* [PATCH net-next v1 03/12] amd-xgbe: Add check to be sure amd-xgbe-phy driver is used
From: Tom Lendacky @ 2015-01-16 18:46 UTC (permalink / raw)
To: netdev; +Cc: David Miller
In-Reply-To: <20150116184618.18809.319.stgit@tlendack-t1.amdoffice.net>
The amd-xgbe driver relies on the amd-xgbe-phy phylib driver. Add a
check to be sure that if any errors occur during probing of the
amd-xgbe-phy driver then the amd-xgbe driver returns an error.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
---
drivers/net/ethernet/amd/xgbe/xgbe-mdio.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-mdio.c b/drivers/net/ethernet/amd/xgbe/xgbe-mdio.c
index 7eeb03d..41e29e2 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-mdio.c
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-mdio.c
@@ -260,10 +260,14 @@ int xgbe_mdio_register(struct xgbe_prv_data *pdata)
of_node_put(phy_node);
goto err_phy_device;
}
+ if (!phydev->dev.driver) {
+ dev_err(pdata->dev, "phy driver probe failed\n");
+ ret = -EIO;
+ goto err_phy_device;
+ }
/* Add a reference to the PHY driver so it can't be unloaded */
- pdata->phy_module = phydev->dev.driver ?
- phydev->dev.driver->owner : NULL;
+ pdata->phy_module = phydev->dev.driver->owner;
if (!try_module_get(pdata->phy_module)) {
dev_err(pdata->dev, "try_module_get failed\n");
ret = -EIO;
^ permalink raw reply related
* [PATCH net-next v1 12/12] amd-xgbe-phy: Allow certain PHY settings to be set by UEFI
From: Tom Lendacky @ 2015-01-16 18:47 UTC (permalink / raw)
To: netdev; +Cc: David Miller
In-Reply-To: <20150116184618.18809.319.stgit@tlendack-t1.amdoffice.net>
Certain PHY settings need to be configurable by UEFI depending on the
platform being used. Add new device tree / ACPI properties that, if
present, will override the pre-determined values currently used.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
---
.../devicetree/bindings/net/amd-xgbe-phy.txt | 16 ++
drivers/net/phy/amd-xgbe-phy.c | 159 +++++++++++++++++---
2 files changed, 152 insertions(+), 23 deletions(-)
diff --git a/Documentation/devicetree/bindings/net/amd-xgbe-phy.txt b/Documentation/devicetree/bindings/net/amd-xgbe-phy.txt
index d54fe1a..33df393 100644
--- a/Documentation/devicetree/bindings/net/amd-xgbe-phy.txt
+++ b/Documentation/devicetree/bindings/net/amd-xgbe-phy.txt
@@ -16,6 +16,18 @@ Optional properties:
0 - 1GbE and 10GbE (default)
1 - 2.5GbE and 10GbE
+The following optional properties are represented by an array with each
+value corresponding to a particular speed. The first array value represents
+the setting for the 1GbE speed, the second value for the 2.5GbE speed and
+the third value for the 10GbE speed. All three values are required if the
+property is used.
+- amd,serdes-blwc: Baseline wandering correction enablement
+ 0 - Off
+ 1 - On
+- amd,serdes-cdr-rate: CDR rate speed selection
+- amd,serdes-pq-skew: PQ (data sampling) skew
+- amd,serdes-tx-amp: TX amplitude boost
+
Example:
xgbe_phy@e1240800 {
compatible = "amd,xgbe-phy-seattle-v1a", "ethernet-phy-ieee802.3-c45";
@@ -25,4 +37,8 @@ Example:
interrupt-parent = <&gic>;
interrupts = <0 323 4>;
amd,speed-set = <0>;
+ amd,serdes-blwc = <1>, <1>, <0>;
+ amd,serdes-cdr-rate = <2>, <2>, <7>;
