Assertion of the MSI Enable bit of RC's MSI CAP is mandatory required to
trigger MSI on i.MX6 PCIe.
This bit would be asserted when CONFIG_PCIEPORTBUS=y.
Thus, the MSI works fine on i.MX6 PCIe before the commit "f3fdfc4".
Assert it unconditionally when MSI is enabled.
Otherwise, the MSI wouldn't be triggered although the EP is present and
the MSIs are assigned.
Signed-off-by: Richard Zhu <[email protected]>
Reviewed-by: Lucas Stach <[email protected]>
---
Changes v1 -> v2:
* Assert the MSI_EN unconditionally when MSI is supported.
Changes v2 -> v3:
* Remove the IS_ENABLED(CONFIG_PCI_MSI) since the driver depends on
PCI_MSI_IRQ_DOMAIN
* Extended with a check for pci_msi_enabled() to see if the user
explicitly want legacy IRQs
Changes v3 -> v4:
* Refer to Bjorn's comments, refine the subject and commit log and change
the PCI_MSI_CAP to PCIE_RC_IMX6_MSI_CAP.
---
drivers/pci/controller/dwc/pci-imx6.c | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/drivers/pci/controller/dwc/pci-imx6.c b/drivers/pci/controller/dwc/pci-imx6.c
index 26087b3..639bb27 100644
--- a/drivers/pci/controller/dwc/pci-imx6.c
+++ b/drivers/pci/controller/dwc/pci-imx6.c
@@ -74,6 +74,7 @@ struct imx6_pcie {
#define PHY_PLL_LOCK_WAIT_USLEEP_MAX 200
/* PCIe Root Complex registers (memory-mapped) */
+#define PCIE_RC_IMX6_MSI_CAP 0x50
#define PCIE_RC_LCR 0x7c
#define PCIE_RC_LCR_MAX_LINK_SPEEDS_GEN1 0x1
#define PCIE_RC_LCR_MAX_LINK_SPEEDS_GEN2 0x2
@@ -926,6 +927,7 @@ static int imx6_pcie_probe(struct platform_device *pdev)
struct resource *dbi_base;
struct device_node *node = dev->of_node;
int ret;
+ u16 val;
imx6_pcie = devm_kzalloc(dev, sizeof(*imx6_pcie), GFP_KERNEL);
if (!imx6_pcie)
@@ -1071,6 +1073,14 @@ static int imx6_pcie_probe(struct platform_device *pdev)
if (ret < 0)
return ret;
+ if (pci_msi_enabled()) {
+ val = dw_pcie_readw_dbi(pci, PCIE_RC_IMX6_MSI_CAP +
+ PCI_MSI_FLAGS);
+ val |= PCI_MSI_FLAGS_ENABLE;
+ dw_pcie_writew_dbi(pci, PCIE_RC_MSI_IMX6_CAP +
+ PCI_MSI_FLAGS, val);
+ }
+
return 0;
}
--
2.7.4
Hi Richard,
I love your patch! Yet something to improve:
[auto build test ERROR on pci/next]
[also build test ERROR on v4.20-rc7 next-20181218]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]
url: https://github.com/0day-ci/linux/commits/Richard-Zhu/PCI-imx-make-msi-work-without-CONFIG_PCIEPORTBUS-y/20181217-004814
base: https://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci.git next
config: arm-allmodconfig (attached as .config)
compiler: arm-linux-gnueabi-gcc (Debian 7.2.0-11) 7.2.0
reproduce:
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
GCC_VERSION=7.2.0 make.cross ARCH=arm
All errors (new ones prefixed by >>):
drivers/pci/controller/dwc/pci-imx6.c: In function 'imx6_pcie_probe':
>> drivers/pci/controller/dwc/pci-imx6.c:1158:27: error: 'PCIE_RC_MSI_IMX6_CAP' undeclared (first use in this function); did you mean 'PCIE_RC_IMX6_MSI_CAP'?
