From: Quentin Colombet <qcolombet@apple.com>
Date: Wed, 16 Nov 2016 01:07:12 +0000 (+0000)
Subject: [RegAllocGreedy] Record missed hint for late recoloring.
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=8934fe31f3cbfa1543b98dc101005a2142bb3fe0;p=llvm

[RegAllocGreedy] Record missed hint for late recoloring.

In https://reviews.llvm.org/D25347, Geoff noticed that we still have
useless copy that we can eliminate after register allocation. At the
time the allocation is chosen for those copies, they are not useless
but, because of changes in the surrounding code, later on they might
become useless.
The Greedy allocator already has a mechanism to deal with such cases
with a late recoloring. However, we missed to record the some of the
missed hints.

This commit fixes that.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287070 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/lib/CodeGen/RegAllocGreedy.cpp b/lib/CodeGen/RegAllocGreedy.cpp
index 9a0fc66a96b..4a381e0aa83 100644
--- a/lib/CodeGen/RegAllocGreedy.cpp
+++ b/lib/CodeGen/RegAllocGreedy.cpp
@@ -646,6 +646,9 @@ unsigned RAGreedy::tryAssign(LiveInterval &VirtReg,
         evictInterference(VirtReg, Hint, NewVRegs);
         return Hint;
       }
+      // Record the missed hint, we may be able to recover
+      // at the end if the surrounding allocation changed.
+      SetOfBrokenHints.insert(&VirtReg);
     }
 
   // Try to evict interference from a cheaper alternative.
diff --git a/test/CodeGen/X86/mul-i1024.ll b/test/CodeGen/X86/mul-i1024.ll
index 60933b1e1fa..93d55a00128 100644
--- a/test/CodeGen/X86/mul-i1024.ll
+++ b/test/CodeGen/X86/mul-i1024.ll
@@ -4460,8 +4460,7 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    mulq %r9
 ; X64-NEXT:    movq %rax, -{{[0-9]+}}(%rsp) # 8-byte Spill
 ; X64-NEXT:    movq %rdx, -{{[0-9]+}}(%rsp) # 8-byte Spill
-; X64-NEXT:    movq %r14, %rdi
-; X64-NEXT:    movq %rdi, %r9
+; X64-NEXT:    movq %r14, %r9
 ; X64-NEXT:    addq %rax, %r9
 ; X64-NEXT:    movq %rcx, %rax
 ; X64-NEXT:    adcq %rdx, %rax
@@ -4470,7 +4469,7 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    movq %rax, %rbp
 ; X64-NEXT:    movq %r11, %rax
 ; X64-NEXT:    movq %r11, -{{[0-9]+}}(%rsp) # 8-byte Spill
-; X64-NEXT:    addq %rdi, %rax
+; X64-NEXT:    addq %r14, %rax
 ; X64-NEXT:    adcq %rcx, %r15
 ; X64-NEXT:    movq %r15, {{[0-9]+}}(%rsp) # 8-byte Spill
 ; X64-NEXT:    movq (%r10), %rax
@@ -4478,8 +4477,9 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    xorl %r15d, %r15d
 ; X64-NEXT:    mulq %r15
 ; X64-NEXT:    movq %rdx, %rsi
