shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions 6 Packages Projects Releases Wiki Activity
llvm-project/llvm/test/Transforms/LoopUnroll/runtime-exit-phi-scev-invalidation.ll
Joel E. Denny 6d44b9082e
[LoopUnroll] Skip remainder loop guard if skip unrolled loop (#156549) 
The original loop (OL) that serves as input to LoopUnroll has basic
blocks that are arranged as follows:

```
OLPreHeader
OLHeader <-.
...        |
OLLatch ---'
OLExit
```

In this depiction, every block has an implicit edge to the next block
below, so any explicit edge indicates a conditional branch.

Given OL and unroll count N, LoopUnroll sometimes creates an unrolled
loop (UL) with a remainder loop (RL) epilogue arranged like this:

```
,-- ULGuard
|   ULPreHeader
|   ULHeader <-.
|   ...        |
|   ULLatch ---'
|   ULExit
`-> RLGuard -----.
    RLPreHeader  |
,-> RLHeader     |
|   ...          |
`-- RLLatch      |
    RLExit       |
    OLExit <-----'
```

Each UL iteration executes N OL iterations, but each RL iteration
executes 1 OL iteration. ULGuard or RLGuard checks whether the first
iteration of UL or RL should execute, respectively. If so, ULLatch or
RLLatch checks whether to execute each subsequent iteration.

Once reached, OL always executes its first iteration but not necessarily
the next N-1 iterations. Thus, ULGuard is always required before the
first UL iteration. However, when control flows from ULGuard directly to
RLGuard, the first OL iteration has yet to execute, so RLGuard is then
redundant before the first RL iteration.

Thus, this patch makes the following changes:
- Adjust ULGuard to branch to RLPreHeader instead of RLGuard, thus
eliminating RLGuard's unnecessary branch instruction for that path.
- Eliminate the creation of RLGuard phi node poison values. Without this
patch, RLGuard has such a phi node for each value that is defined by any
OL iteration and used in OLExit. The poison value is required where
ULGuard is the predecessor. The poison value indicates that control flow
from ULGuard to RLGuard to Exit has no counterpart in OL because the
first OL iteration must execute either in UL or RL.
- Simplify the CFG by not splitting ULExit and RLGuard because, without
the ULGuard predecessor, the single block can now be a dedicated UL
exit.
- To RLPreHeader, add an `llvm.assume` call that asserts the RL trip
count is non-zero. Without this patch, RLPreHeader is reachable only
when RLGuard guarantees that assertion is true. With this patch, RLGuard
guarantees it only when RLGuard is the predecessor, and the OL structure
guarantees it when ULGuard is the predecessor. If RL itself is unrolled
later, this guarantee somehow prevents ScalarEvolution from giving up
when trying to compute a maximum trip count for RL. That maximum trip
count enables the branch instruction in the final unrolled instance of
RLLatch to be eliminated. Without the `llvm.assume` call, some existing
unroll tests start to fail because that instruction is not eliminated.

The original motivation for this patch is to facilitate later patches
that fix LoopUnroll's computation of branch weights so that they
maintain the block frequency of OL's body (see #135812). Specifically,
this patch ensures RLGuard's branch weights do not affect RL's
contribution to the block frequency of OL's body in the case that
ULGuard skips UL.
2025-10-07 10:45:49 -04:00

356 lines
18 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -passes='loop-unroll<runtime;partial>' -S %s | FileCheck %s
target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128-ni:1-p2:32:8:8:32-ni:2"
; Make sure SCEVs for phis are properly invalidated after phis are modified.
declare void @llvm.experimental.deoptimize.isVoid(...)
