[PATCH] D102901: [HWASAN] Update pointer tag for X86_64

[Matt Morehouse via Phabricator via llvm-commits]

morehouse added inline comments.


================
Comment at: llvm/lib/Transforms/Instrumentation/HWAddressSanitizer.cpp:190
+                                    cl::desc("untag mem operate"), cl::Hidden,
+                                    cl::init(false));
+
----------------
xiangzhangllvm wrote:
> morehouse wrote:
> > Can we avoid creating `ClUntagPointer` for now?  I am able to test locally with QEMU, and I'm also setting up a buildbot to ensure new patches don't break the LAM functionality.
> Yes, of course, we can. But here ClUntagPointer is disabled in default (false), no affect our current tests.
> I and back this option, because I find It is very helpful for me in developing HWASAN without hardware or simulator supported. I can run most simple test on my local machine.
It may be inconvenient for now, but as part of setting up a buildbot I'm also automating the process of setting up the custom QEMU and kernel required to test LAM.  If all goes well, it should land later this week or next week.  Then you can use that instead of the `ClUntagPointer` option.

For now, could you keep the `ClUntagPointer` patch locally for your testing purposes, and remove it from the patch we actually submit?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D102901/new/

https://reviews.llvm.org/D102901