Commit a9e60cf0 authored by Grygorii Strashko's avatar Grygorii Strashko Committed by Jakub Kicinski

net: ethernet: ti: am65-cpsw: prepare xmit/rx path for multi-port devices in mac-only mode

This patch adds multi-port support to TI AM65x CPSW driver xmit/rx path in
preparation for adding support for multi-port devices, like Main CPSW0 on
K3 J721E SoC or future CPSW3g on K3 AM64x SoC.
Hence DMA channels are common/shared for all ext Ports and the RX/TX NAPI
and DMA processing going to be assigned to first available netdev this patch:
 - ensures all RX descriptors fields are initialized;
 - adds synchronization for TX DMA push/pop operation (locking) as
Networking core locks are not enough any more;
 - updates TX bql processing for every packet in
am65_cpsw_nuss_tx_compl_packets() as every completed TX skb can have
different ndev assigned (come from different netdevs).

To avoid performance issues for existing one-port CPSW2g devices the above
changes are done only for multi-port devices by splitting xmit path for
one-port and multi-port devices.
Signed-off-by: default avatarGrygorii Strashko <grygorii.strashko@ti.com>
Reviewed-by: default avatarJesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: default avatarJakub Kicinski <kuba@kernel.org>
parent 97067aaf
...@@ -375,7 +375,7 @@ static int am65_cpsw_nuss_rx_push(struct am65_cpsw_common *common, ...@@ -375,7 +375,7 @@ static int am65_cpsw_nuss_rx_push(struct am65_cpsw_common *common,
cppi5_hdesc_init(desc_rx, CPPI5_INFO0_HDESC_EPIB_PRESENT, cppi5_hdesc_init(desc_rx, CPPI5_INFO0_HDESC_EPIB_PRESENT,
AM65_CPSW_NAV_PS_DATA_SIZE); AM65_CPSW_NAV_PS_DATA_SIZE);
cppi5_hdesc_attach_buf(desc_rx, 0, 0, buf_dma, skb_tailroom(skb)); cppi5_hdesc_attach_buf(desc_rx, buf_dma, skb_tailroom(skb), buf_dma, skb_tailroom(skb));
swdata = cppi5_hdesc_get_swdata(desc_rx); swdata = cppi5_hdesc_get_swdata(desc_rx);
*((void **)swdata) = skb; *((void **)swdata) = skb;
...@@ -911,10 +911,57 @@ static void am65_cpsw_nuss_tx_cleanup(void *data, dma_addr_t desc_dma) ...@@ -911,10 +911,57 @@ static void am65_cpsw_nuss_tx_cleanup(void *data, dma_addr_t desc_dma)
dev_kfree_skb_any(skb); dev_kfree_skb_any(skb);
} }
static struct sk_buff *
am65_cpsw_nuss_tx_compl_packet(struct am65_cpsw_tx_chn *tx_chn,
dma_addr_t desc_dma)
{
struct am65_cpsw_ndev_priv *ndev_priv;
struct am65_cpsw_ndev_stats *stats;
struct cppi5_host_desc_t *desc_tx;
struct net_device *ndev;
struct sk_buff *skb;
void **swdata;
desc_tx = k3_cppi_desc_pool_dma2virt(tx_chn->desc_pool,
desc_dma);
swdata = cppi5_hdesc_get_swdata(desc_tx);
skb = *(swdata);
am65_cpsw_nuss_xmit_free(tx_chn, tx_chn->common->dev, desc_tx);
ndev = skb->dev;
am65_cpts_tx_timestamp(tx_chn->common->cpts, skb);
ndev_priv = netdev_priv(ndev);
stats = this_cpu_ptr(ndev_priv->stats);
u64_stats_update_begin(&stats->syncp);
stats->tx_packets++;
stats->tx_bytes += skb->len;
u64_stats_update_end(&stats->syncp);
return skb;
}
static void am65_cpsw_nuss_tx_wake(struct am65_cpsw_tx_chn *tx_chn, struct net_device *ndev,
struct netdev_queue *netif_txq)
{
if (netif_tx_queue_stopped(netif_txq)) {
/* Check whether the queue is stopped due to stalled
* tx dma, if the queue is stopped then wake the queue
* as we have free desc for tx
*/
__netif_tx_lock(netif_txq, smp_processor_id());
if (netif_running(ndev) &&
