| Message ID | 20180321105005.18426-2-u.kleine-koenig@pengutronix.de |
|---|---|
| State | New |
| Headers | show |
| Series | [1/3] arm64: dts: marvell: armada-37xx: mark the gpio controllers as irq controller | expand |
On Wed, Mar 21, 2018 at 11:50:04AM +0100, Uwe Kleine-König wrote: > The switch's INTn line is connected to the CPU's MPP2_23 pad. Wire this > up in the device tree. Note however that up to now the > marvell,armada3710-sb-pinctrl driver doesn't support level irqs, so the > switch driver doesn't make use of this feature. Hi Uwe So that actually happens here, if the interrupt cannot be used? I assume the mv88e6xxx driver does not fail the probe because the IRQ cannot be configured? What does happen? Thanks Andrew
Hello Andrew, On Wed, Mar 21, 2018 at 02:15:36PM +0100, Andrew Lunn wrote: > On Wed, Mar 21, 2018 at 11:50:04AM +0100, Uwe Kleine-König wrote: > > The switch's INTn line is connected to the CPU's MPP2_23 pad. Wire this > > up in the device tree. Note however that up to now the > > marvell,armada3710-sb-pinctrl driver doesn't support level irqs, so the > > switch driver doesn't make use of this feature. > > So that actually happens here, if the interrupt cannot be used? I > assume the mv88e6xxx driver does not fail the probe because the IRQ > cannot be configured? What does happen? Grml, I just tried with all patches applied and it results in: [ 4.777524] mv88e6085 d0032004.mdio-mii:01: switch 0x3410 detected: Marvell 88E6341, revision 0 [ 4.794723] genirq: Setting trigger mode 8 for irq 41 failed (armada_37xx_irq_set_type+0x0/0x160) [ 4.804470] mv88e6085: probe of d0032004.mdio-mii:01 failed with error -22 . It (kind of) worked before your patch net: dsa: mv88e6xxx: Use the DT IRQ trigger mode . So I guess this has to change to + interrupt-parent = <&gpiosb>; + /* actually the irq is active low, but gpiosb doesn't support that and falling edge seems to work just fine */ + interrupts = <23 IRQ_TYPE_EDGE_FALLING>; Best regards Uwe
> . So I guess this has to change to > > + interrupt-parent = <&gpiosb>; > + /* actually the irq is active low, but gpiosb doesn't support that and falling edge seems to work just fine */ > + interrupts = <23 IRQ_TYPE_EDGE_FALLING>; Or wait until you have simulated level interrupts? At the moment, without this patch, you should have driver polled interrupts. That already gives you a good speedup over phylib polling the PHYs. Andrew
diff --git a/arch/arm64/boot/dts/marvell/armada-3720-espressobin.dts b/arch/arm64/boot/dts/marvell/armada-3720-espressobin.dts index 6d070b267abe..a88f053d2f28 100644 --- a/arch/arm64/boot/dts/marvell/armada-3720-espressobin.dts +++ b/arch/arm64/boot/dts/marvell/armada-3720-espressobin.dts @@ -133,6 +133,11 @@ #size-cells = <0>; reg = <1>; + interrupt-parent = <&gpiosb>; + interrupts = <23 IRQ_TYPE_LEVEL_LOW>; + interrupt-controller; + #interrupt-cells = <2>; + dsa,member = <0 0>; ports { @@ -171,12 +176,20 @@ switch0phy0: switch0phy0@11 { reg = <0x11>; + interrupt-parent = <&switch0>; + interrupts = <1 IRQ_TYPE_LEVEL_HIGH>; }; + switch0phy1: switch0phy1@12 { reg = <0x12>; + interrupt-parent = <&switch0>; + interrupts = <2 IRQ_TYPE_LEVEL_HIGH>; }; + switch0phy2: switch0phy2@13 { reg = <0x13>; + interrupt-parent = <&switch0>; + interrupts = <3 IRQ_TYPE_LEVEL_HIGH>; }; }; };
The switch's INTn line is connected to the CPU's MPP2_23 pad. Wire this up in the device tree. Note however that up to now the marvell,armada3710-sb-pinctrl driver doesn't support level irqs, so the switch driver doesn't make use of this feature. Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> --- arch/arm64/boot/dts/marvell/armada-3720-espressobin.dts | 13 +++++++++++++ 1 file changed, 13 insertions(+)