Am Mittwoch, 19. Januar 2005 16:42 schrieb Pete Zaitcev:
> On Tue, 18 Jan 2005 22:14:24 -0800, David Brownell <[EMAIL PROTECTED]> wrote:
>
> > > > Also, I don't like the idea of scattering knowledge all over the place
> > > > that the root hub is always given address 1 ...
> >
> > which you didn't address yet.
>
> Yes, I have to look why you do not like using the pipe. Relying on pipe makes
> tests dependant on URB only. No references to bus or HCD, therefore no
> extra refcounts or worries about oopses. Also, HC drivers zero out the
> urb->dev in giveback sequence which is a royal pain when trying to identify
> a root hub. I thought about adding an extra flag like URB_ROOT_HUB to split
That idea was good. It is simple and will simplify the code cleanly.
> this use from the abuse of URB_NO_TRANSFER_DMA_MAP, but pipe looks better
> all around. If you look at it from the angle I did, it stands to reason
> that excessive encapsulation only masks _why_ it was safer, e.g. if one sees
> something like urb_is_root_hub(urb), one must look up the implementation
> to know if it uses urb->dev or not. Relying on address 1 without any symbolic
> constant is obviously a bad idea though, I'll fix that.
True, but pipe must die. It has no real basis.
Regards
Oliver
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
