KumiSystems/openwrtv4
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity
openwrtv4/target/linux/mediatek/patches-4.9/0059-eth-fixes.patch
John Crispin 97a6ef513f mediatek: various additional ethernet fixes 
* fixes default affinity
* adds a napi watchdog - we were seeing stalled TX queues
* adds up/down locking

Signed-off-by: John Crispin <john@phrozen.org>
2017-08-30 17:05:10 +02:00

515 lines
15 KiB
Diff
Raw Blame History

Index: linux-4.9.44/drivers/net/ethernet/mediatek/mtk_eth_soc.c
===================================================================
--- linux-4.9.44.orig/drivers/net/ethernet/mediatek/mtk_eth_soc.c
+++ linux-4.9.44/drivers/net/ethernet/mediatek/mtk_eth_soc.c
@@ -24,6 +24,7 @@
#include <linux/tcp.h>
#if defined(CONFIG_NET_MEDIATEK_HW_QOS)
+
struct mtk_ioctl_reg {
unsigned int off;
unsigned int val;
@@ -32,6 +33,13 @@ struct mtk_ioctl_reg {
#define REG_HQOS_MAX 0x3FFF
#define RAETH_QDMA_REG_READ 0x89F8
#define RAETH_QDMA_REG_WRITE 0x89F9
+#define RAETH_QDMA_QUEUE_MAPPING 0x89FA
+
+unsigned int M2Q_table[16] = {0};
+unsigned int lan_wan_separate = 0;
+
+EXPORT_SYMBOL_GPL(M2Q_table);
+
#endif
#if defined(CONFIG_NET_MEDIATEK_HNAT) || defined(CONFIG_NET_MEDIATEK_HNAT_MODULE)
@@ -225,7 +233,7 @@ static void mtk_phy_link_adjust(struct n
if (flowctrl & FLOW_CTRL_RX)
mcr |= MAC_MCR_FORCE_RX_FC;
- netif_dbg(mac->hw, link, dev, "rx pause %s, tx pause %s\n",
+ netif_info(mac->hw, link, dev, "rx pause %s, tx pause %s\n",
flowctrl & FLOW_CTRL_RX ? "enabled" : "disabled",
flowctrl & FLOW_CTRL_TX ? "enabled" : "disabled");
}
@@ -508,9 +516,9 @@ static struct rtnl_link_stats64 * mtk_ge
unsigned int start;
if (netif_running(dev) && netif_device_present(dev)) {
- if (spin_trylock_bh(&hw_stats->stats_lock)) {
+ if (spin_trylock(&hw_stats->stats_lock)) {
mtk_stats_update_mac(mac);
- spin_unlock_bh(&hw_stats->stats_lock);
+ spin_unlock(&hw_stats->stats_lock);
}
}
@@ -690,6 +698,7 @@ static int mtk_tx_map(struct sk_buff *sk
txd3 |= skb->mark & 0x7;
if (mac->id)
txd3 += 8;
+ txd3 = 0;
#endif
mapped_addr = dma_map_single(eth->dev, skb->data,
@@ -760,16 +769,7 @@ static int mtk_tx_map(struct sk_buff *sk
WRITE_ONCE(itxd->txd3, (TX_DMA_SWC | TX_DMA_PLEN0(skb_headlen(skb)) |
(!nr_frags * TX_DMA_LS0)));
- /* we have a single DMA ring so BQL needs to be updated for all devices
- * sitting on this ring
- */
- for (i = 0; i < MTK_MAC_COUNT; i++) {
