code_unit);
|
||||
|
@ -480,7 +480,7 @@ JS_DEFINE_NATIVE_FUNCTION(GlobalObject::escape)
|
|||
StringBuilder escaped;
|
||||
for (auto code_point : utf8_to_utf16(string)) {
|
||||
if (code_point < 256) {
|
||||
if ("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789@*_+-./"sv.contains(code_point))
|
||||
if ("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789@*_+-./"sv.contains(static_cast<char>(code_point)))
|
||||
escaped.append(code_point);
|
||||
else
|
||||
escaped.appendff("%{:02X}", code_point);
|
||||
|
|
|
@ -402,7 +402,7 @@ bool is_cors_safelisted_request_header(Header const& header)
|
|||
else if (name.is_one_of_ignoring_case("accept-language"sv, "content-language"sv)) {
|
||||
// If value contains a byte that is not in the range 0x30 (0) to 0x39 (9), inclusive, is not in the range 0x41 (A) to 0x5A (Z), inclusive, is not in the range 0x61 (a) to 0x7A (z), inclusive, and is not 0x20 (SP), 0x2A (*), 0x2C (,), 0x2D (-), 0x2E (.), 0x3B (;), or 0x3D (=), then return false.
|
||||
if (any_of(value.span(), [](auto byte) {
|
||||
return !(is_ascii_digit(byte) || is_ascii_alpha(byte) || " *,-.;="sv.contains(byte));
|
||||
return !(is_ascii_digit(byte) || is_ascii_alpha(byte) || " *,-.;="sv.contains(static_cast<char>(byte)));
|
||||
}))
|
||||
return false;
|
||||
}
|
||||
|
|
|
@ -258,7 +258,7 @@ ErrorOr<int> serenity_main(Main::Arguments args)
|
|||
// Human-readable indexes start at 1, so it's fine to increment already.
|
||||
line_number += 1;
|
||||
StringView line_view(line, nread);
|
||||
bool is_binary = line_view.contains(0);
|
||||
bool is_binary = line_view.contains('\0');
|
||||
|
||||
if (is_binary && binary_mode == BinaryFileMode::Skip)
|
||||
return 1;
|
||||
|
|
Loading…
Reference in a new issue