On Fri, Dec 6, 2019 at 10:14 AM Marek Szyprowski <m.szyprowski@xxxxxxxxxxx> wrote: > On 06.12.2019 08:55, Marek Szyprowski wrote: > > NAK. > > > > Sorry, but this patch breaks USB3503 HUB operation on Arndale5250 > > board. A brief scan through the code reveals that the whole control > > logic for the 'intn' gpio is lost. > > Well, I've checked further and 'intn' logic is there. The issue with > Arndale5250 board is something different. Changing the gpio active > values in arch/arm/boot/dts/exynos5250-arndale.dts from GPIO_ACTIVE_LOW > to GPIO_ACTIVE_HIGH fixed operation of usb3503 HUB. I really wonder why > it worked fine with non-descriptor code and the ACTIVE_LOW DT flags... > > I'm not sure how to handle this. Old code works also fine with DT flags > changed to GPIO_ACTIVE_HIGH, which seems to be a proper value for those > gpio lines. We should of course fix up the device trees so the polarity in them is correct. If the compatibility with elder device trees is mandatory I will make a quirk into the gpiolib-of.c that enforce active high on this specific GPIO line. This is pretty straight-forward, I can just use the compatible of the board and usb3503 in combination to enforce it. Yours, Linus Walleij
