[NVPTX] Promote v2i8 to v2i16 (#111189)

[Artem-B]

Promote v2i8 to v2i16, fixes a crash.
Re-enable a test in NVPTX/vector-returns.ll

Partial cherry-pick of fda2fea w/o the test which does not exist in release/19.x
#104864

[llvmbot]

@llvm/pr-subscribers-backend-nvptx

Author: Artem Belevich (Artem-B)

Changes

Promote v2i8 to v2i16, fixes a crash.
Re-enable a test in NVPTX/vector-returns.ll

Partial cherry-pick of fda2fea w/o the test which does not exist in release/19.x
#104864

---

Full diff: https://github.com/llvm/llvm-project/pull/115081.diff

1 Files Affected:

• (modified) llvm/lib/Target/NVPTX/NVPTXISelLowering.cpp (+4)

diff --git a/llvm/lib/Target/NVPTX/NVPTXISelLowering.cpp b/llvm/lib/Target/NVPTX/NVPTXISelLowering.cpp
index 6975412ce5d35b..b2153a7afe7365 100644
--- a/llvm/lib/Target/NVPTX/NVPTXISelLowering.cpp
+++ b/llvm/lib/Target/NVPTX/NVPTXISelLowering.cpp
@@ -229,6 +229,10 @@ static void ComputePTXValueVTs(const TargetLowering &TLI, const DataLayout &DL,
         // v*i8 are formally lowered as v4i8
         EltVT = MVT::v4i8;
         NumElts = (NumElts + 3) / 4;
+      } else if (EltVT.getSimpleVT() == MVT::i8 && NumElts == 2) {
+        // v2i8 is promoted to v2i16
+        NumElts = 1;
+        EltVT = MVT::v2i16;
       }
       for (unsigned j = 0; j != NumElts; ++j) {
         ValueVTs.push_back(EltVT);

[tru]

This seems to have some CI errors, Can you rebase on latest and see if the windows build errors are legit?

[Artem-B]

The CI check was unhappy because PR was based on the tree before 19.1.4. I've rebased it on top of the most recent commit.

[tru]

Who can review this?

[Artem-B]

That would be me -- I did the review of the original patch and just doing the legwork here to cherry-pick it into the 19.x branch.

If you need somebody else, I'd ask @AlexMaclean

[tru]

Can you back out the merge commit? It needs to be rebased in order to go in cleanly with our scripts.

[Artem-B]

Done.