mirror of
https://github.com/c64scene-ar/llvm-6502.git
synced 2024-12-16 11:30:51 +00:00
1afcace3a3
patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@134829 91177308-0d34-0410-b5e6-96231b3b80d8
133 lines
4.2 KiB
LLVM
133 lines
4.2 KiB
LLVM
; RUN: opt < %s -basicaa -memcpyopt -dse -S | FileCheck %s
|
|
|
|
target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
|
|
target triple = "i686-apple-darwin9"
|
|
|
|
%0 = type { x86_fp80, x86_fp80 }
|
|
%1 = type { i32, i32 }
|
|
|
|
define void @test1(%0* sret %agg.result, x86_fp80 %z.0, x86_fp80 %z.1) nounwind {
|
|
entry:
|
|
%tmp2 = alloca %0
|
|
%memtmp = alloca %0, align 16
|
|
%tmp5 = fsub x86_fp80 0xK80000000000000000000, %z.1
|
|
call void @ccoshl(%0* sret %memtmp, x86_fp80 %tmp5, x86_fp80 %z.0) nounwind
|
|
%tmp219 = bitcast %0* %tmp2 to i8*
|
|
%memtmp20 = bitcast %0* %memtmp to i8*
|
|
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %tmp219, i8* %memtmp20, i32 32, i32 16, i1 false)
|
|
%agg.result21 = bitcast %0* %agg.result to i8*
|
|
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %agg.result21, i8* %tmp219, i32 32, i32 16, i1 false)
|
|
ret void
|
|
|
|
; Check that one of the memcpy's are removed.
|
|
;; FIXME: PR 8643 We should be able to eliminate the last memcpy here.
|
|
|
|
; CHECK: @test1
|
|
; CHECK: call void @ccoshl
|
|
; CHECK: call void @llvm.memcpy
|
|
; CHECK-NOT: llvm.memcpy
|
|
; CHECK: ret void
|
|
}
|
|
|
|
declare void @ccoshl(%0* sret , x86_fp80, x86_fp80) nounwind
|
|
|
|
|
|
; The intermediate alloca and one of the memcpy's should be eliminated, the
|
|
; other should be related with a memmove.
|
|
define void @test2(i8* %P, i8* %Q) nounwind {
|
|
%memtmp = alloca %0, align 16
|
|
%R = bitcast %0* %memtmp to i8*
|
|
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %R, i8* %P, i32 32, i32 16, i1 false)
|
|
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %Q, i8* %R, i32 32, i32 16, i1 false)
|
|
ret void
|
|
|
|
; CHECK: @test2
|
|
; CHECK-NEXT: call void @llvm.memmove{{.*}}(i8* %Q, i8* %P
|
|
; CHECK-NEXT: ret void
|
|
}
|
|
|
|
|
|
|
|
|
|
@x = external global %0
|
|
|
|
define void @test3(%0* noalias sret %agg.result) nounwind {
|
|
%x.0 = alloca %0
|
|
%x.01 = bitcast %0* %x.0 to i8*
|
|
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %x.01, i8* bitcast (%0* @x to i8*), i32 32, i32 16, i1 false)
|
|
%agg.result2 = bitcast %0* %agg.result to i8*
|
|
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %agg.result2, i8* %x.01, i32 32, i32 16, i1 false)
|
|
ret void
|
|
; CHECK: @test3
|
|
; CHECK-NEXT: %agg.result2 = bitcast
|
|
; CHECK-NEXT: call void @llvm.memcpy
|
|
; CHECK-NEXT: ret void
|
|
}
|
|
|
|
|
|
; PR8644
|
|
define void @test4(i8 *%P) {
|
|
%A = alloca %1
|
|
%a = bitcast %1* %A to i8*
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8* %a, i8* %P, i64 8, i32 4, i1 false)
|
|
call void @test4a(i8* byval align 1 %a)
|
|
ret void
|
|
; CHECK: @test4
|
|
; CHECK-NEXT: call void @test4a(
|
|
}
|
|
|
|
declare void @test4a(i8* byval align 1)
|
|
declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i32, i1) nounwind
|
|
|
|
%struct.S = type { i128, [4 x i8]}
|
|
|
|
@sS = external global %struct.S, align 16
|
|
|
|
declare void @test5a(%struct.S* byval align 16) nounwind ssp
|
|
|
|
|
|
; rdar://8713376 - This memcpy can't be eliminated.
|
|
define i32 @test5(i32 %x) nounwind ssp {
|
|
entry:
|
|
%y = alloca %struct.S, align 16
|
|
%tmp = bitcast %struct.S* %y to i8*
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp, i8* bitcast (%struct.S* @sS to i8*), i64 32, i32 16, i1 false)
|
|
%a = getelementptr %struct.S* %y, i64 0, i32 1, i64 0
|
|
store i8 4, i8* %a
|
|
call void @test5a(%struct.S* byval align 16 %y)
|
|
ret i32 0
|
|
; CHECK: @test5(
|
|
; CHECK: store i8 4
|
|
; CHECK: call void @test5a(%struct.S* byval align 16 %y)
|
|
}
|
|
|
|
;; Noop memcpy should be zapped.
|
|
define void @test6(i8 *%P) {
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8* %P, i8* %P, i64 8, i32 4, i1 false)
|
|
ret void
|
|
; CHECK: @test6
|
|
; CHECK-NEXT: ret void
|
|
}
|
|
|
|
|
|
; PR9794 - Should forward memcpy into byval argument even though the memcpy
|
|
; isn't itself 8 byte aligned.
|
|
%struct.p = type { i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32 }
|
|
|
|
define i32 @test7(%struct.p* nocapture byval align 8 %q) nounwind ssp {
|
|
entry:
|
|
%agg.tmp = alloca %struct.p, align 4
|
|
%tmp = bitcast %struct.p* %agg.tmp to i8*
|
|
%tmp1 = bitcast %struct.p* %q to i8*
|
|
call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp, i8* %tmp1, i64 48, i32 4, i1 false)
|
|
%call = call i32 @g(%struct.p* byval align 8 %agg.tmp) nounwind
|
|
ret i32 %call
|
|
; CHECK: @test7
|
|
; CHECK: call i32 @g(%struct.p* byval align 8 %q) nounwind
|
|
}
|
|
|
|
declare i32 @g(%struct.p* byval align 8)
|
|
|
|
declare void @llvm.memcpy.p0i8.p0i8.i32(i8* nocapture, i8* nocapture, i32, i32, i1) nounwind
|
|
|