declare i32 @get()
define void @pr56282() {
; CHECK-LABEL: @pr56282(
; CHECK-NEXT: entry:
; CHECK-NEXT: br label [[OUTER_HEADER:%.*]]
; CHECK: outer.header:
; CHECK-NEXT: [[OUTER_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[OUTER_IV_NEXT:%.*]], [[INNER_2:%.*]] ]
; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[OUTER_IV]], 1
; CHECK-NEXT: [[TMP1:%.*]] = freeze i64 [[TMP0]]
; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[TMP1]], -1
; CHECK-NEXT: [[XTRAITER:%.*]] = and i64 [[TMP1]], 7
; CHECK-NEXT: [[TMP3:%.*]] = icmp ult i64 [[TMP2]], 7
; CHECK-NEXT: br i1 [[TMP3]], label [[INNER_1_HEADER_EPIL_PREHEADER:%.*]], label [[OUTER_HEADER_NEW:%.*]]
; CHECK: outer.header.new:
; CHECK-NEXT: [[UNROLL_ITER:%.*]] = sub i64 [[TMP1]], [[XTRAITER]]
; CHECK-NEXT: br label [[INNER_1_HEADER:%.*]]
; CHECK: inner.1.header:
; CHECK-NEXT: [[INNER_1_IV:%.*]] = phi i64 [ 0, [[OUTER_HEADER_NEW]] ], [ [[INNER_1_IV_NEXT_7:%.*]], [[INNER_1_LATCH_7:%.*]] ]
; CHECK-NEXT: [[NITER:%.*]] = phi i64 [ 0, [[OUTER_HEADER_NEW]] ], [ [[NITER_NEXT_7:%.*]], [[INNER_1_LATCH_7]] ]
; CHECK-NEXT: [[V:%.*]] = call i32 @get()
; CHECK-NEXT: [[C_1:%.*]] = icmp ugt i32 [[V]], 0
; CHECK-NEXT: br i1 [[C_1]], label [[INNER_1_LATCH:%.*]], label [[EXIT_DEOPT_LOOPEXIT:%.*]]
; CHECK: inner.1.latch:
; CHECK-NEXT: [[V_1:%.*]] = call i32 @get()
; CHECK-NEXT: [[C_1_1:%.*]] = icmp ugt i32 [[V_1]], 0
; CHECK-NEXT: br i1 [[C_1_1]], label [[INNER_1_LATCH_1:%.*]], label [[EXIT_DEOPT_LOOPEXIT]]
; CHECK: inner.1.latch.1:
; CHECK-NEXT: [[V_2:%.*]] = call i32 @get()
; CHECK-NEXT: [[C_1_2:%.*]] = icmp ugt i32 [[V_2]], 0
; CHECK-NEXT: br i1 [[C_1_2]], label [[INNER_1_LATCH_2:%.*]], label [[EXIT_DEOPT_LOOPEXIT]]
; CHECK: inner.1.latch.2:
; CHECK-NEXT: [[V_3:%.*]] = call i32 @get()
; CHECK-NEXT: [[C_1_3:%.*]] = icmp ugt i32 [[V_3]], 0
; CHECK-NEXT: br i1 [[C_1_3]], label [[INNER_1_LATCH_3:%.*]], label [[EXIT_DEOPT_LOOPEXIT]]
; CHECK: inner.1.latch.3:
; CHECK-NEXT: [[V_4:%.*]] = call i32 @get()
; CHECK-NEXT: [[C_1_4:%.*]] = icmp ugt i32 [[V_4]], 0
; CHECK-NEXT: br i1 [[C_1_4]], label [[INNER_1_LATCH_4:%.*]], label [[EXIT_DEOPT_LOOPEXIT]]
; CHECK: inner.1.latch.4:
; CHECK-NEXT: [[V_5:%.*]] = call i32 @get()
; CHECK-NEXT: [[C_1_5:%.*]] = icmp ugt i32 [[V_5]], 0
; CHECK-NEXT: br i1 [[C_1_5]], label [[INNER_1_LATCH_5:%.*]], label [[EXIT_DEOPT_LOOPEXIT]]
; CHECK: inner.1.latch.5:
