mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-10-24 05:23:45 +02:00
81eb2d6bae
32-bit x86 Windows targets use a linked-list of nodes allocated on the stack, referenced to via thread-local storage. The personality routine interprets one of the fields in the node as a 'state number' which indicates where the personality routine should transfer control. State transitions are possible only before call-sites which may throw exceptions. Our previous scheme had us update the state number before all call-sites which may throw. Instead, we can try to minimize the number of times we need to store by reasoning about the nearest store which dominates the current call-site. If the last store agrees with the current call-site, then we know that the state-update is redundant and can be elided. This is largely straightforward: an RPO walk of the blocks allows us to correctly forward propagate the information when the function is a DAG. Currently, loops are not handled optimally and may trigger superfluous state stores. Differential Revision: http://reviews.llvm.org/D16763 llvm-svn: 261122
69 lines
2.3 KiB
LLVM
69 lines
2.3 KiB
LLVM
; RUN: llc < %s | FileCheck %s
|
|
|
|
; Based on this C++:
|
|
; struct A {
|
|
; int x;
|
|
; A();
|
|
; A(const A &a);
|
|
; ~A();
|
|
; };
|
|
; extern "C" void takes_two(A a1, A a2);
|
|
; extern "C" void passes_two() { takes_two(A(), A()); }
|
|
|
|
target datalayout = "e-m:x-p:32:32-i64:64-f80:32-n8:16:32-a:0:32-S32"
|
|
target triple = "i686--windows-msvc"
|
|
|
|
%struct.A = type { i32 }
|
|
|
|
define void @passes_two() #0 personality i8* bitcast (i32 (...)* @__CxxFrameHandler3 to i8*) {
|
|
entry:
|
|
%argmem = alloca inalloca <{ %struct.A, %struct.A }>, align 4
|
|
%0 = getelementptr inbounds <{ %struct.A, %struct.A }>, <{ %struct.A, %struct.A }>* %argmem, i32 0, i32 1
|
|
%call = call x86_thiscallcc %struct.A* @"\01??0A@@QAE@XZ"(%struct.A* %0)
|
|
%1 = getelementptr inbounds <{ %struct.A, %struct.A }>, <{ %struct.A, %struct.A }>* %argmem, i32 0, i32 0
|
|
%call1 = invoke x86_thiscallcc %struct.A* @"\01??0A@@QAE@XZ"(%struct.A* %1)
|
|
to label %invoke.cont unwind label %ehcleanup
|
|
|
|
invoke.cont: ; preds = %entry
|
|
call void @takes_two(<{ %struct.A, %struct.A }>* inalloca nonnull %argmem)
|
|
ret void
|
|
|
|
ehcleanup: ; preds = %entry
|
|
%2 = cleanuppad within none []
|
|
call x86_thiscallcc void @"\01??1A@@QAE@XZ"(%struct.A* %0) [ "funclet"(token %2) ]
|
|
cleanupret from %2 unwind to caller
|
|
}
|
|
|
|
; CHECK: _passes_two:
|
|
; CHECK: pushl %ebp
|
|
; CHECK: movl %esp, %ebp
|
|
; CHECK: subl ${{[0-9]+}}, %esp
|
|
; CHECK: movl $8, %eax
|
|
; CHECK: calll __chkstk
|
|
; CHECK: calll "??0A@@QAE@XZ"
|
|
; CHECK: calll "??0A@@QAE@XZ"
|
|
; CHECK: calll _takes_two
|
|
; ESP must be restored via EBP due to "dynamic" alloca.
|
|
; CHECK: leal -{{[0-9]+}}(%ebp), %esp
|
|
; CHECK: popl %ebp
|
|
; CHECK: retl
|
|
|
|
; CHECK: "?dtor$2@?0?passes_two@4HA":
|
|
; CHECK: pushl %ebp
|
|
; CHECK: subl $8, %esp
|
|
; CHECK: addl $16, %ebp
|
|
; CHECK: {{movl|leal}} -{{[0-9]+}}(%ebp), %ecx
|
|
; CHECK: calll "??1A@@QAE@XZ"
|
|
; CHECK: addl $8, %esp
|
|
; CHECK: retl
|
|
|
|
declare void @takes_two(<{ %struct.A, %struct.A }>* inalloca) #0
|
|
|
|
declare x86_thiscallcc %struct.A* @"\01??0A@@QAE@XZ"(%struct.A* returned) #0
|
|
|
|
declare i32 @__CxxFrameHandler3(...)
|
|
|
|
declare x86_thiscallcc void @"\01??1A@@QAE@XZ"(%struct.A*) #0
|
|
|
|
attributes #0 = { "disable-tail-calls"="false" "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
|