File: branch-fold.ll

package info (click to toggle)
llvm-toolchain-3.8 1%3A3.8.1-24
  • links: PTS, VCS
  • area: main
  • in suites: stretch
  • size: 379,280 kB
  • ctags: 388,501
  • sloc: cpp: 2,309,705; ansic: 477,070; objc: 100,918; asm: 97,974; python: 95,911; sh: 18,634; makefile: 7,294; perl: 5,584; ml: 5,460; pascal: 4,661; lisp: 2,548; xml: 686; cs: 350; php: 212; csh: 117
file content (40 lines) | stat: -rw-r--r-- 995 bytes parent folder | download | duplicates (16)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
; RUN: llc < %s -march=nvptx64 -mcpu=sm_35 -disable-cgp | FileCheck %s
; Disable CGP which also folds branches, so that only BranchFolding is under
; the spotlight.

target datalayout = "e-i64:64-v16:16-v32:32-n16:32:64"
target triple = "nvptx64-nvidia-cuda"

define void @foo(i32 %x, float* %output) {
; CHECK-LABEL: .visible .func foo(
; CHECK-NOT: bra.uni
; CHECK-NOT: LBB0_
  %1 = icmp eq i32 %x, 1
  br i1 %1, label %then, label %else

then:
  br label %merge

else:
  br label %merge

merge:
  store float 2.0, float* %output
  ret void
}

; PR24299. no crash
define ptx_kernel void @hoge() #0 {
; CHECK-LABEL: .visible .entry hoge(
bb:
  br i1 undef, label %bb1, label %bb4

bb1:                                              ; preds = %bb1, %bb
  %tmp = phi i64 [ %tmp2, %bb1 ], [ undef, %bb ]
  %tmp2 = add nsw i64 %tmp, 1
  %tmp3 = icmp sle i64 %tmp, 0
  br i1 %tmp3, label %bb1, label %bb4

bb4:                                              ; preds = %bb4, %bb1, %bb
  br label %bb4
}