+ amd,serdes-pq-skew = <10>, <10>, <30>;
+ amd,serdes-tx-amp = <15>, <15>, <10>;
};
diff --git a/drivers/net/phy/amd-xgbe-phy.c b/drivers/net/phy/amd-xgbe-phy.c
index 2f210743..9e3af54 100644
--- a/drivers/net/phy/amd-xgbe-phy.c
+++ b/drivers/net/phy/amd-xgbe-phy.c
@@ -88,6 +88,15 @@ MODULE_DESCRIPTION("AMD 10GbE (amd-xgbe) PHY driver");
#define XGBE_PHY_MASK 0xfffffff0
#define XGBE_PHY_SPEEDSET_PROPERTY "amd,speed-set"
+#define XGBE_PHY_BLWC_PROPERTY "amd,serdes-blwc"
+#define XGBE_PHY_CDR_RATE_PROPERTY "amd,serdes-cdr-rate"
+#define XGBE_PHY_PQ_SKEW_PROPERTY "amd,serdes-pq-skew"
+#define XGBE_PHY_TX_AMP_PROPERTY "amd,serdes-tx-amp"
+
+#define XGBE_PHY_SPEEDS 3
+#define XGBE_PHY_SPEED_1000 0
+#define XGBE_PHY_SPEED_2500 1
+#define XGBE_PHY_SPEED_10000 2
#define XGBE_AN_INT_CMPLT 0x01
#define XGBE_AN_INC_LINK 0x02
@@ -152,10 +161,10 @@ MODULE_DESCRIPTION("AMD 10GbE (amd-xgbe) PHY driver");
#define SIR0_STATUS_RX_READY_WIDTH 1
#define SIR0_STATUS_TX_READY_INDEX 8
#define SIR0_STATUS_TX_READY_WIDTH 1
+#define SIR1_SPEED_CDR_RATE_INDEX 12
+#define SIR1_SPEED_CDR_RATE_WIDTH 4
#define SIR1_SPEED_DATARATE_INDEX 4
#define SIR1_SPEED_DATARATE_WIDTH 2
-#define SIR1_SPEED_PI_SPD_SEL_INDEX 12
-#define SIR1_SPEED_PI_SPD_SEL_WIDTH 4
#define SIR1_SPEED_PLLSEL_INDEX 3
#define SIR1_SPEED_PLLSEL_WIDTH 1
#define SIR1_SPEED_RATECHANGE_INDEX 6
@@ -165,20 +174,26 @@ MODULE_DESCRIPTION("AMD 10GbE (amd-xgbe) PHY driver");
#define SIR1_SPEED_WORDMODE_INDEX 0
#define SIR1_SPEED_WORDMODE_WIDTH 3
+#define SPEED_10000_BLWC 0
#define SPEED_10000_CDR 0x7
#define SPEED_10000_PLL 0x1
+#define SPEED_10000_PQ 0x1e
#define SPEED_10000_RATE 0x0
#define SPEED_10000_TXAMP 0xa
#define SPEED_10000_WORD 0x7
+#define SPEED_2500_BLWC 1
#define SPEED_2500_CDR 0x2
#define SPEED_2500_PLL 0x0
+#define SPEED_2500_PQ 0xa
#define SPEED_2500_RATE 0x1
#define SPEED_2500_TXAMP 0xf
#define SPEED_2500_WORD 0x1
+#define SPEED_1000_BLWC 1
#define SPEED_1000_CDR 0x2
#define SPEED_1000_PLL 0x0
+#define SPEED_1000_PQ 0xa
#define SPEED_1000_RATE 0x3
#define SPEED_1000_TXAMP 0xf
#define SPEED_1000_WORD 0x1
@@ -193,15 +208,6 @@ MODULE_DESCRIPTION("AMD 10GbE (amd-xgbe) PHY driver");
#define RXTX_REG114_PQ_REG_INDEX 9
#define RXTX_REG114_PQ_REG_WIDTH 7
-#define RXTX_10000_BLWC 0
-#define RXTX_10000_PQ 0x1e
-
-#define RXTX_2500_BLWC 1
-#define RXTX_2500_PQ 0xa
-
-#define RXTX_1000_BLWC 1
-#define RXTX_1000_PQ 0xa
-
/* Bit setting and getting macros
* The get macro will extract the current bit field value from within
* the variable
@@ -303,6 +309,30 @@ do { \
XRXTX_IOWRITE((_priv), _reg, reg_val); \
} while (0)
+static const u32 amd_xgbe_phy_serdes_blwc[] = {
+ SPEED_1000_BLWC,
+ SPEED_2500_BLWC,
+ SPEED_10000_BLWC,
+};
+
+static const u32 amd_xgbe_phy_serdes_cdr_rate[] = {
+ SPEED_1000_CDR,
+ SPEED_2500_CDR,
+ SPEED_10000_CDR,
+};
+
+static const u32 amd_xgbe_phy_serdes_pq_skew[] = {
+ SPEED_1000_PQ,
+ SPEED_2500_PQ,
+ SPEED_10000_PQ,
+};
+
+static const u32 amd_xgbe_phy_serdes_tx_amp[] = {
+ SPEED_1000_TXAMP,
+ SPEED_2500_TXAMP,
+ SPEED_10000_TXAMP,
+};
+
enum amd_xgbe_phy_an {
AMD_XGBE_AN_READY = 0,
AMD_XGBE_AN_PAGE_RECEIVED,
@@ -353,6 +383,17 @@ struct amd_xgbe_phy_priv {
unsigned int speed_set;
+ /* SerDes UEFI configurable settings.