; CHECK-NEXT: [[V_6:%.*]] = call i32 @get()
; CHECK-NEXT: [[C_1_6:%.*]] = icmp ugt i32 [[V_6]], 0
; CHECK-NEXT: br i1 [[C_1_6]], label [[INNER_1_LATCH_6:%.*]], label [[EXIT_DEOPT_LOOPEXIT]]
; CHECK: inner.1.latch.6:
; CHECK-NEXT: [[INNER_1_IV_NEXT_7]] = add nuw nsw i64 [[INNER_1_IV]], 8
; CHECK-NEXT: [[V_7:%.*]] = call i32 @get()
; CHECK-NEXT: [[C_1_7:%.*]] = icmp ugt i32 [[V_7]], 0
; CHECK-NEXT: br i1 [[C_1_7]], label [[INNER_1_LATCH_7]], label [[EXIT_DEOPT_LOOPEXIT]]
; CHECK: inner.1.latch.7:
; CHECK-NEXT: [[NITER_NEXT_7]] = add i64 [[NITER]], 8
; CHECK-NEXT: [[NITER_NCMP_7:%.*]] = icmp ne i64 [[NITER_NEXT_7]], [[UNROLL_ITER]]
; CHECK-NEXT: br i1 [[NITER_NCMP_7]], label [[INNER_1_HEADER]], label [[OUTER_MIDDLE_UNR_LCSSA:%.*]]
; CHECK: outer.middle.unr-lcssa:
; CHECK-NEXT: [[V_LCSSA1_PH_PH:%.*]] = phi i32 [ [[V_7]], [[INNER_1_LATCH_7]] ]
; CHECK-NEXT: [[INNER_1_IV_UNR_PH:%.*]] = phi i64 [ [[INNER_1_IV_NEXT_7]], [[INNER_1_LATCH_7]] ]
; CHECK-NEXT: [[LCMP_MOD:%.*]] = icmp ne i64 [[XTRAITER]], 0
; CHECK-NEXT: br i1 [[LCMP_MOD]], label [[INNER_1_HEADER_EPIL_PREHEADER]], label [[OUTER_MIDDLE:%.*]]
; CHECK: inner.1.header.epil.preheader:
; CHECK-NEXT: [[INNER_1_IV_UNR:%.*]] = phi i64 [ 0, [[OUTER_HEADER]] ], [ [[INNER_1_IV_UNR_PH]], [[OUTER_MIDDLE_UNR_LCSSA]] ]
; CHECK-NEXT: [[LCMP_MOD3:%.*]] = icmp ne i64 [[XTRAITER]], 0
; CHECK-NEXT: call void @llvm.assume(i1 [[LCMP_MOD3]])
; CHECK-NEXT: br label [[INNER_1_HEADER_EPIL:%.*]]
; CHECK: inner.1.header.epil:
; CHECK-NEXT: [[INNER_1_IV_EPIL:%.*]] = phi i64 [ [[INNER_1_IV_UNR]], [[INNER_1_HEADER_EPIL_PREHEADER]] ], [ [[INNER_1_IV_NEXT_EPIL:%.*]], [[INNER_1_LATCH_EPIL:%.*]] ]
; CHECK-NEXT: [[EPIL_ITER:%.*]] = phi i64 [ 0, [[INNER_1_HEADER_EPIL_PREHEADER]] ], [ [[EPIL_ITER_NEXT:%.*]], [[INNER_1_LATCH_EPIL]] ]
; CHECK-NEXT: [[INNER_1_IV_NEXT_EPIL]] = add nuw nsw i64 [[INNER_1_IV_EPIL]], 1
; CHECK-NEXT: [[V_EPIL:%.*]] = call i32 @get()
; CHECK-NEXT: [[C_1_EPIL:%.*]] = icmp ugt i32 [[V_EPIL]], 0
; CHECK-NEXT: br i1 [[C_1_EPIL]], label [[INNER_1_LATCH_EPIL]], label [[EXIT_DEOPT_LOOPEXIT3:%.*]]
; CHECK: inner.1.latch.epil:
; CHECK-NEXT: [[C_2_EPIL:%.*]] = icmp ult i64 [[INNER_1_IV_EPIL]], [[OUTER_IV]]
; CHECK-NEXT: [[EPIL_ITER_NEXT]] = add i64 [[EPIL_ITER]], 1
; CHECK-NEXT: [[EPIL_ITER_CMP:%.*]] = icmp ne i64 [[EPIL_ITER_NEXT]], [[XTRAITER]]