dw_pcie_writew_dbi(pci, PCIE_RC_MSI_IMX6_CAP +
^~~~~~~~~~~~~~~~~~~~
PCIE_RC_IMX6_MSI_CAP
drivers/pci/controller/dwc/pci-imx6.c:1158:27: note: each undeclared identifier is reported only once for each function it appears in
vim +1158 drivers/pci/controller/dwc/pci-imx6.c
994
995 static int imx6_pcie_probe(struct platform_device *pdev)
996 {
997 struct device *dev = &pdev->dev;
998 struct dw_pcie *pci;
999 struct imx6_pcie *imx6_pcie;
1000 struct resource *dbi_base;
1001 struct device_node *node = dev->of_node;
1002 int ret;
1003 u16 val;
1004
1005 imx6_pcie = devm_kzalloc(dev, sizeof(*imx6_pcie), GFP_KERNEL);
1006 if (!imx6_pcie)
1007 return -ENOMEM;
1008
1009 pci = devm_kzalloc(dev, sizeof(*pci), GFP_KERNEL);
1010 if (!pci)
1011 return -ENOMEM;
1012
1013 pci->dev = dev;
1014 pci->ops = &dw_pcie_ops;
1015
1016 imx6_pcie->pci = pci;
1017 imx6_pcie->variant =
1018 (enum imx6_pcie_variants)of_device_get_match_data(dev);
1019
1020 if (of_property_read_u32(node, "fsl,controller-id",
1021 &imx6_pcie->controller_id))
1022 imx6_pcie->controller_id = 0;
1023
1024 dbi_base = platform_get_resource(pdev, IORESOURCE_MEM, 0);
1025 pci->dbi_base = devm_ioremap_resource(dev, dbi_base);
1026 if (IS_ERR(pci->dbi_base))
1027 return PTR_ERR(pci->dbi_base);
1028
1029 /* Fetch GPIOs */
1030 imx6_pcie->reset_gpio = of_get_named_gpio(node, "reset-gpio", 0);
1031 imx6_pcie->gpio_active_high = of_property_read_bool(node,
1032 "reset-gpio-active-high");
1033 if (gpio_is_valid(imx6_pcie->reset_gpio)) {
1034 ret = devm_gpio_request_one(dev, imx6_pcie->reset_gpio,
1035 imx6_pcie->gpio_active_high ?
1036 GPIOF_OUT_INIT_HIGH :
1037 GPIOF_OUT_INIT_LOW,
1038 "PCIe reset");
1039 if (ret) {
1040 dev_err(dev, "unable to get reset gpio\n");
1041 return ret;
1042 }
1043 } else if (imx6_pcie->reset_gpio == -EPROBE_DEFER) {
1044 return imx6_pcie->reset_gpio;
1045 }
1046
1047 /* Fetch clocks */
1048 imx6_pcie->pcie_phy = devm_clk_get(dev, "pcie_phy");
1049 if (IS_ERR(imx6_pcie->pcie_phy)) {
1050 dev_err(dev, "pcie_phy clock source missing or invalid\n");
1051 return PTR_ERR(imx6_pcie->pcie_phy);
1052 }
1053
1054 imx6_pcie->pcie_bus = devm_clk_get(dev, "pcie_bus");
1055 if (IS_ERR(imx6_pcie->pcie_bus)) {
1056 dev_err(dev, "pcie_bus clock source missing or invalid\n");
1057 return PTR_ERR(imx6_pcie->pcie_bus);
1058 }
1059
1060 imx6_pcie->pcie = devm_clk_get(dev, "pcie");
1061 if (IS_ERR(imx6_pcie->pcie)) {
1062 dev_err(dev, "pcie clock source missing or invalid\n");
1063 return PTR_ERR(imx6_pcie->pcie);
1064 }
1065
1066 switch (imx6_pcie->variant) {
1067 case IMX6SX:
1068 imx6_pcie->pcie_inbound_axi = devm_clk_get(dev,
1069 "pcie_inbound_axi");
1070 if (IS_ERR(imx6_pcie->pcie_inbound_axi)) {
1071 dev_err(dev, "pcie_inbound_axi clock missing or invalid\n");
1072 return PTR_ERR(imx6_pcie->pcie_inbound_axi);
1073 }
1074 break;
1075 case IMX8MQ:
1076 case IMX7D: /* FALLTHROUGH */
1077 imx6_pcie->pciephy_reset = devm_reset_control_get_exclusive(dev,
1078 "pciephy");
1079 if (IS_ERR(imx6_pcie->pciephy_reset)) {
1080 dev_err(dev, "Failed to get PCIEPHY reset control\n");
1081 return PTR_ERR(imx6_pcie->pciephy_reset);
1082 }
1083
1084 imx6_pcie->apps_reset = devm_reset_control_get_exclusive(dev,
1085 "apps");
1086 if (IS_ERR(imx6_pcie->apps_reset)) {
1087 dev_err(dev, "Failed to get PCIE APPS reset control\n");
1088 return PTR_ERR(imx6_pcie->apps_reset);
1089 }
1090 break;
1091 default:
1092 break;
1093 }
1094
1095 /* Grab turnoff reset */
1096 imx6_pcie->turnoff_reset = devm_reset_control_get_optional_exclusive(dev, "turnoff");
1097 if (IS_ERR(imx6_pcie->turnoff_reset)) {
1098 dev_err(dev, "Failed to get TURNOFF reset control\n");
1099 return PTR_ERR(imx6_pcie->turnoff_reset);
1100 }
1101
1102 /* Grab GPR config register range */
1103 imx6_pcie->iomuxc_gpr =
1104 syscon_regmap_lookup_by_compatible("fsl,imx6q-iomuxc-gpr");
1105 if (IS_ERR(imx6_pcie->iomuxc_gpr)) {
1106 dev_err(dev, "unable to find iomuxc registers\n");
1107 return PTR_ERR(imx6_pcie->iomuxc_gpr);
1108 }
1109
1110 /* Grab PCIe PHY Tx Settings */
1111 if (of_property_read_u32(node, "fsl,tx-deemph-gen1",
1112 &imx6_pcie->tx_deemph_gen1))
1113 imx6_pcie->tx_deemph_gen1 = 0;
1114
1115 if (of_property_read_u32(node, "fsl,tx-deemph-gen2-3p5db",
1116 &imx6_pcie->tx_deemph_gen2_3p5db))
1117 imx6_pcie->tx_deemph_gen2_3p5db = 0;
1118
1119 if (of_property_read_u32(node, "fsl,tx-deemph-gen2-6db",
1120 &imx6_pcie->tx_deemph_gen2_6db))
1121 imx6_pcie->tx_deemph_gen2_6db = 20;
1122
1123 if (of_property_read_u32(node, "fsl,tx-swing-full",
1124 &imx6_pcie->tx_swing_full))
1125 imx6_pcie->tx_swing_full = 127;
1126
1127 if (of_property_read_u32(node, "fsl,tx-swing-low",
1128 &imx6_pcie->tx_swing_low))
1129 imx6_pcie->tx_swing_low = 127;
1130
1131 /* Limit link speed */
1132 ret = of_property_read_u32(node, "fsl,max-link-speed",
1133 &imx6_pcie->link_gen);
1134 if (ret)
1135 imx6_pcie->link_gen = 1;
1136
1137 imx6_pcie->vpcie = devm_regulator_get_optional(&pdev->dev, "vpcie");
1138 if (IS_ERR(imx6_pcie->vpcie)) {
1139 if (PTR_ERR(imx6_pcie->vpcie) == -EPROBE_DEFER)
1140 return -EPROBE_DEFER;
1141 imx6_pcie->vpcie = NULL;
1142 }
1143
1144 platform_set_drvdata(pdev, imx6_pcie);
1145
1146 ret = imx6_pcie_attach_pd(dev);
1147 if (ret)
1148 return ret;
1149
1150 ret = imx6_add_pcie_port(imx6_pcie, pdev);
1151 if (ret < 0)
1152 return ret;
1153
1154 if (pci_msi_enabled()) {
1155 val = dw_pcie_readw_dbi(pci, PCIE_RC_IMX6_MSI_CAP +
1156 PCI_MSI_FLAGS);
1157 val |= PCI_MSI_FLAGS_ENABLE;
> 1158 dw_pcie_writew_dbi(pci, PCIE_RC_MSI_IMX6_CAP +
1159 PCI_MSI_FLAGS, val);
1160 }
1161
1162 return 0;
1163 }
1164
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
Am Mittwoch, den 19.12.2018, 08:12 -0600 schrieb Bjorn Helgaas:
> [+cc Sven, Trent, et al from related report:
> https://lore.kernel.org/linux-pci/[email protected]]
>
> On Fri, Dec 14, 2018 at 06:44:15AM +0000, Richard Zhu wrote:
> > Assertion of the MSI Enable bit of RC's MSI CAP is mandatory required to
> > trigger MSI on i.MX6 PCIe.