+ * Switching between modes/speeds requires new values for some
+ * SerDes settings. The values can be supplied as device
+ * properties in array format. The first array entry is for
+ * 1GbE, second for 2.5GbE and third for 10GbE
+ */
+ u32 serdes_blwc[XGBE_PHY_SPEEDS];
+ u32 serdes_cdr_rate[XGBE_PHY_SPEEDS];
+ u32 serdes_pq_skew[XGBE_PHY_SPEEDS];
+ u32 serdes_tx_amp[XGBE_PHY_SPEEDS];
+
/* Auto-negotiation state machine support */
struct mutex an_mutex;
enum amd_xgbe_phy_an an_result;
@@ -483,12 +524,16 @@ static int amd_xgbe_phy_xgmii_mode(struct phy_device *phydev)
XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, DATARATE, SPEED_10000_RATE);
XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, WORDMODE, SPEED_10000_WORD);
- XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, TXAMP, SPEED_10000_TXAMP);
XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, PLLSEL, SPEED_10000_PLL);
- XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, PI_SPD_SEL, SPEED_10000_CDR);
- XRXTX_IOWRITE_BITS(priv, RXTX_REG20, BLWC_ENA, RXTX_10000_BLWC);
- XRXTX_IOWRITE_BITS(priv, RXTX_REG114, PQ_REG, RXTX_10000_PQ);
+ XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, CDR_RATE,
+ priv->serdes_cdr_rate[XGBE_PHY_SPEED_10000]);
+ XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, TXAMP,
+ priv->serdes_tx_amp[XGBE_PHY_SPEED_10000]);
+ XRXTX_IOWRITE_BITS(priv, RXTX_REG20, BLWC_ENA,
+ priv->serdes_blwc[XGBE_PHY_SPEED_10000]);
+ XRXTX_IOWRITE_BITS(priv, RXTX_REG114, PQ_REG,
+ priv->serdes_pq_skew[XGBE_PHY_SPEED_10000]);
amd_xgbe_phy_serdes_complete_ratechange(phydev);
@@ -531,12 +576,16 @@ static int amd_xgbe_phy_gmii_2500_mode(struct phy_device *phydev)
XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, DATARATE, SPEED_2500_RATE);
XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, WORDMODE, SPEED_2500_WORD);
- XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, TXAMP, SPEED_2500_TXAMP);
XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, PLLSEL, SPEED_2500_PLL);
- XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, PI_SPD_SEL, SPEED_2500_CDR);
- XRXTX_IOWRITE_BITS(priv, RXTX_REG20, BLWC_ENA, RXTX_2500_BLWC);
- XRXTX_IOWRITE_BITS(priv, RXTX_REG114, PQ_REG, RXTX_2500_PQ);
+ XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, CDR_RATE,
+ priv->serdes_cdr_rate[XGBE_PHY_SPEED_2500]);
+ XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, TXAMP,
+ priv->serdes_tx_amp[XGBE_PHY_SPEED_2500]);
+ XRXTX_IOWRITE_BITS(priv, RXTX_REG20, BLWC_ENA,
+ priv->serdes_blwc[XGBE_PHY_SPEED_2500]);
+ XRXTX_IOWRITE_BITS(priv, RXTX_REG114, PQ_REG,
+ priv->serdes_pq_skew[XGBE_PHY_SPEED_2500]);
amd_xgbe_phy_serdes_complete_ratechange(phydev);
@@ -579,12 +628,16 @@ static int amd_xgbe_phy_gmii_mode(struct phy_device *phydev)
XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, DATARATE, SPEED_1000_RATE);
XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, WORDMODE, SPEED_1000_WORD);
- XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, TXAMP, SPEED_1000_TXAMP);
XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, PLLSEL, SPEED_1000_PLL);
- XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, PI_SPD_SEL, SPEED_1000_CDR);
- XRXTX_IOWRITE_BITS(priv, RXTX_REG20, BLWC_ENA, RXTX_1000_BLWC);
- XRXTX_IOWRITE_BITS(priv, RXTX_REG114, PQ_REG, RXTX_1000_PQ);
+ XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, CDR_RATE,
+ priv->serdes_cdr_rate[XGBE_PHY_SPEED_1000]);
+ XSIR1_IOWRITE_BITS(priv, SIR1_SPEED, TXAMP,
+ priv->serdes_tx_amp[XGBE_PHY_SPEED_1000]);
+ XRXTX_IOWRITE_BITS(priv, RXTX_REG20, BLWC_ENA,
+ priv->serdes_blwc[XGBE_PHY_SPEED_1000]);
+ XRXTX_IOWRITE_BITS(priv, RXTX_REG114, PQ_REG,
+ priv->serdes_pq_skew[XGBE_PHY_SPEED_1000]);
amd_xgbe_phy_serdes_complete_ratechange(phydev);
@@ -1555,6 +1608,66 @@ static int amd_xgbe_phy_probe(struct phy_device *phydev)
goto err_sir1;
}
+ if (device_property_present(phy_dev, XGBE_PHY_BLWC_PROPERTY)) {
+ ret = device_property_read_u32_array(phy_dev,
+ XGBE_PHY_BLWC_PROPERTY,
+ priv->serdes_blwc,
+ XGBE_PHY_SPEEDS);
+ if (ret) {
+ dev_err(dev, "invalid %s property\n",
+ XGBE_PHY_BLWC_PROPERTY);
+ goto err_sir1;
