Re: middle-end/3919: __builtin_constant_p false on long long compile-time-known variable converted to long on 32-bit platform

Re: middle-end/3919: __builtin_constant_p false on long long compile-time-known variable converted to long on 32-bit platform

[bangerth]

Synopsis: __builtin_constant_p false on long long compile-time-known variable converted to long on 32-bit platform

State-Changed-From-To: feedback->closed
State-Changed-By: bangerth
State-Changed-When: Wed Nov 20 18:50:46 2002
State-Changed-Why:
    No feedback in a long time and user error anyway.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=3919

Re: middle-end/3919: __builtin_constant_p false on long long compile-time-known variable converted to long on 32-bit platform

[rth]

Synopsis: __builtin_constant_p false on long long compile-time-known variable converted to long on 32-bit platform

State-Changed-From-To: open->feedback
State-Changed-By: rth
State-Changed-When: Thu Apr  4 04:23:59 2002
State-Changed-Why:
    Err.. why do you believe this should return 3 and not 0?
    What is it about "var" that can be considered constant?
    Seems to me if you got anything besides 0 you should have
    been submitting _that_ as a bug...

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=3919

middle-end/3919: __builtin_constant_p false on long long compile-time-known variable converted to long on 32-bit platform

[lkb]

>Number:         3919
>Category:       middle-end
>Synopsis:       __builtin_constant_p false on long long compile-time-known variable converted to long on 32-bit platform
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          pessimizes-code
>Submitter-Id:   net
>Arrival-Date:   Thu Aug 02 19:46:01 PDT 2001
>Closed-Date:
>Last-Modified:
>Originator:     Luca Barbieri
>Release:        3.0
>Organization:
>Environment:
i686-pc-cygwin CYGWIN_NT-5.0 1.3.2(0.39/3/2) i686 unknown
>Description:
__builtin_constant_p returns false when it's called on a global long long variable, known to be constant (__builtin_constant_p returns true on it without typecasts), typecasted to long (probably happens for all 32-bit platforms, but only tested on i386).
This bug isn't present for local variables.
This bug is present at any optimization level.
>How-To-Repeat:
// Compile this (as either C or C++) and check the return value from the shell
// Returns: 2
// Should return: 3
long long var;

int main(int argc, char** argv)
{
	var = 0;

	return (__builtin_constant_p(var) ? 2 : 0)
		| (__builtin_constant_p((long)var) ? 1 : 0);
}
>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted: