Skip to content

Instantly share code, notes, and snippets.

@bjacob

bjacob/a.ll

Created March 20, 2022 02:38
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save bjacob/b52d34f1d04cb80e44c85f936a5d780e to your computer and use it in GitHub Desktop.
Save bjacob/b52d34f1d04cb80e44c85f936a5d780e to your computer and use it in GitHub Desktop.
Download ZIP
Raw
a.ll
running ScalarizerPass on function:
; Function Attrs: nofree nosync nounwind sanitize_thread
define internal i32 @_sort3D_dispatch_0(%iree_hal_executable_environment_v0_t* noalias nocapture nonnull readnone byref(%iree_hal_executable_environment_v0_t) align 16 %0, %iree_hal_executable_dispatch_state_v0_t* noalias nocapture nonnull readonly byref(%iree_hal_executable_dispatch_state_v0_t) align 16 %1, %iree_hal_executable_workgroup_state_v0_t* noalias nocapture nonnull readonly byref(%iree_hal_executable_workgroup_state_v0_t) align 16 %2) #0 !dbg !17 {
%4 = call i8* @llvm.returnaddress(i32 0), !dbg !18
call void @__tsan_func_entry(i8* %4), !dbg !18
%.elt11 = getelementptr inbounds %iree_hal_executable_dispatch_state_v0_t, %iree_hal_executable_dispatch_state_v0_t* %1, i64 0, i32 4, !dbg !18
%5 = bitcast i32* %.elt11 to i8*, !dbg !18
call void @__tsan_read4(i8* %5), !dbg !18
%.unpack12 = load i32, i32* %.elt11, align 4, !dbg !18
%.elt13 = getelementptr inbounds %iree_hal_executable_dispatch_state_v0_t, %iree_hal_executable_dispatch_state_v0_t* %1, i64 0, i32 5, !dbg !18
%6 = bitcast i32* %.elt13 to i8*, !dbg !18
call void @__tsan_read4(i8* %6), !dbg !18
%.unpack14 = load i32, i32* %.elt13, align 16, !dbg !18
%.elt21 = getelementptr inbounds %iree_hal_executable_dispatch_state_v0_t, %iree_hal_executable_dispatch_state_v0_t* %1, i64 0, i32 9, !dbg !18
%7 = bitcast i8*** %.elt21 to i32***, !dbg !18
%8 = bitcast i32*** %7 to i8*, !dbg !18
call void @__tsan_read8(i8* %8), !dbg !18
%.unpack2225 = load i32**, i32*** %7, align 16, !dbg !18
%9 = bitcast i32** %.unpack2225 to i8*, !dbg !18
call void @__tsan_read8(i8* %9), !dbg !18
%10 = load i32*, i32** %.unpack2225, align 8, !dbg !18
%11 = ptrtoint i32* %10 to i64, !dbg !18
%12 = and i64 %11, 63, !dbg !18
%13 = icmp eq i64 %12, 0, !dbg !18
tail call void @llvm.assume(i1 %13), !dbg !18
%14 = load %iree_hal_executable_workgroup_state_v0_t, %iree_hal_executable_workgroup_state_v0_t* %2, align 16, !dbg !20
%15 = extractvalue %iree_hal_executable_workgroup_state_v0_t %14, 1, !dbg !20
%16 = icmp eq i32 %15, 0, !dbg !20
br i1 %16, label %.lr.ph30, label %._crit_edge, !dbg !20
.lr.ph30: ; preds = %3
%17 = zext i32 %.unpack12 to i64, !dbg !20
%18 = extractvalue %iree_hal_executable_workgroup_state_v0_t %14, 0, !dbg !20
%19 = zext i32 %18 to i64, !dbg !20
%20 = shl nuw nsw i64 %19, 6
%21 = shl nuw nsw i64 %17, 6
%22 = icmp eq i32 %18, 0
%23 = icmp eq i32 %.unpack14, 0
br i1 %22, label %.lr.ph30.split.us, label %.lr.ph30.split, !dbg !20
.lr.ph30.split.us: ; preds = %.lr.ph30
%24 = sub nsw i64 1, %20, !dbg !20
%25 = mul nsw i64 %17, -64, !dbg !20
%26 = shl nuw nsw i64 %19, 8, !dbg !20
%27 = or i64 %26, 3, !dbg !20
%28 = shl nuw nsw i64 %17, 8, !dbg !20
%29 = or i64 %26, 2, !dbg !20
%30 = or i64 %26, 1, !dbg !20
%31 = sub nsw i64 2, %20, !dbg !20
%32 = mul nsw i64 %17, -64, !dbg !20
br i1 %23, label %.lr.ph.us.us, label %.lr.ph.us, !dbg !20
.lr.ph.us.us: ; preds = %.split29.us31.us, %.lr.ph30.split.us
br label %33, !dbg !20
33: ; preds = %.split29.us31.us, %.lr.ph.us.us
%indvar75 = phi i64 [ %indvar.next76, %.split29.us31.us ], [ 0, %.lr.ph.us.us ]
%34 = phi i64 [ %104, %.split29.us31.us ], [ %20, %.lr.ph.us.us ]
%35 = mul i64 %32, %indvar75, !dbg !20
%36 = add i64 %31, %35, !dbg !20
%37 = mul i64 %25, %indvar75, !dbg !20
%38 = add i64 %24, %37, !dbg !20
%39 = mul i64 %28, %indvar75, !dbg !20
%40 = add i64 %27, %39, !dbg !20
%scevgep77 = getelementptr i32, i32* %10, i64 %40, !dbg !20
%scevgep7781 = bitcast i32* %scevgep77 to i8*, !dbg !20
%41 = add i64 %29, %39, !dbg !20
%scevgep82 = getelementptr i32, i32* %10, i64 %41, !dbg !20
%scevgep8286 = bitcast i32* %scevgep82 to i8*, !dbg !20
%42 = add i64 %30, %39, !dbg !20
%scevgep87 = getelementptr i32, i32* %10, i64 %42, !dbg !20
%scevgep8791 = bitcast i32* %scevgep87 to i8*, !dbg !20
%43 = add i64 %26, %39, !dbg !20
%scevgep92 = getelementptr i32, i32* %10, i64 %43, !dbg !20
%scevgep9296 = bitcast i32* %scevgep92 to i8*, !dbg !20
%44 = sub i64 2, %34, !dbg !20
%45 = icmp sgt i64 %44, 0
br i1 %45, label %.preheader26.us.us.us.preheader, label %.split29.us31.us, !dbg !20
.preheader26.us.us.us.preheader: ; preds = %33
%min.iters.check100 = icmp ult i64 %36, 16, !dbg !20
br i1 %min.iters.check100, label %.preheader26.us.us.us.preheader138, label %vector.scevcheck74, !dbg !20
.preheader26.us.us.us.preheader138: ; preds = %vector.body99, %vector.scevcheck74, %.preheader26.us.us.us.preheader
%.ph = phi i64 [ 0, %vector.scevcheck74 ], [ 0, %.preheader26.us.us.us.preheader ], [ %n.vec103, %vector.body99 ]
br label %.preheader26.us.us.us, !dbg !20
vector.scevcheck74: ; preds = %.preheader26.us.us.us.preheader
%mul78 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 %38, i64 16), !dbg !20
