iwlwifi: pcie: implement GRO without NAPI
Use the new NAPI infrastructure added to mac80211 to get
GRO. We don't really implement NAPI since we don't have
a real poll function and we never schedule a NAPI poll.
Instead of this, we collect all the packets we got from a
single interrupt and then call napi_gro_flush().
This allows us to benefit from GRO. In half duplex medium
like WiFi, its main advantage is that it reduces the number
of TCP Acks, hence improving the TCP Rx performance.
Since we call the Rx path with a spinlock held, remove
the might_sleep mention from the op_mode's API.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Reviewed-by: Ido Yariv <ido@wizery.com>
[Squash different patches and rewrite the commit message]
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
diff --git a/drivers/net/wireless/iwlwifi/pcie/rx.c b/drivers/net/wireless/iwlwifi/pcie/rx.c
index fdfa396..e8e5afc 100644
--- a/drivers/net/wireless/iwlwifi/pcie/rx.c
+++ b/drivers/net/wireless/iwlwifi/pcie/rx.c
@@ -673,7 +673,6 @@
/* Reuse the page if possible. For notification packets and
* SKBs that fail to Rx correctly, add them back into the
* rx_free list for reuse later. */
- spin_lock(&rxq->lock);
if (rxb->page != NULL) {
rxb->page_dma =
dma_map_page(trans->dev, rxb->page, 0,
@@ -694,7 +693,6 @@
}
} else
list_add_tail(&rxb->list, &rxq->rx_used);
- spin_unlock(&rxq->lock);
}
/*
@@ -709,6 +707,8 @@
u32 count = 8;
int total_empty;
+restart:
+ spin_lock(&rxq->lock);
/* uCode's read index (stored in shared DRAM) indicates the last Rx
* buffer that the driver may process (last buffer filled by ucode). */
r = le16_to_cpu(ACCESS_ONCE(rxq->rb_stts->closed_rb_num)) & 0x0FFF;
@@ -743,18 +743,25 @@
count++;
if (count >= 8) {
rxq->read = i;
+ spin_unlock(&rxq->lock);
iwl_pcie_rx_replenish_now(trans);
count = 0;
+ goto restart;
}
}
}
/* Backtrack one entry */
rxq->read = i;
+ spin_unlock(&rxq->lock);
+
if (fill_rx)
iwl_pcie_rx_replenish_now(trans);
else
iwl_pcie_rxq_restock(trans);
+
+ if (trans_pcie->napi.poll)
+ napi_gro_flush(&trans_pcie->napi, false);
}
/*
@@ -1068,8 +1075,6 @@
iwl_write8(trans, CSR_INT_PERIODIC_REG,
CSR_INT_PERIODIC_DIS);
- iwl_pcie_rx_handle(trans);
-
/*
* Enable periodic interrupt in 8 msec only if we received
* real RX interrupt (instead of just periodic int), to catch
@@ -1082,6 +1087,10 @@
CSR_INT_PERIODIC_ENA);
isr_stats->rx++;
+
+ local_bh_disable();
+ iwl_pcie_rx_handle(trans);
+ local_bh_enable();
}
/* This "Tx" DMA channel is used only for loading uCode */