At Wed, 05 Nov 2014 10:45:09 +0100, Guillaume Gardet wrote:
Le 05/11/2014 10:31, Guillaume Gardet a écrit :
Le 05/11/2014 10:22, Takashi Iwai a écrit :
At Wed, 05 Nov 2014 10:17:10 +0100, Guillaume Gardet wrote:
Le 05/11/2014 10:01, Takashi Iwai a écrit :
At Wed, 05 Nov 2014 09:51:42 +0100, Guillaume Gardet wrote:
Le 05/11/2014 09:45, Takashi Iwai a écrit : > At Wed, 05 Nov 2014 09:39:27 +0100, > Guillaume Gardet wrote: >> Le 05/11/2014 08:08, Takashi Iwai a écrit : >>> At Tue, 04 Nov 2014 18:04:14 +0100, >>> Guillaume Gardet wrote: >>>> Le 04/11/2014 17:00, Takashi Iwai a écrit : >>>>> At Tue, 04 Nov 2014 16:39:41 +0100, >>>>> Guillaume Gardet wrote: >>>>>> Hi, >>>>>> >>>>>> the "patches.rpmify/chipidea-clean-up-dependencies" patch in openSUSE git repo is not working because IMX (and PCI) bits needs to be built as modules, at least when CONFIG_USB_CHIPIDEA is also a module. >>>>>> >>>>>> I replaced the following lines: >>>>>> +obj-$(CONFIG_USB_CHIPIDEA_PCI) += ci_hdrc_pci.o >>>>>> +obj-$(CONFIG_USB_CHIPIDEA_IMX) += usbmisc_imx.o ci_hdrc_imx.o >>>>>> >>>>>> By: >>>>>> +ifneq ($(CONFIG_USB_CHIPIDEA_PCI),) >>>>>> + obj-$(CONFIG_USB_CHIPIDEA) += ci_hdrc_pci.o >>>>>> +endif >>>>>> +ifneq ($(CONFIG_USB_CHIPIDEA_IMX),) >>>>>> + obj-$(CONFIG_USB_CHIPIDEA) += usbmisc_imx.o ci_hdrc_imx.o >>>>>> +endif >>>>>> >>>>>> That way USB is working fine on iMX6 boards (tested on Sabre Lite board), and we are more in line with upstream which uses: >>>>>> # PCI doesn't provide stubs, need to check >>>>>> ifneq ($(CONFIG_PCI),) >>>>>> obj-$(CONFIG_USB_CHIPIDEA) += ci_hdrc_pci.o >>>>>> endif >>>>>> >>>>>> ifneq ($(CONFIG_OF),) >>>>>> obj-$(CONFIG_USB_CHIPIDEA) += usbmisc_imx.o ci_hdrc_imx.o >>>>>> endif >>>>>> >>>>>> >>>>>> Any comment on how I did it? I guess we could do it better, so if you have any suggestion, please share. >>>>>> I would like to send a patch for 13.2 and master ASAP. >>>>> It'd be much easier to change CONFIG_USB_CHIPIDEA_PCI and >>>>> CONFIG_USB_CHIPIDEA_IMX from bool to tristate. >>>> Right. :) >>>> >>>> I will update the patch tomorrow. >>> Reconsidering again, I wonder whether we can just add a proper >>> dependency, e.g. depends on CONFIG_ARCH_MXS, instead of only >>> CONFIG_OF, if the faulty build on powerpc is the problem. >> Why not. (It would be CONFIG_ARCH_MXC instead of CONFIG_ARCH_MXS). >> >> Does the build problem still exist at all on powerpc? > Yeah, that's a primary question. IMO, we should begin with dropping > the patch. If anything gets broken by that, let's fix again. Ok, then I will submit a patch to remove it for master, see what's happen and make a new patch if needed. Oh, no need for that. Removing a patch is easier done by hand. Maybe better to disable the patch by guard at first, then remove it later. Do you mean to comment it out in series.conf or something else?
Add either - or + guard to the patch in series.conf. This is the standard way to disable/enable the patch while keeping it for a while.
Ok. I have done it locally (OBS branch). I am waiting for PPC builds. It could speed up the process, if it breaks.
I just noticed that CONFIG_USB_CHIPIDEA is only enable on armv7 configs! So, there will be no problem on PPC.
Do you want to keep the patch and just disable it in series.conf, or could I remove it?
I already disabled and pushed the git branch. Takashi -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org