From: Manoj Panicker <manoj.panicker2@amd.com>
Add TPH support to the Broadcom BNXT device driver. This allows the
driver to utilize TPH functions for retrieving and configuring Steering
Tags when changing interrupt affinity. With compatible NIC firmware,
network traffic will be tagged correctly with Steering Tags, leading to
significant memory bandwidth savings and other benefits as demonstrated
by real network benchmarks on TPH-capable platforms.
Co-developed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
Co-developed-by: Wei Huang <wei.huang2@amd.com>
Signed-off-by: Wei Huang <wei.huang2@amd.com>
Signed-off-by: Manoj Panicker <manoj.panicker2@amd.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
---
drivers/net/ethernet/broadcom/bnxt/bnxt.c | 83 +++++++++++++++++++++++
drivers/net/ethernet/broadcom/bnxt/bnxt.h | 7 ++
net/core/netdev_rx_queue.c | 1 +
3 files changed, 91 insertions(+)
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
index 6e422e24750a..23ad2b6e70c7 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
@@ -55,6 +55,8 @@
#include <net/page_pool/helpers.h>
#include <linux/align.h>
#include <net/netdev_queues.h>
+#include <net/netdev_rx_queue.h>
+#include <linux/pci-tph.h>
#include "bnxt_hsi.h"
#include "bnxt.h"
@@ -10865,6 +10867,61 @@ int bnxt_reserve_rings(struct bnxt *bp, bool irq_re_init)
return 0;
}
+static void __bnxt_irq_affinity_notify(struct irq_affinity_notify *notify,
+ const cpumask_t *mask)
+{
+ struct bnxt_irq *irq;
+ u16 tag;
+ int err;
+
+ irq = container_of(notify, struct bnxt_irq, affinity_notify);
+ cpumask_copy(irq->cpu_mask, mask);
+
+ if (pcie_tph_get_cpu_st(irq->bp->pdev, TPH_MEM_TYPE_VM,
+ cpumask_first(irq->cpu_mask), &tag))
+ return;
+
+ if (pcie_tph_set_st_entry(irq->bp->pdev, irq->msix_nr, tag))
+ return;
+
+ if (netif_running(irq->bp->dev)) {
+ rtnl_lock();
+ err = netdev_rx_queue_restart(irq->bp->dev, irq->ring_nr);
+ if (err)
+ netdev_err(irq->bp->dev,
+ "rx queue restart failed: err=%d\n", err);
+ rtnl_unlock();
+ }
+}
+
+static void __bnxt_irq_affinity_release(struct kref __always_unused *ref)
+{
+}
+
+static void bnxt_release_irq_notifier(struct bnxt_irq *irq)
+{
+ irq_set_affinity_notifier(irq->vector, NULL);
+}
+
+static void bnxt_register_irq_notifier(struct bnxt *bp, struct bnxt_irq *irq)
+{
+ struct irq_affinity_notify *notify;
+
+ irq->bp = bp;
+
+ /* Nothing to do if TPH is not enabled */
+ if (!bp->tph_mode)
+ return;
+
+ /* Register IRQ affinity notifier */
+ notify = &irq->affinity_notify;
+ notify->irq = irq->vector;
+ notify->notify = __bnxt_irq_affinity_notify;
+ notify->release = __bnxt_irq_affinity_release;
+
+ irq_set_affinity_notifier(irq->vector, notify);
+}
+
static void bnxt_free_irq(struct bnxt *bp)
{
struct bnxt_irq *irq;
@@ -10887,11 +10944,18 @@ static void bnxt_free_irq(struct bnxt *bp)
free_cpumask_var(irq->cpu_mask);
irq->have_cpumask = 0;
}
+
+ bnxt_release_irq_notifier(irq);
