Re: [PATCH 00/10] usb: phy: cleanups to Kconfig and directories

[Felipe Balbi <balbi@ti.com>]

[-- Attachment #1: Type: text/plain, Size: 3718 bytes --]

On Fri, Mar 08, 2013 at 11:37:17AM -0700, Stephen Warren wrote:
> On 03/08/2013 11:26 AM, Felipe Balbi wrote:
> > On Fri, Mar 08, 2013 at 10:14:11AM -0700, Stephen Warren wrote:
> >> On 03/08/2013 12:14 AM, Felipe Balbi wrote:
> >>> Hi,
> >>> 
> >>> On Thu, Mar 07, 2013 at 02:20:36PM -0700, Stephen Warren
> >>> wrote:
> >>>> On 03/07/2013 02:35 AM, Felipe Balbi wrote:
> >>>>> Hi folks,
> >>>>> 
> >>>>> inspired by Paul's DWC2 patchset which added 
> >>>>> usb_otg_state_string() (a copy of otg_state_string()) I
> >>>>> have now renamed otg_state_string() to
> >>>>> usb_otg_state_string(), moved it to usb-common, then moved
> >>>>> all phy drivers to drivers/usb/phy/ and completely deleted
> >>>>> the otg directory.
> >>>>> 
> >>>>> We're also removing CONFIG_USB_OTG_UTILS since that has
> >>>>> lots its meaning long ago.
> >>>>> 
> >>>>> I have compiled all patches with allyes, allno and allmod 
> >>>>> configs, but please make sure to test on your platforms to
> >>>>> make sure we're not leaking any more problems to mainline.
> >>>> 
> >>>> What branch do the patches apply to? They didn't "git am" for
> >>>> me on either next-20130305, nor 
> >>>> git://git.kernel.org/pub/scm/linux/kernel/git/balbi/usb.git 
> >>>> next.
> >>> 
> >>> they're on top of my testing branch.
> >> 
> >> Ah, thanks. I took that whole branch, built ARM's
> >> tegra_defconfig, and see:
> >> 
> >>> warning: (ARCH_TEGRA_2x_SOC && ARCH_TEGRA_3x_SOC) selects
> >>> USB_ULPI which has unmet direct dependencies (USB_SUPPORT &&
> >>> USB_PHY && ARM) warning: (ARCH_TEGRA_2x_SOC &&
> >>> ARCH_TEGRA_3x_SOC) selects USB_ULPI_VIEWPORT which has unmet
> >>> direct dependencies (USB_SUPPORT && USB_PHY && USB_ULPI)
> >>> warning: (ARCH_TEGRA_2x_SOC && ARCH_TEGRA_3x_SOC) selects
> >>> USB_ULPI which has unmet direct dependencies (USB_SUPPORT &&
> >>> USB_PHY && ARM) warning: (ARCH_TEGRA_2x_SOC &&
> >>> ARCH_TEGRA_3x_SOC) selects USB_ULPI_VIEWPORT which has unmet
> >>> direct dependencies (USB_SUPPORT && USB_PHY && USB_ULPI)
> >> 
> >> Manually enabling USB_PHY fixes this. However, this highlights
> >> an issue with your removal of all selects (as mentioned in your
> >> other email) - it will break perhaps any defconfig that has USB
> >> enabled.
> >> 
> >> After enabling USB_PHY, the code builds and runs without issue.
> > 
> > fair enough, but then I'm just exposing the trouble. ARCH
> > shouldn't select USB_ULTI or any of the phy drivers, for that
> > matter.
> 
> Yes, I think it should instead work like:
> 
> ARCH_TEGRA* selects nothing in particular related to USB.
> 
> The Tegra EHCI controller Kconfig depends on ARCH_TEGRA so it doesn't
> show up for other builds. I hope it's OK for the EHCI controller to
> select USB_ARCH_HAS_EHCI?

that's something the ARCH should select but it has very little value. I
guess there has been some discussions about dropping that and I support
it.

> The Tegra EHCI controller Kconfig selects everything needed for it to
> be useful, i.e. PHY support and the Tegra PHY, and I guess the ULPI
> viewport options.

should it ? I don't know... There's no way to select something as
module, but there's no eas\(y\|ier\) way to make sure PHY is enabled
when building EHCI-tegra.

> The Tegra PHY Kconfig probably shouldn't be user-visible (relying on
> being selected by the Tegra EHCI controller) and itself selects
> anything it relies on.

That I really don't like. I think the driver should be selectable and
build in anything architecture. That's the easier way to:

a) build test a driver when applying patches; and
b) make sure drivers won't include <mach/*> or <asm/*>.

-- 
balbi

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]