+ }
+ } else {
+ memcpy(priv->serdes_blwc, amd_xgbe_phy_serdes_blwc,
+ sizeof(priv->serdes_blwc));
+ }
+
+ if (device_property_present(phy_dev, XGBE_PHY_CDR_RATE_PROPERTY)) {
+ ret = device_property_read_u32_array(phy_dev,
+ XGBE_PHY_CDR_RATE_PROPERTY,
+ priv->serdes_cdr_rate,
+ XGBE_PHY_SPEEDS);
+ if (ret) {
+ dev_err(dev, "invalid %s property\n",
+ XGBE_PHY_CDR_RATE_PROPERTY);
+ goto err_sir1;
+ }
+ } else {
+ memcpy(priv->serdes_cdr_rate, amd_xgbe_phy_serdes_cdr_rate,
+ sizeof(priv->serdes_cdr_rate));
+ }
+
+ if (device_property_present(phy_dev, XGBE_PHY_PQ_SKEW_PROPERTY)) {
+ ret = device_property_read_u32_array(phy_dev,
+ XGBE_PHY_PQ_SKEW_PROPERTY,
+ priv->serdes_pq_skew,
+ XGBE_PHY_SPEEDS);
+ if (ret) {
+ dev_err(dev, "invalid %s property\n",
+ XGBE_PHY_PQ_SKEW_PROPERTY);
+ goto err_sir1;
+ }
+ } else {
+ memcpy(priv->serdes_pq_skew, amd_xgbe_phy_serdes_pq_skew,
+ sizeof(priv->serdes_pq_skew));
+ }
+
+ if (device_property_present(phy_dev, XGBE_PHY_TX_AMP_PROPERTY)) {
+ ret = device_property_read_u32_array(phy_dev,
+ XGBE_PHY_TX_AMP_PROPERTY,
+ priv->serdes_tx_amp,
+ XGBE_PHY_SPEEDS);
+ if (ret) {
+ dev_err(dev, "invalid %s property\n",
+ XGBE_PHY_TX_AMP_PROPERTY);
+ goto err_sir1;
+ }
+ } else {
+ memcpy(priv->serdes_tx_amp, amd_xgbe_phy_serdes_tx_amp,
+ sizeof(priv->serdes_tx_amp));
+ }
+
phydev->priv = priv;
if (!priv->adev || acpi_disabled)
^ permalink raw reply related
* [PATCH net-next v1 06/12] amd-xgbe: Simplify the Rx desciptor ring tracking
From: Tom Lendacky @ 2015-01-16 18:46 UTC (permalink / raw)
To: netdev; +Cc: David Miller
In-Reply-To: <20150116184618.18809.319.stgit@tlendack-t1.amdoffice.net>
Make the Rx descriptor ring processing similar to the Tx descriptor
ring processing. Remove the realloc_index and realloc_threshold
variables and base everything on the current index counter and the
dirty index counter.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
---
drivers/net/ethernet/amd/xgbe/xgbe-desc.c | 32 +----------------------------
drivers/net/ethernet/amd/xgbe/xgbe-drv.c | 25 +++++++++++++++++++----
drivers/net/ethernet/amd/xgbe/xgbe.h | 11 +++-------
3 files changed, 25 insertions(+), 43 deletions(-)
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-desc.c b/drivers/net/ethernet/amd/xgbe/xgbe-desc.c
index a50891f..d81fc6b 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-desc.c
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-desc.c
@@ -422,7 +422,6 @@ static void xgbe_wrapper_rx_descriptor_init(struct xgbe_prv_data *pdata)
ring->cur = 0;
ring->dirty = 0;
- memset(&ring->rx, 0, sizeof(ring->rx));
hw_if->rx_desc_init(channel);
}
@@ -621,35 +620,6 @@ err_out:
return 0;
}
-static void xgbe_realloc_rx_buffer(struct xgbe_channel *channel)
-{
- struct xgbe_prv_data *pdata = channel->pdata;
- struct xgbe_hw_if *hw_if = &pdata->hw_if;
- struct xgbe_ring *ring = channel->rx_ring;
- struct xgbe_ring_data *rdata;
- int i;
-
- DBGPR("-->xgbe_realloc_rx_buffer: rx_ring->rx.realloc_index = %u\n",
- ring->rx.realloc_index);
-
- for (i = 0; i < ring->dirty; i++) {
- rdata = XGBE_GET_DESC_DATA(ring, ring->rx.realloc_index);
-
- /* Reset rdata values */
- xgbe_unmap_rdata(pdata, rdata);
-
- if (xgbe_map_rx_buffer(pdata, ring, rdata))
- break;
-
- hw_if->rx_desc_reset(rdata);
-
- ring->rx.realloc_index++;
- }
- ring->dirty = 0;
-
- DBGPR("<--xgbe_realloc_rx_buffer\n");
-}
-
void xgbe_init_function_ptrs_desc(struct xgbe_desc_if *desc_if)
{
DBGPR("-->xgbe_init_function_ptrs_desc\n");
@@ -657,7 +627,7 @@ void xgbe_init_function_ptrs_desc(struct xgbe_desc_if *desc_if)
desc_if->alloc_ring_resources = xgbe_alloc_ring_resources;
desc_if->free_ring_resources = xgbe_free_ring_resources;
desc_if->map_tx_skb = xgbe_map_tx_skb;
- desc_if->realloc_rx_buffer = xgbe_realloc_rx_buffer;
+ desc_if->map_rx_buffer = xgbe_map_rx_buffer;
desc_if->unmap_rdata = xgbe_unmap_rdata;
desc_if->wrapper_tx_desc_init = xgbe_wrapper_tx_descriptor_init;
desc_if->wrapper_rx_desc_init = xgbe_wrapper_rx_descriptor_init;
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-drv.c b/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