%mul.result79 = extractvalue { i64, i1 } %mul78, 0, !dbg !20
%mul.overflow80 = extractvalue { i64, i1 } %mul78, 1, !dbg !20
%46 = getelementptr i8, i8* %scevgep7781, i64 %mul.result79, !dbg !20
%47 = icmp ult i8* %46, %scevgep7781, !dbg !20
%48 = or i1 %47, %mul.overflow80, !dbg !20
%mul83 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 %38, i64 16), !dbg !20
%mul.result84 = extractvalue { i64, i1 } %mul83, 0, !dbg !20
%mul.overflow85 = extractvalue { i64, i1 } %mul83, 1, !dbg !20
%49 = getelementptr i8, i8* %scevgep8286, i64 %mul.result84, !dbg !20
%50 = icmp ult i8* %49, %scevgep8286, !dbg !20
%51 = or i1 %50, %mul.overflow85, !dbg !20
%mul88 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 %38, i64 16), !dbg !20
%mul.result89 = extractvalue { i64, i1 } %mul88, 0, !dbg !20
%mul.overflow90 = extractvalue { i64, i1 } %mul88, 1, !dbg !20
%52 = getelementptr i8, i8* %scevgep8791, i64 %mul.result89, !dbg !20
%53 = icmp ult i8* %52, %scevgep8791, !dbg !20
%54 = or i1 %53, %mul.overflow90, !dbg !20
%mul93 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 %38, i64 16), !dbg !20
%mul.result94 = extractvalue { i64, i1 } %mul93, 0, !dbg !20
%mul.overflow95 = extractvalue { i64, i1 } %mul93, 1, !dbg !20
%55 = getelementptr i8, i8* %scevgep9296, i64 %mul.result94, !dbg !20
%56 = icmp ult i8* %55, %scevgep9296, !dbg !20
%57 = or i1 %56, %mul.overflow95, !dbg !20
%58 = or i1 %48, %51, !dbg !20
%59 = or i1 %58, %54, !dbg !20
%60 = or i1 %59, %57, !dbg !20
br i1 %60, label %.preheader26.us.us.us.preheader138, label %vector.ph101
vector.ph101: ; preds = %vector.scevcheck74
%n.vec103 = add nsw i64 %36, -2, !dbg !20
%broadcast.splatinsert109 = insertelement <16 x i64> poison, i64 %34, i64 0, !dbg !20
%broadcast.splat110 = shufflevector <16 x i64> %broadcast.splatinsert109, <16 x i64> poison, <16 x i32> zeroinitializer, !dbg !20
br label %vector.body99, !dbg !20
vector.body99: ; preds = %vector.body99, %vector.ph101
%index106 = phi i64 [ 0, %vector.ph101 ], [ %index.next136, %vector.body99 ], !dbg !20
%vec.ind107 = phi <16 x i64> [ <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15>, %vector.ph101 ], [ %vec.ind.next108, %vector.body99 ]
%61 = add <16 x i64> %broadcast.splat110, %vec.ind107
%62 = shl <16 x i64> %61, <i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2>
%63 = getelementptr i32, i32* %10, <16 x i64> %62
%bc137 = bitcast <16 x i32*> %63 to <16 x <64 x i32>*>
%64 = extractelement <16 x <64 x i32>*> %bc137, i64 0
%wide.vec111 = load <64 x i32>, <64 x i32>* %64, align 4, !dbg !20
%strided.vec112 = shufflevector <64 x i32> %wide.vec111, <64 x i32> poison, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24, i32 28, i32 32, i32 36, i32 40, i32 44, i32 48, i32 52, i32 56, i32 60>, !dbg !20
%strided.vec113 = shufflevector <64 x i32> %wide.vec111, <64 x i32> poison, <16 x i32> <i32 1, i32 5, i32 9, i32 13, i32 17, i32 21, i32 25, i32 29, i32 33, i32 37, i32 41, i32 45, i32 49, i32 53, i32 57, i32 61>, !dbg !20
%strided.vec114 = shufflevector <64 x i32> %wide.vec111, <64 x i32> poison, <16 x i32> <i32 2, i32 6, i32 10, i32 14, i32 18, i32 22, i32 26, i32 30, i32 34, i32 38, i32 42, i32 46, i32 50, i32 54, i32 58, i32 62>, !dbg !20
%strided.vec115 = shufflevector <64 x i32> %wide.vec111, <64 x i32> poison, <16 x i32> <i32 3, i32 7, i32 11, i32 15, i32 19, i32 23, i32 27, i32 31, i32 35, i32 39, i32 43, i32 47, i32 51, i32 55, i32 59, i32 63>, !dbg !20
%65 = or <16 x i64> %62, <i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1>, !dbg !20
%66 = getelementptr i32, i32* %10, <16 x i64> %65, !dbg !20
%67 = icmp sge <16 x i32> %strided.vec112, %strided.vec113, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %strided.vec113, <16 x i32*> %63, i32 16, <16 x i1> %67), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %strided.vec112, <16 x i32*> %66, i32 4, <16 x i1> %67), !dbg !20
%68 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %strided.vec112, <16 x i32> %strided.vec113)
%69 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %strided.vec112, <16 x i32> %strided.vec113), !dbg !20
%70 = or <16 x i64> %62, <i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2>, !dbg !20
%71 = getelementptr i32, i32* %10, <16 x i64> %70, !dbg !20
%72 = icmp sge <16 x i32> %69, %strided.vec114, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %strided.vec114, <16 x i32*> %66, i32 4, <16 x i1> %72), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %69, <16 x i32*> %71, i32 8, <16 x i1> %72), !dbg !20
%73 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %69, <16 x i32> %strided.vec114)
%74 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %69, <16 x i32> %strided.vec114), !dbg !20
%75 = or <16 x i64> %62, <i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3>, !dbg !20
%76 = getelementptr i32, i32* %10, <16 x i64> %75, !dbg !20
%77 = icmp sge <16 x i32> %74, %strided.vec115, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %strided.vec115, <16 x i32*> %71, i32 8, <16 x i1> %77), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %74, <16 x i32*> %76, i32 4, <16 x i1> %77), !dbg !20