-; X64-NEXT:    movq %rax, %r14
-; X64-NEXT:    addq %rdi, %rax
+; X64-NEXT:    movq %rax, %rbx
+; X64-NEXT:    addq %r14, %rax
+; X64-NEXT:    movq %r14, %rdi
 ; X64-NEXT:    movq %rsi, %rax
 ; X64-NEXT:    adcq %rcx, %rax
 ; X64-NEXT:    movq %rax, {{[0-9]+}}(%rsp) # 8-byte Spill
@@ -4489,7 +4489,8 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    xorl %r8d, %r8d
 ; X64-NEXT:    movq %rax, {{[0-9]+}}(%rsp) # 8-byte Spill
 ; X64-NEXT:    movq %rdx, -{{[0-9]+}}(%rsp) # 8-byte Spill
-; X64-NEXT:    movq %r14, %rcx
+; X64-NEXT:    movq %rbx, %rcx
+; X64-NEXT:    movq %rbx, %r14
 ; X64-NEXT:    addq %rax, %rcx
 ; X64-NEXT:    movq %rsi, %rax
 ; X64-NEXT:    adcq %rdx, %rax
@@ -4611,8 +4612,8 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    movq %rax, %r9
 ; X64-NEXT:    movq %rdx, %rbx
 ; X64-NEXT:    movq {{[0-9]+}}(%rsp), %rax # 8-byte Reload
-; X64-NEXT:    movq 56(%rax), %rax
-; X64-NEXT:    movq %rax, %rsi
+; X64-NEXT:    movq 56(%rax), %rsi
+; X64-NEXT:    movq %rsi, %rax
 ; X64-NEXT:    movq %rsi, -{{[0-9]+}}(%rsp) # 8-byte Spill
 ; X64-NEXT:    mulq %rdi
 ; X64-NEXT:    movq %rdi, %r10
@@ -4776,8 +4777,8 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    movq %rax, %r11
 ; X64-NEXT:    movq %rdx, %rcx
 ; X64-NEXT:    movq {{[0-9]+}}(%rsp), %rax # 8-byte Reload
-; X64-NEXT:    movq 24(%rax), %rax
-; X64-NEXT:    movq %rax, %r8
+; X64-NEXT:    movq 24(%rax), %r8
+; X64-NEXT:    movq %r8, %rax
 ; X64-NEXT:    movq %r8, {{[0-9]+}}(%rsp) # 8-byte Spill
 ; X64-NEXT:    mulq %rdi
 ; X64-NEXT:    movq %rdi, %r13
@@ -4978,8 +4979,8 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    mulq %rdi
 ; X64-NEXT:    movq %rdx, %r8
 ; X64-NEXT:    movq %rax, {{[0-9]+}}(%rsp) # 8-byte Spill
-; X64-NEXT:    movq {{[0-9]+}}(%rsp), %rsi # 8-byte Reload
-; X64-NEXT:    movq %rsi, %rax
+; X64-NEXT:    movq {{[0-9]+}}(%rsp), %r11 # 8-byte Reload
+; X64-NEXT:    movq %r11, %rax
 ; X64-NEXT:    mulq %rdi
 ; X64-NEXT:    movq %rdx, %rbx
 ; X64-NEXT:    movq %rax, %rbp
@@ -4995,8 +4996,7 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    addq %rbx, %rcx
 ; X64-NEXT:    sbbq %rbp, %rbp
 ; X64-NEXT:    andl $1, %ebp
-; X64-NEXT:    movq %rsi, %rax
-; X64-NEXT:    movq %rsi, %rdi
+; X64-NEXT:    movq %r11, %rax
 ; X64-NEXT:    mulq %r14
 ; X64-NEXT:    movq %rdx, %rsi
 ; X64-NEXT:    movq %rax, %rbx
@@ -5014,8 +5014,7 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    mulq %rcx
 ; X64-NEXT:    movq %rdx, %r14
 ; X64-NEXT:    movq %rax, %r8
-; X64-NEXT:    movq %rdi, %rax
-; X64-NEXT:    movq %rdi, %r11
+; X64-NEXT:    movq %r11, %rax
 ; X64-NEXT:    mulq %rcx
 ; X64-NEXT:    movq %rdx, %r12
 ; X64-NEXT:    movq %rax, %rcx
@@ -5184,8 +5183,7 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    addq %rsi, %rcx
 ; X64-NEXT:    sbbq %rdi, %rdi
 ; X64-NEXT:    andl $1, %edi
-; X64-NEXT:    movq %rbp, %rbx
-; X64-NEXT:    movq %rbx, %rax
+; X64-NEXT:    movq %rbp, %rax
 ; X64-NEXT:    mulq %r11
 ; X64-NEXT:    movq %rdx, %r11
 ; X64-NEXT:    movq %rax, %rsi
@@ -5204,8 +5202,8 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    mulq %rcx
 ; X64-NEXT:    movq %rdx, %r8
 ; X64-NEXT:    movq %rax, %r13
-; X64-NEXT:    movq %rbx, %rax
-; X64-NEXT:    movq %rbx, %r9
+; X64-NEXT:    movq %rbp, %rax
+; X64-NEXT:    movq %rbp, %r9
 ; X64-NEXT:    mulq %rcx
 ; X64-NEXT:    movq %rdx, %rbp
 ; X64-NEXT:    movq %rax, %rcx
@@ -5483,8 +5481,8 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    addq %rdx, %rbp
 ; X64-NEXT:    movq -{{[0-9]+}}(%rsp), %rax # 8-byte Reload
 ; X64-NEXT:    movq %rax, %rsi
-; X64-NEXT:    movq -{{[0-9]+}}(%rsp), %rbx # 8-byte Reload
-; X64-NEXT:    imulq %rbx, %rsi
+; X64-NEXT:    movq -{{[0-9]+}}(%rsp), %r12 # 8-byte Reload
+; X64-NEXT:    imulq %r12, %rsi
 ; X64-NEXT:    movq {{[0-9]+}}(%rsp), %rcx # 8-byte Reload
 ; X64-NEXT:    mulq %rcx
 ; X64-NEXT:    movq %rax, %r10
@@ -5500,8 +5498,7 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    mulq %r8
 ; X64-NEXT:    movq %rdx, %rcx
 ; X64-NEXT:    movq %rax, -{{[0-9]+}}(%rsp) # 8-byte Spill
-; X64-NEXT:    movq %rbx, %rax
-; X64-NEXT:    movq %rbx, %r12
+; X64-NEXT:    movq %r12, %rax
 ; X64-NEXT:    mulq %r8
 ; X64-NEXT:    movq %rdx, %rsi
 ; X64-NEXT:    movq %rax, %rbx
@@ -5588,15 +5585,15 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    adcq -{{[0-9]+}}(%rsp), %rdx # 8-byte Folded Reload
 ; X64-NEXT:    movq %rdx, {{[0-9]+}}(%rsp) # 8-byte Spill
 ; X64-NEXT:    movq {{[0-9]+}}(%rsp), %rsi # 8-byte Reload
-; X64-NEXT:    movq 80(%rsi), %rax
-; X64-NEXT:    movq %rax, %rbp
+; X64-NEXT:    movq 80(%rsi), %r9
+; X64-NEXT:    movq %r9, %rax
 ; X64-NEXT:    movq {{[0-9]+}}(%rsp), %rdi # 8-byte Reload
 ; X64-NEXT:    mulq %rdi
 ; X64-NEXT:    movq %rax, {{[0-9]+}}(%rsp) # 8-byte Spill
 ; X64-NEXT:    movq %rdx, %rcx
-; X64-NEXT:    movq 88(%rsi), %rax
+; X64-NEXT:    movq 88(%rsi), %r8
 ; X64-NEXT:    movq %rsi, %r11
-; X64-NEXT:    movq %rax, %r8
+; X64-NEXT:    movq %r8, %rax
 ; X64-NEXT:    movq %r8, -{{[0-9]+}}(%rsp) # 8-byte Spill
 ; X64-NEXT:    mulq %rdi
 ; X64-NEXT:    movq %rdi, %rbx
@@ -5604,8 +5601,7 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    movq %rax, %rdi
