Hi,
(don't send me mails directly or I'll miss them, keep linux-usb in Cc)
On Tue, Feb 05, 2013 at 04:46:17PM +0100, Florian Fainelli wrote:
> Hello Felipe,
>
> On 01/29/2013 12:05 PM, Florian Fainelli wrote:
> >Hello Felipe,
> >
> >On 01/29/2013 08:37 AM, Felipe Balbi wrote:
> >>
> >>>Ok, but then I won't be able to use the generic OHCI and EHCI
> >>>platform drivers
> >>>because they are not yet aware of clocks, PHY slave device etc... For
> >>>now I
> >>>would like to stick with that since this is also very BCM63xx centric.
> >>>
> >>>Would that be ok with you?
> >>sure, but we need to see a move towards making all of this generic and,
> >>perhaps more importantly (at least to me), compilable on all arches so
> >>we can make proper use of linux-next and Fengguang's build systems.
> >
> >After checking a bit more what is being done with other PHY / host
> >drivers, it seems like my concerns about ehci-platform not yet being
> >aware of USB PHYs is not much of an issue. My only remaining concern now
> >is, how can I model the bcm63xx_select_pullup() function within the USB
> >PHY layer? Would it be acceptable to (ab)use usb_set_phy_{power,suspend}
> >to do that, or would you prefer a new set of operations like
> >usb_set_phy_pullup() being added?
>
> I did not get an answer regarding these two questions, what do you think?
so your pullups are really controlled by the PHY ? In that case, we
would have to add some new function pointers to the phy structure to
handle that, but make sure that your only way to control pullups is
directly at the PHY layer.
--
balbi
Attachment:
signature.asc
Description: Digital signature
