To all architecture maintainers: RFC: mmap2()

[Russell King]

Hi,

Looking through the various implementations of mmap2(), it occurs that
at least the sparc implementation is byggy.

Sparc's mmap2() consists of:

        /* Make sure the shift for mmap2 is constant (12), no matter what PAGE_SIZE
           we have. */
        return do_mmap2(addr, len, prot, flags, fd, pgoff >> (PAGE_SHIFT - 12));

What happens if you have an 8K page size (PAGE_SHIFT == 13), and the
user supplies an odd number in pgoff?

You are unable to represent the requested offset into the do_mmap2()
call.  Surely, in this case, mmap2() should return EINVAL.

I haven't looked at all the architectures, but I know that ARM was
more buggy than this before tonight. ;)
   _____
  |_____| ------------------------------------------------- ---+---+-
  |   |         Russell King        [e-mail]      --- ---
  | | | |   http://www.arm.linux.org.uk/~rmk/aboutme.html    /  /  |
  | +-+-+                                                     --- -+-
  /   |               THE developer of ARM Linux              |+| /|\
 /  | | |                                                     ---  |
    +-+-+ -------------------------------------------------  /\\\  |