%78 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %74, <16 x i32> %strided.vec115)
%79 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %74, <16 x i32> %strided.vec115)
%80 = icmp sge <16 x i32> %68, %73, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %73, <16 x i32*> %63, i32 16, <16 x i1> %80), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %68, <16 x i32*> %66, i32 4, <16 x i1> %80), !dbg !20
%81 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %68, <16 x i32> %73)
%82 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %68, <16 x i32> %73), !dbg !20
%83 = icmp sge <16 x i32> %82, %79, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %79, <16 x i32*> %66, i32 4, <16 x i1> %83), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %82, <16 x i32*> %71, i32 8, <16 x i1> %83), !dbg !20
%84 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %82, <16 x i32> %79)
%85 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %82, <16 x i32> %79), !dbg !20
%86 = icmp sge <16 x i32> %85, %78, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %78, <16 x i32*> %71, i32 8, <16 x i1> %86), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %85, <16 x i32*> %76, i32 4, <16 x i1> %86), !dbg !20
%87 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %85, <16 x i32> %78)
%88 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %85, <16 x i32> %78)
%89 = icmp sge <16 x i32> %81, %84, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %84, <16 x i32*> %63, i32 16, <16 x i1> %89), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %81, <16 x i32*> %66, i32 4, <16 x i1> %89), !dbg !20
%90 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %81, <16 x i32> %84)
%91 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %81, <16 x i32> %84), !dbg !20
%92 = icmp sge <16 x i32> %91, %88, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %88, <16 x i32*> %66, i32 4, <16 x i1> %92), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %91, <16 x i32*> %71, i32 8, <16 x i1> %92), !dbg !20
%93 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %91, <16 x i32> %88)
%94 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %91, <16 x i32> %88), !dbg !20
%95 = icmp sge <16 x i32> %94, %87, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %87, <16 x i32*> %71, i32 8, <16 x i1> %95), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %94, <16 x i32*> %76, i32 4, <16 x i1> %95), !dbg !20
%96 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %94, <16 x i32> %87)
%97 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %94, <16 x i32> %87)
%98 = icmp sge <16 x i32> %90, %93, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %93, <16 x i32*> %63, i32 16, <16 x i1> %98), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %90, <16 x i32*> %66, i32 4, <16 x i1> %98), !dbg !20
%99 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %90, <16 x i32> %93), !dbg !20
%100 = icmp sge <16 x i32> %99, %97, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %97, <16 x i32*> %66, i32 4, <16 x i1> %100), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %99, <16 x i32*> %71, i32 8, <16 x i1> %100), !dbg !20
%101 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %99, <16 x i32> %97), !dbg !20
%102 = icmp sge <16 x i32> %101, %96, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %96, <16 x i32*> %71, i32 8, <16 x i1> %102), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %101, <16 x i32*> %76, i32 4, <16 x i1> %102), !dbg !20
%index.next136 = add nuw i64 %index106, 16, !dbg !20
%vec.ind.next108 = add <16 x i64> %vec.ind107, <i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16>
%103 = icmp eq i64 %index.next136, %n.vec103, !dbg !20
br i1 %103, label %.preheader26.us.us.us.preheader138, label %vector.body99, !dbg !20, !llvm.loop !22
.split29.us31.us: ; preds = %189, %33
%104 = add i64 %34, %21, !dbg !20
%105 = icmp slt i64 %104, 2, !dbg !20
%indvar.next76 = add i64 %indvar75, 1, !dbg !20
br i1 %105, label %33, label %.lr.ph.us.us, !dbg !20
106: ; preds = %.preheader26.us.us.us
%107 = bitcast i32* %195 to i8*, !dbg !20
call void @__tsan_write4(i8* %107), !dbg !20
store i32 %201, i32* %195, align 16, !dbg !20
%108 = bitcast i32* %199 to i8*, !dbg !20
call void @__tsan_write4(i8* %108), !dbg !20
store i32 %197, i32* %199, align 4, !dbg !20
br label %109, !dbg !20
109: ; preds = %106, %.preheader26.us.us.us
%110 = phi i32 [ %201, %106 ], [ %197, %.preheader26.us.us.us ]
%111 = phi i32 [ %197, %106 ], [ %201, %.preheader26.us.us.us ], !dbg !20
%112 = or i64 %194, 2, !dbg !20
%113 = getelementptr i32, i32* %10, i64 %112, !dbg !20
%114 = bitcast i32* %113 to i8*, !dbg !20
call void @__tsan_read4(i8* %114), !dbg !20
%115 = load i32, i32* %113, align 8, !dbg !20
%116 = icmp slt i32 %111, %115, !dbg !21
br i1 %116, label %120, label %117, !dbg !20
117: ; preds = %109
%118 = bitcast i32* %199 to i8*, !dbg !20
call void @__tsan_write4(i8* %118), !dbg !20
store i32 %115, i32* %199, align 4, !dbg !20
%119 = bitcast i32* %113 to i8*, !dbg !20
call void @__tsan_write4(i8* %119), !dbg !20
