[PATCH v9 0/4] PCI EP driver support MSI doorbell from host

Frank Li Frank.Li at nxp.com
Tue Sep 6 20:48:52 PDT 2022


                  ┌───────┐          ┌──────────┐
                  │       │          │          │
┌─────────────┐   │       │          │ PCI Host │
│ MSI         │◄┐ │       │          │          │
│ Controller  │ │ │       │          │          │
└─────────────┘ └─┼───────┼──────────┼─Bar0     │
                  │ PCI   │          │ Bar1     │
                  │ Func  │          │ Bar2     │
                  │       │          │ Bar3     │
                  │       │          │ Bar4     │
                  │       ├─────────►│          │
                  └───────┘          └──────────┘

Many PCI controllers provided Endpoint functions.
Generally PCI endpoint is hardware, which is not running a rich OS,
like linux.

But Linux also supports endpoint functions.  PCI Host write BAR<n> space
like write to memory. The EP side can't know memory changed by the Host
driver. 

PCI Spec has not defined a standard method to do that.  Only define
MSI(x) to let EP notified RC status change. 

The basic idea is to trigger an IRQ when PCI RC writes to a memory
address. That's what MSI controller provided.  EP drivers just need to
request a platform MSI interrupt, struct MSI_msg *msg will pass down a
memory address and data.  EP driver will map such memory address to
one of PCI BAR<n>.  Host just writes such an address to trigger EP side
IRQ.

If system have gic-its, only need update PCI EP side driver. But i.MX
have not chip support gic-its yet. So we have to use MU to simulate a
MSI controller. Although only 4 MSI IRQs are simulated, it matched
vntb(pci-epf-vntb) network requirement.

After enable MSI, ping delay reduce < 1ms from ~8ms

IRQchip: imx mu worked as MSI controller: 
     let imx mu worked as MSI controllers. Although IP is not design
as MSI controller, we still can use it if limited IRQ number to 4.

pcie: endpoint: pci-epf-vntb: add endpoint MSI support
	 Based on ntb-next branch. https://github.com/jonmason/ntb/commits/ntb-next
	 Using MSI as door bell registers
	 This patch is totally independent on previous on. It can be
applied to ntb-next seperately.

i.MX EP function driver is upstreaming by Richard Zhu.
Some dts change missed at this patches. below is reference dts change

--- a/arch/arm64/boot/dts/freescale/imx8-ss-hsio.dtsi
+++ b/arch/arm64/boot/dts/freescale/imx8-ss-hsio.dtsi
@@ -160,5 +160,6 @@ pcieb_ep: pcie_ep at 5f010000 {
                num-ib-windows = <6>;
                num-ob-windows = <6>;
                status = "disabled";
+               MSI-parent = <&lsio_mu12>;
        };

--- a/arch/arm64/boot/dts/freescale/imx8-ss-lsio.dtsi
+++ b/arch/arm64/boot/dts/freescale/imx8-ss-lsio.dtsi
@@ -172,6 +172,19 @@ lsio_mu6: mailbox at 5d210000 {
                status = "disabled";
        };

+       lsio_mu12: mailbox at 5d270000 {
+               compatible = "fsl,imx6sx-mu-MSI";
+               msi-controller;
+               interrupt-controller;
+               reg = <0x5d270000 0x10000>,     /* A side */
+                     <0x5d300000 0x10000>;     /* B side */
+               reg-names = "a", "b";
+               interrupts = <GIC_SPI 191 IRQ_TYPE_LEVEL_HIGH>;
+               power-domains = <&pd IMX_SC_R_MU_12A>,
+                               <&pd IMX_SC_R_MU_12B>;
+               power-domain-names = "a", "b";
+       };
+

Change Log
- Change from v8 to v9
  fix dt_bind_check error

- Change from v7 to v8
  irqchip: using name process-a-side as resource bind name
  pcie: endpoint:
     - fix build error reported by kernel test robot <lkp at intel.com>
     - rename epf_db_phy to epf_db_phys
     - rework error message
     - rework commit message
     - change ntb to vtb at apply irq.
     - kept name msi_virqbase because it is msi irq base number,
	not base address. 
		
- Change from v6 to v7
  pcie: endpoint: add endpoint MSI support
  Fine tuning commit message
  Fixed issues, reviewed by Bjorn Helgaas

- Change from v5 to v6
  Fixed build error found by kernel test robot

- Change from v4 to v5
  Fixed dt-binding document
        add msi-cell
        add interrupt max number
	update naming reg-names and power-domain-names.
  Fixed irqchip-Add-IMX-MU-MSI-controller-driver.patch
        rework commit message
        remove some field in struct imx_mu_dcfg
	error handle when link power domain failure.
	add irq_domain_update_bus_token

- Change from v3 to v4
  Fixed dt-binding document according to Krzysztof Kozlowski's feedback
  Fixed irqchip-imx-mu-worked-as-msi-controller according to Marc Zyngier's
        comments.

	There are still two important points, which I am not sure.
	1. clean irq_set_affinity after platform_msi_create_irq_domain.
	   Some function, like platform_msi_write_msg() is static.
	   so I have to set MSI_FLAG_USE_DEF_CHIP_OPS flags, which will
	   set irq_set_affinity to default one.
	2. about comments

	> +	msi_data->msi_domain = platform_msi_create_irq_domain(
	> +				of_node_to_fwnode(msi_data->pdev->dev.of_node),
	> +				&imx_mu_msi_domain_info,
	> +				msi_data->parent);

	"And you don't get an error due to the fact that you use the same
	fwnode for both domains without overriding the domain bus token?"

 	I did not understand yet. 

  Fixed static check warning, reported by Dan Carpenter
	pcie: endpoint: pci-epf-vntb: add endpoint MSI support

- Change from v2 to v3
  Fixed dt-binding docment check failure
  Fixed typo a cover letter.
  Change according Bjorn's comments at patch 
	pcie: endpoint: pci-epf-vntb: add endpoint MSI support
	 

- from V1 to V2
  Fixed fsl,mu-msi.yaml's problem
  Fixed irq-imx-mu-msi.c problem according Marc Zyngier's feeback 
  Added a new patch to allow pass down .pm by IRQCHIP_PLATFORM_DRIVER_END

-- 
2.35.1




More information about the linux-arm-kernel mailing list