Integrate llvm-project and bump dependencies. (iree-org#11741)

LLVM_COMMIT="7504e9a19346b33fd2e0fcdcbd43ecee3d676b86"
MHLO_COMMIT="ae3d50cd6999ab217de5093a3359ffab0a0f806a"
TF_COMMIT="27bbc6fb4049ff97b11c2222c7032004eb2ed330"

compiler/src/iree/compiler/Codegen/Common/FoldAffineMinInDistributedLoops.cpp

@@ -103,9 +103,7 @@ struct FoldAffineMinOverDistributedLoopInductionVariable final
       return success();
     };
 
-    return scf::canonicalizeMinMaxOpInLoop(
-        rewriter, minOp, minOp.getAffineMap(), minOp.getOperands(),
-        /*isMin=*/true, loopMatcher);
+    return scf::canonicalizeMinMaxOpInLoop(rewriter, minOp, loopMatcher);
   }
 };
 

integrations/tensorflow/WORKSPACE

@@ -7,7 +7,7 @@
 
 load("@bazel_tools//tools/build_defs/repo:git.bzl", "git_repository")
 
-TENSORFLOW_COMMIT = "3afa34d76e61cdd1de0d70376879e3f45d7fab05"
+TENSORFLOW_COMMIT = "27bbc6fb4049ff97b11c2222c7032004eb2ed330"
 
 git_repository(
     name = "org_tensorflow",

tests/e2e/tosa_ops/max_pool.mlir

@@ -1,27 +1,27 @@
 func.func @tensor_i8() {
   %0 = util.unfoldable_constant dense<[[[[1], [2], [3], [4]], [[5], [6], [7], [8]]]]> : tensor<1x2x4x1xi8>
-  %result = "tosa.max_pool2d"(%0) {kernel = [2, 2], stride = [1, 1], pad = [0, 0, 0, 0]} : (tensor<1x2x4x1xi8>) -> tensor<1x1x3x1xi8>
+  %result = "tosa.max_pool2d"(%0) {kernel = array<i64: 2, 2>, stride = array<i64: 1, 1>, pad = array<i64: 0, 0, 0, 0>} : (tensor<1x2x4x1xi8>) -> tensor<1x1x3x1xi8>
   check.expect_eq_const(%result, dense<[[[[6], [7], [8]]]]> : tensor<1x1x3x1xi8>) : tensor<1x1x3x1xi8>
   return
 }
 
 func.func @tensor_i16() {
   %0 = util.unfoldable_constant dense<[[[[1], [2], [3], [4]], [[5], [6], [7], [8]]]]> : tensor<1x2x4x1xi16>
-  %result = "tosa.max_pool2d"(%0) {kernel = [2, 2], stride = [1, 1], pad = [0, 0, 0, 0]} : (tensor<1x2x4x1xi16>) -> tensor<1x1x3x1xi16>
+  %result = "tosa.max_pool2d"(%0) {kernel = array<i64: 2, 2>, stride = array<i64: 1, 1>, pad = array<i64: 0, 0, 0, 0>} : (tensor<1x2x4x1xi16>) -> tensor<1x1x3x1xi16>
   check.expect_eq_const(%result, dense<[[[[6], [7], [8]]]]> : tensor<1x1x3x1xi16>) : tensor<1x1x3x1xi16>
   return
 }
 
 func.func @tensor_i32() {
   %0 = util.unfoldable_constant dense<[[[[1], [2], [3], [4]], [[5], [6], [7], [8]]]]> : tensor<1x2x4x1xi32>
-  %result = "tosa.max_pool2d"(%0) {kernel = [2, 2], stride = [1, 1], pad = [0, 0, 0, 0]} : (tensor<1x2x4x1xi32>) -> tensor<1x1x3x1xi32>
+  %result = "tosa.max_pool2d"(%0) {kernel = array<i64: 2, 2>, stride = array<i64: 1, 1>, pad = array<i64: 0, 0, 0, 0>} : (tensor<1x2x4x1xi32>) -> tensor<1x1x3x1xi32>
   check.expect_eq_const(%result, dense<[[[[6], [7], [8]]]]> : tensor<1x1x3x1xi32>) : tensor<1x1x3x1xi32>
   return
 }
 
 func.func @tensor_f32() {
   %0 = util.unfoldable_constant dense<[[[[1.], [2.], [3.], [4.]], [[5.], [6.], [7.], [8.]]]]> : tensor<1x2x4x1xf32>
-  %result = "tosa.max_pool2d"(%0) {kernel = [2, 2], stride = [1, 1], pad = [0, 0, 0, 0]} : (tensor<1x2x4x1xf32>) -> tensor<1x1x3x1xf32>
+  %result = "tosa.max_pool2d"(%0) {kernel = array<i64: 2, 2>, stride = array<i64: 1, 1>, pad = array<i64: 0, 0, 0, 0>} : (tensor<1x2x4x1xf32>) -> tensor<1x1x3x1xf32>
   check.expect_eq_const(%result, dense<[[[[6.], [7.], [8.]]]]> : tensor<1x1x3x1xf32>) : tensor<1x1x3x1xf32>
   return
 }