Hi, On Tue, Oct 11, 2011 at 01:17:03PM -0600, Paul Walmsley wrote: > > > On Tue, 11 Oct 2011, Felipe Balbi wrote: > > > > > > > maybe, but let's go with EXPORT_SYMBOL_GPL(). My only concern is that > > > > this IP comes from synopsys, and I'm not really keen on adding > > > > OMAP-specific, integration-related knowledge on a driver which was > > > > supposed to be reusable ;-) > > > > > > > > But at least for the time being, I guess that's what we need to do ;-) > > > > > > Which IP is coming from Synopsys - the TLL ? The EXPORT_SYMBOL approach > > > > TLL and UHH are OMAP-only wrappers. EHCI and OHCI are Synopsys IPs for > > all I know. > > Ah, I see. > > > > shouldn't add anything OMAP-specific to the TLL driver. So the exported > > > symbols should still be usable by drivers on other SoCs. But perhaps I am > > > misunderstanding your point? > > > > A bit :-) > > > > EHCI/OHCI will need to ask TLL to enable TLL clock, because it's > > EHCI/OHCI who configure the port. Unless I misunderstood what Keshava > > did (and no, I didn't just read the code, too lazy :-) > > The calls to the TLL driver only need to go into the OMAP-specific > ehci-omap.c and ohci-omap3.c wrappers, as far as I can see. So I don't > think there's a need to modify the generic OHCI/EHCI drivers... those where the ones I was talking about. That ehci-omap.c and ohci-omap3.c refer to Synopsys IPs, so (in theory at least) that same file can be used by any other SoC who uses a Synopsys IP. -- balbi
Attachment:
signature.asc
Description: Digital signature
