Merge "Regression test for scudo crash in resizeTaggedChunk."
diff --git a/benchmarks/stdlib_benchmark.cpp b/benchmarks/stdlib_benchmark.cpp
index b6ea58d..14b380a 100644
--- a/benchmarks/stdlib_benchmark.cpp
+++ b/benchmarks/stdlib_benchmark.cpp
@@ -155,88 +155,71 @@
BIONIC_BENCHMARK_WITH_ARG(BM_stdlib_malloc_multiple_8192_allocs_decay1, "AT_SMALL_SIZES");
#endif
-static void BM_stdlib_mbstowcs(benchmark::State& state) {
- const size_t buf_alignment = state.range(0);
- const size_t widebuf_alignment = state.range(1);
-
- std::vector<char> buf;
- std::vector<wchar_t> widebuf;
-
- setlocale(LC_CTYPE, "C.UTF-8")
- || setlocale(LC_CTYPE, "en_US.UTF-8")
- || setlocale(LC_CTYPE, "en_GB.UTF-8")
- || setlocale(LC_CTYPE, "en.UTF-8")
- || setlocale(LC_CTYPE, "de_DE-8")
- || setlocale(LC_CTYPE, "fr_FR-8");
- if (strcmp(nl_langinfo(CODESET), "UTF-8")) {
- errx(1, "ERROR: unable to set locale in BM_stdlib_mbstowcs");
- }
-
- char* buf_aligned = GetAlignedPtr(&buf, buf_alignment, 500000);
- wchar_t* widebuf_aligned = GetAlignedPtr(&widebuf, widebuf_alignment, 500000);
- size_t i, j, k, l;
- l = 0;
- for (i=0xc3; i<0xe0; i++)
- for (j=0x80; j<0xc0; j++)
- buf[l++] = i, buf[l++] = j;
- for (i=0xe1; i<0xed; i++)
- for (j=0x80; j<0xc0; j++)
- for (k=0x80; k<0xc0; k++)
- buf[l++] = i, buf[l++] = j, buf[l++] = k;
- for (i=0xf1; i<0xf4; i++)
- for (j=0x80; j<0xc0; j++)
- for (k=0x80; k<0xc0; k++)
- buf[l++] = i, buf[l++] = j, buf[l++] = 0x80, buf[l++] = k;
- buf[l++] = 0;
+static void BM_stdlib_mbstowcs_ascii(benchmark::State& state) {
+ // It doesn't really matter what ASCII character we pick.
+ // The flow through the fast path is the same regardless.
+ const size_t count = 500000;
+ std::vector<char> mbs(count, 'e');
+ std::vector<wchar_t> wcs(count);
for (auto _ : state) {
- benchmark::DoNotOptimize(mbstowcs(widebuf_aligned, buf_aligned, 500000));
+ benchmark::DoNotOptimize(mbstowcs(&wcs[0], &mbs[0], wcs.size()));
}
- state.SetBytesProcessed(uint64_t(state.iterations()) * uint64_t(500000));
+ state.SetBytesProcessed(uint64_t(state.iterations()) * uint64_t(wcs.size()));
}
-BIONIC_BENCHMARK_WITH_ARG(BM_stdlib_mbstowcs, "0 0");
+BIONIC_BENCHMARK_WITH_ARG(BM_stdlib_mbstowcs_ascii, "");
-static void BM_stdlib_mbrtowc(benchmark::State& state) {
- const size_t buf_alignment = state.range(0);
-
- std::vector<char> buf;
-
- setlocale(LC_CTYPE, "C.UTF-8")
- || setlocale(LC_CTYPE, "en_US.UTF-8")
- || setlocale(LC_CTYPE, "en_GB.UTF-8")
- || setlocale(LC_CTYPE, "en.UTF-8")
- || setlocale(LC_CTYPE, "de_DE-8")
- || setlocale(LC_CTYPE, "fr_FR-8");
- if (strcmp(nl_langinfo(CODESET), "UTF-8")) {
- errx(1, "ERROR: unable to set locale in BM_stdlib_mbrtowc");
+static void BM_stdlib_mbstowcs_wide(benchmark::State& state) {
+ // It doesn't matter much what wide character we pick.
