linux-stable/net/core/gro_cells.c
Eric Dumazet 2543a6000e gro_cells: reduce number of synchronize_net() calls
After cited commit, gro_cells_destroy() became damn slow
on hosts with a lot of cores.

This is because we have one additional synchronize_net() per cpu as
stated in the changelog.

gro_cells_init() is setting NAPI_STATE_NO_BUSY_POLL, and this was enough
to not have one synchronize_net() call per netif_napi_del()

We can factorize all the synchronize_net() to a single one,
right before freeing per-cpu memory.

Fixes: 5198d545db ("net: remove napi_hash_del() from driver-facing API")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20201124203822.1360107-1-eric.dumazet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2020-11-25 11:28:12 -08:00

113 lines
2.3 KiB
C

// SPDX-License-Identifier: GPL-2.0
#include <linux/skbuff.h>
#include <linux/slab.h>
#include <linux/netdevice.h>
#include <net/gro_cells.h>
struct gro_cell {
struct sk_buff_head napi_skbs;
struct napi_struct napi;
};
int gro_cells_receive(struct gro_cells *gcells, struct sk_buff *skb)
{
struct net_device *dev = skb->dev;
struct gro_cell *cell;
int res;
rcu_read_lock();
if (unlikely(!(dev->flags & IFF_UP)))
goto drop;
if (!gcells->cells || skb_cloned(skb) || netif_elide_gro(dev)) {
res = netif_rx(skb);
goto unlock;
}
cell = this_cpu_ptr(gcells->cells);
if (skb_queue_len(&cell->napi_skbs) > netdev_max_backlog) {
drop:
atomic_long_inc(&dev->rx_dropped);
kfree_skb(skb);
res = NET_RX_DROP;
goto unlock;
}
__skb_queue_tail(&cell->napi_skbs, skb);
if (skb_queue_len(&cell->napi_skbs) == 1)
napi_schedule(&cell->napi);
res = NET_RX_SUCCESS;
unlock:
rcu_read_unlock();
return res;
}
EXPORT_SYMBOL(gro_cells_receive);
/* called under BH context */
static int gro_cell_poll(struct napi_struct *napi, int budget)
{
struct gro_cell *cell = container_of(napi, struct gro_cell, napi);
struct sk_buff *skb;
int work_done = 0;
while (work_done < budget) {
skb = __skb_dequeue(&cell->napi_skbs);
if (!skb)
break;
napi_gro_receive(napi, skb);
work_done++;
}
if (work_done < budget)
napi_complete_done(napi, work_done);
return work_done;
}
int gro_cells_init(struct gro_cells *gcells, struct net_device *dev)
{
int i;
gcells->cells = alloc_percpu(struct gro_cell);
if (!gcells->cells)
return -ENOMEM;
for_each_possible_cpu(i) {
struct gro_cell *cell = per_cpu_ptr(gcells->cells, i);
__skb_queue_head_init(&cell->napi_skbs);
set_bit(NAPI_STATE_NO_BUSY_POLL, &cell->napi.state);
netif_napi_add(dev, &cell->napi, gro_cell_poll,
NAPI_POLL_WEIGHT);
napi_enable(&cell->napi);
}
return 0;
}
EXPORT_SYMBOL(gro_cells_init);
void gro_cells_destroy(struct gro_cells *gcells)
{
int i;
if (!gcells->cells)
return;
for_each_possible_cpu(i) {
struct gro_cell *cell = per_cpu_ptr(gcells->cells, i);
napi_disable(&cell->napi);
__netif_napi_del(&cell->napi);
__skb_queue_purge(&cell->napi_skbs);
}
/* This barrier is needed because netpoll could access dev->napi_list
* under rcu protection.
*/
synchronize_net();
free_percpu(gcells->cells);
gcells->cells = NULL;
}
EXPORT_SYMBOL(gro_cells_destroy);