[PATCH] D78619: [mlir] Extended Alloc and Dealloc operations with memory-effect traits.

[Mehdi AMINI via Phabricator via llvm-commits]

mehdi_amini added inline comments.


================
Comment at: mlir/include/mlir/Dialect/StandardOps/IR/Ops.td:138
                    Confined<OptionalAttr<I64Attr>, [IntMinValue<0>]>:$alignment);
-  let results = (outs AnyMemRef);
+  let results = (outs Arg<AnyMemRef, "", resultDecorators>);
 
----------------
dfki-jugr wrote:
> mehdi_amini wrote:
> > dfki-jugr wrote:
> > > mehdi_amini wrote:
> > > > mehdi_amini wrote:
> > > > > Any reason this is not just always tagged "MemAlloc"?
> > > > Seems like we could take an optional string to tag the resource / pool for the alloc, but always mark the result as MemAlloc?
> > > Good point, this makes sense. We will address it in an follow up PR.
> > I'm curious why didn't address it here in the first place instead?
> We wanted to provide a more generalized approach for this. Follow this link to get to the follow up PR:  https://reviews.llvm.org/D78917
I still don't understand why the current revision needed to land first...


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D78619/new/

https://reviews.llvm.org/D78619