Hi Ulf, On Thu, Jul 25, 2019 at 3:44 PM Ulf Hansson <ulf.hansson@xxxxxxxxxx> wrote: > On Tue, 16 Jul 2019 at 17:01, Ulrich Hecht <uli+renesas@xxxxxxxx> wrote: > > This fixes a clock imbalance that occurs because the SD clock is handled > > by both PM and the hardware driver. > > See https://www.spinics.net/lists/linux-mmc/msg44431.html for details. > > This is a generic problem, when a device are being attached to a genpd > and when the genpd has got the ->stop|start() callbacks assigned, as > to manage device clocks. > > Can you try to describe this problem a little bit more in detail, as I > think that's important to carry in the change log. > > > > > This patch removes the clock handling from the driver's PM callbacks and > > runtime PM callbacks and/or system PM callbacks? > > > turns the clock off after probing. > > > > Reported-by: Geert Uytterhoeven <geert+renesas@xxxxxxxxx> > > Signed-off-by: Ulrich Hecht <uli+renesas@xxxxxxxx> > > --- > > drivers/mmc/host/tmio_mmc_core.c | 24 ++++++------------------ > > 1 file changed, 6 insertions(+), 18 deletions(-) > > > > diff --git a/drivers/mmc/host/tmio_mmc_core.c b/drivers/mmc/host/tmio_mmc_core.c > > index 31ffcc3..26dcbba 100644 > > --- a/drivers/mmc/host/tmio_mmc_core.c > > +++ b/drivers/mmc/host/tmio_mmc_core.c > > @@ -1260,9 +1260,14 @@ int tmio_mmc_host_probe(struct tmio_mmc_host *_host) > > /* See if we also get DMA */ > > tmio_mmc_request_dma(_host, pdata); > > > > - pm_runtime_set_active(&pdev->dev); > > +#ifdef CONFIG_PM > > + /* PM handles the clock; disable it so it won't be enabled twice. */ > > + if (_host->clk_disable) > > + _host->clk_disable(_host); > > The clock managed here, is that the same clock as being managed by > genpd's ->stop|start() callbacks? > > > + pm_runtime_get_sync(&pdev->dev); > > pm_runtime_set_autosuspend_delay(&pdev->dev, 50); > > pm_runtime_use_autosuspend(&pdev->dev); > > +#endif > > So what happens if you have CONFIG_PM set, but the device doesn't have > a genpd attached? That's OK for all SoCs served by renesas_sdhi_internal_dmac.c and renesas_sdhi_sys_dmac.c, as they all have their clock domain described in DT... > I am guessing the driver should handle the clock in such scenario, right? ... but it's not for (non-Renesas) systems served by tmio_mmc.c. Gr{oetje,eeting}s, Geert -- Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that. -- Linus Torvalds
