Messages in this thread | | | Subject | Re: [PATCH v3 4/4] usb: dwc2: refactor common low-level hw code to platform.c | From | Marek Szyprowski <> | Date | Fri, 02 Oct 2015 09:47:54 +0200 |
| |
Hello,
On 2015-10-02 00:21, John Youn wrote: > On 10/1/2015 3:04 PM, Felipe Balbi wrote: >> On Thu, Oct 01, 2015 at 09:04:59PM +0000, John Youn wrote: >>> On 10/1/2015 8:50 AM, Felipe Balbi wrote: >>>> On Mon, Sep 21, 2015 at 12:16:12PM +0200, Marek Szyprowski wrote: >>>>> DWC2 module on some platforms needs three additional hardware >>>>> resources: phy controller, clock and power supply. All of them must be >>>>> enabled/activated to properly initialize and operate. This was initially >>>>> handled in s3c-hsotg driver, which has been converted to 'gadget' part >>>>> of dwc2 driver. Unfortunately, not all of this code got moved to common >>>>> platform code, what resulted in accessing DWC2 registers without >>>>> enabling low-level hardware resources. This fails for example on Exynos >>>>> SoCs. This patch moves all the code for managing those resources to >>>>> common platform.c file and provides convenient wrappers for controlling >>>>> them. >>>>> >>>>> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> >>>> I just caught several build errors which this patch. I hope you can >>>> send me a follow-up fix (which I can amend to $subject) otherwise >>>> I'll have to drop this series >>>> >>> I forgot that this was initially part of a larger >>> patch-set. Maybe that is causing issues? If this wasn't intended >>> to go through Felipe's tree then my bad. >>> >>> Also, I noticed this patch causes deadlock warnings with lockdep >>> enabled. >>> >>> Can you look into that also while you investigate this? >> John, do you want me to drop the series meanwhile ? I can do that no problems. > You can drop this one. I think leaving in patch 1-3 is fine as > they are small cleanup patches. Unless you or Marek prefer to > drop all of them.
Patches 1-3 are fixes, so please keep them. I've also just posted an updated patch no 4, I've missed the fact that dwc2_hsotg structure content was defined conditionally.
Best regards -- Marek Szyprowski, PhD Samsung R&D Institute Poland
| |