This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: 1.1.2 Prerelease


>>>>> Jeffrey A Law writes:

 > I've put a pre-release of egcs-1.1.2 in the pub/egcs/snapshots directory
 > "egcs-1.1.2-prerelease" on the egcs ftp server.

 > I suspect we'll have a second pre-release later this week to wrap up the few
 > remaining open issues (cpp script, file initializer scopes, doc updates, etc).

 > I'd like to be sure that egcs-1.1.2 will correctly compile both glibc-2.1
 > and linux-2.2.1 on the major Linux architectures (x86, alpha, powerpc, sparc).

 > So if folks could make a point to build & test glibc-2.1 and linux-2.2.1 
 > kernels on those architectures with the prerelease it would be greatly
 > appreciated.

I've compiled and installed the prerelease successully but noticed the 
following warning compiling glibc 2.1:
../crypt/sysdeps/unix/ufc-crypt.h:25: warning: integer constant out of range

This can be reproduced with the following small file:
# define UINT_FAST32_MAX        (4294967295U)
# define UINT_FAST64_MAX        (18446744073709551615ULL)

#if UINT_FAST32_MAX == UINT_FAST64_MAX
# define _UFC_64_
#else
# define _UFC_32_
#endif

With egcs 1.1.1 (egcs-2.91.60 19981201):
$ /opt/egcs/bin/gcc  -V egcs-2.91.60 -dM -E test-egcs.c  |grep UFC
#define _UFC_32_  

With the prerelease (egcs-2.91.61 19990216)
$ /opt/egcs/bin/gcc  -V egcs-2.91.61 -dM -E test-egcs.c  |grep UFC
test-egcs.c:4: warning: integer constant out of range
#define _UFC_64_  

The current egcs 1.2 CVS version (egcs-2.93.08 19990216) [1]

$ /opt/egcs-1.2/bin/gcc   -dM -E test-egcs.c  
test-egcs.c:4:39: warning: integer constant out of range

With gcc 2.7.2.3:
$ gcc   -dM -E test-egcs.c  |grep UFC
test-egcs.c:4: two `l's in integer constant
#define _UFC_32_  


Please note that egcs 1.1.2 defines _UFC_64_ but the other compilers
define _UFC_32_.

I don't know if this is correct.  Since this is a change of behaviour
in the test environment (glibc 2.1), I'd like to know if glibc needs
to be fixed and if egcs is ok.

Andreas

Footnotes: 
[1]  Using --enable-c-cpp-lib which might be the reason that the
     define is not output.  I'm not sure if this counts as an error.

-- 
 Andreas Jaeger   aj@arthur.rhein-neckar.de    jaeger@informatik.uni-kl.de
  for pgp-key finger ajaeger@aixd1.rhrk.uni-kl.de


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]