> > This bit would be asserted when CONFIG_PCIEPORTBUS=y.
> > Thus, the MSI works fine on i.MX6 PCIe before the commit "f3fdfc4".
> >
> > Assert it unconditionally when MSI is enabled.
> > Otherwise, the MSI wouldn't be triggered although the EP is present and
> > the MSIs are assigned.
>
> OK, I think I finally understand most of what's going on. Please
> check the following possible changelog text:
>
> The MSI Enable bit in the MSI Capability (PCIe r4.0, sec 7.7.1.2)
> controls whether a Function can request service using MSI.
>
> i.MX6 Root Ports implement the MSI Capability and may use MSI to
> request service for events like PME, hotplug, AER, etc. In
> addition, on i.MX6, the MSI Enable bit controls delivery of MSI
> interrupts from components below the Root Port.
>
> Prior to f3fdfc4ac3a2 ("PCI: Remove host driver Kconfig selection of
> CONFIG_PCIEPORTBUS"), enabling CONFIG_PCI_IMX6 automatically also
> enabled CONFIG_PCIEPORTBUS, and when portdrv claimed the Root Ports,
> it set the MSI Enable bit so it could use PME, hotplug, AER, etc.
> As a side effect, that also enabled delivery of MSI interrupts from
> downstream components.
>
> After f3fdfc4ac3a2, the imx6q-pcie driver can operate without
> portdrv, but that means imx6q-pcie must set the MSI Enable bit
> itself if downstream components use MSI.
>
> Fixes: f3fdfc4ac3a2 ("PCI: Remove host driver Kconfig selection of CONFIG_PCIEPORTBUS")
>
> I still don't understand exactly *how* MSI Enable affects MSI from
> downstream components, since the downstream component just does a DMA
> write, and the Root Port can't tell whether the write is to memory or
> interrupt controller unless the Root Port knows where the MSI targets
> are, e.g., if the interrupt controller is actually part of the RC.
The controller terminating the MSI write is part of the DWC PCIe host
controller on i.MX6, which is questionable at least when you think
about how a MSI should be self-synchronizing to memory writes, but
that's reality...
As to why the controller needs the MSI Enable bit set, I have no idea.
But then the DWC controller is known to have some funky design
limitations regarding MSI, like not forwarding legacy PCI interrupts
anymore when MSI is enabled, so it's not totally surprising that we
need some quirky setup here.
Regards,
Lucas
[+cc Sven, Trent, et al from related report:
https://lore.kernel.org/linux-pci/[email protected]]
On Fri, Dec 14, 2018 at 06:44:15AM +0000, Richard Zhu wrote:
> Assertion of the MSI Enable bit of RC's MSI CAP is mandatory required to
> trigger MSI on i.MX6 PCIe.
> This bit would be asserted when CONFIG_PCIEPORTBUS=y.
> Thus, the MSI works fine on i.MX6 PCIe before the commit "f3fdfc4".
>
> Assert it unconditionally when MSI is enabled.
> Otherwise, the MSI wouldn't be triggered although the EP is present and
> the MSIs are assigned.
OK, I think I finally understand most of what's going on. Please
check the following possible changelog text:
The MSI Enable bit in the MSI Capability (PCIe r4.0, sec 7.7.1.2)
controls whether a Function can request service using MSI.
i.MX6 Root Ports implement the MSI Capability and may use MSI to
request service for events like PME, hotplug, AER, etc. In
addition, on i.MX6, the MSI Enable bit controls delivery of MSI
interrupts from components below the Root Port.
Prior to f3fdfc4ac3a2 ("PCI: Remove host driver Kconfig selection of
CONFIG_PCIEPORTBUS"), enabling CONFIG_PCI_IMX6 automatically also
enabled CONFIG_PCIEPORTBUS, and when portdrv claimed the Root Ports,
it set the MSI Enable bit so it could use PME, hotplug, AER, etc.
As a side effect, that also enabled delivery of MSI interrupts from
downstream components.
