[sanitizers] Restore internal_readlink for x32

[hjl.tools]

Summary

r316591 has

@@ -389,13 +383,11 @@ uptr internal_dup2(int oldfd, int newfd) {
 }

 uptr internal_readlink(const char *path, char *buf, uptr bufsize) {
-#if SANITIZER_NETBSD
-  return internal_syscall_ptr(SYSCALL(readlink), path, buf, bufsize);
-#elif SANITIZER_USES_CANONICAL_LINUX_SYSCALLS
+#if SANITIZER_USES_CANONICAL_LINUX_SYSCALLS
   return internal_syscall(SYSCALL(readlinkat), AT_FDCWD,
                           (uptr)path, (uptr)buf, bufsize);
 #else
-  return internal_syscall(SYSCALL(readlink), (uptr)path, (uptr)buf, bufsize);
+  return internal_syscall_ptr(SYSCALL(readlink), path, buf, bufsize);
 #endif
 }

which dropped the (uptr) cast and broke x32. This patch puts back the
(uptr) cast to restore x32 and fixes:

https://bugs.llvm.org/show_bug.cgi?id=40783

[vitalybuka]

lib/sanitizer_common/sanitizer_linux.cc

403 ↗

(On Diff #187462)

Do we have a test which fails because of this?

[hjl.tools]

See:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89409

[vitalybuka]

Thanks,
I see, this very indirect test. Could you please add something into compiler-rt/lib/sanitizer_common/tests/sanitizer_common_test.cc
Maybe test ReadBinaryNameCached() which uses this function

In D58413#1403146, @hjl.tools wrote:

See:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89409

[vitalybuka]

LGTM

[morehouse]

compiler-rt/trunk/lib/sanitizer_common/tests/sanitizer_common_test.cc

444

This test is failing on the Windows sanitizer bot: http://lab.llvm.org:8011/builders/sanitizer-windows/builds/42268

Please take a look.

[morehouse]

compiler-rt/trunk/lib/sanitizer_common/tests/sanitizer_common_test.cc

444

Reverted in https://reviews.llvm.org/rCRT354601.

[hjl.tools]

It looks like that ReadBinaryNameCached doesn't work on Windows.
Someone who is familiar with Windows should take a look.

[morehouse]

@rnk: Could you help out here? I am clueless with Windows.

[rnk]

In D58413#1406000, @morehouse wrote:

@rnk: Could you help out here? I am clueless with Windows.

The answer is very straightforward, this code in sanitizer_win.cc:

uptr ReadBinaryName(/*out*/char *buf, uptr buf_len) {
  // FIXME: Actually implement this function.
  CHECK_GT(buf_len, 0);
  buf[0] = 0;
  return 0;
}

Of course the test fails, this always returns 0. Recommitting with the test disabled on Windows seems reasonable.

[hjl.tools]

In D58413#1406201, @rnk wrote:

Of course the test fails, this always returns 0. Recommitting with the test disabled on Windows seems reasonable.

Could someone do it for me and test it on Windows? Thanks in advance.

compiler-rt/trunk/lib/sanitizer_common/sanitizer_linux.cc

 uptr internal_readlink(const char *path, char *buf, uptr bufsize) {
 #if SANITIZER_USES_CANONICAL_LINUX_SYSCALLS
   return internal_syscall(SYSCALL(readlinkat), AT_FDCWD, (uptr)path, (uptr)buf,
                           bufsize);
 #elif SANITIZER_OPENBSD
   return internal_syscall(SYSCALL(readlinkat), AT_FDCWD, (uptr)path, (uptr)buf,
                           bufsize);
 #else
-  return internal_syscall(SYSCALL(readlink), path, buf, bufsize);
+  return internal_syscall(SYSCALL(readlink), (uptr)path, (uptr)buf, bufsize);
 #endif
 }
 
 uptr internal_unlink(const char *path) {
 #if SANITIZER_USES_CANONICAL_LINUX_SYSCALLS || SANITIZER_OPENBSD
   return internal_syscall(SYSCALL(unlinkat), AT_FDCWD, (uptr)path, 0);
 #else
   return internal_syscall(SYSCALL(unlink), (uptr)path);

compiler-rt/trunk/lib/sanitizer_common/tests/sanitizer_common_test.cc

TEST(SanitizerCommon, ReservedAddressRangeUnmap) {
   address_range.Unmap(new_start, PageSize);
 
   #endif
 
   // Unmapping in the middle of the ReservedAddressRange should fail.
   EXPECT_DEATH(address_range.Unmap(base_addr + (PageSize * 2), PageSize), ".*");
 }
 
+TEST(SanitizerCommon, ReadBinaryNameCached) {
+  char buf[256];
+  EXPECT_NE((uptr)0, ReadBinaryNameCached(buf, sizeof(buf)));
+}
+
 }  // namespace __sanitizer