[PATCH] D69765: [compiler-rt] Support more CPUs in LSan Allocator Address Space

Christos Zoulas via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Sun Nov 3 17:27:47 PST 2019


christos marked an inline comment as done.
christos added inline comments.


================
Comment at: compiler-rt/lib/lsan/lsan_allocator.h:55
+    defined(__sh3__) || defined(__vax__) ||                           \
+    (defined(__mips64) && !defined(_LP64)) ||                         \
+    (defined(__sparc__) && !defined(_LP64)) ||                        \
----------------
krytarowski wrote:
> arichardson wrote:
> > krytarowski wrote:
> > > arichardson wrote:
> > > > Maybe change this to `&& !defined(__mips_n32)`?
> > > > For our CHERI CPU we use n64 but don't have _LP64 defined (since pointers are 128 bits).
> > > There is also o32. I don't know CHERI specifics and it is not supported today on NetBSD. (And there is no LSan for FreeBSD either).
> > I actually meant to add this to the change below not this one. Sorry.
> > 
> > However, the MIPS change seems wrong since the initial `#if defined(__mips64)` condition means the newly added condition never triggers.
> > 
> @christos what is the proper MIPS combination here?
I think that all this complex mess of ifdefs is really answering the question "Are pointers 32 bits?" so why don't we switch the ifdef to that (for example: #if \_\_UINTPTR_MAX\_\_ == 0xffffffffU)

PS: h at e markup. How does one make double underscore not become underline?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D69765/new/

https://reviews.llvm.org/D69765





More information about the llvm-commits mailing list