preprocessor/8139: Bad new INT64_C macro

preprocessor/8139: Bad new INT64_C macro

[philippe_ribet]

>Number:         8139
>Category:       preprocessor
>Synopsis:       Bad new INT64_C macro
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Fri Oct 04 02:26:01 PDT 2002
>Closed-Date:
>Last-Modified:
>Originator:     philippe_ribet@hotmail.com
>Release:        3.2
>Organization:
>Environment:

>Description:
Old INT64_C macro definition was value ## LL.

New definition is probably better but then the compiler has to be modified accordingly.

Note that all values are valid 64 bit numbers.
The bug occured in the range -2^32+1 .. -2^31.
>How-To-Repeat:
#include <inttypes.h>

main(){
  int64_t v;

  v = INT64_C(-2147483648);
  v = INT64_C(-3000000000);
  v = INT64_C(-4000000000);
  v = INT64_C(-5000000000);
}

$ gcc t.c
t.c: In function `main':
t.c:6: warning: decimal constant is so large that it is unsigned
t.c:7: warning: decimal constant is so large that it is unsigned
t.c:8: warning: decimal constant is so large that it is unsigned

>Fix:

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

Re: preprocessor/8139: Bad new INT64_C macro

[Richard Henderson]

The following reply was made to PR preprocessor/8139; it has been noted by GNATS.

From: Richard Henderson <rth@redhat.com>
To: Neil Booth <neil@daikokuya.co.uk>
Cc: Philippe RIBET <philippe_ribet@hotmail.com>, gcc-gnats@gcc.gnu.org,
   gcc-prs@gcc.gnu.org, gcc-bugs@gcc.gnu.org, nobody@gcc.gnu.org
Subject: Re: preprocessor/8139: Bad new INT64_C macro
Date: Mon, 7 Oct 2002 04:57:15 -0700

 On Mon, Oct 07, 2002 at 11:17:48AM +0100, Neil Booth wrote:
 > But what is defining this macro?  GCC, or your system?
 
 Gcc does not provide <stdint.h>; glibc does.
 
 > I don't think it really matters, though.
 
 It does, because...
 
 > Remember that with the new macro the token does not have L or LL
 > appended, and therefore has type int (before promotions) in an
 > arithmetic expression.  This is what is causing the warning -
 > the integer is too big.
 
 ... the whole point of this excercise is to get proper typing
 of these constants.  If "int" is 64-bit, for instance, then you
 most certainly do not want L or LL appended.
 
 It's a glibc bug.
 
 
 r~

Re: preprocessor/8139: Bad new INT64_C macro

[Neil Booth]

The following reply was made to PR preprocessor/8139; it has been noted by GNATS.

From: Neil Booth <neil@daikokuya.co.uk>
To: Philippe RIBET <philippe_ribet@hotmail.com>
Cc: gcc-gnats@gcc.gnu.org, gcc-prs@gcc.gnu.org, gcc-bugs@gcc.gnu.org,
	nobody@gcc.gnu.org
Subject: Re: preprocessor/8139: Bad new INT64_C macro
Date: Mon, 7 Oct 2002 11:17:48 +0100

 Philippe RIBET wrote:-
 
 > >What is the new definition?  What is this macro?
 > 
 > Old macro was:
 > 
 > # if __WORDSIZE == 64
 > #  define INT64_C(c)    c ## L
 > # else
 > #  define INT64_C(c)    c ## LL
 > # endif
 > 
 > New macro is:
 > #define INT64_C(val) (INT_LEAST64_MAX-INT_LEAST64_MAX+(val))
 
 But what is defining this macro?  GCC, or your system?  I don't
 think it really matters, though.
 
 Remember that with the new macro the token does not have L or LL
 appended, and therefore has type int (before promotions) in an
 arithmetic expression.  This is what is causing the warning -
 the integer is too big.  You should probably add L or LL as
 appropriate to your constants.
 
 Neil.

Re: preprocessor/8139: Bad new INT64_C macro

[Neil Booth]

The following reply was made to PR preprocessor/8139; it has been noted by GNATS.

