Be careful about scheduling nodes above previous calls. It increase usages of

more callee-saved registers and introduce copies. Only allows it if scheduling
a node above calls would end up lessen register pressure.

Call operands also has added ABI restrictions for register allocation, so be
extra careful with hoisting them above calls.

rdar://9329627


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@130245 91177308-0d34-0410-b5e6-96231b3b80d8

5 files changed, 73 insertions, 23 deletions

diff --git a/test/CodeGen/ARM/2011-04-26-SchedTweak.ll b/test/CodeGen/ARM/2011-04-26-SchedTweak.ll
new file mode 100644
index 0000000000..ed7dd03320
--- /dev/null
+++ b/test/CodeGen/ARM/2011-04-26-SchedTweak.ll
@@ -0,0 +1,70 @@
+; RUN: llc < %s -mtriple=thumbv7-apple-ios -relocation-model=pic -mcpu=cortex-a8 | FileCheck %s
+
+; Do not move the umull above previous call which would require use of
+; more callee-saved registers and introduce copies.
+; rdar://9329627
+
+%struct.FF = type { i32 (i32*)*, i32 (i32*, i32*, i32, i32, i32, i32)*, i32 (i32, i32, i8*)*, void ()*, i32 (i32, i8*, i32*)*, i32 ()* }
+%struct.BD = type { %struct.BD*, i32, i32, i32, i32, i64, i32 (%struct.BD*, i8*, i64, i32)*, i32 (%struct.BD*, i8*, i32, i32)*, i32 (%struct.BD*, i8*, i64, i32)*, i32 (%struct.BD*, i8*, i32, i32)*, i32 (%struct.BD*, i64, i32)*, [16 x i8], i64, i64 }
+
+@FuncPtr = external hidden unnamed_addr global %struct.FF*
+@.str1 = external hidden unnamed_addr constant [6 x i8], align 4
+@G = external unnamed_addr global i32
+@.str2 = external hidden unnamed_addr constant [58 x i8], align 4
+@.str3 = external hidden unnamed_addr constant [58 x i8], align 4
+
+define i32 @test() nounwind optsize ssp {
+entry:
+; CHECK: test:
+; CHECK: push
+; CHECK-NOT: push
+  %block_size = alloca i32, align 4
+  %block_count = alloca i32, align 4
+  %index_cache = alloca i32, align 4
+  store i32 0, i32* %index_cache, align 4
+  %tmp = load i32* @G, align 4
+  %tmp1 = call i32 @bar(i32 0, i32 0, i32 %tmp) nounwind
+  switch i32 %tmp1, label %bb8 [
+    i32 0, label %bb
+    i32 536870913, label %bb4
+    i32 536870914, label %bb6
+  ]
+
+bb:
+  %tmp2 = load i32* @G, align 4
+  %tmp4 = icmp eq i32 %tmp2, 0
+  br i1 %tmp4, label %bb1, label %bb8
+
+bb1:
+; CHECK: %bb1
+; CHECK-NOT: umull
+; CHECK: blx _Get
+; CHECK: umull
+; CHECK: blx _foo
+  %tmp5 = load i32* %block_size, align 4
+  %tmp6 = load i32* %block_count, align 4
+  %tmp7 = call %struct.FF* @Get() nounwind
+  store %struct.FF* %tmp7, %struct.FF** @FuncPtr, align 4
+  %tmp10 = zext i32 %tmp6 to i64
+  %tmp11 = zext i32 %tmp5 to i64
+  %tmp12 = mul nsw i64 %tmp10, %tmp11
+  %tmp13 = call i32 @foo(i8* getelementptr inbounds ([6 x i8]* @.str1, i32 0, i32 0), i64 %tmp12, i32 %tmp5) nounwind
+  br label %bb8
+
+bb4:
+  ret i32 0
+
+bb6:
+  ret i32 1
+
+bb8:
+  ret i32 -1
+}
+
+declare i32 @printf(i8*, ...)
+
+declare %struct.FF* @Get()
+
+declare i32 @foo(i8*, i64, i32)
+
+declare i32 @bar(i32, i32, i32)
diff --git a/test/CodeGen/ARM/shifter_operand.ll b/test/CodeGen/ARM/shifter_operand.ll
index be891bc76d..f0e2d10261 100644
--- a/test/CodeGen/ARM/shifter_operand.ll
+++ b/test/CodeGen/ARM/shifter_operand.ll
@@ -58,7 +58,7 @@ entry:
 ; A8: str r2, [r0, r1, lsl #2]
 
 ; A9: test4:
-; A9: add r0, r0, r4, lsl #2
+; A9: add r0, r0, r{{[0-9]+}}, lsl #2
 ; A9: ldr r1, [r0]
 ; A9: str r1, [r0]
   %0 = tail call i8* (...)* @malloc(i32 undef) nounwind
diff --git a/test/CodeGen/Thumb/2010-01-15-local-alloc-spill-physical.ll b/test/CodeGen/Thumb/2010-01-15-local-alloc-spill-physical.ll
deleted file mode 100644
index fad26693e7..0000000000
--- a/test/CodeGen/Thumb/2010-01-15-local-alloc-spill-physical.ll
+++ /dev/null
@@ -1,20 +0,0 @@
-; RUN: llc < %s -regalloc=fast -relocation-model=pic | FileCheck %s
-
-target triple = "thumbv6-apple-darwin10"
-
-@fred = internal global i32 0              ; <i32*> [#uses=1]
-
-define void @foo() nounwind {
-entry:
-; CHECK: str r0, [sp
-  %0 = call  i32 (...)* @bar() nounwind ; <i32> [#uses=1]
-; CHECK: blx _bar
-; CHECK: ldr r1, [sp
-  store i32 %0, i32* @fred, align 4
-  br label %return
-
-return:                                           ; preds = %entry
-  ret void
-}
-
-declare i32 @bar(...)
diff --git a/test/CodeGen/Thumb2/2009-10-15-ITBlockBranch.ll b/test/CodeGen/Thumb2/2009-10-15-ITBlockBranch.ll
index 789a891697..3594424e29 100644
--- a/test/CodeGen/Thumb2/2009-10-15-ITBlockBranch.ll
+++ b/test/CodeGen/Thumb2/2009-10-15-ITBlockBranch.ll
@@ -13,7 +13,7 @@ define weak arm_aapcs_vfpcc i32 @_ZNKSs7compareERKSs(%"struct.std::basic_string<
 ; CHECK: _ZNKSs7compareERKSs:
 ; CHECK:      it  eq
 ; CHECK-NEXT: subeq r0, r{{[0-9]+}}, r{{[0-9]+}}
-; CHECK-NEXT: ldmia.w sp!, {r4, r5, r6, r7, r8, pc}
+; CHECK-NEXT: ldmia.w sp!,
 entry:
   %0 = tail call arm_aapcs_vfpcc  i32 @_ZNKSs4sizeEv(%"struct.std::basic_string<char,std::char_traits<char>,std::allocator<char> >"* %this) ; <i32> [#uses=3]
   %1 = tail call arm_aapcs_vfpcc  i32 @_ZNKSs4sizeEv(%"struct.std::basic_string<char,std::char_traits<char>,std::allocator<char> >"* %__str) ; <i32> [#uses=3]
diff --git a/test/CodeGen/X86/pic.ll b/test/CodeGen/X86/pic.ll
index dc5fcd78dc..fb60ac2a60 100644
--- a/test/CodeGen/X86/pic.ll
+++ b/test/CodeGen/X86/pic.ll
@@ -79,8 +79,8 @@ entry:
 ; LINUX-NEXT: .L3$pb:
 ; LINUX: 	popl
 ; LINUX: 	addl	$_GLOBAL_OFFSET_TABLE_+(.L{{.*}}-.L3$pb), %[[REG3:e..]]
-; LINUX: 	movl	pfoo@GOT(%[[REG3]]),
 ; LINUX: 	calll	afoo@PLT
+; LINUX: 	movl	pfoo@GOT(%[[REG3]]),
 ; LINUX: 	calll	*
 }