After f3fdfc4ac3a2, the imx6q-pcie driver can operate without
portdrv, but that means imx6q-pcie must set the MSI Enable bit
itself if downstream components use MSI.
Fixes: f3fdfc4ac3a2 ("PCI: Remove host driver Kconfig selection of CONFIG_PCIEPORTBUS")
I still don't understand exactly *how* MSI Enable affects MSI from
downstream components, since the downstream component just does a DMA
write, and the Root Port can't tell whether the write is to memory or
interrupt controller unless the Root Port knows where the MSI targets
are, e.g., if the interrupt controller is actually part of the RC.
Bjorn
> Signed-off-by: Richard Zhu <[email protected]>
> Reviewed-by: Lucas Stach <[email protected]>
> ---
> Changes v1 -> v2:
> * Assert the MSI_EN unconditionally when MSI is supported.
> Changes v2 -> v3:
> * Remove the IS_ENABLED(CONFIG_PCI_MSI) since the driver depends on
> PCI_MSI_IRQ_DOMAIN
> * Extended with a check for pci_msi_enabled() to see if the user
> explicitly want legacy IRQs
> Changes v3 -> v4:
> * Refer to Bjorn's comments, refine the subject and commit log and change
> the PCI_MSI_CAP to PCIE_RC_IMX6_MSI_CAP.
> ---
> drivers/pci/controller/dwc/pci-imx6.c | 10 ++++++++++
> 1 file changed, 10 insertions(+)
>
> diff --git a/drivers/pci/controller/dwc/pci-imx6.c b/drivers/pci/controller/dwc/pci-imx6.c
> index 26087b3..639bb27 100644
> --- a/drivers/pci/controller/dwc/pci-imx6.c
> +++ b/drivers/pci/controller/dwc/pci-imx6.c
> @@ -74,6 +74,7 @@ struct imx6_pcie {
> #define PHY_PLL_LOCK_WAIT_USLEEP_MAX 200
>
> /* PCIe Root Complex registers (memory-mapped) */
> +#define PCIE_RC_IMX6_MSI_CAP 0x50
> #define PCIE_RC_LCR 0x7c
> #define PCIE_RC_LCR_MAX_LINK_SPEEDS_GEN1 0x1
> #define PCIE_RC_LCR_MAX_LINK_SPEEDS_GEN2 0x2
> @@ -926,6 +927,7 @@ static int imx6_pcie_probe(struct platform_device *pdev)
> struct resource *dbi_base;
> struct device_node *node = dev->of_node;
> int ret;
> + u16 val;
>
> imx6_pcie = devm_kzalloc(dev, sizeof(*imx6_pcie), GFP_KERNEL);
> if (!imx6_pcie)
> @@ -1071,6 +1073,14 @@ static int imx6_pcie_probe(struct platform_device *pdev)
> if (ret < 0)
> return ret;
>
> + if (pci_msi_enabled()) {
> + val = dw_pcie_readw_dbi(pci, PCIE_RC_IMX6_MSI_CAP +
> + PCI_MSI_FLAGS);
> + val |= PCI_MSI_FLAGS_ENABLE;
> + dw_pcie_writew_dbi(pci, PCIE_RC_MSI_IMX6_CAP +
> + PCI_MSI_FLAGS, val);
> + }
> +
> return 0;
> }
>
> --
> 2.7.4
>
>
> _______________________________________________
> linux-arm-kernel mailing list
> [email protected]
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Hi Bjorn:
> -----Original Message-----
> From: Lucas Stach [mailto:[email protected]]
> Sent: 2018年12月19日 23:52
> To: Bjorn Helgaas <[email protected]>; Richard Zhu
> <[email protected]>
> Cc: Lorenzo Pieralisi <[email protected]>; Andrew Smirnov
> <[email protected]>; [email protected];
> [email protected]; [email protected]; Sven Van
> Asbroeck <[email protected]>; Niklas Cassel <[email protected]>;
> Kishon Vijay Abraham I <[email protected]>; Gustavo Pimentel
> <[email protected]>; Shawn Lin <[email protected]>;
> Trent Piepho <[email protected]>
> Subject: Re: [v4] PCI: imx: make msi work without CONFIG_PCIEPORTBUS=y
>
> Am Mittwoch, den 19.12.2018, 08:12 -0600 schrieb Bjorn Helgaas:
> > [+cc Sven, Trent, et al from related report:
> > https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flor
> >
> e.kernel.org%2Flinux-pci%2F20181218210444.5950-1-TheSven73%40google
> mai
> >
> l.com&data=02%7C01%7Chongxing.zhu%40nxp.com%7C05ba3b5bb42c
> 4db9bf3b
> >
> 08d665c9f17b%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C636
> 808315347
> >
> 780466&sdata=ta7ffj9EPGoqAjDiS1ORvR65A2IhMUpb2Endz9%2B6zM4
> %3D&
> > reserved=0]
> >
> > On Fri, Dec 14, 2018 at 06:44:15AM +0000, Richard Zhu wrote:
> > > Assertion of the MSI Enable bit of RC's MSI CAP is mandatory
> > > required to trigger MSI on i.MX6 PCIe.
> > > This bit would be asserted when CONFIG_PCIEPORTBUS=y.
> > > Thus, the MSI works fine on i.MX6 PCIe before the commit "f3fdfc4".
> > >
> > > Assert it unconditionally when MSI is enabled.
> > > Otherwise, the MSI wouldn't be triggered although the EP is present
> > > and the MSIs are assigned.
> >
> > OK, I think I finally understand most of what's going on. Please
> > check the following possible changelog text:
> >
> > The MSI Enable bit in the MSI Capability (PCIe r4.0, sec 7.7.1.2)
> > controls whether a Function can request service using MSI.
> >
> > i.MX6 Root Ports implement the MSI Capability and may use MSI to
> > request service for events like PME, hotplug, AER, etc. In
> > addition, on i.MX6, the MSI Enable bit controls delivery of MSI
> > interrupts from components below the Root Port.
> >
> > Prior to f3fdfc4ac3a2 ("PCI: Remove host driver Kconfig selection of
> > CONFIG_PCIEPORTBUS"), enabling CONFIG_PCI_IMX6 automatically
> also
> > enabled CONFIG_PCIEPORTBUS, and when portdrv claimed the Root
> Ports,
> > it set the MSI Enable bit so it could use PME, hotplug, AER, etc.
> > As a side effect, that also enabled delivery of MSI interrupts from
> > downstream components.
> >
> > After f3fdfc4ac3a2, the imx6q-pcie driver can operate without
> > portdrv, but that means imx6q-pcie must set the MSI Enable bit
> > itself if downstream components use MSI.
> >
> > Fixes: f3fdfc4ac3a2 ("PCI: Remove host driver Kconfig selection of
> > CONFIG_PCIEPORTBUS")
> >
> > I still don't understand exactly *how* MSI Enable affects MSI from
> > downstream components, since the downstream component just does a
> DMA
> > write, and the Root Port can't tell whether the write is to memory or
> > interrupt controller unless the Root Port knows where the MSI targets
> > are, e.g., if the interrupt controller is actually part of the RC.
>
[Richard Zhu] Thanks a lot for your kindly help to craft the commit log.
I'm totally fine with it.
Just like Lucas said below, the MSI trigger conditions are part of the glue logic
When integrate this IP into iMX6 SOC.
The MSI Enable bit set of RC is one of the mandatory required condition to
trigger MSI from EP.
Best Regards
Richard Zhu
> The controller terminating the MSI write is part of the DWC PCIe host
> controller on i.MX6, which is questionable at least when you think about how
> a MSI should be self-synchronizing to memory writes, but that's reality...
>
> As to why the controller needs the MSI Enable bit set, I have no idea.
> But then the DWC controller is known to have some funky design limitations
> regarding MSI, like not forwarding legacy PCI interrupts anymore when MSI is
> enabled, so it's not totally surprising that we need some quirky setup here.