; CHECK-NEXT: br i1 [[EPIL_ITER_CMP]], label [[INNER_1_HEADER_EPIL]], label [[OUTER_MIDDLE_EPILOG_LCSSA:%.*]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: outer.middle.epilog-lcssa:
; CHECK-NEXT: [[V_LCSSA1_PH2:%.*]] = phi i32 [ [[V_EPIL]], [[INNER_1_LATCH_EPIL]] ]
; CHECK-NEXT: br label [[OUTER_MIDDLE]]
; CHECK: outer.middle:
; CHECK-NEXT: [[V_LCSSA1:%.*]] = phi i32 [ [[V_LCSSA1_PH_PH]], [[OUTER_MIDDLE_UNR_LCSSA]] ], [ [[V_LCSSA1_PH2]], [[OUTER_MIDDLE_EPILOG_LCSSA]] ]
; CHECK-NEXT: [[C_3:%.*]] = icmp ugt i32 [[V_LCSSA1]], 0
; CHECK-NEXT: br i1 [[C_3]], label [[INNER_2_PREHEADER:%.*]], label [[EXIT:%.*]]
; CHECK: inner.2.preheader:
; CHECK-NEXT: br label [[INNER_2]]
; CHECK: inner.2:
; CHECK-NEXT: [[OUTER_IV_NEXT]] = add nuw i64 [[OUTER_IV]], 1
; CHECK-NEXT: br label [[OUTER_HEADER]]
; CHECK: exit:
; CHECK-NEXT: ret void
; CHECK: exit.deopt.loopexit:
; CHECK-NEXT: br label [[EXIT_DEOPT:%.*]]
; CHECK: exit.deopt.loopexit4:
; CHECK-NEXT: br label [[EXIT_DEOPT]]
; CHECK: exit.deopt:
; CHECK-NEXT: call void (...) @llvm.experimental.deoptimize.isVoid(i32 0) [ "deopt"() ]
; CHECK-NEXT: ret void
;
entry:
br label %outer.header
outer.header:
%outer.iv = phi i64 [ 0, %entry ], [ %outer.iv.next, %outer.latch ]
br label %inner.1.header
inner.1.header:
%inner.1.iv = phi i64 [ 0, %outer.header ], [ %inner.1.iv.next, %inner.1.latch ]
%inner.1.iv.next = add nuw nsw i64 %inner.1.iv, 1
%v = call i32 @get()
%c.1 = icmp ugt i32 %v, 0
br i1 %c.1, label %inner.1.latch, label %exit.deopt
inner.1.latch: ; preds = %inner.1.header
%c.2 = icmp ult i64 %inner.1.iv, %outer.iv
br i1 %c.2, label %inner.1.header, label %outer.middle
outer.middle:
%c.3 = icmp ugt i32 %v, 0
br i1 %c.3, label %inner.2, label %exit
inner.2:
%inner.2.iv = phi i64 [ 0, %outer.middle ], [ %inner.2.iv.next, %inner.2 ]
%inner.2.iv.next = add nsw i64 %inner.2.iv, -1
%iv.trunc = trunc i64 %inner.2.iv to i32
%c.4 = icmp ult i32 %v, %iv.trunc
br i1 %c.4, label %inner.2, label %outer.latch
outer.latch:
%outer.iv.next = add nuw nsw i64 %outer.iv, 1
br label %outer.header
exit:
ret void
exit.deopt:
call void (...) @llvm.experimental.deoptimize.isVoid(i32 0) [ "deopt"() ]
ret void
}
declare void @bar()
declare void @use.2(ptr, i32)
define void @pr56286(i64 %x, ptr %src, ptr %dst, ptr %ptr.src) !prof !0 {
; CHECK-LABEL: @pr56286(
; CHECK-NEXT: bb:
; CHECK-NEXT: [[SMAX:%.*]] = call i64 @llvm.smax.i64(i64 [[X:%.*]], i64 1)
; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[SMAX]], 1
; CHECK-NEXT: [[TMP1:%.*]] = sub i64 [[TMP0]], [[X]]
; CHECK-NEXT: br label [[OUTER_HEADER:%.*]]
; CHECK: outer.header:
; CHECK-NEXT: [[OUTER_P:%.*]] = phi i32 [ 0, [[BB:%.*]] ], [ [[L_1_LCSSA:%.*]], [[OUTER_LATCH:%.*]] ]
; CHECK-NEXT: [[TMP2:%.*]] = freeze i64 [[TMP1]]
; CHECK-NEXT: [[TMP3:%.*]] = add i64 [[TMP2]], -1
; CHECK-NEXT: [[XTRAITER:%.*]] = and i64 [[TMP2]], 7
; CHECK-NEXT: [[LCMP_MOD:%.*]] = icmp ne i64 [[XTRAITER]], 0
; CHECK-NEXT: br i1 [[LCMP_MOD]], label [[INNER_1_HEADER_PROL_PREHEADER:%.*]], label [[INNER_1_HEADER_PROL_LOOPEXIT:%.*]], !prof [[PROF3:![0-9]+]]
; CHECK: inner.1.header.prol.preheader:
; CHECK-NEXT: br label [[INNER_1_HEADER_PROL:%.*]]
; CHECK: inner.1.header.prol:
; CHECK-NEXT: [[INNER_1_IV_PROL:%.*]] = phi i64 [ [[X]], [[INNER_1_HEADER_PROL_PREHEADER]] ], [ [[INNER_1_IV_NEXT_PROL:%.*]], [[INNER_1_LATCH_PROL:%.*]] ]
; CHECK-NEXT: [[PROL_ITER:%.*]] = phi i64 [ 0, [[INNER_1_HEADER_PROL_PREHEADER]] ], [ [[PROL_ITER_NEXT:%.*]], [[INNER_1_LATCH_PROL]] ]
; CHECK-NEXT: [[CMP_1_PROL:%.*]] = icmp sgt i32 [[OUTER_P]], 0
; CHECK-NEXT: br i1 [[CMP_1_PROL]], label [[EXIT_DEOPT_LOOPEXIT1:%.*]], label [[INNER_1_LATCH_PROL]]
; CHECK: inner.1.latch.prol:
; CHECK-NEXT: [[L_1_PROL:%.*]] = load i32, ptr [[SRC:%.*]], align 4
; CHECK-NEXT: store i32 [[L_1_PROL]], ptr [[DST:%.*]], align 8
; CHECK-NEXT: [[INNER_1_IV_NEXT_PROL]] = add i64 [[INNER_1_IV_PROL]], 1
; CHECK-NEXT: [[CMP_2_PROL:%.*]] = icmp sgt i64 [[INNER_1_IV_PROL]], 0
; CHECK-NEXT: [[PROL_ITER_NEXT]] = add i64 [[PROL_ITER]], 1
; CHECK-NEXT: [[PROL_ITER_CMP:%.*]] = icmp ne i64 [[PROL_ITER_NEXT]], [[XTRAITER]]
; CHECK-NEXT: br i1 [[PROL_ITER_CMP]], label [[INNER_1_HEADER_PROL]], label [[INNER_1_HEADER_PROL_LOOPEXIT_UNR_LCSSA:%.*]], !prof [[PROF4:![0-9]+]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK: inner.1.header.prol.loopexit.unr-lcssa:
; CHECK-NEXT: [[L_1_LCSSA_UNR_PH:%.*]] = phi i32 [ [[L_1_PROL]], [[INNER_1_LATCH_PROL]] ]
; CHECK-NEXT: [[INNER_1_IV_UNR_PH:%.*]] = phi i64 [ [[INNER_1_IV_NEXT_PROL]], [[INNER_1_LATCH_PROL]] ]
; CHECK-NEXT: br label [[INNER_1_HEADER_PROL_LOOPEXIT]]
; CHECK: inner.1.header.prol.loopexit:
; CHECK-NEXT: [[L_1_LCSSA_UNR:%.*]] = phi i32 [ poison, [[OUTER_HEADER]] ], [ [[L_1_LCSSA_UNR_PH]], [[INNER_1_HEADER_PROL_LOOPEXIT_UNR_LCSSA]] ]