 ; X64-NEXT:    addq %rcx, %rdi
 ; X64-NEXT:    adcq $0, %rsi
-; X64-NEXT:    movq %rbp, %rax
-; X64-NEXT:    movq %rbp, %r9
+; X64-NEXT:    movq %r9, %rax
 ; X64-NEXT:    movq %r9, -{{[0-9]+}}(%rsp) # 8-byte Spill
 ; X64-NEXT:    movq {{[0-9]+}}(%rsp), %r15 # 8-byte Reload
 ; X64-NEXT:    mulq %r15
@@ -5636,8 +5632,8 @@ define void @test_1024(i1024* %a, i1024* %b, i1024* %out) nounwind {
 ; X64-NEXT:    addq %rcx, %r12
 ; X64-NEXT:    adcq %rsi, %r8
 ; X64-NEXT:    movq %r11, %rsi
-; X64-NEXT:    movq 64(%rsi), %rax
-; X64-NEXT:    movq %rax, %r11
+; X64-NEXT:    movq 64(%rsi), %r11
+; X64-NEXT:    movq %r11, %rax
 ; X64-NEXT:    movq %rbx, %rdi
 ; X64-NEXT:    mulq %rdi
 ; X64-NEXT:    movq %rax, -{{[0-9]+}}(%rsp) # 8-byte Spill
diff --git a/test/CodeGen/X86/mul-i512.ll b/test/CodeGen/X86/mul-i512.ll
index f9cea6d2ba3..d26040059e6 100644
--- a/test/CodeGen/X86/mul-i512.ll
+++ b/test/CodeGen/X86/mul-i512.ll
@@ -912,11 +912,11 @@ define void @test_512(i512* %a, i512* %b, i512* %out) nounwind {
 ; X64-NEXT:    pushq %rax
 ; X64-NEXT:    movq %rdx, (%rsp) # 8-byte Spill
 ; X64-NEXT:    movq 24(%rdi), %rbp
-; X64-NEXT:    movq 16(%rdi), %rcx
+; X64-NEXT:    movq 16(%rdi), %r11
 ; X64-NEXT:    movq %rsi, -{{[0-9]+}}(%rsp) # 8-byte Spill
 ; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    movq 8(%rsi), %r8
-; X64-NEXT:    movq %rcx, %rax
+; X64-NEXT:    movq %r11, %rax
 ; X64-NEXT:    movq %rdx, %rsi
 ; X64-NEXT:    mulq %rsi
 ; X64-NEXT:    movq %rdx, %r10
@@ -929,8 +929,7 @@ define void @test_512(i512* %a, i512* %b, i512* %out) nounwind {
 ; X64-NEXT:    movq %rax, %rsi
 ; X64-NEXT:    addq %r10, %rsi
 ; X64-NEXT:    adcq $0, %rbx
-; X64-NEXT:    movq %rcx, %rax
-; X64-NEXT:    movq %rcx, %r11
+; X64-NEXT:    movq %r11, %rax
 ; X64-NEXT:    movq %r11, -{{[0-9]+}}(%rsp) # 8-byte Spill
 ; X64-NEXT:    mulq %r8
 ; X64-NEXT:    movq %rdx, %rcx
@@ -964,13 +963,13 @@ define void @test_512(i512* %a, i512* %b, i512* %out) nounwind {
 ; X64-NEXT:    addq %rbp, %r15
 ; X64-NEXT:    adcq %rsi, %r9
 ; X64-NEXT:    movq %rdi, -{{[0-9]+}}(%rsp) # 8-byte Spill
-; X64-NEXT:    movq (%rdi), %rax
-; X64-NEXT:    movq %rax, %r14
+; X64-NEXT:    movq (%rdi), %r14
+; X64-NEXT:    movq %r14, %rax
 ; X64-NEXT:    mulq %rbx
 ; X64-NEXT:    movq %rax, -{{[0-9]+}}(%rsp) # 8-byte Spill
 ; X64-NEXT:    movq %rdx, %r11
-; X64-NEXT:    movq 8(%rdi), %rax
-; X64-NEXT:    movq %rax, %rcx
+; X64-NEXT:    movq 8(%rdi), %rcx
+; X64-NEXT:    movq %rcx, %rax
 ; X64-NEXT:    movq %rcx, -{{[0-9]+}}(%rsp) # 8-byte Spill
 ; X64-NEXT:    mulq %rbx
 ; X64-NEXT:    movq %rdx, %rbp