[NET]: Make NAPI polling independent of struct net_device objects.
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
diff --git a/drivers/net/forcedeth.c b/drivers/net/forcedeth.c
index 1938d6d..24c1294 100644
--- a/drivers/net/forcedeth.c
+++ b/drivers/net/forcedeth.c
@@ -159,6 +159,8 @@
#define dprintk(x...) do { } while (0)
#endif
+#define TX_WORK_PER_LOOP 64
+#define RX_WORK_PER_LOOP 64
/*
* Hardware access:
@@ -745,6 +747,9 @@
struct fe_priv {
spinlock_t lock;
+ struct net_device *dev;
+ struct napi_struct napi;
+
/* General data:
* Locking: spin_lock(&np->lock); */
struct net_device_stats stats;
@@ -1586,9 +1591,10 @@
static void nv_do_rx_refill(unsigned long data)
{
struct net_device *dev = (struct net_device *) data;
+ struct fe_priv *np = netdev_priv(dev);
/* Just reschedule NAPI rx processing */
- netif_rx_schedule(dev);
+ netif_rx_schedule(dev, &np->napi);
}
#else
static void nv_do_rx_refill(unsigned long data)
@@ -2997,7 +3003,7 @@
#ifdef CONFIG_FORCEDETH_NAPI
if (events & NVREG_IRQ_RX_ALL) {
- netif_rx_schedule(dev);
+ netif_rx_schedule(dev, &np->napi);
/* Disable furthur receive irq's */
spin_lock(&np->lock);
@@ -3010,7 +3016,7 @@
spin_unlock(&np->lock);
}
#else
- if (nv_rx_process(dev, dev->weight)) {
+ if (nv_rx_process(dev, RX_WORK_PER_LOOP)) {
if (unlikely(nv_alloc_rx(dev))) {
spin_lock(&np->lock);
if (!np->in_shutdown)
@@ -3079,8 +3085,6 @@
return IRQ_RETVAL(i);
}
-#define TX_WORK_PER_LOOP 64
-#define RX_WORK_PER_LOOP 64
/**
* All _optimized functions are used to help increase performance
* (reduce CPU and increase throughput). They use descripter version 3,
@@ -3114,7 +3118,7 @@
#ifdef CONFIG_FORCEDETH_NAPI
if (events & NVREG_IRQ_RX_ALL) {
- netif_rx_schedule(dev);
+ netif_rx_schedule(dev, &np->napi);
/* Disable furthur receive irq's */
spin_lock(&np->lock);
@@ -3127,7 +3131,7 @@
spin_unlock(&np->lock);
}
#else
- if (nv_rx_process_optimized(dev, dev->weight)) {
+ if (nv_rx_process_optimized(dev, RX_WORK_PER_LOOP)) {
if (unlikely(nv_alloc_rx_optimized(dev))) {
spin_lock(&np->lock);
if (!np->in_shutdown)
@@ -3245,19 +3249,19 @@
}
#ifdef CONFIG_FORCEDETH_NAPI
-static int nv_napi_poll(struct net_device *dev, int *budget)
+static int nv_napi_poll(struct napi_struct *napi, int budget)
{
- int pkts, limit = min(*budget, dev->quota);
- struct fe_priv *np = netdev_priv(dev);
+ struct fe_priv *np = container_of(napi, struct fe_priv, napi);
+ struct net_device *dev = np->dev;
u8 __iomem *base = get_hwbase(dev);
unsigned long flags;
- int retcode;
+ int pkts, retcode;
if (np->desc_ver == DESC_VER_1 || np->desc_ver == DESC_VER_2) {
- pkts = nv_rx_process(dev, limit);
+ pkts = nv_rx_process(dev, budget);
retcode = nv_alloc_rx(dev);
} else {
- pkts = nv_rx_process_optimized(dev, limit);
+ pkts = nv_rx_process_optimized(dev, budget);
retcode = nv_alloc_rx_optimized(dev);
}
@@ -3268,13 +3272,12 @@
spin_unlock_irqrestore(&np->lock, flags);
}
- if (pkts < limit) {
- /* all done, no more packets present */
- netif_rx_complete(dev);
-
+ if (pkts < budget) {
/* re-enable receive interrupts */
spin_lock_irqsave(&np->lock, flags);
+ __netif_rx_complete(dev, napi);
+
np->irqmask |= NVREG_IRQ_RX_ALL;
if (np->msi_flags & NV_MSI_X_ENABLED)
writel(NVREG_IRQ_RX_ALL, base + NvRegIrqMask);
@@ -3282,13 +3285,8 @@
writel(np->irqmask, base + NvRegIrqMask);
spin_unlock_irqrestore(&np->lock, flags);
- return 0;
- } else {
- /* used up our quantum, so reschedule */
- dev->quota -= pkts;
- *budget -= pkts;
- return 1;
}
+ return pkts;
}
#endif
@@ -3296,6 +3294,7 @@
static irqreturn_t nv_nic_irq_rx(int foo, void *data)
{
struct net_device *dev = (struct net_device *) data;
+ struct fe_priv *np = netdev_priv(dev);
u8 __iomem *base = get_hwbase(dev);
u32 events;
@@ -3303,7 +3302,7 @@
writel(NVREG_IRQ_RX_ALL, base + NvRegMSIXIrqStatus);
if (events) {
- netif_rx_schedule(dev);
+ netif_rx_schedule(dev, &np->napi);
/* disable receive interrupts on the nic */
writel(NVREG_IRQ_RX_ALL, base + NvRegIrqMask);
pci_push(base);
@@ -3329,7 +3328,7 @@
if (!(events & np->irqmask))
break;
- if (nv_rx_process_optimized(dev, dev->weight)) {
+ if (nv_rx_process_optimized(dev, RX_WORK_PER_LOOP)) {
if (unlikely(nv_alloc_rx_optimized(dev))) {
spin_lock_irqsave(&np->lock, flags);
if (!np->in_shutdown)
@@ -4620,7 +4619,9 @@
if (test->flags & ETH_TEST_FL_OFFLINE) {
if (netif_running(dev)) {
netif_stop_queue(dev);
- netif_poll_disable(dev);
+#ifdef CONFIG_FORCEDETH_NAPI
+ napi_disable(&np->napi);
+#endif
netif_tx_lock_bh(dev);
spin_lock_irq(&np->lock);
nv_disable_hw_interrupts(dev, np->irqmask);
@@ -4679,7 +4680,9 @@
nv_start_rx(dev);
nv_start_tx(dev);
netif_start_queue(dev);
- netif_poll_enable(dev);
+#ifdef CONFIG_FORCEDETH_NAPI
+ napi_enable(&np->napi);
+#endif
nv_enable_hw_interrupts(dev, np->irqmask);
}
}
@@ -4911,7 +4914,9 @@
nv_start_rx(dev);
nv_start_tx(dev);
netif_start_queue(dev);
- netif_poll_enable(dev);
+#ifdef CONFIG_FORCEDETH_NAPI
+ napi_enable(&np->napi);
+#endif
if (ret) {
netif_carrier_on(dev);
@@ -4942,7 +4947,9 @@
spin_lock_irq(&np->lock);
np->in_shutdown = 1;
spin_unlock_irq(&np->lock);
- netif_poll_disable(dev);
+#ifdef CONFIG_FORCEDETH_NAPI
+ napi_disable(&np->napi);
+#endif
synchronize_irq(dev->irq);
del_timer_sync(&np->oom_kick);
@@ -4994,6 +5001,7 @@
goto out;
np = netdev_priv(dev);
+ np->dev = dev;
np->pci_dev = pci_dev;
spin_lock_init(&np->lock);
SET_MODULE_OWNER(dev);
@@ -5155,9 +5163,8 @@
#ifdef CONFIG_NET_POLL_CONTROLLER
dev->poll_controller = nv_poll_controller;
#endif
- dev->weight = RX_WORK_PER_LOOP;
#ifdef CONFIG_FORCEDETH_NAPI
- dev->poll = nv_napi_poll;
+ netif_napi_add(dev, &np->napi, nv_napi_poll, RX_WORK_PER_LOOP);
#endif
SET_ETHTOOL_OPS(dev, &ops);
dev->tx_timeout = nv_tx_timeout;