+ // A three-byte character seems pretty representative, and all three byte
+ // characters are the same from the code's perspective.
+ const size_t count = 500000;
+ std::string mbs;
+ for (size_t i = 0; i < count; i++) {
+ mbs += "\xe5\xb1\xb1";
}
+ std::vector<wchar_t> wcs(count);
- char* buf_aligned = GetAlignedPtr(&buf, buf_alignment, 500000);
- size_t i, j, k, l;
- l = 0;
- for (i=0xc3; i<0xe0; i++)
- for (j=0x80; j<0xc0; j++)
- buf[l++] = i, buf[l++] = j;
- for (i=0xe1; i<0xed; i++)
- for (j=0x80; j<0xc0; j++)
- for (k=0x80; k<0xc0; k++)
- buf[l++] = i, buf[l++] = j, buf[l++] = k;
- for (i=0xf1; i<0xf4; i++)
- for (j=0x80; j<0xc0; j++)
- for (k=0x80; k<0xc0; k++)
- buf[l++] = i, buf[l++] = j, buf[l++] = 0x80, buf[l++] = k;
- buf[l++] = 0;
-
- wchar_t wc = 0;
for (auto _ : state) {
- for (j = 0; buf_aligned[j]; j+=mbrtowc(&wc, buf_aligned + j, 4, nullptr)) {
- }
+ benchmark::DoNotOptimize(mbstowcs(&wcs[0], &mbs[0], wcs.size()));
}
- state.SetBytesProcessed(uint64_t(state.iterations()) * uint64_t(500000));
+ state.SetBytesProcessed(uint64_t(state.iterations()) * uint64_t(wcs.size()));
}
-BIONIC_BENCHMARK_WITH_ARG(BM_stdlib_mbrtowc, "0");
+BIONIC_BENCHMARK_WITH_ARG(BM_stdlib_mbstowcs_wide, "");
+
+static void BM_stdlib_mbrtowc_1(benchmark::State& state) {
+ wchar_t wc;
+ for (auto _ : state) {
+ benchmark::DoNotOptimize(mbrtowc(&wc, "e", 1, nullptr));
+ }
+}
+BIONIC_BENCHMARK_WITH_ARG(BM_stdlib_mbrtowc_1, "");
+
+static void BM_stdlib_mbrtowc_2(benchmark::State& state) {
+ wchar_t wc;
+ for (auto _ : state) {
+ benchmark::DoNotOptimize(mbrtowc(&wc, "\xc3\x9f", 3, nullptr));
+ }
+}
+BIONIC_BENCHMARK_WITH_ARG(BM_stdlib_mbrtowc_2, "");
+
+static void BM_stdlib_mbrtowc_3(benchmark::State& state) {
+ wchar_t wc;
+ for (auto _ : state) {
+ benchmark::DoNotOptimize(mbrtowc(&wc, "\xe5\xb1\xb1", 3, nullptr));
+ }
+}
+BIONIC_BENCHMARK_WITH_ARG(BM_stdlib_mbrtowc_3, "");
+
+static void BM_stdlib_mbrtowc_4(benchmark::State& state) {
+ wchar_t wc;
+ for (auto _ : state) {
+ benchmark::DoNotOptimize(mbrtowc(&wc, "\xf0\xa4\xad\xa2", 4, nullptr));
+ }
+}
+BIONIC_BENCHMARK_WITH_ARG(BM_stdlib_mbrtowc_4, "");
BIONIC_TRIVIAL_BENCHMARK(BM_stdlib_atoi, atoi(" -123"));
BIONIC_TRIVIAL_BENCHMARK(BM_stdlib_atol, atol(" -123"));
diff --git a/libc/bionic/c16rtomb.cpp b/libc/bionic/c16rtomb.cpp
index 2d6ae93..e052d04 100644
--- a/libc/bionic/c16rtomb.cpp
+++ b/libc/bionic/c16rtomb.cpp
@@ -43,7 +43,7 @@
size_t c16rtomb(char* s, char16_t c16, mbstate_t* ps) {
static mbstate_t __private_state;
