On Mon, Sep 26, 2011 at 12:17:53PM -0700, Franky Lin wrote:

On 09/26/2011 11:26 AM, Arend van Spriel wrote:

quoted

On 09/26/2011 11:37 AM, Johannes Berg wrote:

quoted

On Sat, 2011-09-24 at 14:34 +0200, Arend van Spriel wrote:

quoted

On 09/24/2011 12:38 PM, Johannes Berg wrote:

quoted

On Fri, 2011-09-23 at 19:08 -0700, Franky Lin wrote:

quoted

   struct d11rxhdr {
   	u16 RxFrameSize;
   	u16 PAD;
+	union {
+		struct d11rxhdr_le rxh_le;
+		struct d11rxhdr rxh_cpu;
+	};

This seems a little strange. Why would it be both in LE and CPU byte
order?

Indeed. When we receive it from the device it is in LE and we convert it
to CPU order for further processing using rxh_cpu.

That seems a confusing and error-prone -- you'll have to remember
whether you're before or after conversion. Would it be possible to have
two versions of the outer structure and change the pointer type at that
point?

johannes

For me knowing the driver design (a little ;-) it is not difficult to
remember. Your feedback has valid arguments so I will reconsider. Franky
is looking whether dropping it will affect the other patches submitted
to Greg.

Gr. AvS

Dropping this one will affect some following patches in the series.
Since it's not a bug, shall we keep this one and change it as
Johannes suggested in future commit?

How about just resend the whole series, that way I don't accidentally
apply it, or the other one you wanted dropped?

I've dropped this whole series from my to-apply queue now.

thanks,

greg k-h