store i32 %111, i32* %113, align 8, !dbg !20
br label %120, !dbg !20
120: ; preds = %117, %109
%121 = phi i32 [ %115, %117 ], [ %111, %109 ]
%122 = phi i32 [ %111, %117 ], [ %115, %109 ], !dbg !20
%123 = or i64 %194, 3, !dbg !20
%124 = getelementptr i32, i32* %10, i64 %123, !dbg !20
%125 = bitcast i32* %124 to i8*, !dbg !20
call void @__tsan_read4(i8* %125), !dbg !20
%126 = load i32, i32* %124, align 4, !dbg !20
%127 = icmp slt i32 %122, %126, !dbg !21
br i1 %127, label %.preheader.us.us.us.1, label %128, !dbg !20
128: ; preds = %120
%129 = bitcast i32* %113 to i8*, !dbg !20
call void @__tsan_write4(i8* %129), !dbg !20
store i32 %126, i32* %113, align 8, !dbg !20
%130 = bitcast i32* %124 to i8*, !dbg !20
call void @__tsan_write4(i8* %130), !dbg !20
store i32 %122, i32* %124, align 4, !dbg !20
br label %.preheader.us.us.us.1, !dbg !20
.preheader.us.us.us.1: ; preds = %128, %120
%131 = phi i32 [ %122, %128 ], [ %126, %120 ]
%132 = phi i32 [ %126, %128 ], [ %122, %120 ]
%133 = icmp slt i32 %110, %121, !dbg !21
br i1 %133, label %137, label %134, !dbg !20
134: ; preds = %.preheader.us.us.us.1
%135 = bitcast i32* %195 to i8*, !dbg !20
call void @__tsan_write4(i8* %135), !dbg !20
store i32 %121, i32* %195, align 16, !dbg !20
%136 = bitcast i32* %199 to i8*, !dbg !20
call void @__tsan_write4(i8* %136), !dbg !20
store i32 %110, i32* %199, align 4, !dbg !20
br label %137, !dbg !20
137: ; preds = %134, %.preheader.us.us.us.1
%138 = phi i32 [ %121, %134 ], [ %110, %.preheader.us.us.us.1 ]
%139 = phi i32 [ %110, %134 ], [ %121, %.preheader.us.us.us.1 ], !dbg !20
%140 = icmp slt i32 %139, %132, !dbg !21
br i1 %140, label %144, label %141, !dbg !20
141: ; preds = %137
%142 = bitcast i32* %199 to i8*, !dbg !20
call void @__tsan_write4(i8* %142), !dbg !20
store i32 %132, i32* %199, align 4, !dbg !20
%143 = bitcast i32* %113 to i8*, !dbg !20
call void @__tsan_write4(i8* %143), !dbg !20
store i32 %139, i32* %113, align 8, !dbg !20
br label %144, !dbg !20
144: ; preds = %141, %137
%145 = phi i32 [ %132, %141 ], [ %139, %137 ]
%146 = phi i32 [ %139, %141 ], [ %132, %137 ], !dbg !20
%147 = icmp slt i32 %146, %131, !dbg !21
br i1 %147, label %.preheader.us.us.us.2, label %148, !dbg !20
148: ; preds = %144
%149 = bitcast i32* %113 to i8*, !dbg !20
call void @__tsan_write4(i8* %149), !dbg !20
store i32 %131, i32* %113, align 8, !dbg !20
%150 = bitcast i32* %124 to i8*, !dbg !20
call void @__tsan_write4(i8* %150), !dbg !20
store i32 %146, i32* %124, align 4, !dbg !20
br label %.preheader.us.us.us.2, !dbg !20
.preheader.us.us.us.2: ; preds = %148, %144
%151 = phi i32 [ %146, %148 ], [ %131, %144 ]
%152 = phi i32 [ %131, %148 ], [ %146, %144 ]
%153 = icmp slt i32 %138, %145, !dbg !21
br i1 %153, label %157, label %154, !dbg !20
154: ; preds = %.preheader.us.us.us.2
%155 = bitcast i32* %195 to i8*, !dbg !20
call void @__tsan_write4(i8* %155), !dbg !20
store i32 %145, i32* %195, align 16, !dbg !20
%156 = bitcast i32* %199 to i8*, !dbg !20
call void @__tsan_write4(i8* %156), !dbg !20
store i32 %138, i32* %199, align 4, !dbg !20
br label %157, !dbg !20
157: ; preds = %154, %.preheader.us.us.us.2
%158 = phi i32 [ %145, %154 ], [ %138, %.preheader.us.us.us.2 ]
%159 = phi i32 [ %138, %154 ], [ %145, %.preheader.us.us.us.2 ], !dbg !20
%160 = icmp slt i32 %159, %152, !dbg !21
br i1 %160, label %164, label %161, !dbg !20
161: ; preds = %157
%162 = bitcast i32* %199 to i8*, !dbg !20
call void @__tsan_write4(i8* %162), !dbg !20
store i32 %152, i32* %199, align 4, !dbg !20
%163 = bitcast i32* %113 to i8*, !dbg !20
call void @__tsan_write4(i8* %163), !dbg !20
store i32 %159, i32* %113, align 8, !dbg !20
br label %164, !dbg !20
164: ; preds = %161, %157
%165 = phi i32 [ %152, %161 ], [ %159, %157 ]
%166 = phi i32 [ %159, %161 ], [ %152, %157 ], !dbg !20
%167 = icmp slt i32 %166, %151, !dbg !21
br i1 %167, label %.preheader.us.us.us.3, label %168, !dbg !20
168: ; preds = %164
%169 = bitcast i32* %113 to i8*, !dbg !20
call void @__tsan_write4(i8* %169), !dbg !20
store i32 %151, i32* %113, align 8, !dbg !20
%170 = bitcast i32* %124 to i8*, !dbg !20
call void @__tsan_write4(i8* %170), !dbg !20
store i32 %166, i32* %124, align 4, !dbg !20
br label %.preheader.us.us.us.3, !dbg !20
.preheader.us.us.us.3: ; preds = %168, %164
%171 = phi i32 [ %166, %168 ], [ %151, %164 ]
%172 = phi i32 [ %151, %168 ], [ %166, %164 ]
%173 = icmp slt i32 %158, %165, !dbg !21
br i1 %173, label %177, label %174, !dbg !20
174: ; preds = %.preheader.us.us.us.3
%175 = bitcast i32* %195 to i8*, !dbg !20
call void @__tsan_write4(i8* %175), !dbg !20
store i32 %165, i32* %195, align 16, !dbg !20
%176 = bitcast i32* %199 to i8*, !dbg !20
call void @__tsan_write4(i8* %176), !dbg !20
store i32 %158, i32* %199, align 4, !dbg !20
br label %177, !dbg !20
177: ; preds = %174, %.preheader.us.us.us.3
%178 = phi i32 [ %158, %174 ], [ %165, %.preheader.us.us.us.3 ], !dbg !20
%179 = icmp slt i32 %178, %172, !dbg !21
br i1 %179, label %183, label %180, !dbg !20
180: ; preds = %177
%181 = bitcast i32* %199 to i8*, !dbg !20
call void @__tsan_write4(i8* %181), !dbg !20
store i32 %172, i32* %199, align 4, !dbg !20
