Re: Linux/IA-64 byte order

Jeremy Fitzhardinge (jeremy@goop.org)
Tue, 09 Mar 1999 18:58:43 -0800 (PST)


On 10-Mar-99 Richard Gooch wrote:
> Jeremy Fitzhardinge writes:
>> On 09-Mar-99 Vladimir Dergachev wrote:
>> >
>> > Is it too stupid to suggest an "ENDIANNESS" flag for mmap ? The overhead
>> > for loading a page from disk is much bigger than the overhead for
>> > rearranging the bytes. (Perhaps it should be not a flag but an ioctl..)
>>
>> That would inhibit sharing between processes unless all mappings are
>> for the same endianness.
>
> No, it wouldn't. Each process can map it as it pleases.

Vladmir is proposing having the kernel rearrange the bytes manually, rather
than relying on hardware support. This wouldn't work for two reasons:
- sharing wouldn't work with mappings of different endianness
- the kernel can't do the required reversal, since it depends on the type of
the data, but the kernel only sees raw bytes.

If the hardware has per-page endianness settings then neither of these apply.

J

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/