Re: ffs problem in windows64

2014-09-23 Thread Christian Franke
Christian Franke wrote: This segfaults if gcc -fno-builtin option is used: ffs(0x8000) These work: ffs(0xc000) ffs(0x4000) Likely a subtle problem in table based ffs() implementation in winsup/cygwin/syscalls.cc if only upper bit is set. The newlib version of ffs() works.

Re: ffs problem in windows64

2014-09-23 Thread Christian Franke
Eliot Moss wrote: On 9/23/2014 8:47 AM, Paulo César Pereira de Andrade wrote: Hi, I am almost done porting https://savannah.gnu.org/projects/lightning to cygwin and windows64 abi. But I found a problem with the ffs function. The test case is simple: $ cat x.c extern int ffs(int); i

Re: ffs problem in windows64

2014-09-23 Thread Eliot Moss
On 9/23/2014 8:47 AM, Paulo César Pereira de Andrade wrote: Hi, I am almost done porting https://savannah.gnu.org/projects/lightning to cygwin and windows64 abi. But I found a problem with the ffs function. The test case is simple: $ cat x.c extern int ffs(int); int main(void) { long

ffs problem in windows64

2014-09-23 Thread Paulo César Pereira de Andrade
Hi, I am almost done porting https://savannah.gnu.org/projects/lightning to cygwin and windows64 abi. But I found a problem with the ffs function. The test case is simple: $ cat x.c extern int ffs(int); int main(void) { long l = 0x8000; int i = ffs((int)l); return 0; } $ gcc -O0 -g3 x