index bdb373c..e2f560f 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
@@ -225,6 +225,11 @@ static inline unsigned int xgbe_tx_avail_desc(struct xgbe_ring *ring)
return (ring->rdesc_count - (ring->cur - ring->dirty));
}
+static inline unsigned int xgbe_rx_dirty_desc(struct xgbe_ring *ring)
+{
+ return (ring->cur - ring->dirty);
+}
+
static int xgbe_maybe_stop_tx_queue(struct xgbe_channel *channel,
struct xgbe_ring *ring, unsigned int count)
{
@@ -1775,15 +1780,28 @@ struct net_device_ops *xgbe_get_netdev_ops(void)
static void xgbe_rx_refresh(struct xgbe_channel *channel)
{
struct xgbe_prv_data *pdata = channel->pdata;
+ struct xgbe_hw_if *hw_if = &pdata->hw_if;
struct xgbe_desc_if *desc_if = &pdata->desc_if;
struct xgbe_ring *ring = channel->rx_ring;
struct xgbe_ring_data *rdata;
- desc_if->realloc_rx_buffer(channel);
+ while (ring->dirty != ring->cur) {
+ rdata = XGBE_GET_DESC_DATA(ring, ring->dirty);
+
+ /* Reset rdata values */
+ desc_if->unmap_rdata(pdata, rdata);
+
+ if (desc_if->map_rx_buffer(pdata, ring, rdata))
+ break;
+
+ hw_if->rx_desc_reset(rdata);
+
+ ring->dirty++;
+ }
/* Update the Rx Tail Pointer Register with address of
* the last cleaned entry */
- rdata = XGBE_GET_DESC_DATA(ring, ring->rx.realloc_index - 1);
+ rdata = XGBE_GET_DESC_DATA(ring, ring->dirty - 1);
XGMAC_DMA_IOWRITE(channel, DMA_CH_RDTR_LO,
lower_32_bits(rdata->rdesc_dma));
}
@@ -1933,7 +1951,7 @@ static int xgbe_rx_poll(struct xgbe_channel *channel, int budget)
read_again:
rdata = XGBE_GET_DESC_DATA(ring, ring->cur);
- if (ring->dirty > (XGBE_RX_DESC_CNT >> 3))
+ if (xgbe_rx_dirty_desc(ring) > (XGBE_RX_DESC_CNT >> 3))
xgbe_rx_refresh(channel);
if (hw_if->dev_read(channel))
@@ -1941,7 +1959,6 @@ read_again:
received++;
ring->cur++;
- ring->dirty++;
incomplete = XGMAC_GET_BITS(packet->attributes,
RX_PACKET_ATTRIBUTES,
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe.h b/drivers/net/ethernet/amd/xgbe/xgbe.h
index 2af6aff..e6ee64e 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe.h
+++ b/drivers/net/ethernet/amd/xgbe/xgbe.h
@@ -361,8 +361,7 @@ struct xgbe_ring {
* cur - Tx: index of descriptor to be used for current transfer
* Rx: index of descriptor to check for packet availability
* dirty - Tx: index of descriptor to check for transfer complete
- * Rx: count of descriptors in which a packet has been received
- * (used with skb_realloc_index to refresh the ring)
+ * Rx: index of descriptor to check for buffer reallocation
*/
unsigned int cur;
unsigned int dirty;
@@ -377,11 +376,6 @@ struct xgbe_ring {
unsigned short cur_mss;
unsigned short cur_vlan_ctag;
} tx;
-
- struct {
- unsigned int realloc_index;
- unsigned int realloc_threshold;
- } rx;
};
} ____cacheline_aligned;
@@ -596,7 +590,8 @@ struct xgbe_desc_if {
int (*alloc_ring_resources)(struct xgbe_prv_data *);
void (*free_ring_resources)(struct xgbe_prv_data *);
int (*map_tx_skb)(struct xgbe_channel *, struct sk_buff *);
- void (*realloc_rx_buffer)(struct xgbe_channel *);
+ int (*map_rx_buffer)(struct xgbe_prv_data *, struct xgbe_ring *,
+ struct xgbe_ring_data *);
void (*unmap_rdata)(struct xgbe_prv_data *, struct xgbe_ring_data *);
void (*wrapper_tx_desc_init)(struct xgbe_prv_data *);
void (*wrapper_rx_desc_init)(struct xgbe_prv_data *);
^ permalink raw reply related
* [PATCH net-next v1 07/12] amd-xgbe: Remove need for Tx path spinlock
From: Tom Lendacky @ 2015-01-16 18:46 UTC (permalink / raw)
To: netdev; +Cc: David Miller
In-Reply-To: <20150116184618.18809.319.stgit@tlendack-t1.amdoffice.net>
Since the Tx ring cleanup can run at the same time that data is being
transmitted, a spin lock was used to protect the ring. This patch
eliminates the need for Tx spinlocks by updating the current ring
position only after all ownership bits for data being transmitted have
been set. This will insure that ring operations in the Tx cleanup path
do not interfere with the ring operations in the Tx transmit path.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
---
drivers/net/ethernet/amd/xgbe/xgbe-dev.c | 17 +++++++++--------
drivers/net/ethernet/amd/xgbe/xgbe-drv.c | 21 +--------------------
2 files changed, 10 insertions(+), 28 deletions(-)
