This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
fwd: operator& narrowing bug
- To: gcc-bugs at gcc dot gnu dot org
- Subject: fwd: operator& narrowing bug
- From: Jason Merrill <jason at cygnus dot com>
- Date: 28 Jan 2000 17:37:09 -0800
This seems to be caused by doing the & in the narrowest suitable mode; 0xf0
becomes (signed char)-16 with overflow set, so the result is 0 with
overflow, which doesn't satisfy integer_zerop.
------- Start of forwarded message -------
From: "John H. Spicer" <jhs@edg.com>
To: brendan@cygnus.com
Subject: g++/gcc bug
Date: Fri, 28 Jan 2000 09:46:55 -0500 (EST)
Message-Id: <200001281446.JAA03862@edg3.edg.com>
I'm not sure how one goes about reporting gcc/g++ bugs, so I thought I'd
pass this one along to you.
On Intel, 2.95.2 computes "i" as 7 instead of 8. The sparc version, and
earlier version on Intel produced the right result (8).
If there is someplace else that you would prefer I send this, please let
me know.
Thanks!
John.
#ifdef __cplusplus
extern "C"
#endif
int printf(char *, ...);
int main()
{
int i;
(i=7)&0xf0 || ++i;
if (i != 8) {
printf("i=%d\n", i);
return 1;
}
}
------- End of forwarded message -------