mbstate_t* state = (ps == nullptr) ? &__private_state : ps;
- if (mbsinit(state)) {
+ if (mbstate_is_initial(state)) {
if (is_high_surrogate(c16)) {
char32_t c32 = (c16 & ~0xd800) << 10;
mbstate_set_byte(state, 3, (c32 & 0xff0000) >> 16);
diff --git a/libc/bionic/c32rtomb.cpp b/libc/bionic/c32rtomb.cpp
index 2909d8b..d2519b9 100644
--- a/libc/bionic/c32rtomb.cpp
+++ b/libc/bionic/c32rtomb.cpp
@@ -50,7 +50,7 @@
return mbstate_reset_and_return(1, state);
}
- if (!mbsinit(state)) {
+ if (!mbstate_is_initial(state)) {
return mbstate_reset_and_return_illegal(EILSEQ, state);
}
diff --git a/libc/bionic/mbrtoc32.cpp b/libc/bionic/mbrtoc32.cpp
index 644e542..21603a1 100644
--- a/libc/bionic/mbrtoc32.cpp
+++ b/libc/bionic/mbrtoc32.cpp
@@ -55,7 +55,7 @@
}
uint8_t ch;
- if (mbsinit(state) && (((ch = static_cast<uint8_t>(*s)) & ~0x7f) == 0)) {
+ if (mbstate_is_initial(state) && (((ch = static_cast<uint8_t>(*s)) & ~0x7f) == 0)) {
// Fast path for plain ASCII characters.
if (pc32 != nullptr) {
*pc32 = ch;
@@ -105,7 +105,7 @@
size_t bytes_wanted = length - bytes_so_far;
size_t i;
for (i = 0; i < MIN(bytes_wanted, n); i++) {
- if (!mbsinit(state) && ((*s & 0xc0) != 0x80)) {
+ if (!mbstate_is_initial(state) && ((*s & 0xc0) != 0x80)) {
// Malformed input; bad characters in the middle of a character.
return mbstate_reset_and_return_illegal(EILSEQ, state);
}
diff --git a/libc/bionic/wchar.cpp b/libc/bionic/wchar.cpp
index dabe824..bd9a45e 100644
--- a/libc/bionic/wchar.cpp
+++ b/libc/bionic/wchar.cpp
@@ -54,7 +54,7 @@
//
int mbsinit(const mbstate_t* ps) {
- return (ps == nullptr || (*(reinterpret_cast<const uint32_t*>(ps->__seq)) == 0));
+ return ps == nullptr || mbstate_is_initial(ps);
}
size_t mbrtowc(wchar_t* pwc, const char* s, size_t n, mbstate_t* ps) {
@@ -148,7 +148,7 @@
static mbstate_t __private_state;
mbstate_t* state = (ps == nullptr) ? &__private_state : ps;
- if (!mbsinit(state)) {
+ if (!mbstate_is_initial(state)) {
return mbstate_reset_and_return_illegal(EILSEQ, state);
}
diff --git a/libc/include/stdlib.h b/libc/include/stdlib.h
index 583287f..252e73a 100644
--- a/libc/include/stdlib.h
+++ b/libc/include/stdlib.h
@@ -161,11 +161,11 @@
void setprogname(const char* __name) __INTRODUCED_IN(21);
int mblen(const char* __s, size_t __n) __INTRODUCED_IN_NO_GUARD_FOR_NDK(26);
-size_t mbstowcs(wchar_t* __dst, const char* __src, size_t __n);
+size_t mbstowcs(wchar_t* __dst, const char* __src, size_t __n) __INTRODUCED_IN(21) __VERSIONER_NO_GUARD;