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-dev.c b/drivers/net/ethernet/amd/xgbe/xgbe-dev.c
index e424997..5dff127 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-dev.c
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-dev.c
@@ -1359,6 +1359,7 @@ static void xgbe_dev_xmit(struct xgbe_channel *channel)
unsigned int tso_context, vlan_context;
unsigned int tx_set_ic;
int start_index = ring->cur;
+ int cur_index = ring->cur;
int i;
DBGPR("-->xgbe_dev_xmit\n");
@@ -1401,7 +1402,7 @@ static void xgbe_dev_xmit(struct xgbe_channel *channel)
else
tx_set_ic = 0;
- rdata = XGBE_GET_DESC_DATA(ring, ring->cur);
+ rdata = XGBE_GET_DESC_DATA(ring, cur_index);
rdesc = rdata->rdesc;
/* Create a context descriptor if this is a TSO packet */
@@ -1444,8 +1445,8 @@ static void xgbe_dev_xmit(struct xgbe_channel *channel)
ring->tx.cur_vlan_ctag = packet->vlan_ctag;
}
- ring->cur++;
- rdata = XGBE_GET_DESC_DATA(ring, ring->cur);
+ cur_index++;
+ rdata = XGBE_GET_DESC_DATA(ring, cur_index);
rdesc = rdata->rdesc;
}
@@ -1473,7 +1474,7 @@ static void xgbe_dev_xmit(struct xgbe_channel *channel)
XGMAC_SET_BITS_LE(rdesc->desc3, TX_NORMAL_DESC3, CTXT, 0);
/* Set OWN bit if not the first descriptor */
- if (ring->cur != start_index)
+ if (cur_index != start_index)
XGMAC_SET_BITS_LE(rdesc->desc3, TX_NORMAL_DESC3, OWN, 1);
if (tso) {
@@ -1497,9 +1498,9 @@ static void xgbe_dev_xmit(struct xgbe_channel *channel)
packet->length);
}
- for (i = ring->cur - start_index + 1; i < packet->rdesc_count; i++) {
- ring->cur++;
- rdata = XGBE_GET_DESC_DATA(ring, ring->cur);
+ for (i = cur_index - start_index + 1; i < packet->rdesc_count; i++) {
+ cur_index++;
+ rdata = XGBE_GET_DESC_DATA(ring, cur_index);
rdesc = rdata->rdesc;
/* Update buffer address */
@@ -1551,7 +1552,7 @@ static void xgbe_dev_xmit(struct xgbe_channel *channel)
/* Make sure ownership is written to the descriptor */
wmb();
- ring->cur++;
+ ring->cur = cur_index + 1;
if (!packet->skb->xmit_more ||
netif_xmit_stopped(netdev_get_tx_queue(pdata->netdev,
channel->queue_index)))
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-drv.c b/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
index e2f560f..c036a0e 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
@@ -415,17 +415,13 @@ static enum hrtimer_restart xgbe_tx_timer(struct hrtimer *timer)
struct xgbe_channel *channel = container_of(timer,
struct xgbe_channel,
tx_timer);
- struct xgbe_ring *ring = channel->tx_ring;
struct xgbe_prv_data *pdata = channel->pdata;
struct napi_struct *napi;
- unsigned long flags;
DBGPR("-->xgbe_tx_timer\n");
napi = (pdata->per_channel_irq) ? &channel->napi : &pdata->napi;
- spin_lock_irqsave(&ring->lock, flags);
-
if (napi_schedule_prep(napi)) {
/* Disable Tx and Rx interrupts */
if (pdata->per_channel_irq)
@@ -439,8 +435,6 @@ static enum hrtimer_restart xgbe_tx_timer(struct hrtimer *timer)
channel->tx_timer_active = 0;
- spin_unlock_irqrestore(&ring->lock, flags);
-
DBGPR("<--xgbe_tx_timer\n");
return HRTIMER_NORESTART;
@@ -1450,7 +1444,6 @@ static int xgbe_xmit(struct sk_buff *skb, struct net_device *netdev)
struct xgbe_ring *ring;
struct xgbe_packet_data *packet;
struct netdev_queue *txq;
- unsigned long flags;
int ret;
DBGPR("-->xgbe_xmit: skb->len = %d\n", skb->len);
@@ -1462,8 +1455,6 @@ static int xgbe_xmit(struct sk_buff *skb, struct net_device *netdev)
ret = NETDEV_TX_OK;
- spin_lock_irqsave(&ring->lock, flags);
-
if (skb->len == 0) {
netdev_err(netdev, "empty skb received from stack\n");
dev_kfree_skb_any(skb);
@@ -1510,10 +1501,6 @@ static int xgbe_xmit(struct sk_buff *skb, struct net_device *netdev)
ret = NETDEV_TX_OK;
tx_netdev_return:
- spin_unlock_irqrestore(&ring->lock, flags);
-
- DBGPR("<--xgbe_xmit\n");
-
return ret;
}
@@ -1841,7 +1828,6 @@ static int xgbe_tx_poll(struct xgbe_channel *channel)
struct xgbe_ring_desc *rdesc;
struct net_device *netdev = pdata->netdev;
struct netdev_queue *txq;
- unsigned long flags;
int processed = 0;
unsigned int tx_packets = 0, tx_bytes = 0;
@@ -1853,8 +1839,6 @@ static int xgbe_tx_poll(struct xgbe_channel *channel)
txq = netdev_get_tx_queue(netdev, channel->queue_index);
