| Message ID | 1422056653-11271-1-git-send-email-soren.brinkmann@xilinx.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Fri, 2015-01-23 at 03:44PM -0800, Soren Brinkmann wrote: > The CPU_2X clock does not have a classical in kernel user, but is, > amongst other things, required for OCM and debug access. Make sure this > clock does not mistakenly disabled during boot up by enabling it in the > platforms clock driver. > > Signed-off-by: Soren Brinkmann <soren.brinkmann@xilinx.com> > --- > I just noticed this on Linus' latest tip. I can't remember having had issues > connecting the debugger before, but it might be a candidate for stable in case > people report something like that. I bisected this down to commit: ae43b3289186480f81c78bb63d788a85a3631f47 is the first bad commit commit ae43b3289186480f81c78bb63d788a85a3631f47 Author: Krzysztof Kozlowski <k.kozlowski@samsung.com> Date: Fri Nov 14 09:48:57 2014 +0100 ARM: 8202/1: dmaengine: pl330: Add runtime Power Management support v12 This patch adds runtime PM support to pl330 DMA engine driver. The runtime power management for pl330 DMA driver allows gating of AMBA clock (PDMA) in FSYS clock domain, when the device is not processing any requests. This is necessary to enter low power modes on Exynos SoCs (e.g. LPA on Exynos4x12 or W-AFTR on Exynos3250). Runtime PM resuming of the device may happen in atomic context (during call device_issue_pending()) so pm_runtime_irq_safe() is used. This will lead only to disabling/enabling of the clock but this is sufficient for gating the clock and for reducing energy usage. Driver uses runtime PM callbacks from amba/bus.c driver only. Suggested-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com> Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org> Acked-by: Vinod Koul <vinod.koul@intel.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> :040000 040000 16e0cb02587254ea396e9d913d7997ac63bd3d49 97c1ee366f0790e65df00e85d06b2bcd07aa9f58 M drivers which makes sense, since the PL330 clock in Zynq is the mentioned CPU_2X_CLK. I guess the additional runtime_pm in the driver causes the clock to be gated which causes the issues for the mentioned other users of this clock. Looks like this patch got introduced in the 3.19 merge window. Hence, stable should be off the table. Sören
On Sun, 2015-01-25 at 07:22PM -0800, Sören Brinkmann wrote: > On Fri, 2015-01-23 at 03:44PM -0800, Soren Brinkmann wrote: > > The CPU_2X clock does not have a classical in kernel user, but is, > > amongst other things, required for OCM and debug access. Make sure this > > clock does not mistakenly disabled during boot up by enabling it in the > > platforms clock driver. > > > > Signed-off-by: Soren Brinkmann <soren.brinkmann@xilinx.com> > > --- > > I just noticed this on Linus' latest tip. I can't remember having had issues > > connecting the debugger before, but it might be a candidate for stable in case > > people report something like that. > > I bisected this down to commit: > > ae43b3289186480f81c78bb63d788a85a3631f47 is the first bad commit > commit ae43b3289186480f81c78bb63d788a85a3631f47 > Author: Krzysztof Kozlowski <k.kozlowski@samsung.com> > Date: Fri Nov 14 09:48:57 2014 +0100 > > ARM: 8202/1: dmaengine: pl330: Add runtime Power Management support v12 > > This patch adds runtime PM support to pl330 DMA engine driver. > > The runtime power management for pl330 DMA driver allows gating of AMBA > clock (PDMA) in FSYS clock domain, when the device is not processing any > requests. This is necessary to enter low power modes on Exynos SoCs > (e.g. LPA on Exynos4x12 or W-AFTR on Exynos3250). > > Runtime PM resuming of the device may happen in atomic context (during > call device_issue_pending()) so pm_runtime_irq_safe() is used. This will > lead only to disabling/enabling of the clock but this is sufficient for > gating the clock and for reducing energy usage. > > Driver uses runtime PM callbacks from amba/bus.c driver only. > > Suggested-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> > Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com> > Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org> > Acked-by: Vinod Koul <vinod.koul@intel.com> > Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> > > :040000 040000 16e0cb02587254ea396e9d913d7997ac63bd3d49 97c1ee366f0790e65df00e85d06b2bcd07aa9f58 M drivers > > which makes sense, since the PL330 clock in Zynq is the mentioned > CPU_2X_CLK. I guess the additional runtime_pm in the driver causes the > clock to be gated which causes the issues for the mentioned other users > of this clock. > Looks like this patch got introduced in the 3.19 merge window. Hence, > stable should be off the table. That statement may have been a little premature. Turns out, it actually is the PL330 driver that is responsible for activating the CPU2X clock on Zynq. Things work as expected when that driver is in the kernel, but break as seen on 3.19 when that driver is not part of the kernel. Hence, this should probably go into stable, all the way back to when we merged Zynq's CCF code, which seems to have happened in 3.11. Soren
diff --git a/drivers/clk/zynq/clkc.c b/drivers/clk/zynq/clkc.c index c363106ed726..f8469315161d 100644 --- a/drivers/clk/zynq/clkc.c +++ b/drivers/clk/zynq/clkc.c @@ -346,6 +346,7 @@ static void __init zynq_clk_setup(struct device_node *np) clks[cpu_2x] = clk_register_gate(NULL, clk_output_name[cpu_2x], "cpu_2x_div", CLK_IGNORE_UNUSED, SLCR_ARM_CLK_CTRL, 26, 0, &armclk_lock); + clk_prepare_enable(clks[cpu_2x]); clk = clk_register_fixed_factor(NULL, "cpu_1x_div", "cpu_div", 0, 1, 4 + 2 * tmp);
The CPU_2X clock does not have a classical in kernel user, but is, amongst other things, required for OCM and debug access. Make sure this clock does not mistakenly disabled during boot up by enabling it in the platforms clock driver. Signed-off-by: Soren Brinkmann <soren.brinkmann@xilinx.com> --- I just noticed this on Linus' latest tip. I can't remember having had issues connecting the debugger before, but it might be a candidate for stable in case people report something like that. Sören drivers/clk/zynq/clkc.c | 1 + 1 file changed, 1 insertion(+)