int mbtowc(wchar_t* __wc_ptr, const char* __s, size_t __n) __INTRODUCED_IN_NO_GUARD_FOR_NDK(21);
int wctomb(char* __dst, wchar_t __wc) __INTRODUCED_IN_NO_GUARD_FOR_NDK(21);
-size_t wcstombs(char* __dst, const wchar_t* __src, size_t __n);
+size_t wcstombs(char* __dst, const wchar_t* __src, size_t __n) __INTRODUCED_IN(21) __VERSIONER_NO_GUARD;
#if __ANDROID_API__ >= 21
size_t __ctype_get_mb_cur_max(void) __INTRODUCED_IN(21);
diff --git a/libc/libc.map.txt b/libc/libc.map.txt
index 00b25d3..7397b68 100644
--- a/libc/libc.map.txt
+++ b/libc/libc.map.txt
@@ -631,7 +631,7 @@
mbsinit;
mbsnrtowcs; # introduced=21
mbsrtowcs;
- mbstowcs;
+ mbstowcs; # introduced=21
mbtowc; # introduced=21
memalign;
memccpy;
@@ -1194,7 +1194,7 @@
wcstold_l; # introduced=21
wcstoll; # introduced=21
wcstoll_l; # introduced=21
- wcstombs;
+ wcstombs; # introduced=21
wcstoul;
wcstoull; # introduced=21
wcstoull_l; # introduced=21
diff --git a/libc/private/bionic_mbstate.h b/libc/private/bionic_mbstate.h
index 352115a..243b220 100644
--- a/libc/private/bionic_mbstate.h
+++ b/libc/private/bionic_mbstate.h
@@ -44,6 +44,10 @@
#define __MB_IS_ERR(rv) (rv == __MB_ERR_ILLEGAL_SEQUENCE || \
rv == __MB_ERR_INCOMPLETE_SEQUENCE)
+static inline __wur bool mbstate_is_initial(const mbstate_t* ps) {
+ return *(reinterpret_cast<const uint32_t*>(ps->__seq)) == 0;
+}
+
static inline __wur size_t mbstate_bytes_so_far(const mbstate_t* ps) {
return
(ps->__seq[2] != 0) ? 3 :
diff --git a/tests/libs/Android.bp b/tests/libs/Android.bp
index cef95cd..0046ef6 100644
--- a/tests/libs/Android.bp
+++ b/tests/libs/Android.bp
@@ -1181,14 +1181,6 @@
defaults: ["bionic_testlib_defaults"],
srcs: ["dl_df_1_global.cpp"],
ldflags: ["-Wl,-z,global"],
-
- target: {
- host: {
- // TODO (dimitry): host ld.gold does not yet support -z global
- // remove this line once it is updated.
- ldflags: ["-fuse-ld=bfd"],
- },
- },
}
// -----------------------------------------------------------------------------
@@ -1209,14 +1201,6 @@
defaults: ["bionic_testlib_defaults"],
srcs: ["dl_df_1_global_dummy.cpp"],
ldflags: ["-Wl,-z,global"],
-
- target: {
- host: {
- // TODO (dimitry): host ld.gold does not yet support -z global
- // remove this line once it is updated.
- ldflags: ["-fuse-ld=bfd"],
- },
- },
}
// -----------------------------------------------------------------------------
diff --git a/tests/uchar_test.cpp b/tests/uchar_test.cpp
index 48c500d..4dc6314 100644
--- a/tests/uchar_test.cpp
+++ b/tests/uchar_test.cpp
@@ -36,6 +36,7 @@
// Any non-initial state is invalid when calling c32rtomb.
