[PATCH] D55506: [RFC v2] Allow target to handle STRICT floating-point nodes

[Ulrich Weigand via Phabricator via llvm-commits]

uweigand added a comment.

In D55506#1558492 <https://reviews.llvm.org/D55506#1558492>, @wuzish wrote:

> Is there any further step to enable the rounding-mode infra to handle rounding mode metadata in intrinsics? For example, covert rounding mode metadata to a new constant operand of strict_* series op. If some target/platform does not have related machine instruction for different static rounding-mode, then ignore this constant in td selection.


If this is actually needed by some target, I agree that there are further steps required to propagate that information.  Instead of using a constant operand, I think it would probably be better to create a new subclass of SDNode (e.g. StrictFPSDNode or the like), and store this information in its associated bits, like e.g. access bits are stored for MemSDNodes.


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D55506/new/

https://reviews.llvm.org/D55506