[PATCH] D40754: [sanitizer] 64-bit allocator's PopulateFreeArray partial refactor

[Aleksey Shlyapnikov via Phabricator via llvm-commits]

alekseyshl added inline comments.


================
Comment at: lib/sanitizer_common/sanitizer_allocator_primary64.h:723
+      region->mapped_meta += meta_map_size;
     }
 
----------------
cryptoad wrote:
> alekseyshl wrote:
> > I'd suggest to wrap the entire metadata allocating block (from // Calculate the required space for metadata) with:
> > 
> >   if (kMetadataSize) {
> >     ...
> >   }
> So I thought about it, but I was wondering about something: currently the check for exhaustion will be done whether or not we use metadata.
> I wanted to leave it like that at first, because it sort of made sense.
> But then I figure that if we are not using metadata, and are exhausted, then the initial mmap would fail anyway since we'd hit the free array.
> It seems to be safe to place all of this within `if (kMetadataSize)`, but I wanted to raise this to your attention and get your feedback.
Yep, now we will return false due to MapWithCallback failure and the OOM error handling will kick in. I think, it's just what we want.


https://reviews.llvm.org/D40754