%182 = bitcast i32* %113 to i8*, !dbg !20
call void @__tsan_write4(i8* %182), !dbg !20
store i32 %178, i32* %113, align 8, !dbg !20
br label %183, !dbg !20
183: ; preds = %180, %177
%184 = phi i32 [ %178, %180 ], [ %172, %177 ], !dbg !20
%185 = icmp slt i32 %184, %171, !dbg !21
br i1 %185, label %189, label %186, !dbg !20
186: ; preds = %183
%187 = bitcast i32* %113 to i8*, !dbg !20
call void @__tsan_write4(i8* %187), !dbg !20
store i32 %171, i32* %113, align 8, !dbg !20
%188 = bitcast i32* %124 to i8*, !dbg !20
call void @__tsan_write4(i8* %188), !dbg !20
store i32 %184, i32* %124, align 4, !dbg !20
br label %189, !dbg !20
189: ; preds = %186, %183
%190 = add nuw nsw i64 %192, 1, !dbg !20
%191 = icmp slt i64 %190, %44, !dbg !20
br i1 %191, label %.preheader26.us.us.us, label %.split29.us31.us, !dbg !20, !llvm.loop !24
.preheader26.us.us.us: ; preds = %.preheader26.us.us.us.preheader138, %189
%192 = phi i64 [ %190, %189 ], [ %.ph, %.preheader26.us.us.us.preheader138 ]
%193 = add i64 %34, %192
%194 = shl i64 %193, 2
%195 = getelementptr i32, i32* %10, i64 %194, !dbg !20
%196 = bitcast i32* %195 to i8*, !dbg !20
call void @__tsan_read4(i8* %196), !dbg !20
%197 = load i32, i32* %195, align 16, !dbg !20
%198 = or i64 %194, 1, !dbg !20
%199 = getelementptr i32, i32* %10, i64 %198, !dbg !20
%200 = bitcast i32* %199 to i8*, !dbg !20
call void @__tsan_read4(i8* %200), !dbg !20
%201 = load i32, i32* %199, align 4, !dbg !20
%202 = icmp slt i32 %197, %201, !dbg !21
br i1 %202, label %109, label %106, !dbg !20
.lr.ph.us: ; preds = %.lr.ph30.split.us, %.split29.us31
%indvar = phi i64 [ %indvar.next, %.split29.us31 ], [ 0, %.lr.ph30.split.us ]
%203 = phi i64 [ %273, %.split29.us31 ], [ %20, %.lr.ph30.split.us ]
%204 = mul i64 %32, %indvar, !dbg !20
%205 = add i64 %31, %204, !dbg !20
%206 = mul i64 %25, %indvar, !dbg !20
%207 = add i64 %24, %206, !dbg !20
%208 = mul i64 %28, %indvar, !dbg !20
%209 = add i64 %27, %208, !dbg !20
%scevgep = getelementptr i32, i32* %10, i64 %209, !dbg !20
%scevgep36 = bitcast i32* %scevgep to i8*, !dbg !20
%210 = add i64 %29, %208, !dbg !20
%scevgep37 = getelementptr i32, i32* %10, i64 %210, !dbg !20
%scevgep3741 = bitcast i32* %scevgep37 to i8*, !dbg !20
%211 = add i64 %30, %208, !dbg !20
%scevgep42 = getelementptr i32, i32* %10, i64 %211, !dbg !20
%scevgep4246 = bitcast i32* %scevgep42 to i8*, !dbg !20
%212 = add i64 %26, %208, !dbg !20
%scevgep47 = getelementptr i32, i32* %10, i64 %212, !dbg !20
%scevgep4751 = bitcast i32* %scevgep47 to i8*, !dbg !20
%213 = sub i64 2, %203, !dbg !20
%214 = icmp sgt i64 %213, 0
br i1 %214, label %.preheader26.us.us.preheader, label %.split29.us31, !dbg !20
.preheader26.us.us.preheader: ; preds = %.lr.ph.us
%min.iters.check = icmp ult i64 %205, 16, !dbg !20
br i1 %min.iters.check, label %.preheader26.us.us.preheader139, label %vector.scevcheck, !dbg !20
.preheader26.us.us.preheader139: ; preds = %vector.body, %vector.scevcheck, %.preheader26.us.us.preheader
%.ph140 = phi i64 [ 0, %vector.scevcheck ], [ 0, %.preheader26.us.us.preheader ], [ %n.vec, %vector.body ]
br label %.preheader26.us.us, !dbg !20
vector.scevcheck: ; preds = %.preheader26.us.us.preheader
%mul = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 %207, i64 16), !dbg !20
%mul.result = extractvalue { i64, i1 } %mul, 0, !dbg !20
%mul.overflow = extractvalue { i64, i1 } %mul, 1, !dbg !20
%215 = getelementptr i8, i8* %scevgep36, i64 %mul.result, !dbg !20
%216 = icmp ult i8* %215, %scevgep36, !dbg !20
%217 = or i1 %216, %mul.overflow, !dbg !20
%mul38 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 %207, i64 16), !dbg !20
%mul.result39 = extractvalue { i64, i1 } %mul38, 0, !dbg !20
%mul.overflow40 = extractvalue { i64, i1 } %mul38, 1, !dbg !20
%218 = getelementptr i8, i8* %scevgep3741, i64 %mul.result39, !dbg !20
%219 = icmp ult i8* %218, %scevgep3741, !dbg !20
%220 = or i1 %219, %mul.overflow40, !dbg !20
%mul43 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 %207, i64 16), !dbg !20
%mul.result44 = extractvalue { i64, i1 } %mul43, 0, !dbg !20
%mul.overflow45 = extractvalue { i64, i1 } %mul43, 1, !dbg !20
%221 = getelementptr i8, i8* %scevgep4246, i64 %mul.result44, !dbg !20
%222 = icmp ult i8* %221, %scevgep4246, !dbg !20
%223 = or i1 %222, %mul.overflow45, !dbg !20
%mul48 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 %207, i64 16), !dbg !20
%mul.result49 = extractvalue { i64, i1 } %mul48, 0, !dbg !20
%mul.overflow50 = extractvalue { i64, i1 } %mul48, 1, !dbg !20
%224 = getelementptr i8, i8* %scevgep4751, i64 %mul.result49, !dbg !20
%225 = icmp ult i8* %224, %scevgep4751, !dbg !20
%226 = or i1 %225, %mul.overflow50, !dbg !20
%227 = or i1 %217, %220, !dbg !20
%228 = or i1 %227, %223, !dbg !20
%229 = or i1 %228, %226, !dbg !20
br i1 %229, label %.preheader26.us.us.preheader139, label %vector.ph
vector.ph: ; preds = %vector.scevcheck
%n.vec = add nsw i64 %205, -2, !dbg !20
%broadcast.splatinsert = insertelement <16 x i64> poison, i64 %203, i64 0, !dbg !20
%broadcast.splat = shufflevector <16 x i64> %broadcast.splatinsert, <16 x i64> poison, <16 x i32> zeroinitializer, !dbg !20
