On Mon, Jan 14, 2013 at 09:12:43AM +0800, Shawn Guo wrote: > Balbi, > > On Fri, Jan 11, 2013 at 02:50:59PM +0200, Felipe Balbi wrote: > > As I said before, this patch is too big for -rc and is unnecessary > > considering patch I wrote above. Note that there is no problems in > > checking if ULPI PHY clk is 60MHz on all arches and, for the workaround, > > you already have a runtime check. > > > Ok, I did not have these facts on my mind. If these are true, the > cpu_is_xxx() shouldn't be necessary there from the beginning, and we > can simply remove them then. > > > Shawn, it can be broken down into smaller pieces because you can *FIX > > THE COMPILE BREAKAGE* with a very small patch as above (only issue now > > is usage of MX32_IO_ADDRESS()). > > > The MX35_IO_ADDRESS() also seems unnecessary, since as Peter's patch > suggested that pdata->regs can be used instead. pdata->regs is a hack. The 'canonical' way to pass addresses to drivers is via struct resource. -- balbi
Attachment:
signature.asc
Description: Digital signature