; CHECK-NEXT: [[INNER_1_IV_UNR:%.*]] = phi i64 [ [[X]], [[OUTER_HEADER]] ], [ [[INNER_1_IV_UNR_PH]], [[INNER_1_HEADER_PROL_LOOPEXIT_UNR_LCSSA]] ]
; CHECK-NEXT: [[TMP4:%.*]] = icmp ult i64 [[TMP3]], 7
; CHECK-NEXT: br i1 [[TMP4]], label [[OUTER_MIDDLE:%.*]], label [[OUTER_HEADER_NEW:%.*]], !prof [[PROF3]]
; CHECK: outer.header.new:
; CHECK-NEXT: br label [[INNER_1_HEADER:%.*]]
; CHECK: inner.1.header:
; CHECK-NEXT: [[INNER_1_IV:%.*]] = phi i64 [ [[INNER_1_IV_UNR]], [[OUTER_HEADER_NEW]] ], [ [[INNER_1_IV_NEXT_7:%.*]], [[INNER_1_LATCH_7:%.*]] ]
; CHECK-NEXT: [[CMP_1:%.*]] = icmp sgt i32 [[OUTER_P]], 0
; CHECK-NEXT: br i1 [[CMP_1]], label [[EXIT_DEOPT_LOOPEXIT:%.*]], label [[INNER_1_LATCH:%.*]]
; CHECK: inner.1.latch:
; CHECK-NEXT: [[L_1:%.*]] = load i32, ptr [[SRC]], align 4
; CHECK-NEXT: store i32 [[L_1]], ptr [[DST]], align 8
; CHECK-NEXT: br i1 false, label [[EXIT_DEOPT_LOOPEXIT]], label [[INNER_1_LATCH_1:%.*]]
; CHECK: inner.1.latch.1:
; CHECK-NEXT: [[L_1_1:%.*]] = load i32, ptr [[SRC]], align 4
; CHECK-NEXT: store i32 [[L_1_1]], ptr [[DST]], align 8
; CHECK-NEXT: [[CMP_1_2:%.*]] = icmp sgt i32 [[OUTER_P]], 0
; CHECK-NEXT: br i1 [[CMP_1_2]], label [[EXIT_DEOPT_LOOPEXIT]], label [[INNER_1_LATCH_2:%.*]]
; CHECK: inner.1.latch.2:
; CHECK-NEXT: [[L_1_2:%.*]] = load i32, ptr [[SRC]], align 4
; CHECK-NEXT: store i32 [[L_1_2]], ptr [[DST]], align 8
; CHECK-NEXT: br i1 false, label [[EXIT_DEOPT_LOOPEXIT]], label [[INNER_1_LATCH_3:%.*]]
; CHECK: inner.1.latch.3:
; CHECK-NEXT: [[L_1_3:%.*]] = load i32, ptr [[SRC]], align 4
; CHECK-NEXT: store i32 [[L_1_3]], ptr [[DST]], align 8
; CHECK-NEXT: [[CMP_1_4:%.*]] = icmp sgt i32 [[OUTER_P]], 0
; CHECK-NEXT: br i1 [[CMP_1_4]], label [[EXIT_DEOPT_LOOPEXIT]], label [[INNER_1_LATCH_4:%.*]]
; CHECK: inner.1.latch.4:
; CHECK-NEXT: [[L_1_4:%.*]] = load i32, ptr [[SRC]], align 4
; CHECK-NEXT: store i32 [[L_1_4]], ptr [[DST]], align 8
; CHECK-NEXT: br i1 false, label [[EXIT_DEOPT_LOOPEXIT]], label [[INNER_1_LATCH_5:%.*]]
; CHECK: inner.1.latch.5:
; CHECK-NEXT: [[L_1_5:%.*]] = load i32, ptr [[SRC]], align 4
; CHECK-NEXT: store i32 [[L_1_5]], ptr [[DST]], align 8
; CHECK-NEXT: [[CMP_1_6:%.*]] = icmp sgt i32 [[OUTER_P]], 0
; CHECK-NEXT: br i1 [[CMP_1_6]], label [[EXIT_DEOPT_LOOPEXIT]], label [[INNER_1_LATCH_6:%.*]]
; CHECK: inner.1.latch.6:
; CHECK-NEXT: [[L_1_6:%.*]] = load i32, ptr [[SRC]], align 4
