I see your point, I agree the two layers of glue is bad. But this means<br>the double code is spread across many drivers instead of being in one<br>place (fsl_soc.c).<br><br><div><span class="gmail_quote">On 7/7/07, <b class="gmail_sendername">
Grant Likely</b> <<a href="mailto:grant.likely@secretlab.ca">grant.likely@secretlab.ca</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
On 7/7/07, John Rigby <<a href="mailto:jcrigby@gmail.com">jcrigby@gmail.com</a>> wrote:<br>> > This same comment probably goes for the other arch_initcall functions<br>> > in fsl_soc.c which do exactly the same thing for other devices.
<br>><br>> This depends, some devices in fsl_soc.c may exist on non-powerpc SoCs<br>> that do not have OF. The USB core in 8349 for example also is in the<br>> arm based mx27 and mx31. These devices should remain platform devices
<br>> and the glue in fls_soc.c will continure to be needed.<br><br>I disagree; the current method is "glue for the glue". There is no<br>reason why the driver cannot have two bindings; one for<br>platform_device and one for of_platform_device. It's about the same
<br>amount of code, but uses less indirection for the device tree case.<br><br>Cheers,<br>g.<br><br><br>--<br>Grant Likely, B.Sc., P.Eng.<br>Secret Lab Technologies Ltd.<br><a href="mailto:grant.likely@secretlab.ca">grant.likely@secretlab.ca
</a><br>(403) 399-0195<br></blockquote></div><br>