maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/X86/select.ll

251 lines
6.2 KiB
LLVM
Raw Normal View History

fix the rest of the linux miscompares :) llvm-svn: 120933
2010-12-05 10:08:07 +08:00
; RUN: llc < %s -mtriple=x86_64-apple-darwin10 | FileCheck %s
Change this to the correct PR number. llvm-svn: 91148
2009-12-12 04:09:21 +08:00
; PR5757
Fix the result type of SELECT nodes lowered from Select instructions with aggregate return values. This fixes PR5754. llvm-svn: 91145
2009-12-12 03:50:50 +08:00
%0 = type { i64, i32 }
merge some tests into select.ll and make them more specific. llvm-svn: 120928
2010-12-05 09:13:58 +08:00
define i32 @test1(%0* %p, %0* %q, i1 %r) nounwind {
Fix the result type of SELECT nodes lowered from Select instructions with aggregate return values. This fixes PR5754. llvm-svn: 91145
2009-12-12 03:50:50 +08:00
%t0 = load %0* %p
%t1 = load %0* %q
%t4 = select i1 %r, %0 %t0, %0 %t1
%t5 = extractvalue %0 %t4, 1
ret i32 %t5
merge some tests into select.ll and make them more specific. llvm-svn: 120928
2010-12-05 09:13:58 +08:00
; CHECK: test1:
; CHECK: cmovneq %rdi, %rsi
; CHECK: movl (%rsi), %eax
}
; PR2139
define i32 @test2() nounwind {
entry:
%tmp73 = tail call i1 @return_false() ; <i8> [#uses=1]
%g.0 = select i1 %tmp73, i16 0, i16 -480 ; <i16> [#uses=2]
%tmp7778 = sext i16 %g.0 to i32 ; <i32> [#uses=1]
%tmp80 = shl i32 %tmp7778, 3 ; <i32> [#uses=2]
%tmp87 = icmp sgt i32 %tmp80, 32767 ; <i1> [#uses=1]
br i1 %tmp87, label %bb90, label %bb91
bb90: ; preds = %bb84, %bb72
unreachable
bb91: ; preds = %bb84
ret i32 0
; CHECK: test2:
; CHECK: movnew
As Dan pointed out, movzbl, movsbl, and friends are nicer than their alias (movzx/movsx) because they give more information. Revert that part of the patch. llvm-svn: 129498
2011-04-14 09:46:37 +08:00
; CHECK: movswl
merge some tests into select.ll and make them more specific. llvm-svn: 120928
2010-12-05 09:13:58 +08:00
}
declare i1 @return_false()
;; Select between two floating point constants.
define float @test3(i32 %x) nounwind readnone {
entry:
%0 = icmp eq i32 %x, 0 ; <i1> [#uses=1]
%iftmp.0.0 = select i1 %0, float 4.200000e+01, float 2.300000e+01 ; <float> [#uses=1]
ret float %iftmp.0.0
; CHECK: test3:
relax this to handle linux defaulting to -static. llvm-svn: 120930
2010-12-05 09:31:13 +08:00
; CHECK: movss {{.*}},4), %xmm0
merge some tests into select.ll and make them more specific. llvm-svn: 120928
2010-12-05 09:13:58 +08:00
}
define signext i8 @test4(i8* nocapture %P, double %F) nounwind readonly {
entry:
%0 = fcmp olt double %F, 4.200000e+01 ; <i1> [#uses=1]
%iftmp.0.0 = select i1 %0, i32 4, i32 0 ; <i32> [#uses=1]
%1 = getelementptr i8* %P, i32 %iftmp.0.0 ; <i8*> [#uses=1]
%2 = load i8* %1, align 1 ; <i8> [#uses=1]
ret i8 %2
; CHECK: test4:
; CHECK: movsbl ({{.*}},4), %eax
}
define void @test5(i1 %c, <2 x i16> %a, <2 x i16> %b, <2 x i16>* %p) nounwind {
%x = select i1 %c, <2 x i16> %a, <2 x i16> %b
store <2 x i16> %x, <2 x i16>* %p
ret void
; CHECK: test5:
}
define void @test6(i32 %C, <4 x float>* %A, <4 x float>* %B) nounwind {
%tmp = load <4 x float>* %A ; <<4 x float>> [#uses=1]
%tmp3 = load <4 x float>* %B ; <<4 x float>> [#uses=2]
%tmp9 = fmul <4 x float> %tmp3, %tmp3 ; <<4 x float>> [#uses=1]
%tmp.upgrd.1 = icmp eq i32 %C, 0 ; <i1> [#uses=1]
%iftmp.38.0 = select i1 %tmp.upgrd.1, <4 x float> %tmp9, <4 x float> %tmp ; <<4 x float>> [#uses=1]
store <4 x float> %iftmp.38.0, <4 x float>* %A
ret void
; Verify that the fmul gets sunk into the one part of the diamond where it is
; needed.
; CHECK: test6:
Flip the new block-placement pass to be on by default. This is mostly to test the waters. I'd like to get results from FNT build bots and other bots running on non-x86 platforms. This feature has been pretty heavily tested over the last few months by me, and it fixes several of the execution time regressions caused by the inlining work by preventing inlining decisions from radically impacting block layout. I've seen very large improvements in yacr2 and ackermann benchmarks, along with the expected noise across all of the benchmark suite whenever code layout changes. I've analyzed all of the regressions and fixed them, or found them to be impossible to fix. See my email to llvmdev for more details. I'd like for this to be in 3.1 as it complements the inliner changes, but if any failures are showing up or anyone has concerns, it is just a flag flip and so can be easily turned off. I'm switching it on tonight to try and get at least one run through various folks' performance suites in case SPEC or something else has serious issues with it. I'll watch bots and revert if anything shows up. llvm-svn: 154816
2012-04-16 21:49:17 +08:00
; CHECK: je
merge some tests into select.ll and make them more specific. llvm-svn: 120928
2010-12-05 09:13:58 +08:00
; CHECK: ret
Flip the new block-placement pass to be on by default. This is mostly to test the waters. I'd like to get results from FNT build bots and other bots running on non-x86 platforms. This feature has been pretty heavily tested over the last few months by me, and it fixes several of the execution time regressions caused by the inlining work by preventing inlining decisions from radically impacting block layout. I've seen very large improvements in yacr2 and ackermann benchmarks, along with the expected noise across all of the benchmark suite whenever code layout changes. I've analyzed all of the regressions and fixed them, or found them to be impossible to fix. See my email to llvmdev for more details. I'd like for this to be in 3.1 as it complements the inliner changes, but if any failures are showing up or anyone has concerns, it is just a flag flip and so can be easily turned off. I'm switching it on tonight to try and get at least one run through various folks' performance suites in case SPEC or something else has serious issues with it. I'll watch bots and revert if anything shows up. llvm-svn: 154816
2012-04-16 21:49:17 +08:00
; CHECK: mulps
merge some tests into select.ll and make them more specific. llvm-svn: 120928
2010-12-05 09:13:58 +08:00
; CHECK: ret
}
; Select with fp80's
define x86_fp80 @test7(i32 %tmp8) nounwind {
%tmp9 = icmp sgt i32 %tmp8, -1 ; <i1> [#uses=1]
%retval = select i1 %tmp9, x86_fp80 0xK4005B400000000000000, x86_fp80 0xK40078700000000000000
ret x86_fp80 %retval
Improve an integer select optimization in two ways: 1. generalize (select (x == 0), -1, 0) -> (sign_bit (x - 1)) to: (select (x == 0), -1, y) -> (sign_bit (x - 1)) | y 2. Handle the identical pattern that happens with !=: (select (x != 0), y, -1) -> (sign_bit (x - 1)) | y cmov is often high latency and can't fold immediates or memory operands. For example for (x == 0) ? -1 : 1, before we got: < testb %sil, %sil < movl $-1, %ecx < movl $1, %eax < cmovel %ecx, %eax now we get: > cmpb $1, %sil > sbbl %eax, %eax > orl $1, %eax llvm-svn: 120929
2010-12-05 09:23:24 +08:00
; CHECK: test7:
merge some tests into select.ll and make them more specific. llvm-svn: 120928
2010-12-05 09:13:58 +08:00
; CHECK: leaq
; CHECK: fldt (%r{{.}}x,%r{{.}}x)
}
; widening select v6i32 and then a sub
define void @test8(i1 %c, <6 x i32>* %dst.addr, <6 x i32> %src1,<6 x i32> %src2) nounwind {
%x = select i1 %c, <6 x i32> %src1, <6 x i32> %src2
%val = sub <6 x i32> %x, < i32 1, i32 1, i32 1, i32 1, i32 1, i32 1 >
store <6 x i32> %val, <6 x i32>* %dst.addr
ret void
test/CodeGen/X86/select.ll: remove spaces llvm-svn: 155840
2012-05-01 02:54:27 +08:00
Improve an integer select optimization in two ways: 1. generalize (select (x == 0), -1, 0) -> (sign_bit (x - 1)) to: (select (x == 0), -1, y) -> (sign_bit (x - 1)) | y 2. Handle the identical pattern that happens with !=: (select (x != 0), y, -1) -> (sign_bit (x - 1)) | y cmov is often high latency and can't fold immediates or memory operands. For example for (x == 0) ? -1 : 1, before we got: < testb %sil, %sil < movl $-1, %ecx < movl $1, %eax < cmovel %ecx, %eax now we get: > cmpb $1, %sil > sbbl %eax, %eax > orl $1, %eax llvm-svn: 120929
2010-12-05 09:23:24 +08:00
; CHECK: test8:
Fix the result type of SELECT nodes lowered from Select instructions with aggregate return values. This fixes PR5754. llvm-svn: 91145
2009-12-12 03:50:50 +08:00
}
Improve an integer select optimization in two ways: 1. generalize (select (x == 0), -1, 0) -> (sign_bit (x - 1)) to: (select (x == 0), -1, y) -> (sign_bit (x - 1)) | y 2. Handle the identical pattern that happens with !=: (select (x != 0), y, -1) -> (sign_bit (x - 1)) | y cmov is often high latency and can't fold immediates or memory operands. For example for (x == 0) ? -1 : 1, before we got: < testb %sil, %sil < movl $-1, %ecx < movl $1, %eax < cmovel %ecx, %eax now we get: > cmpb $1, %sil > sbbl %eax, %eax > orl $1, %eax llvm-svn: 120929
2010-12-05 09:23:24 +08:00
;; Test integer select between values and constants.
define i64 @test9(i64 %x, i64 %y) nounwind readnone ssp noredzone {
%cmp = icmp ne i64 %x, 0
%cond = select i1 %cmp, i64 %y, i64 -1
ret i64 %cond
; CHECK: test9:
; CHECK: cmpq $1, %rdi
; CHECK: sbbq %rax, %rax
; CHECK: orq %rsi, %rax
; CHECK: ret
}
;; Same as test9
define i64 @test9a(i64 %x, i64 %y) nounwind readnone ssp noredzone {
%cmp = icmp eq i64 %x, 0
%cond = select i1 %cmp, i64 -1, i64 %y
ret i64 %cond
; CHECK: test9a:
; CHECK: cmpq $1, %rdi
; CHECK: sbbq %rax, %rax
; CHECK: orq %rsi, %rax
; CHECK: ret
}
define i64 @test9b(i64 %x, i64 %y) nounwind readnone ssp noredzone {
%cmp = icmp eq i64 %x, 0
%A = sext i1 %cmp to i64
%cond = or i64 %y, %A
ret i64 %cond
; CHECK: test9b:
; CHECK: cmpq $1, %rdi
; CHECK: sbbq %rax, %rax
; CHECK: orq %rsi, %rax
; CHECK: ret
}
;; Select between -1 and 1.
define i64 @test10(i64 %x, i64 %y) nounwind readnone ssp noredzone {
%cmp = icmp eq i64 %x, 0
%cond = select i1 %cmp, i64 -1, i64 1
ret i64 %cond
; CHECK: test10:
; CHECK: cmpq $1, %rdi
; CHECK: sbbq %rax, %rax
; CHECK: orq $1, %rax
; CHECK: ret
}
generalize the previous check to handle -1 on either side of the select, inserting a not to compensate. Add a missing isZero check that I lost somehow. This improves codegen of: void *func(long count) { return new int[count]; } from: __Z4funcl: ## @_Z4funcl movl $4, %ecx ## encoding: [0xb9,0x04,0x00,0x00,0x00] movq %rdi, %rax ## encoding: [0x48,0x89,0xf8] mulq %rcx ## encoding: [0x48,0xf7,0xe1] testq %rdx, %rdx ## encoding: [0x48,0x85,0xd2] movq $-1, %rdi ## encoding: [0x48,0xc7,0xc7,0xff,0xff,0xff,0xff] cmoveq %rax, %rdi ## encoding: [0x48,0x0f,0x44,0xf8] jmp __Znam ## TAILCALL ## encoding: [0xeb,A] to: __Z4funcl: ## @_Z4funcl movl $4, %ecx ## encoding: [0xb9,0x04,0x00,0x00,0x00] movq %rdi, %rax ## encoding: [0x48,0x89,0xf8] mulq %rcx ## encoding: [0x48,0xf7,0xe1] cmpq $1, %rdx ## encoding: [0x48,0x83,0xfa,0x01] sbbq %rdi, %rdi ## encoding: [0x48,0x19,0xff] notq %rdi ## encoding: [0x48,0xf7,0xd7] orq %rax, %rdi ## encoding: [0x48,0x09,0xc7] jmp __Znam ## TAILCALL ## encoding: [0xeb,A] llvm-svn: 120932
2010-12-05 10:00:51 +08:00
define i64 @test11(i64 %x, i64 %y) nounwind readnone ssp noredzone {
%cmp = icmp eq i64 %x, 0
%cond = select i1 %cmp, i64 %y, i64 -1
ret i64 %cond
; CHECK: test11:
; CHECK: cmpq $1, %rdi
; CHECK: sbbq %rax, %rax
; CHECK: notq %rax
; CHECK: orq %rsi, %rax
; CHECK: ret
}
define i64 @test11a(i64 %x, i64 %y) nounwind readnone ssp noredzone {
%cmp = icmp ne i64 %x, 0
%cond = select i1 %cmp, i64 -1, i64 %y
ret i64 %cond
; CHECK: test11a:
; CHECK: cmpq $1, %rdi
; CHECK: sbbq %rax, %rax
; CHECK: notq %rax
; CHECK: orq %rsi, %rax
; CHECK: ret
}
Improve an integer select optimization in two ways: 1. generalize (select (x == 0), -1, 0) -> (sign_bit (x - 1)) to: (select (x == 0), -1, y) -> (sign_bit (x - 1)) | y 2. Handle the identical pattern that happens with !=: (select (x != 0), y, -1) -> (sign_bit (x - 1)) | y cmov is often high latency and can't fold immediates or memory operands. For example for (x == 0) ? -1 : 1, before we got: < testb %sil, %sil < movl $-1, %ecx < movl $1, %eax < cmovel %ecx, %eax now we get: > cmpb $1, %sil > sbbl %eax, %eax > orl $1, %eax llvm-svn: 120929
2010-12-05 09:23:24 +08:00
Teach X86ISelLowering that the second result of X86ISD::UMUL is a flags result. This allows us to compile: void *test12(long count) { return new int[count]; } into: test12: movl $4, %ecx movq %rdi, %rax mulq %rcx movq $-1, %rdi cmovnoq %rax, %rdi jmp __Znam ## TAILCALL instead of: test12: movl $4, %ecx movq %rdi, %rax mulq %rcx seto %cl testb %cl, %cl movq $-1, %rdi cmoveq %rax, %rdi jmp __Znam Of course it would be even better if the regalloc inverted the cmov to 'cmovoq', which would eliminate the need for the 'movq %rdi, %rax'. llvm-svn: 120936
2010-12-05 15:49:54 +08:00
declare noalias i8* @_Znam(i64) noredzone
define noalias i8* @test12(i64 %count) nounwind ssp noredzone {
entry:
%A = tail call { i64, i1 } @llvm.umul.with.overflow.i64(i64 %count, i64 4)
%B = extractvalue { i64, i1 } %A, 1
%C = extractvalue { i64, i1 } %A, 0
%D = select i1 %B, i64 -1, i64 %C
%call = tail call noalias i8* @_Znam(i64 %D) nounwind noredzone
ret i8* %call
; CHECK: test12:
; CHECK: mulq
; CHECK: movq $-1, %rdi
; CHECK: cmovnoq %rax, %rdi
; CHECK: jmp __Znam
}
declare { i64, i1 } @llvm.umul.with.overflow.i64(i64, i64) nounwind readnone