- if (!eth->netdev[i])
- continue;
-
- netdev_sent_queue(eth->netdev[i], skb->len);
- }
-
+ netdev_sent_queue(dev, skb->len);
skb_tx_timestamp(skb);
ring->next_free = mtk_qdma_phys_to_virt(ring, txd->txd2);
@@ -980,20 +980,9 @@ static int mtk_poll_rx(struct napi_struc
if (!(trxd.rxd2 & RX_DMA_DONE))
break;
- /* find out which mac the packet comes from. If the special tag is
- * we can assume that the traffic is coming from the builtin mt7530
- * and the DSA driver has loaded. FPORT will be the physical switch
- * port in this case rather than the FE forward port id. */
- if (!(trxd.rxd4 & RX_DMA_SP_TAG)) {
- /* values start at 1 */
- mac = (trxd.rxd4 >> RX_DMA_FPORT_SHIFT) &
- RX_DMA_FPORT_MASK;
- mac--;
- }
-
- if (unlikely(mac < 0 || mac >= MTK_MAC_COUNT ||
- !eth->netdev[mac]))
- goto release_desc;
+ /* find out which mac the packet come from. values start at 1 */
+ mac = (trxd.rxd4 >> 22) & 0x1;
+ mac = (mac + 1) % 2;
netdev = eth->netdev[mac];
@@ -1017,6 +1006,9 @@ static int mtk_poll_rx(struct napi_struc
}
/* receive data */
+ if (mac < 0 || mac > 2)
+ mac = 0;
+
skb = build_skb(data, ring->frag_size);
if (unlikely(!skb)) {
skb_free_frag(new_data);
@@ -1076,18 +1068,21 @@ static int mtk_poll_tx(struct mtk_eth *e
struct mtk_tx_dma *desc;
struct sk_buff *skb;
struct mtk_tx_buf *tx_buf;
- int total = 0, done = 0;
- unsigned int bytes = 0;
+ unsigned int done[MTK_MAX_DEVS];
+ unsigned int bytes[MTK_MAX_DEVS];
u32 cpu, dma;
static int condition;
- int i;
+ int total = 0, i;
+
+ memset(done, 0, sizeof(done));
+ memset(bytes, 0, sizeof(bytes));
cpu = mtk_r32(eth, MTK_QTX_CRX_PTR);
dma = mtk_r32(eth, MTK_QTX_DRX_PTR);
desc = mtk_qdma_phys_to_virt(ring, cpu);
- while ((cpu != dma) && done < budget) {
+ while ((cpu != dma) && budget) {
u32 next_cpu = desc->txd2;
int mac = 0;
@@ -1106,8 +1101,9 @@ static int mtk_poll_tx(struct mtk_eth *e
}
if (skb != (struct sk_buff *)MTK_DMA_DUMMY_DESC) {
- bytes += skb->len;
- done++;
+ bytes[mac] += skb->len;
+ done[mac]++;
+ budget--;
}
mtk_tx_unmap(eth, tx_buf);
@@ -1119,13 +1115,11 @@ static int mtk_poll_tx(struct mtk_eth *e
mtk_w32(eth, cpu, MTK_QTX_CRX_PTR);
- /* we have a single DMA ring so BQL needs to be updated for all devices
- * sitting on this ring
- */
for (i = 0; i < MTK_MAC_COUNT; i++) {
- if (!eth->netdev[i])
+ if (!eth->netdev[i] || !done[i])
continue;
- netdev_completed_queue(eth->netdev[i], done, bytes);
