On 09/28/2018 12:56 PM, Tony Lindgren wrote: > * Keerthy <j-keerthy@xxxxxx> [180928 11:43]: >> I tried [1]. I see gpio throwing some errors while suspending: >> >> https://pastebin.ubuntu.com/p/92qbWh4rgW/ >> >> [ 64.610205] omap_gpio 44e07000.gpio: sysc_child_suspend_noirq busy at >> 1257: -16 >> [ 64.612509] omap_gpio 48320000.gpio: sysc_child_suspend_noirq busy at >> 1257: -16 >> [ 64.612754] omap_gpio 481ae000.gpio: sysc_child_suspend_noirq busy at >> 1257: -16 > > Thanks for testing :) Yeah I'm seeing those warnings too, the system > came up just fine from resume right? > > With the gpio notifier changes we now only idle gpio interrupts > from the notifier so in general those warnings are harmless. We > should probably just make these warnings dev_dbg now. > > Those warnings should only appear for gpio instances with interrupts > as in grep gpio /proc/interrupts. And we only idle those gpio banks > if a deeper idle state is entered, which is not the case for dra7. > > I'm mostly worried that all the devices work as before on dra7 > though, like LCD, USB, SATA and so on.. This is suspend, not idle and there is difference between omap_device and ti-sysc (not all ti-sysc nodes works in legacy mode as you have hwmod props in DT: sysc_child_suspend_noirq() if (!pm_runtime_status_suspended(dev)) { error = pm_generic_runtime_suspend(dev); if (error) { dev_warn(dev, "%s busy at %i: %i\n", __func__, __LINE__, error); ^^^ with your recent gpio changes this case will be hit for gpio banks with irs enabled always. return 0; } omap_device: ret = pm_generic_suspend_noirq(dev); if (!ret && !pm_runtime_status_suspended(dev)) { if (pm_generic_runtime_suspend(dev) == 0) { ^^ will be silent in the same case omap_device_idle(pdev); od->flags |= OMAP_DEVICE_SUSPENDED; } } So, may be you can change dev_warn() -> dev_dbg(). -- regards, -grygorii