memset(&ps, 0, sizeof(ps));
EXPECT_EQ(static_cast<size_t>(-2), mbrtoc32(nullptr, "\xc2", 1, &ps));
+ errno = 0;
EXPECT_EQ(static_cast<size_t>(-1), c32rtomb(out, 0x00a2, &ps));
EXPECT_EQ(EILSEQ, errno);
@@ -87,11 +88,7 @@
EXPECT_EQ('\xe2', bytes[0]);
EXPECT_EQ('\x82', bytes[1]);
EXPECT_EQ('\xac', bytes[2]);
-}
-
-TEST(uchar, c16rtomb_surrogate) {
- char bytes[MB_LEN_MAX];
-
+ // 4-byte UTF-8 from a surrogate pair...
memset(bytes, 0, sizeof(bytes));
EXPECT_EQ(0U, c16rtomb(bytes, 0xdbea, nullptr));
EXPECT_EQ(4U, c16rtomb(bytes, 0xdfcd, nullptr));
@@ -143,16 +140,16 @@
// 3-byte UTF-8.
ASSERT_EQ(3U, mbrtoc16(&out, "\xe2\x82\xac" "def", 6, nullptr));
ASSERT_EQ(static_cast<char16_t>(0x20ac), out);
-}
-
-TEST(uchar, mbrtoc16_surrogate) {
- char16_t out;
-
+ // 4-byte UTF-8 will be returned as a surrogate pair...
ASSERT_EQ(static_cast<size_t>(-3),
mbrtoc16(&out, "\xf4\x8a\xaf\x8d", 6, nullptr));
ASSERT_EQ(static_cast<char16_t>(0xdbea), out);
ASSERT_EQ(4U, mbrtoc16(&out, "\xf4\x8a\xaf\x8d" "ef", 6, nullptr));
ASSERT_EQ(static_cast<char16_t>(0xdfcd), out);
+ // Illegal 5-byte UTF-8.
+ errno = 0;
+ ASSERT_EQ(static_cast<size_t>(-1), mbrtoc16(&out, "\xf8\xa1\xa2\xa3\xa4", 5, nullptr));
+ ASSERT_EQ(EILSEQ, errno);
}
TEST(uchar, mbrtoc16_reserved_range) {
@@ -194,6 +191,7 @@
// Invalid 2-byte
ASSERT_EQ(static_cast<size_t>(-2), mbrtoc16(&out, "\xc2", 1, ps));
+ errno = 0;
ASSERT_EQ(static_cast<size_t>(-1), mbrtoc16(&out, "\x20" "cdef", 5, ps));
ASSERT_EQ(EILSEQ, errno);
}
@@ -247,6 +245,7 @@
EXPECT_EQ('\xad', bytes[2]);
EXPECT_EQ('\xa2', bytes[3]);
// Invalid code point.
+ errno = 0;
EXPECT_EQ(static_cast<size_t>(-1), c32rtomb(bytes, 0xffffffff, nullptr));
EXPECT_EQ(EILSEQ, errno);
}
@@ -307,10 +306,12 @@
ASSERT_EQ(static_cast<char32_t>(0x24b62), out[0]);
#if defined(__BIONIC__) // glibc allows this.
// Illegal 5-byte UTF-8.
+ errno = 0;
ASSERT_EQ(static_cast<size_t>(-1), mbrtoc32(out, "\xf8\xa1\xa2\xa3\xa4" "f", 6, nullptr));
ASSERT_EQ(EILSEQ, errno);
#endif
// Illegal over-long sequence.
+ errno = 0;
ASSERT_EQ(static_cast<size_t>(-1), mbrtoc32(out, "\xf0\x82\x82\xac" "ef", 6, nullptr));
ASSERT_EQ(EILSEQ, errno);
}
@@ -340,6 +341,7 @@
// Invalid 2-byte
ASSERT_EQ(static_cast<size_t>(-2), mbrtoc32(&out, "\xc2", 1, ps));
+ errno = 0;
ASSERT_EQ(static_cast<size_t>(-1), mbrtoc32(&out, "\x20" "cdef", 5, ps));
ASSERT_EQ(EILSEQ, errno);
}