br label %vector.body, !dbg !20
vector.body: ; preds = %vector.body, %vector.ph
%index = phi i64 [ 0, %vector.ph ], [ %index.next, %vector.body ], !dbg !20
%vec.ind = phi <16 x i64> [ <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15>, %vector.ph ], [ %vec.ind.next, %vector.body ]
%230 = add <16 x i64> %broadcast.splat, %vec.ind
%231 = shl <16 x i64> %230, <i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2>
%232 = getelementptr i32, i32* %10, <16 x i64> %231
%bc = bitcast <16 x i32*> %232 to <16 x <64 x i32>*>
%233 = extractelement <16 x <64 x i32>*> %bc, i64 0
%wide.vec = load <64 x i32>, <64 x i32>* %233, align 4, !dbg !20
%strided.vec = shufflevector <64 x i32> %wide.vec, <64 x i32> poison, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24, i32 28, i32 32, i32 36, i32 40, i32 44, i32 48, i32 52, i32 56, i32 60>, !dbg !20
%strided.vec52 = shufflevector <64 x i32> %wide.vec, <64 x i32> poison, <16 x i32> <i32 1, i32 5, i32 9, i32 13, i32 17, i32 21, i32 25, i32 29, i32 33, i32 37, i32 41, i32 45, i32 49, i32 53, i32 57, i32 61>, !dbg !20
%strided.vec53 = shufflevector <64 x i32> %wide.vec, <64 x i32> poison, <16 x i32> <i32 2, i32 6, i32 10, i32 14, i32 18, i32 22, i32 26, i32 30, i32 34, i32 38, i32 42, i32 46, i32 50, i32 54, i32 58, i32 62>, !dbg !20
%strided.vec54 = shufflevector <64 x i32> %wide.vec, <64 x i32> poison, <16 x i32> <i32 3, i32 7, i32 11, i32 15, i32 19, i32 23, i32 27, i32 31, i32 35, i32 39, i32 43, i32 47, i32 51, i32 55, i32 59, i32 63>, !dbg !20
%234 = or <16 x i64> %231, <i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1>, !dbg !20
%235 = getelementptr i32, i32* %10, <16 x i64> %234, !dbg !20
%236 = icmp sge <16 x i32> %strided.vec, %strided.vec52, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %strided.vec52, <16 x i32*> %232, i32 16, <16 x i1> %236), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %strided.vec, <16 x i32*> %235, i32 4, <16 x i1> %236), !dbg !20
%237 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %strided.vec, <16 x i32> %strided.vec52)
%238 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %strided.vec, <16 x i32> %strided.vec52), !dbg !20
%239 = or <16 x i64> %231, <i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2>, !dbg !20
%240 = getelementptr i32, i32* %10, <16 x i64> %239, !dbg !20
%241 = icmp sge <16 x i32> %238, %strided.vec53, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %strided.vec53, <16 x i32*> %235, i32 4, <16 x i1> %241), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %238, <16 x i32*> %240, i32 8, <16 x i1> %241), !dbg !20
%242 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %238, <16 x i32> %strided.vec53)
%243 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %238, <16 x i32> %strided.vec53), !dbg !20
%244 = or <16 x i64> %231, <i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3>, !dbg !20
%245 = getelementptr i32, i32* %10, <16 x i64> %244, !dbg !20
%246 = icmp sge <16 x i32> %243, %strided.vec54, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %strided.vec54, <16 x i32*> %240, i32 8, <16 x i1> %246), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %243, <16 x i32*> %245, i32 4, <16 x i1> %246), !dbg !20
%247 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %243, <16 x i32> %strided.vec54)
%248 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %243, <16 x i32> %strided.vec54)
%249 = icmp sge <16 x i32> %237, %242, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %242, <16 x i32*> %232, i32 16, <16 x i1> %249), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %237, <16 x i32*> %235, i32 4, <16 x i1> %249), !dbg !20
%250 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %237, <16 x i32> %242)
%251 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %237, <16 x i32> %242), !dbg !20
%252 = icmp sge <16 x i32> %251, %248, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %248, <16 x i32*> %235, i32 4, <16 x i1> %252), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %251, <16 x i32*> %240, i32 8, <16 x i1> %252), !dbg !20
%253 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %251, <16 x i32> %248)
%254 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %251, <16 x i32> %248), !dbg !20
%255 = icmp sge <16 x i32> %254, %247, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %247, <16 x i32*> %240, i32 8, <16 x i1> %255), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %254, <16 x i32*> %245, i32 4, <16 x i1> %255), !dbg !20
%256 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %254, <16 x i32> %247)
%257 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %254, <16 x i32> %247)
%258 = icmp sge <16 x i32> %250, %253, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %253, <16 x i32*> %232, i32 16, <16 x i1> %258), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %250, <16 x i32*> %235, i32 4, <16 x i1> %258), !dbg !20
%259 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %250, <16 x i32> %253)
%260 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %250, <16 x i32> %253), !dbg !20
%261 = icmp sge <16 x i32> %260, %257, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %257, <16 x i32*> %235, i32 4, <16 x i1> %261), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %260, <16 x i32*> %240, i32 8, <16 x i1> %261), !dbg !20
%262 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %260, <16 x i32> %257)
%263 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %260, <16 x i32> %257), !dbg !20
%264 = icmp sge <16 x i32> %263, %256, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %256, <16 x i32*> %240, i32 8, <16 x i1> %264), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %263, <16 x i32*> %245, i32 4, <16 x i1> %264), !dbg !20
%265 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %263, <16 x i32> %256)
%266 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %263, <16 x i32> %256)
%267 = icmp sge <16 x i32> %259, %262, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %262, <16 x i32*> %232, i32 16, <16 x i1> %267), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %259, <16 x i32*> %235, i32 4, <16 x i1> %267), !dbg !20
%268 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %259, <16 x i32> %262), !dbg !20
%269 = icmp sge <16 x i32> %268, %266, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %266, <16 x i32*> %235, i32 4, <16 x i1> %269), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %268, <16 x i32*> %240, i32 8, <16 x i1> %269), !dbg !20
%270 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %268, <16 x i32> %266), !dbg !20
%271 = icmp sge <16 x i32> %270, %265, !dbg !21
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %265, <16 x i32*> %240, i32 8, <16 x i1> %271), !dbg !20
call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> %270, <16 x i32*> %245, i32 4, <16 x i1> %271), !dbg !20
%index.next = add nuw i64 %index, 16, !dbg !20
%vec.ind.next = add <16 x i64> %vec.ind, <i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16>
%272 = icmp eq i64 %index.next, %n.vec, !dbg !20
br i1 %272, label %.preheader26.us.us.preheader139, label %vector.body, !dbg !20, !llvm.loop !25
.split29.us31: ; preds = %358, %.lr.ph.us
%273 = add i64 %203, %21, !dbg !20
%274 = icmp slt i64 %273, 2, !dbg !20
%indvar.next = add i64 %indvar, 1, !dbg !20
br i1 %274, label %.lr.ph.us, label %._crit_edge, !dbg !20
275: ; preds = %.preheader26.us.us
%276 = bitcast i32* %364 to i8*, !dbg !20
call void @__tsan_write4(i8* %276), !dbg !20
store i32 %370, i32* %364, align 16, !dbg !20
%277 = bitcast i32* %368 to i8*, !dbg !20
call void @__tsan_write4(i8* %277), !dbg !20
store i32 %366, i32* %368, align 4, !dbg !20
br label %278, !dbg !20
278: ; preds = %275, %.preheader26.us.us
%279 = phi i32 [ %370, %275 ], [ %366, %.preheader26.us.us ]
%280 = phi i32 [ %366, %275 ], [ %370, %.preheader26.us.us ], !dbg !20
%281 = or i64 %363, 2, !dbg !20
%282 = getelementptr i32, i32* %10, i64 %281, !dbg !20
%283 = bitcast i32* %282 to i8*, !dbg !20
call void @__tsan_read4(i8* %283), !dbg !20
%284 = load i32, i32* %282, align 8, !dbg !20
%285 = icmp slt i32 %280, %284, !dbg !21
br i1 %285, label %289, label %286, !dbg !20
286: ; preds = %278
%287 = bitcast i32* %368 to i8*, !dbg !20
call void @__tsan_write4(i8* %287), !dbg !20
store i32 %284, i32* %368, align 4, !dbg !20
%288 = bitcast i32* %282 to i8*, !dbg !20
call void @__tsan_write4(i8* %288), !dbg !20
store i32 %280, i32* %282, align 8, !dbg !20
br label %289, !dbg !20
289: ; preds = %286, %278
%290 = phi i32 [ %284, %286 ], [ %280, %278 ]
%291 = phi i32 [ %280, %286 ], [ %284, %278 ], !dbg !20
%292 = or i64 %363, 3, !dbg !20
%293 = getelementptr i32, i32* %10, i64 %292, !dbg !20
%294 = bitcast i32* %293 to i8*, !dbg !20
call void @__tsan_read4(i8* %294), !dbg !20
%295 = load i32, i32* %293, align 4, !dbg !20
%296 = icmp slt i32 %291, %295, !dbg !21
br i1 %296, label %.preheader.us.us.1, label %297, !dbg !20
297: ; preds = %289
%298 = bitcast i32* %282 to i8*, !dbg !20
call void @__tsan_write4(i8* %298), !dbg !20
store i32 %295, i32* %282, align 8, !dbg !20
%299 = bitcast i32* %293 to i8*, !dbg !20
call void @__tsan_write4(i8* %299), !dbg !20
store i32 %291, i32* %293, align 4, !dbg !20
br label %.preheader.us.us.1, !dbg !20
.preheader.us.us.1: ; preds = %297, %289
%300 = phi i32 [ %291, %297 ], [ %295, %289 ]
%301 = phi i32 [ %295, %297 ], [ %291, %289 ]
%302 = icmp slt i32 %279, %290, !dbg !21
br i1 %302, label %306, label %303, !dbg !20
303: ; preds = %.preheader.us.us.1
%304 = bitcast i32* %364 to i8*, !dbg !20
call void @__tsan_write4(i8* %304), !dbg !20
store i32 %290, i32* %364, align 16, !dbg !20
%305 = bitcast i32* %368 to i8*, !dbg !20
call void @__tsan_write4(i8* %305), !dbg !20
store i32 %279, i32* %368, align 4, !dbg !20
br label %306, !dbg !20
306: ; preds = %303, %.preheader.us.us.1
%307 = phi i32 [ %290, %303 ], [ %279, %.preheader.us.us.1 ]
%308 = phi i32 [ %279, %303 ], [ %290, %.preheader.us.us.1 ], !dbg !20
%309 = icmp slt i32 %308, %301, !dbg !21
br i1 %309, label %313, label %310, !dbg !20
310: ; preds = %306
%311 = bitcast i32* %368 to i8*, !dbg !20
call void @__tsan_write4(i8* %311), !dbg !20
store i32 %301, i32* %368, align 4, !dbg !20
