[ASan] Do not misrepresent high value address dereferences as null dereferences

Dereferences with addresses above the 48-bit hardware addressable range
produce "invalid instruction" (instead of "invalid access") hardware
exceptions (there is no hardware address decoding logic for those bits),
and the address provided by this exception is the address of the
instruction (not the faulting address).  The kernel maps the "invalid
instruction" to SEGV, but fails to provide the real fault address.

Because of this ASan lies and says that those cases are null
dereferences.  This downgrades the severity of a found bug in terms of
security.  In the ASan signal handler, we can not provide the real
faulting address, but at least we can try not to lie.

rdar://50366151

Reviewed By: vitalybuka

Differential Revision: https://reviews.llvm.org/D68676

git-svn-id: https://llvm.org/svn/llvm-project/compiler-rt/trunk@374265 91177308-0d34-0410-b5e6-96231b3b80d8

1 files changed, 50 insertions, 0 deletions

diff --git a/test/asan/TestCases/Posix/high-address-dereference.c b/test/asan/TestCases/Posix/high-address-dereference.c
new file mode 100644
index 000000000..785033028
--- /dev/null
+++ b/test/asan/TestCases/Posix/high-address-dereference.c
@@ -0,0 +1,50 @@
+// On x86_64, the kernel does not provide the faulting address for dereferences
+// of addresses greater than the 48-bit hardware addressable range, i.e.,
+// `siginfo.si_addr` is zero in ASan's SEGV signal handler. This test checks
+// that ASan does not misrepresent such cases as "NULL dereferences".
+
+// REQUIRES: x86_64-target-arch
+// RUN: %clang_asan %s -o %t
+// RUN: export %env_asan_opts=print_scariness=1
+// RUN: not %run %t 0x0000000000000000 2>&1 | FileCheck %s --check-prefixes=ZERO,HINT-PAGE0
+// RUN: not %run %t 0x0000000000000FFF 2>&1 | FileCheck %s --check-prefixes=LOW1,HINT-PAGE0
+// RUN: not %run %t 0x0000000000001000 2>&1 | FileCheck %s --check-prefixes=LOW2,HINT-NONE
+// RUN: not %run %t 0x4141414141414141 2>&1 | FileCheck %s --check-prefixes=HIGH,HINT-HIGHADDR
+// RUN: not %run %t 0xFFFFFFFFFFFFFFFF 2>&1 | FileCheck %s --check-prefixes=MAX,HINT-HIGHADDR
+
+#include <stdint.h>
+#include <stdlib.h>
+
+int main(int argc, const char *argv[]) {
+  const char *hex = argv[1];
+  uint64_t *addr = (uint64_t *)strtoull(hex, NULL, 16);
+  uint64_t x = *addr;  // segmentation fault
+  return x;
+}
+
+// ZERO:  SEGV on unknown address 0x000000000000 (pc
+// LOW1:  SEGV on unknown address 0x000000000fff (pc
+// LOW2:  SEGV on unknown address 0x000000001000 (pc
+// HIGH:  SEGV on unknown address (pc
+// MAX:   SEGV on unknown address (pc
+
+// HINT-PAGE0-NOT: Hint: this fault was caused by a dereference of a high value address
+// HINT-PAGE0:     Hint: address points to the zero page.
+
+// HINT-NONE-NOT:  Hint: this fault was caused by a dereference of a high value address
+// HINT-NONE-NOT:  Hint: address points to the zero page.
+
+// HINT-HIGHADDR:     Hint: this fault was caused by a dereference of a high value address
+// HINT-HIGHADDR-NOT: Hint: address points to the zero page.
+
+// ZERO:  SCARINESS: 10 (null-deref)
+// LOW1:  SCARINESS: 10 (null-deref)
+// LOW2:  SCARINESS: 20 (wild-addr-read)
+// HIGH:  SCARINESS: 20 (wild-addr-read)
+// MAX:   SCARINESS: 20 (wild-addr-read)
+
+// TODO: Currently, register values are only printed on Mac.  Once this changes,
+//       remove the 'TODO_' prefix in the following lines.
+// TODO_HIGH,TODO_MAX: Register values:
+// TODO_HIGH: = 0x4141414141414141
+// TODO_MAX:  = 0xffffffffffffffff