Re: WIP: PCIe on MSM8998

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 2019-03-20 8:17 pm, Marc Gonzalez wrote:
On 20/03/2019 17:51, Robin Murphy wrote:

On 12/03/2019 15:59, Marc Gonzalez wrote:

I still can't get the ethernet card to play ball. As soon as I set the link up,
the system hangs a few seconds, then reboots.

Here are the relevant DT nodes:
(There's probably an error or 12 lurking in there)

		anoc1_smmu: arm,smmu@1680000 {
			compatible = "qcom,msm8996-smmu-v2", "qcom,smmu-v2";
			reg = <0x01680000 0x10000>;
			#iommu-cells = <0>;

That's almost certainly not what you want - note that the binding is
only defined for 1 or 2 cells. The current Linux driver does actually
happen to handle 0 cells, in which case it will assume every device uses
Stream ID 0, but if the funky mappings in the SDM845 patch are anything
to go by I doubt that's correct.

			#global-interrupts = <0>;

Ouch - not having a global fault interrupt will make this even more
painful to debug. At least with that you can figure out what Stream IDs
are in play by forcing stuff to fault.

Hello Robin,

Your insight is always appreciated!

I can't say I have wrapped my head around iommu-cells, iommu-map,
and iommu-map-mask (yet) or stream IDs.

FWIW, here is the DT node as it appears downstream:
https://source.codeaurora.org/quic/la/kernel/msm-4.4/tree/arch/arm/boot/dts/qcom/msm-arm-smmu-8998.dtsi?h=LE.UM.1.3.r3.25#n18

	anoc1_smmu: arm,smmu-anoc1@1680000 {
		status = "ok";
		compatible = "qcom,smmu-v2";
		reg = <0x1680000 0x10000>;
		#iommu-cells = <0>;
		qcom,register-save;
		qcom,skip-init;
		#global-interrupts = <0>;
		interrupts = <GIC_SPI 364 IRQ_TYPE_EDGE_RISING>,
			   <GIC_SPI 365 IRQ_TYPE_EDGE_RISING>,
			   <GIC_SPI 366 IRQ_TYPE_EDGE_RISING>,
			   <GIC_SPI 367 IRQ_TYPE_EDGE_RISING>,
			   <GIC_SPI 368 IRQ_TYPE_EDGE_RISING>,
			   <GIC_SPI 369 IRQ_TYPE_EDGE_RISING>;
		qcom,msm-bus,name = "smmu-bus-client-anoc1";
		qcom,msm-bus,num-cases = <2>;
		qcom,msm-bus,active-only;
		qcom,msm-bus,num-paths = <1>;
		/* aggre1_noc_clk */
		qcom,msm-bus,vectors-KBps =
				<84 10062 0 0>,
				<84 10062 0 1000>;
	};

There's even a rationale for #iommu-cells=0, albeit for a different board:
https://source.codeaurora.org/quic/la/kernel/msm-4.4/commit/arch/arm/boot/dts/qcom?h=LE.UM.1.3.r3.25&id=f1fa301f977f06dcf990c0452d85e2f67d8cbbf1

There's currently a placeholder stream ID of "1" in the PCIe RC
device tree node. The PCIe RC doesn't actually attach and doesn't
need a stream ID though, the "1" was put there to get around an error
caused by the fact that #iommu-cells was equal to 1 for the
anoc1_smmu, even though it should have actually been 0. Fix all this
by making #iommu-cells = <0> for the anoc1 SMMU and removing the
bogus placeholder stream ID.

I can't say I understand this blurb.

Unfortunately, having looked around the code, I think I do. 4.4 long predates the iommu-map binding, and in the absence of anything other than the hard-coded SID==RID assumption of arm-smmu at the time, they apparently went and did their own wacky thing[1]. AFAICS the Stream ID appears to be pretty much derived from the PCI topology as I would hope, but it looks like it might depend on some sort of lookup table being programmed appropriately as well.

Bear in mind that the this is _The Qualcomm Android Kernel_ we're trying to reason about here - playing true to the stereotype, the diff against the mainline driver is significantly bigger than the entire mainline driver itself; the line count of arm-smmu.c alone is pushing 2-and-a-half times that of the file in 4.4.y ;)

Since the curiosity had set in, I finally got round to dumping the ACPI tables from my Snapdragon 835 laptop, and judging by the IORT it seems like the EFI firmware for Windows machines does provide some set of static ID mappings which could probably transcribe to an iommu-map (if indeed it's valid at all - Windows itself doesn't seem to be even touching PCI here), but I guess the Android BSP might not be so generous. That'll be a question for the Qualcomm folks. FWIW mine interestingly claims that its SMMU instances are all sharing SPI 231 as a global fault interrupt, but whether that's true and/or depends on the runtime firmware, again I really have no idea.

Robin.

[1] https://source.codeaurora.org/quic/la/kernel/msm-4.4/tree/drivers/pci/host/pci-msm.c?h=LE.UM.1.3.r3.25&id=f1fa301f977f06dcf990c0452d85e2f67d8cbbf1#n4687



[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [Linux for Sparc]     [IETF Annouce]     [Security]     [Bugtraq]     [Linux MIPS]     [ECOS]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux