Re: [PATCHv7 00/19] First round in OTG rework

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Tue, Nov 29, 2011 at 11:43:05AM +0000, Chen Peter-B29397 wrote:
> > On Tue, Nov 29, 2011 at 11:43:46AM +0200, Heikki Krogerus wrote:
> > > Hi,
> > >
> > > On Mon, Nov 28, 2011 at 01:14:22PM +0000, Chen Peter-B29397 wrote:
> > > > Hi Heikki,
> > > >
> > > > Any updates for your patch? If possible, I would like to co-work with
> > you
> > > > about this patch, and test it at my platform.
> > >
> > > I'm afraid I don't have any updates for these. I'm stuck with some
> > > other things at work, so any help would be appreciated.
> > >
> > > I think we need to have a new plan how to go forward with these. My
> > > idea was that these would be really only the fist set, followed by
> > > the very basic support for multiple PHYs and some minor reorganizing
> > > of the code. However, Felipe commented about the importance of the
> > > multiple PHYs, and if I understood correctly, he would like to see
> > > it already at this first stage. Felipe, please correct me if I'm
> > > wrong.
> > 
> > I mentioned that I would like to hear at least some plans because it
> > wasn't clear from your patchset what you wanted to do in the long run
> > ;-)
> > 
> > > This is a problem as I newer planned to come up with anything more
> > > then very basic support for multiple transceivers. I don't have a plan
> > > for it, and even though I understand some special cases that need to
> > > be addressed, like RX-51 which has two transceivers, I don't know all
> > > the different platforms/things that we need to consider. I guess OMAP4
> > > and USB3 are most obvious ones. So, we need a plan/design for multi
> > > PHY support and I'm not the correct person to come up with it.
> > 
> > Kishon has been working on that part on top of your patchset. Kishon ?
> > 
> > > My goal on top of have the PHY split from OTG, is to have generic OTG
> > > stage machine as part of the OTG utility. This would be useful
> > > initially with the various ChipIdea UDC drivers that all seem to have
> > > the OTG state machine implemented separately, the Marvell one being
> > > the last to do this. The idea is to implement support for this generic
> > > state machine for the ci13xxx_udc, that we are planning to make the
> > > only driver for all the users of the ChipIdea UDC. Obviously the
> > > generic OTG state machine would be useful with other controllers
> > > besides ChipIdea.
> > 
> > sure.
> > 
> > > So the tasks I see here are:
> > >
> > > 1. Split PHY from OTG (covered by this set)
> > 
> > Kishon had a few comments to some of your patches and so had I (IIRC),
> > you need to sort those out and resend the series, then I can apply them.
> > 
> > > 2. Support for multiple PHYs (I can't take care of this)
> > > 3. Generic OTG sm as part of the utility
> > >
> > > Since these features have raised interest on many, I'm hoping we can
> > > have a discussion about how we want to go forward with these and
> > > perhaps about the workslit. Does the multi PHY support really need to
> > > be part of the first task of splitting PHY from OTG? Felipe, it was
> > > originally your idea to start by simply splitting PHY from OTG, or did
> > > I misunderstood you? If you feel this is not the correct approach from
> > > the beginning, then please comment. How would you guys like to do
> > > this?
> > 
> > let's get those comments sorted out, then I can take your patches and
> > Kishon can send his take on multiple-PHY support. Without multiple-PHY,
> > we can't support easily USB3 peripheral/drd controllers.
> > 
> 
> Balbi, I can also give some help for this.
> Freescale i.mx51 bbg platform has two phys at board, one is on-chip
> UTMI phy for OTG controller, the other is ULPI phy for HOST 1
> controller. I can post a patch using refined heikki's phy utility
> file, related phy drivers, and i.mx host driver ehci-mxc.c. Then we
> can more clearly how separated phy struct works (split from otg), phy
> driver works and how multiple PHYs works, and refine a good PHY
> framework.  Do you think so?

it would be nice to see, but as of today the host stack doesn't even
know about the PHYs, how do you plan to add that ? PHYs have, until now,
only been used by DRD/OTG and peripheral-only devices. The standard *hci
hosts never used them. Anyway, please post your patches and we can start
playing with that side too.

Put Kishon in Cc too for those PHY-related patches as he has been
playing with that lately.

-- 
balbi

Attachment: signature.asc
Description: Digital signature


[Index of Archives]     [Linux Media]     [Linux Input]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Old Linux USB Devel Archive]

  Powered by Linux