File: pr22236.c

package info (click to toggle)
gcc-arm-none-eabi 15%3A8-2019-q3-1
  • links: PTS, VCS
  • area: main
  • in suites: bullseye
  • size: 571,828 kB
  • sloc: ansic: 2,937,651; cpp: 881,644; ada: 597,189; makefile: 65,528; asm: 56,499; xml: 46,621; exp: 24,747; sh: 19,684; python: 7,256; pascal: 4,370; awk: 3,497; perl: 2,695; yacc: 316; ml: 285; f90: 234; lex: 198; objc: 194; haskell: 119
file content (33 lines) | stat: -rw-r--r-- 707 bytes parent folder | download | duplicates (10) 
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
 
/* { dg-do run } */
/* { dg-options "-O1 -fno-tree-vrp -fwrapv" } */

/* PR tree-optimization/22236

    Avoid conversion of (signed char) {(uchar)1, +, (uchar)1}_x when
    it is not possible to prove that the scev does not wrap.  

    In this PR, a sequence 1, 2, ..., 255 has to be converted to
    signed char, but this would wrap: 1, 2, ..., 127, -128, ...  The
    result should not be a linear scev {(schar)1, +, (schar)1}_x.
    The conversion should be kept: (schar) {(uchar)1, +, (uchar)1}_x.
 */

void abort(void);

static inline void
foo (signed char a)
{
  int b = a - 0x7F;
  if (b > 1)
    abort();
}

int main()
{
  unsigned char b;
  for(b = 0; b < 0xFF; b++)
    foo (b);

  return 0;
}