From: Neil Booth <neil@daikokuya.co.uk>
To: Philippe RIBET <philippe_ribet@hotmail.com>
Cc: gcc-gnats@gcc.gnu.org, gcc-prs@gcc.gnu.org, gcc-bugs@gcc.gnu.org,
	nobody@gcc.gnu.org
Subject: Re: preprocessor/8139: Bad new INT64_C macro
Date: Mon, 7 Oct 2002 11:11:06 +0100

 Philippe RIBET wrote:-
 
 > >I suspect the warnings are correct, but new, and that the definition
 > of the macro is unchanged.
 > No, definition changed.
 > No, warnings are not correct as INT64_C(-3000000000) is valid code because 
 > -3000000000 is valid 64bits integer.
 
 No, that's your misunderstanding.  The integer is 300000000; the -
 is an operator, and it is not 64 bit.  It has type int.
  
 > >You should probably add a U to the relevant numbers.
 > ??? 'U' with negative numbers ???
 > 
 > >Anyway, not a GCC bug.
 > Is IS a GCC bug if you consider ANSI C99 standard.
 
 I don't think so.  You can fix your code like I suggested.
 
 Neil.

Re: preprocessor/8139: Bad new INT64_C macro

[Philippe RIBET]

The following reply was made to PR preprocessor/8139; it has been noted by GNATS.

From: "Philippe RIBET" <philippe_ribet@hotmail.com>
To: gcc-gnats@gcc.gnu.org, gcc-prs@gcc.gnu.org, gcc-bugs@gcc.gnu.org, nobody@gcc.gnu.org
Cc: philippe_ribet@hotmail.com
Subject: Re: preprocessor/8139: Bad new INT64_C macro
Date: Mon, 07 Oct 2002 08:59:44 +0000

 >What is the new definition?  What is this macro?
 
 Old macro was:
 
 # if __WORDSIZE == 64
 #  define INT64_C(c)    c ## L
 # else
 #  define INT64_C(c)    c ## LL
 # endif
 
 New macro is:
 #define INT64_C(val) (INT_LEAST64_MAX-INT_LEAST64_MAX+(val))
 
 
 (in file include/stdint.h)
 
 This macro is part of ANSI C99 standard. It allows to specify integer 
 constant size.
 
 >I suspect the warnings are correct, but new, and that the definition
 of the macro is unchanged.
 No, definition changed.
 No, warnings are not correct as INT64_C(-3000000000) is valid code because 
 -3000000000 is valid 64bits integer.
 
 >You should probably add a U to the relevant numbers.
 ??? 'U' with negative numbers ???
 
 >Anyway, not a GCC bug.
 Is IS a GCC bug if you consider ANSI C99 standard.
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=8139
 
 
 --
 Philippe Ribet
 
 
 
 _________________________________________________________________
 Send and receive Hotmail on your mobile device: http://mobile.msn.com
 

Re: preprocessor/8139: Bad new INT64_C macro

[neil]

Synopsis: Bad new INT64_C macro

State-Changed-From-To: open->closed
State-Changed-By: neil
State-Changed-When: Sat Oct  5 02:28:04 2002
State-Changed-Why:
    You should probably add a U to the relevant numbers.  Anyway, not a GCC bug.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=8139

Re: preprocessor/8139: Bad new INT64_C macro

[Neil Booth]

The following reply was made to PR preprocessor/8139; it has been noted by GNATS.

From: Neil Booth <neil@daikokuya.co.uk>
To: philippe_ribet@hotmail.com
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: preprocessor/8139: Bad new INT64_C macro
Date: Fri, 4 Oct 2002 23:05:58 +0100

 philippe_ribet@hotmail.com wrote:-
 
 > Old INT64_C macro definition was value ## LL.
 > 
 > New definition is probably better but then the compiler has to be modified accordingly.
 
 What is the new definition?  What is this macro?
 
 > Note that all values are valid 64 bit numbers.
 > The bug occured in the range -2^32+1 .. -2^31.
 > >How-To-Repeat:
 > #include <inttypes.h>
 > 
 > main(){
 >   int64_t v;
 > 
 >   v = INT64_C(-2147483648);
 >   v = INT64_C(-3000000000);
 >   v = INT64_C(-4000000000);
 >   v = INT64_C(-5000000000);
 > }
 > 
 > $ gcc t.c
 > t.c: In function `main':
 > t.c:6: warning: decimal constant is so large that it is unsigned
 > t.c:7: warning: decimal constant is so large that it is unsigned
 > t.c:8: warning: decimal constant is so large that it is unsigned
 
 I don't think this macro is defined by GCC.  Is it system specific?
 I suspect the warnings are correct, but new, and that the definition
 of the macro is unchanged.
 
 Neil.