[compiler-rt] [hwasan] Fix and re-enable deep-recursion.c (PR #69265)

via llvm-commits llvm-commits at lists.llvm.org
Mon Oct 16 16:21:31 PDT 2023


llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-compiler-rt-sanitizer

Author: Thurston Dang (thurstond)

<details>
<summary>Changes</summary>

deep-recursion.c was disabled
(https://github.com/llvm/llvm-project/commit/c007e0f66ee3f96467fd12f6200218fb4c38c2c9)
because the test may get unlucky and end up with a zero-tagged variable, leading to a false negative (https://github.com/llvm/llvm-project/issues/69221).

This patch re-enables the test and adds a workaround: it checks
if the variable is zero-tagged, and if so, it will instead use the
neighboring variable, which must have a different (hence non-zero)
tag.

Fixing the stack allocation tagging is left as an exercise for the
reader. It is non-trivial because, even if the stackTagBase is
non-zero, tags for subsequent allocations in the stack frame may wrap
around to zero; working around this would require adding multiple
instructions to each alloca.


---
Full diff: https://github.com/llvm/llvm-project/pull/69265.diff


1 Files Affected:

- (modified) compiler-rt/test/hwasan/TestCases/deep-recursion.c (+17-4) 


``````````diff
diff --git a/compiler-rt/test/hwasan/TestCases/deep-recursion.c b/compiler-rt/test/hwasan/TestCases/deep-recursion.c
index bf390d051d472e7..c992f205917ba22 100644
--- a/compiler-rt/test/hwasan/TestCases/deep-recursion.c
+++ b/compiler-rt/test/hwasan/TestCases/deep-recursion.c
@@ -17,9 +17,6 @@
 // Stack histories are currently not recorded on x86.
 // XFAIL: target=x86_64{{.*}}
 
-// Flaky on AArch64 Linux, see https://github.com/llvm/llvm-project/issues/69221.
-// UNSUPPORTED: target=aarch64{{.*}}
-
 #include <stdlib.h>
 // At least -O1 is needed for this function to not have a stack frame on
 // AArch64.
@@ -29,7 +26,23 @@ void USE(void *x) { // pretend_to_do_something(void *x)
 
 volatile int four = 4;
 
-__attribute__((noinline)) void OOB() { int x[4]; x[four] = 0; USE(&x[0]); }
+__attribute__((noinline)) void OOB() {
+  int x[4];
+  int y[4];
+
+  // Tags for stack-allocated variables can occasionally be zero, resulting in
+  // a false negative for this test. This is not easy to fix, hence we work
+  // around it: if the tag is zero, we use the neighboring variable instead,
+  // which must have a different (hence non-zero) tag.
+  // This tag check assumes aarch64.
+  if(((unsigned long)&x) >> 56 == 0) {
+    y[four] = 0;
+  } else {
+    x[four] = 0;
+  }
+  USE(&x[0]);
+  USE(&y[0]);
+}
 __attribute__((noinline)) void FUNC1() { int x; USE(&x); OOB(); }
 __attribute__((noinline)) void FUNC2() { int x; USE(&x); FUNC1(); }
 __attribute__((noinline)) void FUNC3() { int x; USE(&x); FUNC2(); }

``````````

</details>


https://github.com/llvm/llvm-project/pull/69265


More information about the llvm-commits mailing list