This is the mail archive of the
libc-alpha@sourceware.cygnus.com
mailing list for the glibc project.
Re: LFS meets PowerPC - score 0,1
On Fri, Jun 18, 1999 at 02:25:44PM +1000, Geoff Keating wrote:
> Just some suggestions about coding style:
>
> - I don't think the underscores on the variable names are necessary.
> - Use the real off64_t, not the one with the underscores.
I'm a little shaky on this - note that I included <sys/mman.h> (should
I not?). The copy it finds has, among other things, these lines:
extern __ptr_t __mmap __P ((__ptr_t __addr, size_t __len, int __prot,
int __flags, int __fd, __off_t __offset));
extern __ptr_t __mmap64 __P ((__ptr_t __addr, size_t __len, int __prot,
int __flags, int __fd, __off64_t __offset));
Can I then declare the function to take arguments of different names
then those in the prototypes, and is it a good idea?
Also, I seem to recall using off64_t triggering prototype mismatches.
> - Put it in linux/mmap64.c rather than linux/powerpc/mmap64.c, since
> it's not powerpc-specific. If the Intel people find that the
> current version works for them, they can put it in linux/i386/,
> but I'm sure it won't work on sparc or m68k.
I preferred not to hit more architectures than necessary, since
mmap64() _appeared_ to by working on sparc. However, now that I think
about it, I suppose that it was probably just picking up a less garbage
value from the input registers.
> - glibc will never work on pre-ANSI compilers, so there's no reason
> not to write ANSI prototypes in new code.
Just matching the other coding styles I saw when I took a look around.
Dan
/--------------------------------\ /--------------------------------\
| Daniel Jacobowitz |__| SCS Class of 2002 |
| Debian GNU/Linux Developer __ Carnegie Mellon University |
| dan@debian.org | | dmj+@andrew.cmu.edu |
\--------------------------------/ \--------------------------------/