Thread (43 messages) 43 messages, 4 authors, 2021-08-10

Re: [PATCH V7 6/9] PCI: Enable 10-Bit Tag support for PCIe RP devices

From: Bjorn Helgaas <helgaas@kernel.org>
Date: 2021-08-09 17:26:54
Also in: linux-pci, netdev

On Thu, Aug 05, 2021 at 04:25:23PM +0800, Dongdong Liu wrote:
On 2021/8/5 7:38, Bjorn Helgaas wrote:
quoted
On Wed, Aug 04, 2021 at 09:47:05PM +0800, Dongdong Liu wrote:
quoted
PCIe spec 5.0r1.0 section 2.2.6.2 implementation note, In configurations
where a Requester with 10-Bit Tag Requester capability needs to target
multiple Completers, one needs to ensure that the Requester sends 10-Bit
Tag Requests only to Completers that have 10-Bit Tag Completer capability.
So we enable 10-Bit Tag Requester for root port only when the devices
under the root port support 10-Bit Tag Completer.
Fix quoting.  I can't tell what is from the spec and what you wrote.
Will fix.
quoted
quoted
Signed-off-by: Dongdong Liu <redacted>
---
 drivers/pci/pcie/portdrv_pci.c | 69 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 69 insertions(+)
diff --git a/drivers/pci/pcie/portdrv_pci.c b/drivers/pci/pcie/portdrv_pci.c
index c7ff1ee..2382cd2 100644
--- a/drivers/pci/pcie/portdrv_pci.c
+++ b/drivers/pci/pcie/portdrv_pci.c
@@ -90,6 +90,72 @@ static const struct dev_pm_ops pcie_portdrv_pm_ops = {
 #define PCIE_PORTDRV_PM_OPS	NULL
 #endif /* !PM */

+static int pci_10bit_tag_comp_support(struct pci_dev *dev, void *data)
+{
+	bool *support = (bool *)data;
+
+	if (!pci_is_pcie(dev)) {
+		*support = false;
+		return 1;
+	}
+
+	/*
+	 * PCIe spec 5.0r1.0 section 2.2.6.2 implementation note.
+	 * For configurations where a Requester with 10-Bit Tag Requester
+	 * capability targets Completers where some do and some do not have
+	 * 10-Bit Tag Completer capability, how the Requester determines which
+	 * NPRs include 10-Bit Tags is outside the scope of this specification.
+	 * So we do not consider hotplug scenario.
+	 */
+	if (dev->is_hotplug_bridge) {
+		*support = false;
+		return 1;
+	}
+
+	if (!(dev->pcie_devcap2 & PCI_EXP_DEVCAP2_10BIT_TAG_COMP)) {
+		*support = false;
+		return 1;
+	}
+
+	return 0;
+}
+
+static void pci_configure_rp_10bit_tag(struct pci_dev *dev)
+{
+	bool support = true;
+
+	if (dev->subordinate == NULL)
+		return;
+
+	/* If no devices under the root port, no need to enable 10-Bit Tag. */
+	if (list_empty(&dev->subordinate->devices))
+		return;
+
+	pci_10bit_tag_comp_support(dev, &support);
+	if (!support)
+		return;
+
+	/*
+	 * PCIe spec 5.0r1.0 section 2.2.6.2 implementation note.
+	 * In configurations where a Requester with 10-Bit Tag Requester
+	 * capability needs to target multiple Completers, one needs to ensure
+	 * that the Requester sends 10-Bit Tag Requests only to Completers
+	 * that have 10-Bit Tag Completer capability. So we enable 10-Bit Tag
+	 * Requester for root port only when the devices under the root port
+	 * support 10-Bit Tag Completer.
+	 */
+	pci_walk_bus(dev->subordinate, pci_10bit_tag_comp_support, &support);
+	if (!support)
+		return;
+
+	if (!(dev->pcie_devcap2 & PCI_EXP_DEVCAP2_10BIT_TAG_REQ))
+		return;
+
+	pci_dbg(dev, "enabling 10-Bit Tag Requester\n");
+	pcie_capability_set_word(dev, PCI_EXP_DEVCTL2,
+				 PCI_EXP_DEVCTL2_10BIT_TAG_REQ_EN);
+}
+
 /*
  * pcie_portdrv_probe - Probe PCI-Express port devices
  * @dev: PCI-Express port device being probed
@@ -111,6 +177,9 @@ static int pcie_portdrv_probe(struct pci_dev *dev,
 	     (type != PCI_EXP_TYPE_RC_EC)))
 		return -ENODEV;

+	if (type == PCI_EXP_TYPE_ROOT_PORT)
+		pci_configure_rp_10bit_tag(dev);
I don't think this has anything to do with the portdrv, so all this
should go somewhere else.
Yes, any suggestion where to put the code?
It seems similar to pci_configure_ltr(), pci_configure_eetlp_prefix(),
and other things in drivers/pci/probe.c, so maybe there?

Or, if this is more of a theoretical advantage than a demonstrated
performance improvement, we could just hold off on doing it until it
becomes important.  I can't tell if you have a scenario that actually
benefits from this yet.
quoted
Out of curiosity, IIUC this enables 10-bit tags for MMIO transactions
from the root port toward the device, i.e., traffic that originates
from a CPU.  Is that a significant benefit?  I would expect high-speed
devices would primarily operate via DMA with relatively little MMIO
traffic.
The benefits of 10-Bit Tag for EP are obvious.
There are few RP scenarios. Unless there are two:
1. RC has its own DMA.
2. The P2P tag is replaced at the RP when the P2PDMA go through RP.
  
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help