[PATCH] ioctl(2) fixes for 64-bit platforms

[Jung-uk Kim]

On Thursday 24 June 2010 02:38 pm, Jung-uk Kim wrote:
> Python handles ioctl(2) strangely and causes warnings on 64-bit
> platforms.  The analysis was originally posted here:
>
> http://docs.freebsd.org/cgi/mid.cgi?201006231708.40032.jkim
>
> Now I made more complete patch for all Python ports and it is
> available from here:
>
> http://people.freebsd.org/~jkim/python-fcntlmodule.diff

I have updated the patch:

http://people.freebsd.org/~jkim/python-fcntlmodule2.diff

> Note I also fixed a Linuxism of fcntl(2) for python31.  They use
> 'long' for the optional third argument of fcntl() unlike POSIX and
> *BSD.  It's okay for little-endian or 32-bit platforms but I guess
> it is bad for big-endian LP64 machines.

I think it was my mis-understanding as it doesn't pass the pointer, 
withdrawn.

> While I was here, I updated test cases of ioctl() for all Python
> ports.  I also added FreeBSD -CURRENT for python31 but I didn't
> bother test cases of older releases. :-)

Now the new patch is much simpler and it should be upstream-worthy, I 
think.

Jung-uk Kim