Before Version Patches
======================
patch v5
https://patchwork.kernel.org/patch/10493797/
patch v4
https://patchwork.kernel.org/patch/10402399/
patch v3
https://www.spinics.net/lists/linux-pci/msg72322.html
patch v2
https://www.spinics.net/lists/kernel/msg2797610.html
patch v1
https://www.spinics.net/lists/kernel/msg2796410.html
Changes between V6 and V5
=========================
1. Fix the bug pointed by Lorenzo.
(1) The bug of checking the return value of kirin_pcie_add_msi is fixed, which was pointed by Lorenzo.
2. Test the patch based on Hikey960.
(1) the patch is tested on Hikey960 board with COLORFUL CN600 M.2 SSD connected.
(2) As i can not cat /proc/interrupts for "adb shell" is not working, so some log was added to test this patch as following.
--- a/drivers/nvme/host/pci.c
+++ b/drivers/nvme/host/pci.c
@@ -1966,8 +1966,13 @@ static int nvme_setup_io_queues(struct nvme_dev *dev)
pci_free_irq_vectors(pdev);
result = pci_alloc_irq_vectors_affinity(pdev, 1, nr_io_queues + 1,
PCI_IRQ_ALL_TYPES | PCI_IRQ_AFFINITY, &affd);
- if (result <= 0)
+ if (result <= 0) {
+ printk(KERN_ERR "NVMe: alloc irq Fail.\n");
return -EIO;
+ }
+ else
+ printk(KERN_ERR "PCIe Device NVMe enable MSI IRQ Success.\n");
+
--- a/drivers/pci/dwc/pcie-kirin.c
+++ b/drivers/pci/dwc/pcie-kirin.c
@@ -462,6 +462,9 @@ static int kirin_pcie_add_msi(struct dw_pcie *pci,
}
pci->pp.msi_irq = ret;
+
+ dev_err(&pdev->dev,
+ "Kirin PCIe MSI IRQ No.(%d)\n", ret);
}
return ret;
(3) The running log is shown as below:
...
[ 0.567993] kirin-pcie f4000000.pcie: Kirin PCIe MSI IRQ No.(55)
[ 0.568012] PCI: OF: host bridge /soc/pcie@f4000000 ranges:
[ 0.568026] PCI: OF: MEM 0xf6000000..0xf7ffffff -> 0x00000000
[ 0.587310] kirin-pcie f4000000.pcie: PCI host bridge to bus 0000:00
[ 0.587319] pci_bus 0000:00: root bus resource [bus 00-ff]
[ 0.587326] pci_bus 0000:00: root bus resource [mem 0xf6000000-0xf7ffffff] (bus address [0x00000000-0x01ffffff])
[ 0.587362] pci 0000:00:00.0: [19e5:3660] type 01 class 0x060400
[ 0.587409] pci 0000:00:00.0: reg 0x10: [mem 0xf6000000-0xf6ffffff 64bit]
[ 0.587514] pci 0000:00:00.0: supports D1 D2
[ 0.587519] pci 0000:00:00.0: PME# supported from D0 D1 D2 D3hot
[ 0.589065] pci 0000:01:00.0: [10ec:5760] type 00 class 0x010802
[ 0.589643] pci 0000:01:00.0: reg 0x10: [mem 0xf6000000-0xf6003fff 64bit]
[ 0.590142] pci 0000:01:00.0: reg 0x24: [mem 0xf6000000-0xf6001fff]
[ 0.604141] pci 0000:00:00.0: BAR 0: assigned [mem 0xf6000000-0xf6ffffff 64bit]
[ 0.604158] pci 0000:00:00.0: BAR 14: assigned [mem 0xf7000000-0xf70fffff]
[ 0.604166] pci 0000:01:00.0: BAR 0: assigned [mem 0xf7000000-0xf7003fff 64bit]
[ 0.604344] pci 0000:01:00.0: BAR 5: assigned [mem 0xf7004000-0xf7005fff]
[ 0.604400] pci 0000:00:00.0: PCI bridge to [bus 01-ff]
[ 0.604409] pci 0000:00:00.0: bridge window [mem 0xf7000000-0xf70fffff]
[ 0.604750] pcieport 0000:00:00.0: Signaling PME with IRQ 62
[ 0.604853] pcieport 0000:00:00.0: AER enabled with IRQ 62
...
[ 0.623179] nvme nvme0: pci function 0000:01:00.0
[ 0.623252] nvme 0000:01:00.0: enabling device (0000 -> 0002)
...
[ 0.974765] PCIe Device NVMe enable MSI IRQ Success.
...
Changes between V5 and V4
=========================
1. rebase the patch based on Linux next-version branch.
2. fix issues according to review comments from Andy Shevchenko and Lorenzo.
(1) Del MSI irq handler for DW handle it, while Kirin only read the number from DT,
(2) Use the function dw_pcie_msi_init directly if CONFIG_PCI_MSI is enabled.
Changes between V4 and V3
=========================
1. remove DT binding patch, for Wei Xu had applied it.
2. fix issues according to review comments from Andy Shevchenko.
(1) Take the msi number get and request as a seperate function,
(2) Use the result of platform_get_irq as return val.
Changes between V3 and V2
=========================
1. fix issues according to review comments
(1)from Bjorn Helgaas: Check for 'pci->pp.msi_irq < 0'.
(2)from Bjorn Helgaas: Update the message of 'msi irq' in dev_err().
Changes between V2 and V1
=========================
1. seperate DT binding patch.
2. fix issues according to review comments
(1)from Bjorn Helgaas: Update the style of subject and changelog.
(2)from Bjorn Helgaas: Add msi irq number in the message.
(3)from Bjorn Helgaas: Delete unnecessary code 'pci->pp.root_bus_nr = -1'.
(4)from Dmitry Shmidt: Fix typing error. Replace 'interrupts-names' of 'interrupt-names'.
Xiaowei Song (1):
PCI: kirin: Add MSI support
drivers/pci/dwc/pcie-kirin.c | 28 ++++++++++++++++++++++++++++
1 file changed, 28 insertions(+)
--
2.11.GIT
Add support for MSI
Signed-off-by: Xiaowei Song <[email protected]>
Signed-off-by: Yao Chen <[email protected]>
---
drivers/pci/dwc/pcie-kirin.c | 28 ++++++++++++++++++++++++++++
1 file changed, 28 insertions(+)
diff --git a/drivers/pci/dwc/pcie-kirin.c b/drivers/pci/dwc/pcie-kirin.c
index d2970a009eb5..369bf87d2fff 100644
--- a/drivers/pci/dwc/pcie-kirin.c
+++ b/drivers/pci/dwc/pcie-kirin.c
@@ -430,6 +430,9 @@ static int kirin_pcie_host_init(struct pcie_port *pp)
{
kirin_pcie_establish_link(pp);
+ if (IS_ENABLED(CONFIG_PCI_MSI))
+ dw_pcie_msi_init(pp);
+
return 0;
}
@@ -445,9 +448,34 @@ static const struct dw_pcie_host_ops kirin_pcie_host_ops = {
.host_init = kirin_pcie_host_init,
};
+static int kirin_pcie_add_msi(struct dw_pcie *pci,
+ struct platform_device *pdev)
+{
+ int ret = 0;
+
+ if (IS_ENABLED(CONFIG_PCI_MSI)) {
+ ret = platform_get_irq(pdev, 0);
+ if (ret < 0) {
+ dev_err(&pdev->dev,
+ "failed to get MSI IRQ (%d)\n", ret);
+ return ret;
+ }
+
+ pci->pp.msi_irq = ret;
+ }
+
+ return ret;
+}
+
static int __init kirin_add_pcie_port(struct dw_pcie *pci,
struct platform_device *pdev)
{
+ int ret;
+
+ ret = kirin_pcie_add_msi(pci, pdev);
+ if (ret < 0)
+ return ret;
+
pci->pp.ops = &kirin_pcie_host_ops;
return dw_pcie_host_init(&pci->pp);
--
2.11.GIT
On Wed, Jul 11, 2018 at 9:57 AM, Xiaowei Song <[email protected]> wrote:
> Add support for MSI
> +static int kirin_pcie_add_msi(struct dw_pcie *pci,
> + struct platform_device *pdev)
> +{
> + int ret = 0;
> +
> + if (IS_ENABLED(CONFIG_PCI_MSI)) {
> + ret = platform_get_irq(pdev, 0);
> + if (ret < 0) {
> + dev_err(&pdev->dev,
> + "failed to get MSI IRQ (%d)\n", ret);
> + return ret;
> + }
> +
> + pci->pp.msi_irq = ret;
> + }
> +
> + return ret;
> +}
Like someone already noticed, is it really IRQ number you would like to return?
If you rewrite like
int irq;
if (IS_ENABLED(...)) {
irq = ...
if (irq < 0)
return irq;
... = irq;
}
return 0;
It would be slightly more explicit what you do and what you return.
--
With Best Regards,
Andy Shevchenko
> -----邮件原件-----
> 发件人: Andy Shevchenko [mailto:[email protected]]
> 发送时间: 2018年7月11日 15:23
> 收件人: Songxiaowei (Kirin_DRV) <[email protected]>
> 抄送: Wangbinghui <[email protected]>; Bjorn Helgaas
> <[email protected]>; Lorenzo Pieralisi <[email protected]>;
> Rob Herring <[email protected]>; [email protected]; Linux Kernel
> Mailing List <[email protected]>; Suzhuangluan
> <[email protected]>; Kongfei <[email protected]>; chenyao
> (F) <[email protected]>
> 主题: Re: [PATCH v6 1/1] PCI: kirin: Add MSI support
>
> On Wed, Jul 11, 2018 at 9:57 AM, Xiaowei Song <[email protected]>
> wrote:
> > Add support for MSI
>
> > +static int kirin_pcie_add_msi(struct dw_pcie *pci,
> > + struct platform_device *pdev) {
> > + int ret = 0;
> > +
> > + if (IS_ENABLED(CONFIG_PCI_MSI)) {
> > + ret = platform_get_irq(pdev, 0);
> > + if (ret < 0) {
> > + dev_err(&pdev->dev,
> > + "failed to get MSI IRQ (%d)\n", ret);
> > + return ret;
> > + }
> > +
> > + pci->pp.msi_irq = ret;
> > + }
> > +
> > + return ret;
> > +}
>
> Like someone already noticed, is it really IRQ number you would like to
> return?
>
> If you rewrite like
>
> int irq;
>
> if (IS_ENABLED(...)) {
> irq = ...
> if (irq < 0)
> return irq;
>
> ... = irq;
>
> }
>
> return 0;
>
> It would be slightly more explicit what you do and what you return.
>
Hi Andy
I think did not get Lorenzo's meaning before, thank you so much, and I'll modify the patch as soon as possible.
Best Regards,
Xiaowei Song.
> --
> With Best Regards,
> Andy Shevchenko