RE: [PATCH 02/10] PCI: designware-ep: Add the doorbell mode of MSI-X in EP mode

[Xiaowei Bao <xiaowei.bao@nxp.com>]

> -----Original Message-----
> From: Kishon Vijay Abraham I <kishon@ti.com>
> Sent: 2019年8月16日 18:50
> To: Xiaowei Bao <xiaowei.bao@nxp.com>; Andrew Murray
> <andrew.murray@arm.com>
> Cc: jingoohan1@gmail.com; gustavo.pimentel@synopsys.com;
> bhelgaas@google.com; robh+dt@kernel.org; mark.rutland@arm.com;
> shawnguo@kernel.org; Leo Li <leoyang.li@nxp.com>;
> lorenzo.pieralisi@arm.com; arnd@arndb.de; gregkh@linuxfoundation.org;
> M.h. Lian <minghuan.lian@nxp.com>; Mingkai Hu <mingkai.hu@nxp.com>;
> linux-pci@vger.kernel.org; devicetree@vger.kernel.org;
> linux-kernel@vger.kernel.org; linux-arm-kernel@lists.infradead.org;
> linuxppc-dev@lists.ozlabs.org; Z.q. Hou <zhiqiang.hou@nxp.com>
> Subject: Re: [PATCH 02/10] PCI: designware-ep: Add the doorbell mode of
> MSI-X in EP mode
> 
> Hi,
> 
> On 16/08/19 8:28 AM, Xiaowei Bao wrote:
> >
> >
> >> -----Original Message-----
> >> From: Andrew Murray <andrew.murray@arm.com>
> >> Sent: 2019年8月15日 19:54
> >> To: Xiaowei Bao <xiaowei.bao@nxp.com>
> >> Cc: jingoohan1@gmail.com; gustavo.pimentel@synopsys.com;
> >> bhelgaas@google.com; robh+dt@kernel.org; mark.rutland@arm.com;
> >> shawnguo@kernel.org; Leo Li <leoyang.li@nxp.com>; kishon@ti.com;
> >> lorenzo.pieralisi@arm.com; arnd@arndb.de; gregkh@linuxfoundation.org;
> >> M.h. Lian <minghuan.lian@nxp.com>; Mingkai Hu
> <mingkai.hu@nxp.com>;
> >> Roy Zang <roy.zang@nxp.com>; linux-pci@vger.kernel.org;
> >> devicetree@vger.kernel.org; linux-kernel@vger.kernel.org;
> >> linux-arm-kernel@lists.infradead.org; linuxppc-dev@lists.ozlabs.org
> >> Subject: Re: [PATCH 02/10] PCI: designware-ep: Add the doorbell mode
> >> of MSI-X in EP mode
> >>
> >> On Thu, Aug 15, 2019 at 04:37:08PM +0800, Xiaowei Bao wrote:
> >>> Add the doorbell mode of MSI-X in EP mode.
> >>>
> >>> Signed-off-by: Xiaowei Bao <xiaowei.bao@nxp.com>
> >>> ---
> >>>  drivers/pci/controller/dwc/pcie-designware-ep.c | 14
> ++++++++++++++
> >>>  drivers/pci/controller/dwc/pcie-designware.h    | 14
> ++++++++++++++
> >>>  2 files changed, 28 insertions(+)
> >>>
> >>> diff --git a/drivers/pci/controller/dwc/pcie-designware-ep.c
> >>> b/drivers/pci/controller/dwc/pcie-designware-ep.c
> >>> index 75e2955..e3a7cdf 100644
> >>> --- a/drivers/pci/controller/dwc/pcie-designware-ep.c
> >>> +++ b/drivers/pci/controller/dwc/pcie-designware-ep.c
> >>> @@ -454,6 +454,20 @@ int dw_pcie_ep_raise_msi_irq(struct
> dw_pcie_ep
> >> *ep, u8 func_no,
> >>>  	return 0;
> >>>  }
> >>>
> >>> +int dw_pcie_ep_raise_msix_irq_doorbell(struct dw_pcie_ep *ep, u8
> >> func_no,
> >>> +				       u16 interrupt_num)
> >>> +{
> >>> +	struct dw_pcie *pci = to_dw_pcie_from_ep(ep);
> >>> +	u32 msg_data;
> >>> +
> >>> +	msg_data = (func_no << PCIE_MSIX_DOORBELL_PF_SHIFT) |
> >>> +		   (interrupt_num - 1);
> >>> +
> >>> +	dw_pcie_writel_dbi(pci, PCIE_MSIX_DOORBELL, msg_data);
> >>> +
> >>> +	return 0;
> >>> +}
> >>> +
> >>>  int dw_pcie_ep_raise_msix_irq(struct dw_pcie_ep *ep, u8 func_no,
> >>>  			      u16 interrupt_num)
> >>
> >> Have I understood correctly that the hardware provides an alternative
> >> mechanism that allows for raising MSI-X interrupts without the bother
> >> of reading the capabilities registers?
> > Yes, the hardware provide two way to MSI-X, please check the page 492
> > of
> > DWC_pcie_dm_registers_4.30 Menu.
> > MSIX_DOORBELL_OFF on page 492 0x948 Description: MSI-X Doorbell
> > Register....>
> >>
> >> If so is there any good reason to keep dw_pcie_ep_raise_msix_irq?
> >> (And thus use it in dw_plat_pcie_ep_raise_irq also)?
> > I am not sure, but I think the dw_pcie_ep_raise_msix_irq function is
> > not correct, because I think we can't get the MSIX table from the
> > address ep->phys_base + tbl_addr, but I also don't know where I can get the
> correct MSIX table.
> 
> Sometime back when I tried raising MSI-X from EP, it was failing. It's quite
> possible dw_pcie_ep_raise_msix_irq function is not correct.
> 
> MSI-X table can be obtained from the inbound ATU corresponding to the MSIX
> bar.
> IMO MSI-X support in EP mode needs rework. For instance set_msix should
> also take BAR number as input to be configured in the MSI-X capability. The
> function driver (pci-epf-test.c) should allocate memory taking into account the
> MSI-X table.
Hi Kishon,

Thanks a lot for your explain, yes, we can get the MSI-X table from the inbound ATU of
the MSIX BAR.
> 
> Thanks
> Kishon