Hi Pavel, On Thu, May 7, 2020 at 6:44 PM Pavel Machek <pavel@xxxxxx> wrote: > > Hi! > > > > R-Car/RZ-G2x SoC's, this also extends the epf framework to handle multiple windows > > supported by the controller for mapping PCI address locally. > > > > Note: > > The cadence/rockchip/designware endpoint drivers are build tested only. > > > > Changes for v9 (Re-spun this series as there were minimal changes requested): > ... > > * Replaced mdelay(1) with usleep_range(1000, 1001) in rcar_pcie_ep_assert_intx() > > Are you sure that is good idea? You are requesting 1ms sleep time with 1us tolerance, > I dont believe common systems can do that. > Agreed the systems cannot do that, but the main reason of replacing mdelay(1) with usleep_range(1000, 1001) was since pci_epc_raise_irq() calls mutex_lock() and then this function rcar_pcie_ep_assert_intx(), so we can assume this function also can sleep. And, according to Documentation/timers/timers-howto.rst, we should use usleep_range() instead of mdelay(). Cheers, --Prabhakar > Best regards, > Pavel > > -- > (english) http://www.livejournal.com/~pavelmachek > (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