+ netdev_completed_queue(eth->netdev[i], done[i], bytes[i]);
+ total += done[i];
}
if (mtk_queue_stopped(eth) &&
@@ -1286,21 +1280,11 @@ static void mtk_tx_clean(struct mtk_eth
static int mtk_rx_alloc(struct mtk_eth *eth, int ring_no, int rx_flag)
{
- struct mtk_rx_ring *ring;
+ struct mtk_rx_ring *ring = &eth->rx_ring[ring_no];
int rx_data_len, rx_dma_size;
int i;
- u32 offset = 0;
-
- if (rx_flag & MTK_RX_FLAGS_QDMA) {
- if (ring_no)
- return -EINVAL;
- ring = &eth->rx_ring_qdma;
- offset = 0x1000;
- } else {
- ring = &eth->rx_ring[ring_no];
- }
- if (rx_flag & MTK_RX_FLAGS_HWLRO) {
+ if (rx_flag == MTK_RX_FLAGS_HWLRO) {
rx_data_len = MTK_MAX_LRO_RX_LENGTH;
rx_dma_size = MTK_HW_LRO_DMA_SIZE;
} else {
@@ -1348,16 +1332,104 @@ static int mtk_rx_alloc(struct mtk_eth *
*/
wmb();
- mtk_w32(eth, ring->phys, MTK_PRX_BASE_PTR_CFG(ring_no) + offset);
- mtk_w32(eth, rx_dma_size, MTK_PRX_MAX_CNT_CFG(ring_no) + offset);
- mtk_w32(eth, ring->calc_idx, ring->crx_idx_reg + offset);
- mtk_w32(eth, MTK_PST_DRX_IDX_CFG(ring_no), MTK_PDMA_RST_IDX + offset);
+ mtk_w32(eth, ring->phys, MTK_PRX_BASE_PTR_CFG(ring_no));
+ mtk_w32(eth, rx_dma_size, MTK_PRX_MAX_CNT_CFG(ring_no));
+ mtk_w32(eth, ring->calc_idx, ring->crx_idx_reg);
+ mtk_w32(eth, MTK_PST_DRX_IDX_CFG(ring_no), MTK_PDMA_RST_IDX);
return 0;
}
-static void mtk_rx_clean(struct mtk_eth *eth, struct mtk_rx_ring *ring)
+static int mtk_rx_alloc_qdma(struct mtk_eth *eth, int rx_flag)
{
+ struct mtk_rx_ring *ring = &eth->rx_ring_qdma;
+ int rx_data_len, rx_dma_size;
+ int i;
+
+ rx_data_len = ETH_DATA_LEN;
+ rx_dma_size = MTK_DMA_SIZE;
+
+ ring->frag_size = mtk_max_frag_size(rx_data_len);
+ ring->buf_size = mtk_max_buf_size(ring->frag_size);
+ ring->data = kcalloc(rx_dma_size, sizeof(*ring->data),
+ GFP_KERNEL);
+ if (!ring->data)
+ return -ENOMEM;
+
+ for (i = 0; i < rx_dma_size; i++) {
+ ring->data[i] = netdev_alloc_frag(ring->frag_size);
+ if (!ring->data[i])
+ return -ENOMEM;
+ }
+
+ ring->dma = dma_alloc_coherent(eth->dev,
+ rx_dma_size * sizeof(*ring->dma),
+ &ring->phys,
+ GFP_ATOMIC | __GFP_ZERO);
+ if (!ring->dma)
+ return -ENOMEM;
+
+ for (i = 0; i < rx_dma_size; i++) {
+ dma_addr_t dma_addr = dma_map_single(eth->dev,
+ ring->data[i] + NET_SKB_PAD,
+ ring->buf_size,
+ DMA_FROM_DEVICE);
+ if (unlikely(dma_mapping_error(eth->dev, dma_addr)))
+ return -ENOMEM;
+ ring->dma[i].rxd1 = (unsigned int)dma_addr;
+
+ ring->dma[i].rxd2 = RX_DMA_PLEN0(ring->buf_size);
+ }
+ ring->dma_size = rx_dma_size;
+ ring->calc_idx_update = false;
+ ring->calc_idx = rx_dma_size - 1;
+ ring->crx_idx_reg = MTK_QRX_CRX_IDX_CFG(0);
+ /* make sure that all changes to the dma ring are flushed before we
+ * continue
+ */
+ wmb();
+
+ mtk_w32(eth, ring->phys, MTK_QRX_BASE_PTR_CFG(0));
+ mtk_w32(eth, rx_dma_size, MTK_QRX_MAX_CNT_CFG(0));
+ mtk_w32(eth, ring->calc_idx, ring->crx_idx_reg);
+ mtk_w32(eth, MTK_PST_DRX_IDX_CFG(0), MTK_QDMA_RST_IDX);
+
+ return 0;
+}
+
+static void mtk_rx_clean(struct mtk_eth *eth, int ring_no)
+{
+ struct mtk_rx_ring *ring = &eth->rx_ring[ring_no];
+ int i;
+
+ if (ring->data && ring->dma) {
+ for (i = 0; i < ring->dma_size; i++) {
+ if (!ring->data[i])
+ continue;
+ if (!ring->dma[i].rxd1)
+ continue;
+ dma_unmap_single(eth->dev,
+ ring->dma[i].rxd1,
+ ring->buf_size,
+ DMA_FROM_DEVICE);
+ skb_free_frag(ring->data[i]);
+ }
+ kfree(ring->data);
+ ring->data = NULL;
+ }
+
+ if (ring->dma) {
+ dma_free_coherent(eth->dev,
+ ring->dma_size * sizeof(*ring->dma),
+ ring->dma,
+ ring->phys);
+ ring->dma = NULL;
+ }
+}
+
+static void mtk_rx_clean_qdma(struct mtk_eth *eth)
+{
+ struct mtk_rx_ring *ring = &eth->rx_ring_qdma;
int i;
if (ring->data && ring->dma) {
@@ -1683,7 +1755,7 @@ static int mtk_dma_init(struct mtk_eth *
if (err)
return err;
- err = mtk_rx_alloc(eth, 0, MTK_RX_FLAGS_QDMA);
+ err = mtk_rx_alloc_qdma(eth, MTK_RX_FLAGS_NORMAL);
if (err)
return err;
@@ -1702,6 +1774,7 @@ static int mtk_dma_init(struct mtk_eth *
return err;
}
+
/* Enable random early drop and set drop threshold automatically */
mtk_w32(eth, FC_THRES_DROP_MODE | FC_THRES_DROP_EN | FC_THRES_MIN,
MTK_QDMA_FC_THRES);
@@ -1726,13 +1799,13 @@ static void mtk_dma_free(struct mtk_eth
eth->phy_scratch_ring = 0;
}
mtk_tx_clean(eth);
- mtk_rx_clean(eth, &eth->rx_ring[0]);
- mtk_rx_clean(eth, &eth->rx_ring_qdma);
+ mtk_rx_clean(eth, 0);
+ mtk_rx_clean_qdma(eth);
if (eth->hwlro) {
mtk_hwlro_rx_uninit(eth);
for (i = 1; i < MTK_MAX_RX_RING_NUM; i++)
- mtk_rx_clean(eth, &eth->rx_ring[i]);
+ mtk_rx_clean(eth, i);
}
kfree(eth->scratch_head);
@@ -1947,20 +2020,14 @@ static int mtk_hw_init(struct mtk_eth *e
val = mtk_r32(eth, MTK_CDMQ_IG_CTRL);
mtk_w32(eth, val | MTK_CDMQ_STAG_EN, MTK_CDMQ_IG_CTRL);
- /* Indicates CDM to parse the MTK special tag from CPU
- * which also is working out for untag packets.
- */
- val = mtk_r32(eth, MTK_CDMQ_IG_CTRL);
- mtk_w32(eth, val | MTK_CDMQ_STAG_EN, MTK_CDMQ_IG_CTRL);
- val = mtk_r32(eth, MTK_CDMP_IG_CTRL);
- mtk_w32(eth, val | MTK_CDMP_STAG_EN, MTK_CDMP_IG_CTRL);
-
/* Enable RX VLan Offloading */
if (MTK_HW_FEATURES & NETIF_F_HW_VLAN_CTAG_RX)
mtk_w32(eth, 1, MTK_CDMP_EG_CTRL);
else
mtk_w32(eth, 0, MTK_CDMP_EG_CTRL);
+ mtk_w32(eth, 0x81000001, MTK_CDMP_IG_CTRL);
+
/* disable delay and normal interrupt */
#ifdef MTK_IRQ_DLY
mtk_w32(eth, 0x84048404, MTK_PDMA_DELAY_INT);
@@ -1990,6 +2057,9 @@ static int mtk_hw_init(struct mtk_eth *e
/* Enable RX checksum */
val |= MTK_GDMA_ICS_EN | MTK_GDMA_TCS_EN | MTK_GDMA_UCS_EN;
+ if (!i)
+ val |= BIT(24);
+
/* setup the mac dma */
mtk_w32(eth, val, MTK_GDMA_FWD_CFG(i));
}
@@ -2069,7 +2139,18 @@ static int mtk_do_ioctl(struct net_devic
if (reg.off > REG_HQOS_MAX)
return -EINVAL;
mtk_w32(eth, reg.val, 0x1800 + reg.off);
-// printk("write reg off:%x val:%x\n", reg.off, reg.val);
+ printk("write reg off:%x val:%x\n", reg.off, reg.val);
+ return 0;
+
+ case RAETH_QDMA_QUEUE_MAPPING:
+ copy_from_user(&reg, ifr->ifr_data, sizeof(reg));
+ if ((reg.off & 0x100) == 0x100) {
+ lan_wan_separate = 1;
+ reg.off &= 0xff;
+ } else {
+ lan_wan_separate = 0;
+ }
+ M2Q_table[reg.off] = reg.val;
return 0;
#endif
case SIOCGMIIPHY:
@@ -2288,9 +2369,9 @@ static void mtk_get_ethtool_stats(struct
return;
if (netif_running(dev) && netif_device_present(dev)) {
- if (spin_trylock_bh(&hwstats->stats_lock)) {
+ if (spin_trylock(&hwstats->stats_lock)) {
mtk_stats_update_mac(mac);
- spin_unlock_bh(&hwstats->stats_lock);
+ spin_unlock(&hwstats->stats_lock);
}
}
@@ -2443,7 +2524,7 @@ static int mtk_add_mac(struct mtk_eth *e
mac->hw_stats->reg_offset = id * MTK_STAT_OFFSET;
SET_NETDEV_DEV(eth->netdev[id], eth->dev);
- eth->netdev[id]->watchdog_timeo = 30 * HZ;
+ eth->netdev[id]->watchdog_timeo = 15 * HZ;
eth->netdev[id]->netdev_ops = &mtk_netdev_ops;
eth->netdev[id]->base_addr = (unsigned long)eth->base;
Index: linux-4.9.44/drivers/net/ethernet/mediatek/mtk_eth_soc.h
===================================================================
--- linux-4.9.44.orig/drivers/net/ethernet/mediatek/mtk_eth_soc.h
+++ linux-4.9.44/drivers/net/ethernet/mediatek/mtk_eth_soc.h
@@ -80,7 +80,6 @@
/* CDMP Ingress Control Register */
#define MTK_CDMP_IG_CTRL 0x400
-#define MTK_CDMP_STAG_EN BIT(0)
/* CDMP Exgress Control Register */
#define MTK_CDMP_EG_CTRL 0x404
@@ -91,12 +90,27 @@
#define MTK_GDMA_TCS_EN BIT(21)
#define MTK_GDMA_UCS_EN BIT(20)
+/* GDMA Ingress Control Register */
+#define MTK_GDMA1_IG_CTRL(x) (0x500 + (x * 0x1000))
+
/* Unicast Filter MAC Address Register - Low */
#define MTK_GDMA_MAC_ADRL(x) (0x508 + (x * 0x1000))
/* Unicast Filter MAC Address Register - High */
#define MTK_GDMA_MAC_ADRH(x) (0x50C + (x * 0x1000))
+/* QDMA RX Base Pointer Register */
+#define MTK_QRX_BASE_PTR0 0x1900
+#define MTK_QRX_BASE_PTR_CFG(x) (MTK_QRX_BASE_PTR0 + (x * 0x10))
+
+/* QDMA RX Maximum Count Register */
+#define MTK_QRX_MAX_CNT0 0x1904
+#define MTK_QRX_MAX_CNT_CFG(x) (MTK_QRX_MAX_CNT0 + (x * 0x10))
+
+/* QDMA RX CPU Pointer Register */
+#define MTK_QRX_CRX_IDX0 0x1908
+#define MTK_QRX_CRX_IDX_CFG(x) (MTK_QRX_CRX_IDX0 + (x * 0x10))
+
/* PDMA RX Base Pointer Register */
#define MTK_PRX_BASE_PTR0 0x900
#define MTK_PRX_BASE_PTR_CFG(x) (MTK_PRX_BASE_PTR0 + (x * 0x10))
@@ -240,7 +254,10 @@
#define MTK_QDMA_INT_MASK 0x1A1C
/* QDMA Interrupt Mask Register */
+#define MTK_QDMA_HRED1 0x1A40
#define MTK_QDMA_HRED2 0x1A44
+#define MTK_QDMA_SRED1 0x1A48
+#define MTK_QDMA_SRED2 0x1A4c
/* QDMA TX Forward CPU Pointer Register */
#define MTK_QTX_CTX_PTR 0x1B00
@@ -275,6 +292,7 @@
#define TX_DMA_TSO BIT(28)
#define TX_DMA_FPORT_SHIFT 25
#define TX_DMA_FPORT_MASK 0x7
+#define TX_DMA_VQID0 BIT(17)
#define TX_DMA_INS_VLAN BIT(16)
/* QDMA descriptor txd3 */
@@ -294,7 +312,6 @@
/* QDMA descriptor rxd4 */
#define RX_DMA_L4_VALID BIT(24)
-#define RX_DMA_SP_TAG BIT(22)
#define RX_DMA_FPORT_SHIFT 19
#define RX_DMA_FPORT_MASK 0x7
@@ -310,6 +327,7 @@
/* Mac control registers */
#define MTK_MAC_MCR(x) (0x10100 + (x * 0x100))
+#define MTK_MAC_MSR(x) (0x10108 + (x * 0x100))
#define MAC_MCR_MAX_RX_1536 BIT(24)
#define MAC_MCR_IPG_CFG (BIT(18) | BIT(16))
#define MAC_MCR_FORCE_MODE BIT(15)
@@ -495,7 +513,6 @@ struct mtk_tx_ring {
enum mtk_rx_flags {
MTK_RX_FLAGS_NORMAL = 0,
MTK_RX_FLAGS_HWLRO,
- MTK_RX_FLAGS_QDMA,
};
/* struct mtk_rx_ring - This struct holds info describing a RX ring
@@ -539,9 +556,9 @@ struct mtk_rx_ring {
* @pctl: The register map pointing at the range used to setup
* GMAC port drive/slew values
* @dma_refcnt: track how many netdevs are using the DMA engine
- * @tx_ring: Pointer to the memory holding info about the TX ring
- * @rx_ring: Pointer to the memory holding info about the RX ring
- * @rx_ring_qdma: Pointer to the memory holding info about the QDMA RX ring
+ * @tx_ring: Pointer to the memore holding info about the TX ring
+ * @rx_ring: Pointer to the memore holding info about the RX ring
+ * @rx_ring_qdma: Pointer to the memore holding info about the RX ring (QDMA)
* @tx_napi: The TX NAPI struct
* @rx_napi: The RX NAPI struct
* @scratch_ring: Newer SoCs need memory for a second HW managed TX ring
@@ -563,6 +580,7 @@ struct mtk_eth {
struct net_device *netdev[MTK_MAX_DEVS];
struct mtk_mac *mac[MTK_MAX_DEVS];
int irq[3];
+ cpumask_t affinity_mask[3];
u32 msg_enable;
unsigned long sysclk;
struct regmap *ethsys;
@@ -615,4 +633,6 @@ void mtk_stats_update_mac(struct mtk_mac
void mtk_w32(struct mtk_eth *eth, u32 val, unsigned reg);
u32 mtk_r32(struct mtk_eth *eth, unsigned reg);
+extern unsigned int M2Q_table[16];
+
#endif /* MTK_ETH_H */
Reference in a new issue View git blame Copy permalink