In visitSTORE, always use FindBetterChain, rather than only when UseAA is enabled.

    Recommiting after fixing X86 inc/dec chain bug.

    * Simplify Consecutive Merge Store Candidate Search

    Now that address aliasing is much less conservative, push through
    simplified store merging search and chain alias analysis which only
    checks for parallel stores through the chain subgraph. This is cleaner
    as the separation of non-interfering loads/stores from the
    store-merging logic.

    When merging stores search up the chain through a single load, and
    finds all possible stores by looking down from through a load and a
    TokenFactor to all stores visited.

    This improves the quality of the output SelectionDAG and the output
    Codegen (save perhaps for some ARM cases where we correctly constructs
    wider loads, but then promotes them to float operations which appear
    but requires more expensive constant generation).

    Some minor peephole optimizations to deal with improved SubDAG shapes (listed below)

    Additional Minor Changes:

      1. Finishes removing unused AliasLoad code

      2. Unifies the chain aggregation in the merged stores across code
         paths

      3. Re-add the Store node to the worklist after calling
         SimplifyDemandedBits.

      4. Increase GatherAllAliasesMaxDepth from 6 to 18. That number is
         arbitrary, but seems sufficient to not cause regressions in
         tests.

      5. Remove Chain dependencies of Memory operations on CopyfromReg
         nodes as these are captured by data dependence

      6. Forward loads-store values through tokenfactors containing
          {CopyToReg,CopyFromReg} Values.

      7. Peephole to convert buildvector of extract_vector_elt to
         extract_subvector if possible (see
         CodeGen/AArch64/store-merge.ll)

      8. Store merging for the ARM target is restricted to 32-bit as
         some in some contexts invalid 64-bit operations are being
         generated. This can be removed once appropriate checks are
         added.

    This finishes the change Matt Arsenault started in r246307 and
    jyknight's original patch.

    Many tests required some changes as memory operations are now
    reorderable, improving load-store forwarding. One test in
    particular is worth noting:

      CodeGen/PowerPC/ppc64-align-long-double.ll - Improved load-store
      forwarding converts a load-store pair into a parallel store and
      a memory-realized bitcast of the same value. However, because we
      lose the sharing of the explicit and implicit store values we
      must create another local store. A similar transformation
      happens before SelectionDAG as well.

    Reviewers: arsenm, hfinkel, tstellarAMD, jyknight, nhaehnle

llvm-svn: 293893

5 files changed, 36 insertions, 46 deletions

diff --git a/llvm/test/CodeGen/AMDGPU/debugger-insert-nops.ll b/llvm/test/CodeGen/AMDGPU/debugger-insert-nops.ll
index 6638f4e2582..7be7d9486a4 100644
--- a/llvm/test/CodeGen/AMDGPU/debugger-insert-nops.ll
+++ b/llvm/test/CodeGen/AMDGPU/debugger-insert-nops.ll
@@ -1,13 +1,21 @@
-; RUN: llc -O0 -mtriple=amdgcn--amdhsa -mcpu=fiji -mattr=+amdgpu-debugger-insert-nops -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -O0 -mtriple=amdgcn--amdhsa -mcpu=fiji -mattr=+amdgpu-debugger-insert-nops -verify-machineinstrs < %s | FileCheck %s --check-prefix=CHECK
+; RUN: llc -O0 -mtriple=amdgcn--amdhsa -mcpu=fiji -mattr=+amdgpu-debugger-insert-nops -verify-machineinstrs < %s | FileCheck %s --check-prefix=CHECKNOP
 
-; CHECK: test01.cl:2:{{[0-9]+}}
-; CHECK-NEXT: s_nop 0
+; This test expects that we have one instance for each line in some order with "s_nop 0" instances after each.
 
-; CHECK: test01.cl:3:{{[0-9]+}}
-; CHECK-NEXT: s_nop 0
+; Check that each line appears at least once
+; CHECK-DAG: test01.cl:2:3
+; CHECK-DAG: test01.cl:3:3
+; CHECK-DAG: test01.cl:4:3
 
-; CHECK: test01.cl:4:{{[0-9]+}}
-; CHECK-NEXT: s_nop 0
+
+; Check that each of each of the lines consists of the line output, followed by "s_nop 0"
+; CHECKNOP: test01.cl:{{[234]}}:3
+; CHECKNOP-NEXT: s_nop 0
+; CHECKNOP: test01.cl:{{[234]}}:3
+; CHECKNOP-NEXT: s_nop 0
+; CHECKNOP: test01.cl:{{[234]}}:3
+; CHECKNOP-NEXT: s_nop 0
 
 ; CHECK: test01.cl:5:{{[0-9]+}}
 ; CHECK-NEXT: s_nop 0
@@ -21,7 +29,7 @@ entry:
   call void @llvm.dbg.declare(metadata i32 addrspace(1)** %A.addr, metadata !17, metadata !18), !dbg !19
   %0 = load i32 addrspace(1)*, i32 addrspace(1)** %A.addr, align 4, !dbg !20
   %arrayidx = getelementptr inbounds i32, i32 addrspace(1)* %0, i32 0, !dbg !20
-  store i32 1, i32 addrspace(1)* %arrayidx, align 4, !dbg !21
+  store i32 1, i32 addrspace(1)* %arrayidx, align 4, !dbg !20
   %1 = load i32 addrspace(1)*, i32 addrspace(1)** %A.addr, align 4, !dbg !22
   %arrayidx1 = getelementptr inbounds i32, i32 addrspace(1)* %1, i32 1, !dbg !22
   store i32 2, i32 addrspace(1)* %arrayidx1, align 4, !dbg !23
diff --git a/llvm/test/CodeGen/AMDGPU/insert_vector_elt.ll b/llvm/test/CodeGen/AMDGPU/insert_vector_elt.ll
index 65ac693a4f4..08acee0d4be 100644
--- a/llvm/test/CodeGen/AMDGPU/insert_vector_elt.ll
+++ b/llvm/test/CodeGen/AMDGPU/insert_vector_elt.ll
@@ -257,9 +257,8 @@ define void @dynamic_insertelement_v2i8(<2 x i8> addrspace(1)* %out, <2 x i8> %a
 
 ; GCN-DAG: buffer_store_byte v{{[0-9]+}}, off, s{{\[[0-9]+:[0-9]+\]}}, s{{[0-9]+}} offset:2
 ; GCN-DAG: buffer_store_byte v{{[0-9]+}}, off, s{{\[[0-9]+:[0-9]+\]}}, s{{[0-9]+}} offset:1
-; GCN-DAG: buffer_store_byte v{{[0-9]+}}, off, s{{\[[0-9]+:[0-9]+\]}}, s{{[0-9]+}}{{$}}
-
-; GCN: buffer_store_byte v{{[0-9]+}}, v{{[0-9]+}}, s{{\[[0-9]+:[0-9]+\]}}, s{{[0-9]+}} offen{{$}}
+; GCN-DAG: buffer_store_byte v{{[0-9]+}}, v{{[0-9]+}}, s{{\[[0-9]+:[0-9]+\]}}, s{{[0-9]+}} offen{{$}}
+; GCN-DAG: buffer_store_byte v{{[0-9]+}}, v{{[0-9]+}}, s{{\[[0-9]+:[0-9]+\]}}, s{{[0-9]+}} offen{{$}}
 
 ; GCN: buffer_load_ubyte
 ; GCN: buffer_load_ubyte
diff --git a/llvm/test/CodeGen/AMDGPU/merge-stores.ll b/llvm/test/CodeGen/AMDGPU/merge-stores.ll
index 07104ebc8c9..fd2d8ee1189 100644
--- a/llvm/test/CodeGen/AMDGPU/merge-stores.ll
+++ b/llvm/test/CodeGen/AMDGPU/merge-stores.ll
@@ -1,8 +1,5 @@
-; RUN: llc -march=amdgcn -verify-machineinstrs -amdgpu-load-store-vectorizer=0 < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=GCN-NOAA %s
-; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs -amdgpu-load-store-vectorizer=0 < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=GCN-NOAA %s
-
-; RUN: llc -march=amdgcn -verify-machineinstrs -combiner-alias-analysis -amdgpu-load-store-vectorizer=0 < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=GCN-AA %s
-; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs -combiner-alias-analysis -amdgpu-load-store-vectorizer=0 < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=GCN-AA %s
+; RUN: llc -march=amdgcn -verify-machineinstrs -amdgpu-load-store-vectorizer=0 < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=GCN-AA %s
+; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs -amdgpu-load-store-vectorizer=0 < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=GCN-AA %s
 
 ; This test is mostly to test DAG store merging, so disable the vectorizer.
 ; Run with devices with different unaligned load restrictions.
@@ -150,12 +147,7 @@ define void @merge_global_store_4_constants_f32(float addrspace(1)* %out) #0 {
 }
 
 ; GCN-LABEL: {{^}}merge_global_store_4_constants_mixed_i32_f32:
-; GCN-NOAA: buffer_store_dwordx4 v
-
-; GCN-AA: buffer_store_dwordx2
-; GCN-AA: buffer_store_dword v
-; GCN-AA: buffer_store_dword v
-
+; GCN-AA: buffer_store_dwordx4 v
 ; GCN: s_endpgm
 define void @merge_global_store_4_constants_mixed_i32_f32(float addrspace(1)* %out) #0 {
   %out.gep.1 = getelementptr float, float addrspace(1)* %out, i32 1
@@ -474,17 +466,9 @@ define void @merge_global_store_4_adjacent_loads_i8_natural_align(i8 addrspace(1
   ret void
 }
 
-; This works once AA is enabled on the subtarget
 ; GCN-LABEL: {{^}}merge_global_store_4_vector_elts_loads_v4i32:
 ; GCN: buffer_load_dwordx4 [[LOAD:v\[[0-9]+:[0-9]+\]]]
-
-; GCN-NOAA: buffer_store_dword v
-; GCN-NOAA: buffer_store_dword v
-; GCN-NOAA: buffer_store_dword v
-; GCN-NOAA: buffer_store_dword v
-
-; GCN-AA: buffer_store_dwordx4 [[LOAD]]
-
+; GCN: buffer_store_dwordx4 [[LOAD]]
 ; GCN: s_endpgm
 define void @merge_global_store_4_vector_elts_loads_v4i32(i32 addrspace(1)* %out, <4 x i32> addrspace(1)* %in) #0 {
   %out.gep.1 = getelementptr i32, i32 addrspace(1)* %out, i32 1
diff --git a/llvm/test/CodeGen/AMDGPU/private-element-size.ll b/llvm/test/CodeGen/AMDGPU/private-element-size.ll
index de9a8f75512..8bbb72fcda4 100644
--- a/llvm/test/CodeGen/AMDGPU/private-element-size.ll
+++ b/llvm/test/CodeGen/AMDGPU/private-element-size.ll
@@ -32,10 +32,10 @@
 ; HSA-ELT4-DAG: buffer_store_dword {{v[0-9]+}}, off, s[0:3], s9 offset:24{{$}}
 ; HSA-ELT4-DAG: buffer_store_dword {{v[0-9]+}}, off, s[0:3], s9 offset:28{{$}}
 
-; HSA-ELT4: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen{{$}}
-; HSA-ELT4: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen offset:4{{$}}
-; HSA-ELT4: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen offset:8{{$}}
-; HSA-ELT4: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen offset:12{{$}}
+; HSA-ELT4-DAG: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen{{$}}
+; HSA-ELT4-DAG: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen offset:4{{$}}
+; HSA-ELT4-DAG: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen offset:8{{$}}
+; HSA-ELT4-DAG: buffer_load_dword {{v[0-9]+}}, v{{[0-9]+}}, s[0:3], s9 offen offset:12{{$}}
 define void @private_elt_size_v4i32(<4 x i32> addrspace(1)* %out, i32 addrspace(1)* %index.array) #0 {
 entry:
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
@@ -130,8 +130,8 @@ entry:
 ; HSA-ELT8: private_element_size = 2
 ; HSA-ELT4: private_element_size = 1
 
-; HSA-ELTGE8-DAG: buffer_store_dwordx2 {{v\[[0-9]+:[0-9]+\]}}, off, s[0:3], s9{{$}}
-; HSA-ELTGE8-DAG: buffer_store_dwordx2 {{v\[[0-9]+:[0-9]+\]}}, off, s[0:3], s9 offset:8
+; HSA-ELTGE8-DAG: buffer_store_dwordx2 {{v\[[0-9]+:[0-9]+\]}}, {{off|v[0-9]}}, s[0:3], s9{{$}}
+; HSA-ELTGE8-DAG: buffer_store_dwordx2 {{v\[[0-9]+:[0-9]+\]}}, {{off|v[0-9]}}, s[0:3], s9 offset:8
 
 ; HSA-ELTGE8: buffer_load_dwordx2 {{v\[[0-9]+:[0-9]+\]}}, v{{[0-9]+}}, s[0:3], s9 offen
 
diff --git a/llvm/test/CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll b/llvm/test/CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll
index 4beefb047f2..c1d691fcff8 100644
--- a/llvm/test/CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll
+++ b/llvm/test/CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll
@@ -157,9 +157,8 @@ define void @reorder_global_load_local_store_global_load(i32 addrspace(1)* %out,
 
 ; FUNC-LABEL: @reorder_local_offsets
 ; CI: ds_read2_b32 {{v\[[0-9]+:[0-9]+\]}}, {{v[0-9]+}} offset0:100 offset1:102
-; CI: ds_write2_b32 {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}} offset0:3 offset1:100
-; CI: ds_read_b32 {{v[0-9]+}}, {{v[0-9]+}} offset:12
-; CI: ds_write_b32 {{v[0-9]+}}, {{v[0-9]+}} offset:408
+; CI-DAG: ds_write2_b32 {{v[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} offset0:3 offset1:100
+; CI-DAG: ds_write_b32 {{v[0-9]+}}, {{v[0-9]+}} offset:408
 ; CI: buffer_store_dword
 ; CI: s_endpgm
 define void @reorder_local_offsets(i32 addrspace(1)* nocapture %out, i32 addrspace(1)* noalias nocapture readnone %gptr, i32 addrspace(3)* noalias nocapture %ptr0) #0 {
@@ -181,12 +180,12 @@ define void @reorder_local_offsets(i32 addrspace(1)* nocapture %out, i32 addrspa
 }
 
 ; FUNC-LABEL: @reorder_global_offsets
-; CI: buffer_load_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:400
-; CI: buffer_load_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:408
-; CI: buffer_store_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:12
-; CI: buffer_store_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:400
-; CI: buffer_store_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:408
-; CI: buffer_load_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:12
+; CI-DAG: buffer_load_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:400
+; CI-DAG: buffer_load_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:408
+; CI-DAG: buffer_store_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:12
+; CI-DAG: buffer_store_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:400
+; CI-DAG: buffer_store_dword {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 offset:408
+; CI: buffer_store_dword
 ; CI: s_endpgm
 define void @reorder_global_offsets(i32 addrspace(1)* nocapture %out, i32 addrspace(1)* noalias nocapture readnone %gptr, i32 addrspace(1)* noalias nocapture %ptr0) #0 {
   %ptr1 = getelementptr inbounds i32, i32 addrspace(1)* %ptr0, i32 3