[PATCH] D98192: [dfsan] Tracking origins at memory transfer

[Matt Morehouse via Phabricator via llvm-commits]

morehouse added inline comments.


================
Comment at: compiler-rt/test/dfsan/origin_memcpy.c:19
+void fn_g(int a, int b) {
+  xx[idx] = a; xx[idx + 10] = b;
+}
----------------
Why don't we get an origin saved for these stores?


================
Comment at: compiler-rt/test/dfsan/origin_memcpy.c:34
+  int volatile z1;
+  int volatile z2;
+  dfsan_set_label(8, (void *)&z1, sizeof(z1));
----------------
Currently the test triggers read of uninitialized memory, but that isn't necessary for this test.  So let's initialize these two variables here.


================
Comment at: compiler-rt/test/dfsan/origin_unaligned_memtrans.c:44
+  return 0;
+}
+
----------------
This test seems to cover all the cases from origin_memmove.c and origin_memcpy.c.  Do we still need those?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D98192/new/

https://reviews.llvm.org/D98192