On Tue, 16 Sep 2003, Randy.Dunlap wrote:
> On Tue, 16 Sep 2003 17:31:21 -0700 David Brownell <[EMAIL PROTECTED]> wrote:
>
> | But I've got a couple questions about this one, maybe you know the
> | answers to them:
> |
> | > - unsigned no_interrupt : 1,
> | > - zero : 1,
> | > - short_not_ok : 1;
> | > + unsigned no_interrupt:1,
> | > + zero:1,
> | > + short_not_ok:1;
> |
> | I tried this and it made "no_interrupt" appear in the kerneldoc.
> | But NOT the other two bits. Is someone fixing kerneldoc bugs,
> | so that issue can have some useful resolution?
>
> Oh, bad. Sorry about that. Not that I know of.
>
> | Related question, I'm guessing that having each one on a line
> | by itself would make kerneldoc happy. But as I recall, that'd
> | be at the cost of making the bits live in separate words, which
> | is waste I'd rather avoid ... know if that's true?
>
> Yes, if I recall correctly, that would allocate a new <unsigned>
> for each one instead of a series of bits in one <unsigned>....
It may depend on the particular compiler you use. I just tried the
experiment using gcc 2.96 from RedHat 7.2 on a Pentium-class machine.
Given the following structure definition:
struct s {
unsigned i:1, j:1;
unsigned k:1;
unsigned h:1;
};
and invoked with -O -S the compiler placed all four bits in the same byte
(determined by reading the assembler-code output). It did the same when
invoked with -O2 -S although the code generated was different.
Alan Stern
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
