Hi, On 09-12-18 15:44, Andy Shevchenko wrote:
On Thu, Dec 6, 2018 at 8:49 PM Stephan Gerhold <[email protected]> wrote:Commit 211f658b7b40 ("usb: dwc3: pci: Use devm functions to get the phy GPIOs") changed the code to claim the PHY GPIOs permanently for Intel Baytrail devices. This causes issues when the actual PHY driver attempts to claim the same GPIO descriptors. For example, tusb1210 now fails to probe with: tusb1210: probe of dwc3.0.auto.ulpi failed with error -16 (EBUSY) dwc3-pci needs to turn on the PHY once before dwc3 is loaded, but usually the PHY driver will then hold the GPIOs to turn off the PHY when requested (e.g. during suspend). To fix the problem, this reverts the commit to restore the old behavior to put the GPIOs immediately after usage. Link: https://www.spinics.net/lists/linux-usb/msg174681.html Cc: [email protected] Signed-off-by: Stephan Gerhold <[email protected]> --- drivers/usb/dwc3/dwc3-pci.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/drivers/usb/dwc3/dwc3-pci.c b/drivers/usb/dwc3/dwc3-pci.c index 842795856bf4..fdc6e4e403e8 100644 --- a/drivers/usb/dwc3/dwc3-pci.c +++ b/drivers/usb/dwc3/dwc3-pci.c @@ -170,20 +170,20 @@ static int dwc3_pci_quirks(struct dwc3_pci *dwc) * put the gpio descriptors again here because the phy driver * might want to grab them, too. */ - gpio = devm_gpiod_get_optional(&pdev->dev, "cs", - GPIOD_OUT_LOW); + gpio = gpiod_get_optional(&pdev->dev, "cs", GPIOD_OUT_LOW); if (IS_ERR(gpio)) return PTR_ERR(gpio); gpiod_set_value_cansleep(gpio, 1); + gpiod_put(gpio); - gpio = devm_gpiod_get_optional(&pdev->dev, "reset", - GPIOD_OUT_LOW); + gpio = gpiod_get_optional(&pdev->dev, "reset", GPIOD_OUT_LOW); if (IS_ERR(gpio)) return PTR_ERR(gpio); if (gpio) { gpiod_set_value_cansleep(gpio, 1);+ gpiod_put(gpio); usleep_range(10000, 11000);If something happens to GPIO line in between of these lines, the sleep might become obsolete. Shouldn't gpiod_put() be placed after?
The actual reading of the ULPI product and vendor IDs happens a whole bunch of functions after the dwc3_pci_quirks() function runs, so the assumption is that nothing touches the GPIO until the phy driver loads. Also the current version of this patch is a straight revert of my patch, IOW the put allowing potential mucking with the GPIO before the IDs are read is a pre-existing problem. Regards, Hans