+
free_irq(irq->vector, bp->bnapi[i]);
}
irq->requested = 0;
}
+
+ /* Disable TPH support */
+ pcie_disable_tph(bp->pdev);
+ bp->tph_mode = 0;
}
static int bnxt_request_irq(struct bnxt *bp)
@@ -10911,6 +10975,12 @@ static int bnxt_request_irq(struct bnxt *bp)
#ifdef CONFIG_RFS_ACCEL
rmap = bp->dev->rx_cpu_rmap;
#endif
+
+ /* Enable TPH support as part of IRQ request */
+ rc = pcie_enable_tph(bp->pdev, PCI_TPH_ST_IV_MODE);
+ if (!rc)
+ bp->tph_mode = PCI_TPH_ST_IV_MODE;
+
for (i = 0, j = 0; i < bp->cp_nr_rings; i++) {
int map_idx = bnxt_cp_num_to_irq_num(bp, i);
struct bnxt_irq *irq = &bp->irq_tbl[map_idx];
@@ -10934,8 +11004,11 @@ static int bnxt_request_irq(struct bnxt *bp)
if (zalloc_cpumask_var(&irq->cpu_mask, GFP_KERNEL)) {
int numa_node = dev_to_node(&bp->pdev->dev);
+ u16 tag;
irq->have_cpumask = 1;
+ irq->msix_nr = map_idx;
+ irq->ring_nr = i;
cpumask_set_cpu(cpumask_local_spread(i, numa_node),
irq->cpu_mask);
rc = irq_set_affinity_hint(irq->vector, irq->cpu_mask);
@@ -10945,6 +11018,16 @@ static int bnxt_request_irq(struct bnxt *bp)
irq->vector);
break;
}
+
+ bnxt_register_irq_notifier(bp, irq);
+
+ /* Init ST table entry */
+ if (pcie_tph_get_cpu_st(irq->bp->pdev, TPH_MEM_TYPE_VM,
+ cpumask_first(irq->cpu_mask),
+ &tag))
+ continue;
+
+ pcie_tph_set_st_entry(irq->bp->pdev, irq->msix_nr, tag);
}
}
return rc;
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.h b/drivers/net/ethernet/broadcom/bnxt/bnxt.h
index 69231e85140b..641d25646367 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.h
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.h
@@ -1227,6 +1227,11 @@ struct bnxt_irq {
u8 have_cpumask:1;
char name[IFNAMSIZ + BNXT_IRQ_NAME_EXTRA];
cpumask_var_t cpu_mask;
+
+ struct bnxt *bp;
+ int msix_nr;
+ int ring_nr;
+ struct irq_affinity_notify affinity_notify;
};
#define HWRM_RING_ALLOC_TX 0x1
@@ -2183,6 +2188,8 @@ struct bnxt {
struct net_device *dev;
struct pci_dev *pdev;
+ u8 tph_mode;
+
atomic_t intr_sem;
u32 flags;
diff --git a/net/core/netdev_rx_queue.c b/net/core/netdev_rx_queue.c
index e217a5838c87..10e95d7b6892 100644
--- a/net/core/netdev_rx_queue.c
+++ b/net/core/netdev_rx_queue.c
@@ -79,3 +79,4 @@ int netdev_rx_queue_restart(struct net_device *dev, unsigned int rxq_idx)
return err;
}
+EXPORT_SYMBOL_GPL(netdev_rx_queue_restart);
--
2.46.0On Wed, 2 Oct 2024 11:59:53 -0500 Wei Huang wrote:
> + if (netif_running(irq->bp->dev)) {
> + rtnl_lock();
> + err = netdev_rx_queue_restart(irq->bp->dev, irq->ring_nr);
> + if (err)
> + netdev_err(irq->bp->dev,
> + "rx queue restart failed: err=%d\n", err);
> + rtnl_unlock();
> + }
> +}
> +
> +static void __bnxt_irq_affinity_release(struct kref __always_unused *ref)
> +{
> +}
An empty release function is always a red flag.
How is the reference counting used here?
Is irq_set_affinity_notifier() not synchronous?
Otherwise the rtnl_lock() should probably cover the running check.
© 2016 - 2024 Red Hat, Inc.