%312 = bitcast i32* %282 to i8*, !dbg !20
call void @__tsan_write4(i8* %312), !dbg !20
store i32 %308, i32* %282, align 8, !dbg !20
br label %313, !dbg !20
313: ; preds = %310, %306
%314 = phi i32 [ %301, %310 ], [ %308, %306 ]
%315 = phi i32 [ %308, %310 ], [ %301, %306 ], !dbg !20
%316 = icmp slt i32 %315, %300, !dbg !21
br i1 %316, label %.preheader.us.us.2, label %317, !dbg !20
317: ; preds = %313
%318 = bitcast i32* %282 to i8*, !dbg !20
call void @__tsan_write4(i8* %318), !dbg !20
store i32 %300, i32* %282, align 8, !dbg !20
%319 = bitcast i32* %293 to i8*, !dbg !20
call void @__tsan_write4(i8* %319), !dbg !20
store i32 %315, i32* %293, align 4, !dbg !20
br label %.preheader.us.us.2, !dbg !20
.preheader.us.us.2: ; preds = %317, %313
%320 = phi i32 [ %315, %317 ], [ %300, %313 ]
%321 = phi i32 [ %300, %317 ], [ %315, %313 ]
%322 = icmp slt i32 %307, %314, !dbg !21
br i1 %322, label %326, label %323, !dbg !20
323: ; preds = %.preheader.us.us.2
%324 = bitcast i32* %364 to i8*, !dbg !20
call void @__tsan_write4(i8* %324), !dbg !20
store i32 %314, i32* %364, align 16, !dbg !20
%325 = bitcast i32* %368 to i8*, !dbg !20
call void @__tsan_write4(i8* %325), !dbg !20
store i32 %307, i32* %368, align 4, !dbg !20
br label %326, !dbg !20
326: ; preds = %323, %.preheader.us.us.2
%327 = phi i32 [ %314, %323 ], [ %307, %.preheader.us.us.2 ]
%328 = phi i32 [ %307, %323 ], [ %314, %.preheader.us.us.2 ], !dbg !20
%329 = icmp slt i32 %328, %321, !dbg !21
br i1 %329, label %333, label %330, !dbg !20
330: ; preds = %326
%331 = bitcast i32* %368 to i8*, !dbg !20
call void @__tsan_write4(i8* %331), !dbg !20
store i32 %321, i32* %368, align 4, !dbg !20
%332 = bitcast i32* %282 to i8*, !dbg !20
call void @__tsan_write4(i8* %332), !dbg !20
store i32 %328, i32* %282, align 8, !dbg !20
br label %333, !dbg !20
333: ; preds = %330, %326
%334 = phi i32 [ %321, %330 ], [ %328, %326 ]
%335 = phi i32 [ %328, %330 ], [ %321, %326 ], !dbg !20
%336 = icmp slt i32 %335, %320, !dbg !21
br i1 %336, label %.preheader.us.us.3, label %337, !dbg !20
337: ; preds = %333
%338 = bitcast i32* %282 to i8*, !dbg !20
call void @__tsan_write4(i8* %338), !dbg !20
store i32 %320, i32* %282, align 8, !dbg !20
%339 = bitcast i32* %293 to i8*, !dbg !20
call void @__tsan_write4(i8* %339), !dbg !20
store i32 %335, i32* %293, align 4, !dbg !20
br label %.preheader.us.us.3, !dbg !20
.preheader.us.us.3: ; preds = %337, %333
%340 = phi i32 [ %335, %337 ], [ %320, %333 ]
%341 = phi i32 [ %320, %337 ], [ %335, %333 ]
%342 = icmp slt i32 %327, %334, !dbg !21
br i1 %342, label %346, label %343, !dbg !20
343: ; preds = %.preheader.us.us.3
%344 = bitcast i32* %364 to i8*, !dbg !20
call void @__tsan_write4(i8* %344), !dbg !20
store i32 %334, i32* %364, align 16, !dbg !20
%345 = bitcast i32* %368 to i8*, !dbg !20
call void @__tsan_write4(i8* %345), !dbg !20
store i32 %327, i32* %368, align 4, !dbg !20
br label %346, !dbg !20
346: ; preds = %343, %.preheader.us.us.3
%347 = phi i32 [ %327, %343 ], [ %334, %.preheader.us.us.3 ], !dbg !20
%348 = icmp slt i32 %347, %341, !dbg !21
br i1 %348, label %352, label %349, !dbg !20
349: ; preds = %346
%350 = bitcast i32* %368 to i8*, !dbg !20
call void @__tsan_write4(i8* %350), !dbg !20
store i32 %341, i32* %368, align 4, !dbg !20
%351 = bitcast i32* %282 to i8*, !dbg !20
call void @__tsan_write4(i8* %351), !dbg !20
store i32 %347, i32* %282, align 8, !dbg !20
br label %352, !dbg !20
352: ; preds = %349, %346
%353 = phi i32 [ %347, %349 ], [ %341, %346 ], !dbg !20
%354 = icmp slt i32 %353, %340, !dbg !21
br i1 %354, label %358, label %355, !dbg !20
355: ; preds = %352
%356 = bitcast i32* %282 to i8*, !dbg !20
call void @__tsan_write4(i8* %356), !dbg !20
store i32 %340, i32* %282, align 8, !dbg !20
%357 = bitcast i32* %293 to i8*, !dbg !20
call void @__tsan_write4(i8* %357), !dbg !20
store i32 %353, i32* %293, align 4, !dbg !20
br label %358, !dbg !20
358: ; preds = %355, %352
%359 = add nuw nsw i64 %361, 1, !dbg !20
%360 = icmp slt i64 %359, %213, !dbg !20
br i1 %360, label %.preheader26.us.us, label %.split29.us31, !dbg !20, !llvm.loop !26
.preheader26.us.us: ; preds = %.preheader26.us.us.preheader139, %358
%361 = phi i64 [ %359, %358 ], [ %.ph140, %.preheader26.us.us.preheader139 ]
%362 = add i64 %203, %361
%363 = shl i64 %362, 2
%364 = getelementptr i32, i32* %10, i64 %363, !dbg !20
%365 = bitcast i32* %364 to i8*, !dbg !20
call void @__tsan_read4(i8* %365), !dbg !20
%366 = load i32, i32* %364, align 16, !dbg !20
%367 = or i64 %363, 1, !dbg !20
%368 = getelementptr i32, i32* %10, i64 %367, !dbg !20
%369 = bitcast i32* %368 to i8*, !dbg !20
call void @__tsan_read4(i8* %369), !dbg !20
%370 = load i32, i32* %368, align 4, !dbg !20
%371 = icmp slt i32 %366, %370, !dbg !21
br i1 %371, label %278, label %275, !dbg !20
.lr.ph30.split: ; preds = %.lr.ph30
br i1 %23, label %.loopexit, label %._crit_edge, !dbg !20
.loopexit: ; preds = %.lr.ph30.split, %.loopexit
br label %.loopexit
._crit_edge: ; preds = %.split29.us31, %.lr.ph30.split, %3
call void @__tsan_func_exit(), !dbg !20
ret i32 0, !dbg !20
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment