mirror of
https://github.com/c64scene-ar/llvm-6502.git
synced 2024-09-10 22:55:57 +00:00
9e59c64c14
memdep keeps track of how PHIs affect the pointer in dep queries, which allows it to eliminate the load in cases like rle-phi-translate.ll, which basically end up being: BB1: X = load P br BB3 BB2: Y = load Q br BB3 BB3: R = phi [P] [Q] load R turning "load R" into a phi of X/Y. In addition to additional exposed opportunities, this makes memdep safe in many cases that it wasn't before (which is required for load PRE) and also makes it substantially more efficient. For example, consider: bb1: // has many predecessors. P = some_operator() load P In this example, previously memdep would scan all the predecessors of BB1 to see if they had something that would mustalias P. In some cases (e.g. test/Transforms/GVN/rle-must-alias.ll) it would actually find them and end up eliminating something. In many other cases though, it would scan and not find anything useful. MemDep now stops at a block if the pointer is defined in that block and cannot be phi translated to predecessors. This causes it to miss the (rare) cases like rle-must-alias.ll, but makes it faster by not scanning tons of stuff that is unlikely to be useful. For example, this speeds up GVN as a whole from 3.928s to 2.448s (60%)!. IMO, scalar GVN should be enhanced to simplify the rle-must-alias pointer base anyway, which would allow the loads to be eliminated. In the future, this should be enhanced to phi translate through geps and bitcasts as well (as indicated by FIXMEs) making memdep even more powerful. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61022 91177308-0d34-0410-b5e6-96231b3b80d8
48 lines
1.5 KiB
LLVM
48 lines
1.5 KiB
LLVM
; RUN: llvm-as < %s | opt -gvn | llvm-dis | grep {DEAD.rle = phi i32}
|
|
; XFAIL: *
|
|
|
|
; FIXME: GVN should eliminate the fully redundant %9 GEP which
|
|
; allows DEAD to be removed. This is PR3198.
|
|
|
|
; The %7 and %4 loads combine to make %DEAD unneeded.
|
|
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 = "i386-apple-darwin7"
|
|
@H = common global [100 x i32] zeroinitializer, align 32 ; <[100 x i32]*> [#uses=3]
|
|
@G = common global i32 0 ; <i32*> [#uses=2]
|
|
|
|
define i32 @test(i32 %i) nounwind {
|
|
entry:
|
|
%0 = tail call i32 (...)* @foo() nounwind ; <i32> [#uses=1]
|
|
%1 = icmp eq i32 %0, 0 ; <i1> [#uses=1]
|
|
br i1 %1, label %bb1, label %bb
|
|
|
|
bb: ; preds = %entry
|
|
%2 = tail call i32 (...)* @bar() nounwind ; <i32> [#uses=0]
|
|
%3 = getelementptr [100 x i32]* @H, i32 0, i32 %i ; <i32*> [#uses=1]
|
|
%4 = load i32* %3, align 4 ; <i32> [#uses=1]
|
|
store i32 %4, i32* @G, align 4
|
|
br label %bb3
|
|
|
|
bb1: ; preds = %entry
|
|
%5 = tail call i32 (...)* @baz() nounwind ; <i32> [#uses=0]
|
|
%6 = getelementptr [100 x i32]* @H, i32 0, i32 %i ; <i32*> [#uses=1]
|
|
%7 = load i32* %6, align 4 ; <i32> [#uses=2]
|
|
store i32 %7, i32* @G, align 4
|
|
%8 = icmp eq i32 %7, 0 ; <i1> [#uses=1]
|
|
br i1 %8, label %bb3, label %bb4
|
|
|
|
bb3: ; preds = %bb1, %bb
|
|
%9 = getelementptr [100 x i32]* @H, i32 0, i32 %i ; <i32*> [#uses=1]
|
|
%DEAD = load i32* %9, align 4 ; <i32> [#uses=1]
|
|
ret i32 %DEAD
|
|
|
|
bb4: ; preds = %bb1
|
|
ret i32 0
|
|
}
|
|
|
|
declare i32 @foo(...)
|
|
|
|
declare i32 @bar(...)
|
|
|
|
declare i32 @baz(...)
|