X86: Lower a select directly to a setcc_carry if possible. int test(unsigned long a, unsigned long b) { return -(a < b); } compiles to _test: ## @test cmpq %rsi, %rdi ## encoding: [0x48,0x39,0xf7] sbbl %eax, %eax ## encoding: [0x19,0xc0] ret ## encoding: [0xc3] instead of _test: ## @test xorl %ecx, %ecx ## encoding: [0x31,0xc9] cmpq %rsi, %rdi ## encoding: [0x48,0x39,0xf7] movl $-1, %eax ## encoding: [0xb8,0xff,0xff,0xff,0xff] cmovael %ecx, %eax ## encoding: [0x0f,0x43,0xc1] ret ## encoding: [0xc3] llvm-svn: 122451
2010-12-23 07:09:28 +08:00
define i32 @test13(i32 %a, i32 %b) nounwind {
%c = icmp ult i32 %a, %b
%d = sext i1 %c to i32
ret i32 %d
; CHECK: test13:
; CHECK: cmpl
; CHECK-NEXT: sbbl
; CHECK-NEXT: ret
}
Teach X86ISelLowering that the second result of X86ISD::UMUL is a flags result. This allows us to compile: void *test12(long count) { return new int[count]; } into: test12: movl $4, %ecx movq %rdi, %rax mulq %rcx movq $-1, %rdi cmovnoq %rax, %rdi jmp __Znam ## TAILCALL instead of: test12: movl $4, %ecx movq %rdi, %rax mulq %rcx seto %cl testb %cl, %cl movq $-1, %rdi cmoveq %rax, %rdi jmp __Znam Of course it would be even better if the regalloc inverted the cmov to 'cmovoq', which would eliminate the need for the 'movq %rdi, %rax'. llvm-svn: 120936
2010-12-05 15:49:54 +08:00
X86: Lower a select directly to a setcc_carry if possible. int test(unsigned long a, unsigned long b) { return -(a < b); } compiles to _test: ## @test cmpq %rsi, %rdi ## encoding: [0x48,0x39,0xf7] sbbl %eax, %eax ## encoding: [0x19,0xc0] ret ## encoding: [0xc3] instead of _test: ## @test xorl %ecx, %ecx ## encoding: [0x31,0xc9] cmpq %rsi, %rdi ## encoding: [0x48,0x39,0xf7] movl $-1, %eax ## encoding: [0xb8,0xff,0xff,0xff,0xff] cmovael %ecx, %eax ## encoding: [0x0f,0x43,0xc1] ret ## encoding: [0xc3] llvm-svn: 122451
2010-12-23 07:09:28 +08:00
define i32 @test14(i32 %a, i32 %b) nounwind {
%c = icmp uge i32 %a, %b
%d = sext i1 %c to i32
ret i32 %d
; CHECK: test14:
; CHECK: cmpl
; CHECK-NEXT: sbbl
; CHECK-NEXT: notl
; CHECK-NEXT: ret
}
Improve an integer select optimization in two ways: 1. generalize (select (x == 0), -1, 0) -> (sign_bit (x - 1)) to: (select (x == 0), -1, y) -> (sign_bit (x - 1)) | y 2. Handle the identical pattern that happens with !=: (select (x != 0), y, -1) -> (sign_bit (x - 1)) | y cmov is often high latency and can't fold immediates or memory operands. For example for (x == 0) ? -1 : 1, before we got: < testb %sil, %sil < movl $-1, %ecx < movl $1, %eax < cmovel %ecx, %eax now we get: > cmpb $1, %sil > sbbl %eax, %eax > orl $1, %eax llvm-svn: 120929
2010-12-05 09:23:24 +08:00
X86: optimization for -(x != 0) This patch will optimize -(x != 0) on X86 FROM cmpl $0x01,%edi sbbl %eax,%eax notl %eax TO negl %edi sbbl %eax %eax In order to generate negl, I added patterns in Target/X86/X86InstrCompiler.td: def : Pat<(X86sub_flag 0, GR32:$src), (NEG32r GR32:$src)>; rdar: 10961709 llvm-svn: 156312
2012-05-08 02:06:23 +08:00
; rdar://10961709
define i32 @test15(i32 %x) nounwind {
entry:
%cmp = icmp ne i32 %x, 0
%sub = sext i1 %cmp to i32
ret i32 %sub
; CHECK: test15:
; CHECK: negl
; CHECK: sbbl
}
define i64 @test16(i64 %x) nounwind uwtable readnone ssp {
entry:
%cmp = icmp ne i64 %x, 0
%conv1 = sext i1 %cmp to i64
ret i64 %conv1
; CHECK: test16:
; CHECK: negq
; CHECK: sbbq
}
define i16 @test17(i16 %x) nounwind {
entry:
%cmp = icmp ne i16 %x, 0
%sub = sext i1 %cmp to i16
ret i16 %sub
; CHECK: test17:
; CHECK: negw
; CHECK: sbbw
}