mirror of
https://github.com/RPCSX/llvm.git
synced 2025-01-09 13:41:35 +00:00
d380dee4ce
Summary:
**Description**
This makes `WidenIV::widenIVUse` (IndVarSimplify.cpp) fail to widen narrow IV uses in some cases. The latter affects IndVarSimplify which may not eliminate narrow IV's when there actually exists such a possibility, thereby producing ineffective code.
When `WidenIV::widenIVUse` gets a NarrowUse such as `{(-2 + %inc.lcssa),+,1}<nsw><%for.body3>`, it first tries to get a wide recurrence for it via the `getWideRecurrence` call.
`getWideRecurrence` returns recurrence like this: `{(sext i32 (-2 + %inc.lcssa) to i64),+,1}<nsw><%for.body3>`.
Then a wide use operation is generated by `cloneIVUser`. The generated wide use is evaluated to `{(-2 + (sext i32 %inc.lcssa to i64))<nsw>,+,1}<nsw><%for.body3>`, which is different from the `getWideRecurrence` result. `cloneIVUser` sees the difference and returns nullptr.
This patch also fixes the broken LLVM tests by adding missing <nsw> entries introduced by the correction.
**Minimal reproducer:**
```
int foo(int a, int b, int c);
int baz();
void bar()
{
int arr[20];
int i = 0;
for (i = 0; i < 4; ++i)
arr[i] = baz();
for (; i < 20; ++i)
arr[i] = foo(arr[i - 4], arr[i - 3], arr[i - 2]);
}
```
**Clang command line:**
```
clang++ -mllvm -debug -S -emit-llvm -O3 --target=aarch64-linux-elf test.cpp -o test.ir
```
**Expected result:**
The ` -mllvm -debug` log shows that all the IV's for the second `for` loop have been eliminated.
Reviewers: sanjoy
Subscribers: atrick, asl, aemerson, mzolotukhin, llvm-commits
Differential Revision: http://reviews.llvm.org/D20058
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@270695 91177308-0d34-0410-b5e6-96231b3b80d8
84 lines
3.6 KiB
LLVM
84 lines
3.6 KiB
LLVM
; RUN: opt < %s -S -analyze -scalar-evolution | FileCheck %s
|
|
|
|
; ScalarEvolution should be able to fold away the sign-extensions
|
|
; on this loop with a primary induction variable incremented with
|
|
; a nsw add of 2 (this test is derived from the nsw-offset.ll test, but uses an
|
|
; assume instead of a preheader conditional branch to guard the loop).
|
|
|
|
target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128"
|
|
|
|
define void @foo(i32 %no, double* nocapture %d, double* nocapture %q) nounwind {
|
|
entry:
|
|
%n = and i32 %no, 4294967294
|
|
%0 = icmp sgt i32 %n, 0 ; <i1> [#uses=1]
|
|
tail call void @llvm.assume(i1 %0)
|
|
br label %bb.nph
|
|
|
|
bb.nph: ; preds = %entry
|
|
br label %bb
|
|
|
|
bb: ; preds = %bb.nph, %bb1
|
|
%i.01 = phi i32 [ %16, %bb1 ], [ 0, %bb.nph ] ; <i32> [#uses=5]
|
|
|
|
; CHECK: %1 = sext i32 %i.01 to i64
|
|
; CHECK: --> {0,+,2}<nuw><nsw><%bb>
|
|
%1 = sext i32 %i.01 to i64 ; <i64> [#uses=1]
|
|
|
|
; CHECK: %2 = getelementptr inbounds double, double* %d, i64 %1
|
|
; CHECK: --> {%d,+,16}<nsw><%bb>
|
|
%2 = getelementptr inbounds double, double* %d, i64 %1 ; <double*> [#uses=1]
|
|
|
|
%3 = load double, double* %2, align 8 ; <double> [#uses=1]
|
|
%4 = sext i32 %i.01 to i64 ; <i64> [#uses=1]
|
|
%5 = getelementptr inbounds double, double* %q, i64 %4 ; <double*> [#uses=1]
|
|
%6 = load double, double* %5, align 8 ; <double> [#uses=1]
|
|
%7 = or i32 %i.01, 1 ; <i32> [#uses=1]
|
|
|
|
; CHECK: %8 = sext i32 %7 to i64
|
|
; CHECK: --> {1,+,2}<nuw><nsw><%bb>
|
|
%8 = sext i32 %7 to i64 ; <i64> [#uses=1]
|
|
|
|
; CHECK: %9 = getelementptr inbounds double, double* %q, i64 %8
|
|
; CHECK: {(8 + %q)<nsw>,+,16}<nsw><%bb>
|
|
%9 = getelementptr inbounds double, double* %q, i64 %8 ; <double*> [#uses=1]
|
|
|
|
; Artificially repeat the above three instructions, this time using
|
|
; add nsw instead of or.
|
|
%t7 = add nsw i32 %i.01, 1 ; <i32> [#uses=1]
|
|
|
|
; CHECK: %t8 = sext i32 %t7 to i64
|
|
; CHECK: --> {1,+,2}<nuw><nsw><%bb>
|
|
%t8 = sext i32 %t7 to i64 ; <i64> [#uses=1]
|
|
|
|
; CHECK: %t9 = getelementptr inbounds double, double* %q, i64 %t8
|
|
; CHECK: {(8 + %q)<nsw>,+,16}<nsw><%bb>
|
|
%t9 = getelementptr inbounds double, double* %q, i64 %t8 ; <double*> [#uses=1]
|
|
|
|
%10 = load double, double* %9, align 8 ; <double> [#uses=1]
|
|
%11 = fadd double %6, %10 ; <double> [#uses=1]
|
|
%12 = fadd double %11, 3.200000e+00 ; <double> [#uses=1]
|
|
%13 = fmul double %3, %12 ; <double> [#uses=1]
|
|
%14 = sext i32 %i.01 to i64 ; <i64> [#uses=1]
|
|
%15 = getelementptr inbounds double, double* %d, i64 %14 ; <double*> [#uses=1]
|
|
store double %13, double* %15, align 8
|
|
%16 = add nsw i32 %i.01, 2 ; <i32> [#uses=2]
|
|
br label %bb1
|
|
|
|
bb1: ; preds = %bb
|
|
%17 = icmp slt i32 %16, %n ; <i1> [#uses=1]
|
|
br i1 %17, label %bb, label %bb1.return_crit_edge
|
|
|
|
bb1.return_crit_edge: ; preds = %bb1
|
|
br label %return
|
|
|
|
return: ; preds = %bb1.return_crit_edge, %entry
|
|
ret void
|
|
}
|
|
|
|
declare void @llvm.assume(i1) nounwind
|
|
|
|
; Note: Without the preheader assume, there is an 'smax' in the
|
|
; backedge-taken count expression:
|
|
; CHECK: Loop %bb: backedge-taken count is ((-1 + (2 * (%no /u 2))) /u 2)
|
|
; CHECK: Loop %bb: max backedge-taken count is 1073741822
|