Hi, On Thu, Sep 29, 2011 at 01:36:49PM -0700, Paul Zimmerman wrote: > > From: Felipe Balbi [mailto:balbi@xxxxxx] > > Sent: Thursday, September 29, 2011 11:35 AM > > > > On Thu, Sep 29, 2011 at 11:15:37AM -0700, Paul Zimmerman wrote: > > > > > > This looks correct to me. > > > > > > Reviewed-by: Paul Zimmerman <paulz@xxxxxxxxxxxx> > > > > thanks, I have added your Reviewed-by to the patch ;-) > > Hi Felipe, > > Kind of off-topic for this particular patch, but for example: > > + if (usb_endpoint_xfer_bulk(desc) && dep->endpoint.max_streams) { > + params.param1.depcfg.stream_capable = true; > + params.param1.depcfg.stream_event_enable = true; > + dep->stream_capable = true; > + } > > You are using bitfields here to fill in structures that get sent to the > the hardware. This will not work as-is for big-endian systems, unless > the hardware has built-in support for swizzling the bytes into the > correct order. > > As I understand it, using bitfields to access hardware is frowned upon > in the Linux kernel, instead you should use shift-and-mask. Do you plan > to fix this? You are correct :-) I will fix this all tomorrow. Should be fairly simple, though long, patch. Do you have a big endian system to test against ? -- balbi
Attachment:
signature.asc
Description: Digital signature