From mboxrd@z Thu Jan 1 00:00:00 1970 From: clabbe.montjoie@gmail.com (Corentin Labbe) Date: Mon, 7 Nov 2016 11:28:15 +0100 Subject: [BUG] pinctrl: sunxi: sunxi-pinctrl fail to load with CONFIG_DEBUG_TEST_DRIVER_REMOVE In-Reply-To: <20161107101500.ostb5auj2fpxcqe6@lukather> References: <20161107081832.GA23580@Red> <20161107095612.ih5ns24m2qiwujy5@lukather> <20161107095953.GB20037@Red> <20161107101500.ostb5auj2fpxcqe6@lukather> Message-ID: <20161107102815.GA505@Red> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Mon, Nov 07, 2016 at 11:15:00AM +0100, Maxime Ripard wrote: > On Mon, Nov 07, 2016 at 10:59:53AM +0100, Corentin Labbe wrote: > > On Mon, Nov 07, 2016 at 10:56:12AM +0100, Maxime Ripard wrote: > > > On Mon, Nov 07, 2016 at 05:48:43PM +0800, Chen-Yu Tsai wrote: > > > > Hi, > > > > > > > > On Mon, Nov 7, 2016 at 4:18 PM, LABBE Corentin > > > > wrote: > > > > > Hello > > > > > > > > > > With CONFIG_DEBUG_TEST_DRIVER_REMOVE=y pinctrl-sunxi fail to load on the second try. > > > > > > > > > > [ 3.900061] sun8i-h3-pinctrl 1c20800.pinctrl: initialized sunXi PIO driver > > > > > [ 3.916251] gpio gpiochip1: GPIO integer space overlap, cannot add chip > > > > > [ 3.923016] gpiochip_add_data: GPIOs 0..223 (1c20800.pinctrl) failed to register > > > > > [ 3.931099] sun8i-h3-pinctrl: probe of 1c20800.pinctrl failed with error -16 > > > > > [ 3.944709] sun8i-h3-r-pinctrl 1f02c00.pinctrl: initialized sunXi PIO driver > > > > > [ 3.960796] gpio gpiochip2: GPIO integer space overlap, cannot add chip > > > > > [ 3.967594] gpiochip_add_data: GPIOs 352..383 (1f02c00.pinctrl) failed to register > > > > > [ 3.975633] sun8i-h3-r-pinctrl: probe of 1f02c00.pinctrl failed with error -16 > > > > > > > > > > Without it, all subsequent drivers fail to load. > > > > > Tested on Orange PI PC board. > > > > > > > > I don't think the pinctrl drivers were designed to be removed. > > > > And I thought the lack of a .remove callback in the driver blocks > > > > the core from removing the device? Maybe I remember wrong... > > > > > > Using a builtin_platform_driver should be enough to prevent it to be > > > removed. > > > > > > > The problem is that it is already builtin_platform_driver() > > Then there's no way it can be removed in the first place. > I will send a patch for fixing CONFIG_DEBUG_TEST_DRIVER_REMOVE