; CHECK-NEXT: store i32 [[L_1_6]], ptr [[DST]], align 8
; CHECK-NEXT: [[INNER_1_IV_NEXT_6:%.*]] = add i64 [[INNER_1_IV]], 7
; CHECK-NEXT: br i1 false, label [[EXIT_DEOPT_LOOPEXIT]], label [[INNER_1_LATCH_7]]
; CHECK: inner.1.latch.7:
; CHECK-NEXT: [[L_1_7:%.*]] = load i32, ptr [[SRC]], align 4
; CHECK-NEXT: store i32 [[L_1_7]], ptr [[DST]], align 8
; CHECK-NEXT: [[INNER_1_IV_NEXT_7]] = add i64 [[INNER_1_IV]], 8
; CHECK-NEXT: [[CMP_2_7:%.*]] = icmp sgt i64 [[INNER_1_IV_NEXT_6]], 0
; CHECK-NEXT: br i1 [[CMP_2_7]], label [[OUTER_MIDDLE_UNR_LCSSA:%.*]], label [[INNER_1_HEADER]], !prof [[PROF6:![0-9]+]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK: outer.middle.unr-lcssa:
; CHECK-NEXT: [[L_1_LCSSA_PH:%.*]] = phi i32 [ [[L_1_7]], [[INNER_1_LATCH_7]] ]
; CHECK-NEXT: br label [[OUTER_MIDDLE]]
; CHECK: outer.middle:
; CHECK-NEXT: [[L_1_LCSSA]] = phi i32 [ [[L_1_LCSSA_UNR]], [[INNER_1_HEADER_PROL_LOOPEXIT]] ], [ [[L_1_LCSSA_PH]], [[OUTER_MIDDLE_UNR_LCSSA]] ]
; CHECK-NEXT: br label [[INNER_2:%.*]]
; CHECK: inner.2:
; CHECK-NEXT: [[INNER_2_IV:%.*]] = phi i32 [ [[L_1_LCSSA]], [[OUTER_MIDDLE]] ], [ [[INNER_2_IV_NEXT_2:%.*]], [[INNER_2]] ]
; CHECK-NEXT: [[TMP15:%.*]] = phi i32 [ 0, [[OUTER_MIDDLE]] ], [ [[TMP33_2:%.*]], [[INNER_2]] ]
; CHECK-NEXT: [[L_2:%.*]] = load i32, ptr [[SRC]], align 8
; CHECK-NEXT: [[INNER_2_IV_NEXT:%.*]] = add i32 [[INNER_2_IV]], 1
; CHECK-NEXT: [[TMP27:%.*]] = load ptr, ptr [[PTR_SRC:%.*]], align 8
; CHECK-NEXT: [[ADD_1:%.*]] = add i32 [[INNER_2_IV]], [[L_2]]
; CHECK-NEXT: [[TMP281:%.*]] = call i32 @use.2(ptr [[TMP27]], i32 [[ADD_1]])
; CHECK-NEXT: [[TMP31:%.*]] = shl nuw nsw i32 [[TMP15]], 16
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: [[L_2_1:%.*]] = load i32, ptr [[SRC]], align 8
; CHECK-NEXT: [[INNER_2_IV_NEXT_1:%.*]] = add i32 [[INNER_2_IV]], 2
; CHECK-NEXT: [[TMP27_1:%.*]] = load ptr, ptr [[PTR_SRC]], align 8
; CHECK-NEXT: [[ADD_1_1:%.*]] = add i32 [[INNER_2_IV_NEXT]], [[L_2_1]]
; CHECK-NEXT: [[TMP281_1:%.*]] = call i32 @use.2(ptr [[TMP27_1]], i32 [[ADD_1_1]])
; CHECK-NEXT: [[TMP32_1:%.*]] = add nuw i32 [[TMP31]], 524288
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: [[L_2_2:%.*]] = load i32, ptr [[SRC]], align 8
; CHECK-NEXT: [[INNER_2_IV_NEXT_2]] = add i32 [[INNER_2_IV]], 3
; CHECK-NEXT: [[TMP27_2:%.*]] = load ptr, ptr [[PTR_SRC]], align 8
; CHECK-NEXT: [[ADD_1_2:%.*]] = add i32 [[INNER_2_IV_NEXT_1]], [[L_2_2]]
