On Fri, Mar 04, 2011 at 01:09:23AM -0800, per...@pluto.rain.com wrote:
> Erik Trulsson <ertr1...@student.uu.se> wrote:
> > On Thu, Mar 03, 2011 at 11:26:12AM -0500, Frank Solensky wrote:
> > > In sys/netinet/ip.h, the first octet of the ip header structure
> > > tests the byte ordering to determine the ordering of the header
> > > length (ip_hl) and version (ip_v) fields.
> > > 
> > > My question: that always works?  While my reading of the
> > > language specification document leaves both the ordering of
> > > the bits within a byte and the bytes within a longer field as
> > > implementation choices, the two are independent of each other.
> > > 
> > > I haven't run into a CPU where this assumption was proven
> > > incorrect ...
> >
> > Unless you have a CPU where memory is addressed bit-by-bit rather
> > than byte-by-byte the ordering of bits within a byte is not only
> > completely irrelevant, it is also pretty much impossible to
> > determine programatically.
> 
> Agreed it is at least difficult to determine programatically,
> however it is quite important when dealing with hardware that
> converts between a sequence of bytes and a bitstream, e.g.
> serial ports, network interfaces, SATA ports.  Driver writers
> had _better_ know which bit of the byte, as well as which
> byte of a word/longword/quadword, is going on the wire first.

Although it certainly matters for serial I/O devices which bit goes out
first on the wire, you only need to know if the I/O hardware will push
out data with the most significant bit first or the least significant
bit first and possibly adjust the values you write to the I/O hardware
if it does not match the order you want the bits to go out.

You still don't need to know anything about in which order bits are
stored in a byte inside the CPU or in RAM.






-- 
<Insert your favourite quote here.>
Erik Trulsson
ertr1...@student.uu.se
_______________________________________________
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "freebsd-questions-unsubscr...@freebsd.org"

Reply via email to