(k3_cppi_desc_pool_avail(tx_chn->desc_pool) >= MAX_SKB_FRAGS))
netif_tx_wake_queue(netif_txq);
__netif_tx_unlock(netif_txq);
}
}
static int am65_cpsw_nuss_tx_compl_packets(struct am65_cpsw_common *common, static int am65_cpsw_nuss_tx_compl_packets(struct am65_cpsw_common *common,
int chn, unsigned int budget) int chn, unsigned int budget)
{ {
struct cppi5_host_desc_t *desc_tx;
struct device *dev = common->dev; struct device *dev = common->dev;
struct am65_cpsw_tx_chn *tx_chn; struct am65_cpsw_tx_chn *tx_chn;
struct netdev_queue *netif_txq; struct netdev_queue *netif_txq;
...@@ -923,15 +970,13 @@ static int am65_cpsw_nuss_tx_compl_packets(struct am65_cpsw_common *common, ...@@ -923,15 +970,13 @@ static int am65_cpsw_nuss_tx_compl_packets(struct am65_cpsw_common *common,
struct sk_buff *skb; struct sk_buff *skb;
dma_addr_t desc_dma; dma_addr_t desc_dma;
int res, num_tx = 0; int res, num_tx = 0;
void **swdata;
tx_chn = &common->tx_chns[chn]; tx_chn = &common->tx_chns[chn];
while (true) { while (true) {
struct am65_cpsw_ndev_priv *ndev_priv; spin_lock(&tx_chn->lock);
struct am65_cpsw_ndev_stats *stats;
res = k3_udma_glue_pop_tx_chn(tx_chn->tx_chn, &desc_dma); res = k3_udma_glue_pop_tx_chn(tx_chn->tx_chn, &desc_dma);
spin_unlock(&tx_chn->lock);
if (res == -ENODATA) if (res == -ENODATA)
break; break;
...@@ -941,23 +986,52 @@ static int am65_cpsw_nuss_tx_compl_packets(struct am65_cpsw_common *common, ...@@ -941,23 +986,52 @@ static int am65_cpsw_nuss_tx_compl_packets(struct am65_cpsw_common *common,
break; break;
} }
desc_tx = k3_cppi_desc_pool_dma2virt(tx_chn->desc_pool, skb = am65_cpsw_nuss_tx_compl_packet(tx_chn, desc_dma);
desc_dma); total_bytes = skb->len;
swdata = cppi5_hdesc_get_swdata(desc_tx);
skb = *(swdata);
am65_cpsw_nuss_xmit_free(tx_chn, dev, desc_tx);
ndev = skb->dev; ndev = skb->dev;
napi_consume_skb(skb, budget);
num_tx++;
am65_cpts_tx_timestamp(common->cpts, skb); netif_txq = netdev_get_tx_queue(ndev, chn);
ndev_priv = netdev_priv(ndev); netdev_tx_completed_queue(netif_txq, num_tx, total_bytes);
stats = this_cpu_ptr(ndev_priv->stats);
u64_stats_update_begin(&stats->syncp); am65_cpsw_nuss_tx_wake(tx_chn, ndev, netif_txq);
stats->tx_packets++; }
stats->tx_bytes += skb->len;
u64_stats_update_end(&stats->syncp); dev_dbg(dev, "%s:%u pkt:%d\n", __func__, chn, num_tx);
return num_tx;
}
static int am65_cpsw_nuss_tx_compl_packets_2g(struct am65_cpsw_common *common,
int chn, unsigned int budget)
{
struct device *dev = common->dev;
struct am65_cpsw_tx_chn *tx_chn;
struct netdev_queue *netif_txq;
unsigned int total_bytes = 0;
struct net_device *ndev;
struct sk_buff *skb;
dma_addr_t desc_dma;
int res, num_tx = 0;
tx_chn = &common->tx_chns[chn];
while (true) {
res = k3_udma_glue_pop_tx_chn(tx_chn->tx_chn, &desc_dma);
if (res == -ENODATA)
break;
if (cppi5_desc_is_tdcm(desc_dma)) {
if (atomic_dec_and_test(&common->tdown_cnt))
complete(&common->tdown_complete);
break;
}
skb = am65_cpsw_nuss_tx_compl_packet(tx_chn, desc_dma);
ndev = skb->dev;
total_bytes += skb->len; total_bytes += skb->len;
napi_consume_skb(skb, budget); napi_consume_skb(skb, budget);
num_tx++; num_tx++;
...@@ -970,19 +1044,8 @@ static int am65_cpsw_nuss_tx_compl_packets(struct am65_cpsw_common *common, ...@@ -970,19 +1044,8 @@ static int am65_cpsw_nuss_tx_compl_packets(struct am65_cpsw_common *common,
netdev_tx_completed_queue(netif_txq, num_tx, total_bytes); netdev_tx_completed_queue(netif_txq, num_tx, total_bytes);
if (netif_tx_queue_stopped(netif_txq)) { am65_cpsw_nuss_tx_wake(tx_chn, ndev, netif_txq);
/* Check whether the queue is stopped due to stalled tx dma,
* if the queue is stopped then wake the queue as
* we have free desc for tx
*/
__netif_tx_lock(netif_txq, smp_processor_id());
if (netif_running(ndev) &&
(k3_cppi_desc_pool_avail(tx_chn->desc_pool) >=
MAX_SKB_FRAGS))
netif_tx_wake_queue(netif_txq);
__netif_tx_unlock(netif_txq);
}
dev_dbg(dev, "%s:%u pkt:%d\n", __func__, chn, num_tx); dev_dbg(dev, "%s:%u pkt:%d\n", __func__, chn, num_tx);
return num_tx; return num_tx;
...@@ -993,8 +1056,11 @@ static int am65_cpsw_nuss_tx_poll(struct napi_struct *napi_tx, int budget) ...@@ -993,8 +1056,11 @@ static int am65_cpsw_nuss_tx_poll(struct napi_struct *napi_tx, int budget)
struct am65_cpsw_tx_chn *tx_chn = am65_cpsw_napi_to_tx_chn(napi_tx); struct am65_cpsw_tx_chn *tx_chn = am65_cpsw_napi_to_tx_chn(napi_tx);
int num_tx; int num_tx;
num_tx = am65_cpsw_nuss_tx_compl_packets(tx_chn->common, tx_chn->id, if (AM65_CPSW_IS_CPSW2G(tx_chn->common))
budget); num_tx = am65_cpsw_nuss_tx_compl_packets_2g(tx_chn->common, tx_chn->id, budget);
else
num_tx = am65_cpsw_nuss_tx_compl_packets(tx_chn->common, tx_chn->id, budget);
num_tx = min(num_tx, budget); num_tx = min(num_tx, budget);
if (num_tx < budget) { if (num_tx < budget) {
napi_complete(napi_tx); napi_complete(napi_tx);
...@@ -1139,7 +1205,13 @@ static netdev_tx_t am65_cpsw_nuss_ndo_slave_xmit(struct sk_buff *skb, ...@@ -1139,7 +1205,13 @@ static netdev_tx_t am65_cpsw_nuss_ndo_slave_xmit(struct sk_buff *skb,
cppi5_hdesc_set_pktlen(first_desc, pkt_len); cppi5_hdesc_set_pktlen(first_desc, pkt_len);
desc_dma = k3_cppi_desc_pool_virt2dma(tx_chn->desc_pool, first_desc); desc_dma = k3_cppi_desc_pool_virt2dma(tx_chn->desc_pool, first_desc);
if (AM65_CPSW_IS_CPSW2G(common)) {
ret = k3_udma_glue_push_tx_chn(tx_chn->tx_chn, first_desc, desc_dma); ret = k3_udma_glue_push_tx_chn(tx_chn->tx_chn, first_desc, desc_dma);
} else {
spin_lock_bh(&tx_chn->lock);
ret = k3_udma_glue_push_tx_chn(tx_chn->tx_chn, first_desc, desc_dma);
spin_unlock_bh(&tx_chn->lock);
}
if (ret) { if (ret) {
dev_err(dev, "can't push desc %d\n", ret); dev_err(dev, "can't push desc %d\n", ret);
/* inform bql */ /* inform bql */
...@@ -1470,6 +1542,7 @@ static int am65_cpsw_nuss_init_tx_chns(struct am65_cpsw_common *common) ...@@ -1470,6 +1542,7 @@ static int am65_cpsw_nuss_init_tx_chns(struct am65_cpsw_common *common)
snprintf(tx_chn->tx_chn_name, snprintf(tx_chn->tx_chn_name,
sizeof(tx_chn->tx_chn_name), "tx%d", i); sizeof(tx_chn->tx_chn_name), "tx%d", i);
spin_lock_init(&tx_chn->lock);
tx_chn->common = common; tx_chn->common = common;
tx_chn->id = i; tx_chn->id = i;
tx_chn->descs_num = max_desc_num; tx_chn->descs_num = max_desc_num;
......
...@@ -60,6 +60,7 @@ struct am65_cpsw_tx_chn { ...@@ -60,6 +60,7 @@ struct am65_cpsw_tx_chn {
struct am65_cpsw_common *common; struct am65_cpsw_common *common;
struct k3_cppi_desc_pool *desc_pool; struct k3_cppi_desc_pool *desc_pool;
struct k3_udma_glue_tx_channel *tx_chn; struct k3_udma_glue_tx_channel *tx_chn;
spinlock_t lock; /* protect TX rings in multi-port mode */
int irq; int irq;
u32 id; u32 id;
u32 descs_num; u32 descs_num;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment