[Backend] Bump to llvm/llvm-project@ebf5d9e

This bump picks up llvm/llvm-project#169546
which resolves some register spilling in downstream use cases.

Additionally, we also pick up llvm/llvm-project#166648
which converted udiv and sdiv to non Pure attributes. This require changes on
warp_specialize_to_llvm.mlir `remat_subgraph` since the original test found in
triton-lang@8601b39
was meant to test rematerializing some "pure ops" into partition region,
which the tested udiv is no longer.
This PR Replaced it with urem which is similar class of op who still has Pure trait.

Signed-off-by: Stanley Winata <[email protected]>

Author: raikonenfnu

cmake/llvm-hash.txt

@@ -1 +1 @@
-49d5bb0ad0cb31410184c462801c5049ad671517
+ebf5d9ef7de29b55fd9e9d504f83689b4013e0de

test/Conversion/warp_specialize_to_llvm.mlir

@@ -653,7 +653,7 @@ llvm.func @remat_subgraph(%arg0: i32, %arg1: i32) attributes {allocation.offset
   %1 = llvm.getelementptr %0[%arg0] : (!llvm.ptr<3>, i32) -> !llvm.ptr<3>, i32
   %2 = llvm.add %arg0, %arg1 : i32
   %3 = llvm.mul %2, %arg1 : i32
-  %4 = llvm.udiv %2, %3 : i32
+  %4 = llvm.urem %2, %3 : i32
   ttg.warp_specialize(%1, %4) attributes {allocation.offset = 0 : i32, warpGroupStartIds = array<i32: 4>}
   default {
     ttg.warp_yield
@@ -663,9 +663,9 @@ llvm.func @remat_subgraph(%arg0: i32, %arg1: i32) attributes {allocation.offset
     // CHECK-NEXT: "llvm.nvvm.barrier.cta.sync.all"([[C1]])
     // CHECK-NEXT: [[ADD:%.*]] = llvm.add %arg0, %arg1 : i32
     // CHECK-NEXT: [[MUL:%.*]] = llvm.mul [[ADD]], %arg1 : i32
-    // CHECK-NEXT: [[UDIV:%.*]] = llvm.udiv [[ADD]], [[MUL]] : i32
+    // CHECK-NEXT: [[UREM:%.*]] = llvm.urem [[ADD]], [[MUL]] : i32
     // CHECK-NEXT: [[PTR:%.*]] = llvm.getelementptr [[ADDR]][%arg0]
-    // CHECK-NEXT: "use"([[PTR]], [[UDIV]])
+    // CHECK-NEXT: "use"([[PTR]], [[UREM]])
     "use"(%arg2, %arg3) : (!llvm.ptr<3>, i32) -> ()
     // CHECK-NEXT: "llvm.nvvm.barrier.cta.sync.all"([[C1]])
     ttg.warp_return