On Mon, Oct 22, 2001, Alan Cox <[EMAIL PROTECTED]> wrote:
> > > if (td->status & TD_CTRL_ACTIVE) {
> > > - td->status |= TD_CTRL_IOC;
> > > + set_bit(TD_CTRL_IOC_BIT, &td->status);
> > > break;
> > > }
> > > }
> >
> > This is a very bad idea in my book because set_bit sets bits
> > in an architecture specific way (from MSB or from LSB), and
> > it is defined to act on longs, while td->status is u32.
> > Of course, uhci is only used on x86, so all is fine here,
> > and I would only demand this taken down if it was ohci.
> > But a bad code all the same, fosters bad habits.
>
> Pete is right here. And if we do the things we want to do to make set_bit
> error on non ulong one day it'll break completely.
It was Linus' suggestion. Is there another portable way of atomically
setting a bit?
JE
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