; CHECK-NEXT: [[TMP281_2:%.*]] = call i32 @use.2(ptr [[TMP27_2]], i32 [[ADD_1_2]])
; CHECK-NEXT: [[TMP32_2:%.*]] = add nuw i32 [[TMP31]], 786432
; CHECK-NEXT: [[TMP33_2]] = ashr exact i32 [[TMP32_2]], 16
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: call void @bar()
; CHECK-NEXT: [[CMP_3_2:%.*]] = icmp sgt i32 [[TMP32_1]], 2031616
; CHECK-NEXT: br i1 [[CMP_3_2]], label [[OUTER_LATCH]], label [[INNER_2]]
; CHECK: outer.latch:
; CHECK-NEXT: br label [[OUTER_HEADER]]
; CHECK: exit.deopt.loopexit:
; CHECK-NEXT: br label [[EXIT_DEOPT:%.*]]
; CHECK: exit.deopt.loopexit1:
; CHECK-NEXT: br label [[EXIT_DEOPT]]
; CHECK: exit.deopt:
; CHECK-NEXT: call void (...) @llvm.experimental.deoptimize.isVoid(i32 0) [ "deopt"() ]
; CHECK-NEXT: ret void
;
bb:
br label %outer.header
outer.header:
%outer.p = phi i32 [ 0, %bb ], [ %l.1, %outer.latch ]
br label %inner.1.header
inner.1.header:
%inner.1.iv = phi i64 [ %x, %outer.header ], [ %inner.1.iv.next, %inner.1.latch ]
%cmp.1 = icmp sgt i32 %outer.p, 0
br i1 %cmp.1, label %exit.deopt, label %inner.1.latch
inner.1.latch:
%l.1 = load i32, ptr %src, align 4
store i32 %l.1, ptr %dst, align 8
%inner.1.iv.next = add i64 %inner.1.iv, 1
%cmp.2 = icmp sgt i64 %inner.1.iv, 0
br i1 %cmp.2, label %outer.middle, label %inner.1.header, !prof !1
outer.middle:
br label %inner.2
inner.2:
%inner.2.iv = phi i32 [ %l.1, %outer.middle ], [ %inner.2.iv.next, %inner.2 ]
%tmp15 = phi i32 [ 0, %outer.middle ], [ %tmp33, %inner.2 ]
%l.2 = load i32, ptr %src , align 8
%l.3 = load i32, ptr %dst, align 4
%inner.2.iv.next = add i32 %inner.2.iv, 1
%tmp27 = load ptr, ptr %ptr.src
%add.1 = add i32 %inner.2.iv, %l.2
%add.2 = add i32 %add.1, %l.3
%tmp281 = call i32 @use.2(ptr %tmp27, i32 %add.1)
%tmp31 = shl nuw nsw i32 %tmp15, 16
%tmp32 = add nuw i32 %tmp31, 262144
%tmp33 = ashr exact i32 %tmp32, 16
call void @bar()
call void @bar()
call void @bar()
call void @bar()
call void @bar()
call void @bar()
call void @bar()
call void @bar()
%cmp.3 = icmp sgt i32 %tmp31, 2031616
br i1 %cmp.3, label %outer.latch, label %inner.2
outer.latch:
br label %outer.header
exit.deopt:
call void (...) @llvm.experimental.deoptimize.isVoid(i32 0) [ "deopt"() ]
ret void
}
!0 = !{!"function_entry_count", i64 32768}
!1 = !{!"branch_weights", i32 1, i32 32}
Reference in New Issue View Git Blame Copy Permalink