RE: [PATCH v19 04/19] PCI: designware-ep: Add INTx IRQs support

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

 



Hello Serge,

> From: Serge Semin, Sent: Wednesday, August 23, 2023 7:32 PM
> 
> On Wed, Aug 23, 2023 at 06:11:38PM +0900, Yoshihiro Shimoda wrote:
> > Add support for triggering INTx IRQs by using outbound iATU.
> > Outbound iATU is utilized to send assert and de-assert INTA TLPs
> > as simulated edge IRQ for INTA. (Other INT[BCD] are not asserted.)
> > This INTx support is optional (if there is no memory for INTx,
> > probe will not fail).
> >
> > The message is generated based on the payloadless Msg TLP with type
> > 0x14, where 0x4 is the routing code implying the Terminate at
> > Receiver message. The message code is specified as b1000xx for
> > the INTx assertion and b1001xx for the INTx de-assertion.
> >
> > Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@xxxxxxxxxxx>
> > Reviewed-by: Serge Semin <fancer.lancer@xxxxxxxxx>
> > ---
> >  .../pci/controller/dwc/pcie-designware-ep.c   | 70 +++++++++++++++++--
> >  drivers/pci/controller/dwc/pcie-designware.h  |  2 +
> >  2 files changed, 68 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/pci/controller/dwc/pcie-designware-ep.c b/drivers/pci/controller/dwc/pcie-designware-ep.c
> > index 747d5bc07222..4a8c116cdd4b 100644
> > --- a/drivers/pci/controller/dwc/pcie-designware-ep.c
> > +++ b/drivers/pci/controller/dwc/pcie-designware-ep.c
> > @@ -6,9 +6,11 @@
> >   * Author: Kishon Vijay Abraham I <kishon@xxxxxx>
> >   */
> >
> > +#include <linux/delay.h>
> >  #include <linux/of.h>
> >  #include <linux/platform_device.h>
> >
> > +#include "../../pci.h"
> >  #include "pcie-designware.h"
> >  #include <linux/pci-epc.h>
> >  #include <linux/pci-epf.h>
> > @@ -484,14 +486,61 @@ static const struct pci_epc_ops epc_ops = {
> >  	.get_features		= dw_pcie_ep_get_features,
> >  };
> >
> > +static int dw_pcie_ep_send_msg(struct dw_pcie_ep *ep, u8 func_no, u8 code,
> > +			       u8 routing)
> > +{
> > +	struct dw_pcie_ob_atu_cfg atu = { 0 };
> > +	struct pci_epc *epc = ep->epc;
> > +	int ret;
> > +
> > +	atu.func_no = func_no;
> > +	atu.code = code;
> > +	atu.routing = routing;
> > +	atu.type = PCIE_ATU_TYPE_MSG;
> > +	atu.cpu_addr = ep->intx_mem_phys;
> > +	atu.size = epc->mem->window.page_size;
> > +
> > +	ret = dw_pcie_ep_outbound_atu(ep, &atu);
> > +	if (ret)
> > +		return ret;
> > +
> 
> > +	/* A MWr with an effecitive length of '0' is converted to Msg */
> 
> 1. writel() means 4-byte IO-write. What is the "effecitive" length you are
> talking about?
> 2. You don't generate MWr TLP here. It's Msg TLP. See the atu.type you
> specified.
> 3. s/effecitive/effective
> 
> > +	writel(0, ep->intx_mem);
> 
> IMO what is done here is a dummy-write triggering the INTx Msg being
> generated. I bet you could have done it within any address from
> ep->intx_mem to (ep->intx_mem + epc->mem->window.page_size) and by
> writing any value, not only 0 because the Msg TLP doesn't have the
> memory address and data.

You're correct. The following cases could work correctly too.

writel(0xffffffff, ep->intx_mem);
writeb(0xff, ep->intx_mem);
writeb(0xff, ep->intx_mem + 0x100);

So, my adding comment was completely wrong. I should have checked this before.
And then, I'll change the comment like below. But, what do you think?

-----
+	/* A dummy-write ep->intx_mem is converted to a Msg TLP */
+	writel(0, ep->intx_mem);
-----

Best regards,
Yoshihiro Shimoda

> -Serge(y)
> 
> > +
> > +	dw_pcie_ep_unmap_addr(epc, func_no, 0, ep->intx_mem_phys);
> > +
> > +	return 0;
> > +}
> > +
> >  int dw_pcie_ep_raise_legacy_irq(struct dw_pcie_ep *ep, u8 func_no)
> >  {
> >  	struct dw_pcie *pci = to_dw_pcie_from_ep(ep);
> >  	struct device *dev = pci->dev;
> > +	int ret;
> >
> > -	dev_err(dev, "EP cannot trigger legacy IRQs\n");
> > +	if (!ep->intx_mem) {
> > +		dev_err(dev, "legacy IRQs not supported\n");
> > +		return -EOPNOTSUPP;
> > +	}
> >
> > -	return -EINVAL;
> > +	/*
> > +	 * Even though the PCI bus specification implies the level-triggered
> > +	 * INTx interrupts the kernel PCIe endpoint framework has a single
> > +	 * PCI_EPC_IRQ_INTx flag defined for the legacy IRQs simulation. Thus
> > +	 * this function sends the Deassert_INTx PCIe TLP after the Assert_INTx
> > +	 * message with the 50 usec duration basically implementing the
> > +	 * rising-edge triggering IRQ. Hopefully the interrupt controller will
> > +	 * still be able to register the incoming IRQ event...
> > +	 */
> > +	ret = dw_pcie_ep_send_msg(ep, func_no, PCI_MSG_CODE_ASSERT_INTA,
> > +				  PCI_MSG_TYPE_R_ROUTING_LOCAL);
> > +	if (ret)
> > +		return ret;
> > +
> > +	usleep_range(50, 100);
> > +
> > +	return dw_pcie_ep_send_msg(ep, func_no, PCI_MSG_CODE_DEASSERT_INTA,
> > +				   PCI_MSG_TYPE_R_ROUTING_LOCAL);
> >  }
> >  EXPORT_SYMBOL_GPL(dw_pcie_ep_raise_legacy_irq);
> >
> > @@ -622,6 +671,10 @@ void dw_pcie_ep_exit(struct dw_pcie_ep *ep)
> >
> >  	dw_pcie_edma_remove(pci);
> >
> > +	if (ep->intx_mem)
> > +		pci_epc_mem_free_addr(epc, ep->intx_mem_phys, ep->intx_mem,
> > +				      epc->mem->window.page_size);
> > +
> >  	pci_epc_mem_free_addr(epc, ep->msi_mem_phys, ep->msi_mem,
> >  			      epc->mem->window.page_size);
> >
> > @@ -793,9 +846,14 @@ int dw_pcie_ep_init(struct dw_pcie_ep *ep)
> >  		goto err_exit_epc_mem;
> >  	}
> >
> > +	ep->intx_mem = pci_epc_mem_alloc_addr(epc, &ep->intx_mem_phys,
> > +					      epc->mem->window.page_size);
> > +	if (!ep->intx_mem)
> > +		dev_warn(dev, "Failed to reserve memory for INTx\n");
> > +
> >  	ret = dw_pcie_edma_detect(pci);
> >  	if (ret)
> > -		goto err_free_epc_mem;
> > +		goto err_free_epc_mem_intx;
> >
> >  	if (ep->ops->get_features) {
> >  		epc_features = ep->ops->get_features(ep);
> > @@ -812,7 +870,11 @@ int dw_pcie_ep_init(struct dw_pcie_ep *ep)
> >  err_remove_edma:
> >  	dw_pcie_edma_remove(pci);
> >
> > -err_free_epc_mem:
> > +err_free_epc_mem_intx:
> > +	if (ep->intx_mem)
> > +		pci_epc_mem_free_addr(epc, ep->intx_mem_phys, ep->intx_mem,
> > +				      epc->mem->window.page_size);
> > +
> >  	pci_epc_mem_free_addr(epc, ep->msi_mem_phys, ep->msi_mem,
> >  			      epc->mem->window.page_size);
> >
> > diff --git a/drivers/pci/controller/dwc/pcie-designware.h b/drivers/pci/controller/dwc/pcie-designware.h
> > index d31c018a3803..c17e5255fab6 100644
> > --- a/drivers/pci/controller/dwc/pcie-designware.h
> > +++ b/drivers/pci/controller/dwc/pcie-designware.h
> > @@ -365,6 +365,8 @@ struct dw_pcie_ep {
> >  	unsigned long		*ob_window_map;
> >  	void __iomem		*msi_mem;
> >  	phys_addr_t		msi_mem_phys;
> > +	void __iomem		*intx_mem;
> > +	phys_addr_t		intx_mem_phys;
> >  	struct pci_epf_bar	*epf_bar[PCI_STD_NUM_BARS];
> >  };
> >
> > --
> > 2.25.1
> >




[Index of Archives]     [Linux Samsung SOC]     [Linux Wireless]     [Linux Kernel]     [ATH6KL]     [Linux Bluetooth]     [Linux Netdev]     [Kernel Newbies]     [IDE]     [Security]     [Git]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux ATA RAID]     [Samba]     [Device Mapper]

  Powered by Linux