- spin_lock_irqsave(&ring->lock, flags);
-
while ((processed < XGBE_TX_DESC_MAX_PROC) &&
(ring->dirty != ring->cur)) {
rdata = XGBE_GET_DESC_DATA(ring, ring->dirty);
@@ -1885,7 +1869,7 @@ static int xgbe_tx_poll(struct xgbe_channel *channel)
}
if (!processed)
- goto unlock;
+ return 0;
netdev_tx_completed_queue(txq, tx_packets, tx_bytes);
@@ -1897,9 +1881,6 @@ static int xgbe_tx_poll(struct xgbe_channel *channel)
DBGPR("<--xgbe_tx_poll: processed=%d\n", processed);
-unlock:
- spin_unlock_irqrestore(&ring->lock, flags);
-
return processed;
}
^ permalink raw reply related
* [PATCH net-next v1 09/12] amd-xgbe-phy: Properly support the FEC auto-negotiation
From: Tom Lendacky @ 2015-01-16 18:47 UTC (permalink / raw)
To: netdev; +Cc: David Miller
In-Reply-To: <20150116184618.18809.319.stgit@tlendack-t1.amdoffice.net>
Advertise and apply the Forward Error Correction capabilities of the
device based on the FEC ability of the device. Also, remove the use
of some hard coded values related to KR and FEC in preference of some
#defines.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
---
drivers/net/phy/amd-xgbe-phy.c | 44 ++++++++++++++++++++++++++++++----------
1 file changed, 33 insertions(+), 11 deletions(-)
diff --git a/drivers/net/phy/amd-xgbe-phy.c b/drivers/net/phy/amd-xgbe-phy.c
index 7fde508..7207f36 100644
--- a/drivers/net/phy/amd-xgbe-phy.c
+++ b/drivers/net/phy/amd-xgbe-phy.c
@@ -99,10 +99,21 @@ MODULE_DESCRIPTION("AMD 10GbE (amd-xgbe) PHY driver");
#define XGBE_PHY_RATECHANGE_COUNT 500
+#define XGBE_PHY_KR_TRAINING_START 0x01
+#define XGBE_PHY_KR_TRAINING_ENABLE 0x02
+
+#define XGBE_PHY_FEC_ENABLE 0x01
+#define XGBE_PHY_FEC_FORWARD 0x02
+#define XGBE_PHY_FEC_MASK 0x03
+
#ifndef MDIO_PMA_10GBR_PMD_CTRL
#define MDIO_PMA_10GBR_PMD_CTRL 0x0096
#endif
+#ifndef MDIO_PMA_10GBR_FEC_ABILITY
+#define MDIO_PMA_10GBR_FEC_ABILITY 0x00aa
+#endif
+
#ifndef MDIO_PMA_10GBR_FEC_CTRL
#define MDIO_PMA_10GBR_FEC_CTRL 0x00ab
#endif
@@ -345,6 +356,7 @@ struct amd_xgbe_phy_priv {
struct workqueue_struct *an_workqueue;
unsigned int an_supported;
unsigned int parallel_detect;
+ unsigned int fec_ability;
unsigned int lpm_ctrl; /* CTRL1 for resume */
};
@@ -357,7 +369,7 @@ static int amd_xgbe_an_enable_kr_training(struct phy_device *phydev)
if (ret < 0)
return ret;
- ret |= 0x02;
+ ret |= XGBE_PHY_KR_TRAINING_ENABLE;
phy_write_mmd(phydev, MDIO_MMD_PMAPMD, MDIO_PMA_10GBR_PMD_CTRL, ret);
return 0;
@@ -371,7 +383,7 @@ static int amd_xgbe_an_disable_kr_training(struct phy_device *phydev)
if (ret < 0)
return ret;
- ret &= ~0x02;
+ ret &= ~XGBE_PHY_KR_TRAINING_ENABLE;
phy_write_mmd(phydev, MDIO_MMD_PMAPMD, MDIO_PMA_10GBR_PMD_CTRL, ret);
return 0;
@@ -690,10 +702,9 @@ static enum amd_xgbe_phy_an amd_xgbe_an_tx_training(struct phy_device *phydev,
if (ret < 0)
return AMD_XGBE_AN_ERROR;
+ ret &= ~XGBE_PHY_FEC_MASK;
if ((ad_reg & 0xc000) && (lp_reg & 0xc000))
- ret |= 0x01;
- else
- ret &= ~0x01;
+ ret |= priv->fec_ability;
phy_write_mmd(phydev, MDIO_MMD_PMAPMD, MDIO_PMA_10GBR_FEC_CTRL, ret);
@@ -702,12 +713,15 @@ static enum amd_xgbe_phy_an amd_xgbe_an_tx_training(struct phy_device *phydev,
if (ret < 0)
return AMD_XGBE_AN_ERROR;
- XSIR0_IOWRITE_BITS(priv, SIR0_KR_RT_1, RESET, 1);
+ if (ret & XGBE_PHY_KR_TRAINING_ENABLE) {
+ XSIR0_IOWRITE_BITS(priv, SIR0_KR_RT_1, RESET, 1);
- ret |= 0x01;
- phy_write_mmd(phydev, MDIO_MMD_PMAPMD, MDIO_PMA_10GBR_PMD_CTRL, ret);
+ ret |= XGBE_PHY_KR_TRAINING_START;
+ phy_write_mmd(phydev, MDIO_MMD_PMAPMD, MDIO_PMA_10GBR_PMD_CTRL,
+ ret);
- XSIR0_IOWRITE_BITS(priv, SIR0_KR_RT_1, RESET, 0);
+ XSIR0_IOWRITE_BITS(priv, SIR0_KR_RT_1, RESET, 0);
+ }
return AMD_XGBE_AN_PAGE_RECEIVED;
}
@@ -1092,12 +1106,16 @@ static int amd_xgbe_phy_config_init(struct phy_device *phydev)
priv->an_irq_allocated = 1;
}
+ ret = phy_read_mmd(phydev, MDIO_MMD_PMAPMD, MDIO_PMA_10GBR_FEC_ABILITY);
+ if (ret < 0)
+ return ret;
+ priv->fec_ability = ret & XGBE_PHY_FEC_MASK;
+
/* Initialize supported features */
phydev->supported = SUPPORTED_Autoneg;
phydev->supported |= SUPPORTED_Pause | SUPPORTED_Asym_Pause;
phydev->supported |= SUPPORTED_Backplane;
- phydev->supported |= SUPPORTED_10000baseKR_Full |