>
> Regards,
> Lucas
Hi Bjorn:
> -----Original Message-----
> From: Lucas Stach [mailto:[email protected]]
> Sent: 2018年12月19日 23:52
> To: Bjorn Helgaas <[email protected]>; Richard Zhu
> <[email protected]>
> Cc: Lorenzo Pieralisi <[email protected]>; Andrew Smirnov
> <[email protected]>; [email protected];
> [email protected]; [email protected]; Sven Van
> Asbroeck <[email protected]>; Niklas Cassel <[email protected]>;
> Kishon Vijay Abraham I <[email protected]>; Gustavo Pimentel
> <[email protected]>; Shawn Lin <[email protected]>;
> Trent Piepho <[email protected]>
> Subject: Re: [v4] PCI: imx: make msi work without CONFIG_PCIEPORTBUS=y
>
> Am Mittwoch, den 19.12.2018, 08:12 -0600 schrieb Bjorn Helgaas:
> > [+cc Sven, Trent, et al from related report:
> > https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flor
> >
> e.kernel.org%2Flinux-pci%2F20181218210444.5950-1-TheSven73%40google
> mai
> >
> l.com&data=02%7C01%7Chongxing.zhu%40nxp.com%7C05ba3b5bb42c
> 4db9bf3b
> >
> 08d665c9f17b%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C636
> 808315347
> >
> 780466&sdata=ta7ffj9EPGoqAjDiS1ORvR65A2IhMUpb2Endz9%2B6zM4
> %3D&
> > reserved=0]
> >
> > On Fri, Dec 14, 2018 at 06:44:15AM +0000, Richard Zhu wrote:
> > > Assertion of the MSI Enable bit of RC's MSI CAP is mandatory
> > > required to trigger MSI on i.MX6 PCIe.
> > > This bit would be asserted when CONFIG_PCIEPORTBUS=y.
> > > Thus, the MSI works fine on i.MX6 PCIe before the commit "f3fdfc4".
> > >
> > > Assert it unconditionally when MSI is enabled.
> > > Otherwise, the MSI wouldn't be triggered although the EP is present
> > > and the MSIs are assigned.
> >
> > OK, I think I finally understand most of what's going on. Please
> > check the following possible changelog text:
> >
> > The MSI Enable bit in the MSI Capability (PCIe r4.0, sec 7.7.1.2)
> > controls whether a Function can request service using MSI.
> >
> > i.MX6 Root Ports implement the MSI Capability and may use MSI to
> > request service for events like PME, hotplug, AER, etc. In
> > addition, on i.MX6, the MSI Enable bit controls delivery of MSI
> > interrupts from components below the Root Port.
> >
> > Prior to f3fdfc4ac3a2 ("PCI: Remove host driver Kconfig selection of
> > CONFIG_PCIEPORTBUS"), enabling CONFIG_PCI_IMX6 automatically
> also
> > enabled CONFIG_PCIEPORTBUS, and when portdrv claimed the Root
> Ports,
> > it set the MSI Enable bit so it could use PME, hotplug, AER, etc.
> > As a side effect, that also enabled delivery of MSI interrupts from
> > downstream components.
> >
> > After f3fdfc4ac3a2, the imx6q-pcie driver can operate without
> > portdrv, but that means imx6q-pcie must set the MSI Enable bit
> > itself if downstream components use MSI.
> >
> > Fixes: f3fdfc4ac3a2 ("PCI: Remove host driver Kconfig selection of
> > CONFIG_PCIEPORTBUS")
> >
> > I still don't understand exactly *how* MSI Enable affects MSI from
> > downstream components, since the downstream component just does a
> DMA
> > write, and the Root Port can't tell whether the write is to memory or
> > interrupt controller unless the Root Port knows where the MSI targets
> > are, e.g., if the interrupt controller is actually part of the RC.
>
[Richard Zhu] Thanks a lot for your kindly help to craft the commit log.
I'm totally fine with it.
Just like Lucas said below, the MSI trigger conditions are part of the glue logic
When integrate this IP into iMX6 SOC.
The MSI Enable bit set of RC is one of the mandatory required condition to
trigger MSI from EP.
Best Regards
Richard Zhu
> The controller terminating the MSI write is part of the DWC PCIe host
> controller on i.MX6, which is questionable at least when you think about how
> a MSI should be self-synchronizing to memory writes, but that's reality...
>
> As to why the controller needs the MSI Enable bit set, I have no idea.
> But then the DWC controller is known to have some funky design limitations
> regarding MSI, like not forwarding legacy PCI interrupts anymore when MSI is
> enabled, so it's not totally surprising that we need some quirky setup here.
>
> Regards,
> Lucas