On Wed, Oct 30, 2019 at 1:46 PM Peter Geis <pgwipeout@xxxxxxxxx> wrote: > > On Wed, Oct 30, 2019 at 4:15 AM Heiko Stuebner <heiko@xxxxxxxxx> wrote: > > > > Am Montag, 28. Oktober 2019, 19:22:50 CET schrieb Peter Geis: > > > Add the rockchip innosilicon usb3 phy driver, supporting devices such as the rk3328. > > > Pulled from: > > > https://github.com/FireflyTeam/kernel/blob/roc-rk3328-cc/drivers/phy/rockchip/phy-rockchip-inno-usb3.c > > > > > > Signed-off-by: Peter Geis <pgwipeout@xxxxxxxxx> > > > > What happens on plug - unplug - plug? > > > > I remember what kept me from pushing the usb3 stuff for rk3328 > > was the usbphy breaking hotplug after the first unplug. > > > > Did this get fixed? > > > > Thanks > > Heiko > > > > > It seems to have been corrected, at least partially. > The hotplug issue may have been resolved by [0]. Other possibilities are two commits [1], [2]. One disables usb3 autosuspend, the other adds a linestate quirk. [1] https://github.com/FireflyTeam/kernel/commit/1758c75d334c7b321a86708797541e05e9301ec5 [2] https://github.com/FireflyTeam/kernel/commit/65bfd30a169c880da359a6dfef032b670457debd > > On my rk3328-roc-cc board, both USB2 and USB3 work correctly for > hotplugging with this driver. > I did encounter some odd behavior with the USB2HOST port, where it > stopped reading FS devices when this was loaded. > I'm wondering if we have a common clock problem or some other > undocumented shared dependancy. > > Robin encountered some odd behavior when unplugging a USB3 device, the > controller didn't recognize it for a while. > I don't have that problem, but I don't know what board she's using. > > [0] commit fb903392131a324a243c7731389277db1cd9f8df > clk: rockchip: fix wrong clock definitions for rk3328