- SUPPORTED_10000baseR_FEC;
+ phydev->supported |= SUPPORTED_10000baseKR_Full;
switch (priv->speed_set) {
case AMD_XGBE_PHY_SPEEDSET_1000_10000:
phydev->supported |= SUPPORTED_1000baseKX_Full;
@@ -1106,6 +1124,10 @@ static int amd_xgbe_phy_config_init(struct phy_device *phydev)
phydev->supported |= SUPPORTED_2500baseX_Full;
break;
}
+
+ if (priv->fec_ability & XGBE_PHY_FEC_ENABLE)
+ phydev->supported |= SUPPORTED_10000baseR_FEC;
+
phydev->advertising = phydev->supported;
/* Set initial mode - call the mode setting routines
^ permalink raw reply related
* Re: [PATCH net 2/2] sh_eth: Fix ethtool operation crash when net device is down
From: Florian Fainelli @ 2015-01-16 18:45 UTC (permalink / raw)
To: Ben Hutchings
Cc: ct178-internal, David S. Miller, netdev, linux-kernel,
Nobuhiro Iwamatsu, Mitsuhiro Kimura, Hisashi Nakamura,
Yoshihiro Kaneko
In-Reply-To: <1421430685.1222.192.camel@xylophone.i.decadent.org.uk>
2015-01-16 9:51 GMT-08:00 Ben Hutchings <ben.hutchings@codethink.co.uk>:
> The driver connects and disconnects the PHY device whenever the
> net device is brought up and down. The ethtool get_settings,
> set_settings and nway_reset operations will dereference a null
> or dangling pointer if called while it is down.
>
> I think it would be preferable to keep the PHY connected, but there
> may be good reasons not to.
phy_disconnect() is the canonical way to stop the PHY library state
machine, and avoid deferred work to be done and call the driver's
adjust_link function. This also boils down to calling phy_detach()
which can put the PHY in a low-power mode when implemented.
>
> As an immediate fix for this bug:
> - Set the phydev pointer to NULL after disconnecting the PHY
> - Change those three operations to return -ENODEV while the PHY is
> not connected
>
> Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
> ---
> drivers/net/ethernet/renesas/sh_eth.c | 10 ++++++++++
> 1 file changed, 10 insertions(+)
>
> diff --git a/drivers/net/ethernet/renesas/sh_eth.c b/drivers/net/ethernet/renesas/sh_eth.c
> index 0c4d5b5..28e3822 100644
> --- a/drivers/net/ethernet/renesas/sh_eth.c
> +++ b/drivers/net/ethernet/renesas/sh_eth.c
> @@ -1827,6 +1827,9 @@ static int sh_eth_get_settings(struct net_device *ndev,
> unsigned long flags;
> int ret;
>
> + if (!mdp->phydev)
> + return -ENODEV;
Since the PHY is disconnected, would not checking for netif_running()
make sense here, unless there is a good reason to still allow
phy_ethtool_gset() to be called?
> +
> spin_lock_irqsave(&mdp->lock, flags);
> ret = phy_ethtool_gset(mdp->phydev, ecmd);
> spin_unlock_irqrestore(&mdp->lock, flags);
> @@ -1841,6 +1844,9 @@ static int sh_eth_set_settings(struct net_device *ndev,
> unsigned long flags;
> int ret;
>
> + if (!mdp->phydev)
> + return -ENODEV;
> +
> spin_lock_irqsave(&mdp->lock, flags);
>
> /* disable tx and rx */
> @@ -1875,6 +1881,9 @@ static int sh_eth_nway_reset(struct net_device *ndev)
> unsigned long flags;
> int ret;
>
> + if (!mdp->phydev)
> + return -ENODEV;
> +
> spin_lock_irqsave(&mdp->lock, flags);
> ret = phy_start_aneg(mdp->phydev);
> spin_unlock_irqrestore(&mdp->lock, flags);
> @@ -2184,6 +2193,7 @@ static int sh_eth_close(struct net_device *ndev)
> if (mdp->phydev) {
> phy_stop(mdp->phydev);
> phy_disconnect(mdp->phydev);
> + mdp->phydev = NULL;
> }
>
> free_irq(ndev->irq, ndev);
> --
> 1.7.10.4
>
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
Florian
^ permalink raw reply
* Re: [PATCH 7/9] rhashtable: Per bucket locks & deferred expansion/shrinking
From: Patrick McHardy @ 2015-01-16 18:36 UTC (permalink / raw)
To: Thomas Graf
Cc: David Laight, davem@davemloft.net, netdev@vger.kernel.org,
herbert@gondor.apana.org.au, paulmck@linux.vnet.ibm.com,
edumazet@google.com, john.r.fastabend@intel.com,
josh@joshtriplett.org, netfilter-devel@vger.kernel.org
In-Reply-To: <20150116165302.GE15052@casper.infradead.org>
On 16.01, Thomas Graf wrote:
> On 01/16/15 at 04:43pm, David Laight wrote:
> > The walker is unlikely to see items that get inserted early in the hash
> > table even without a resize.
>
> I don't follow, you have to explain this statement.
>
> Walkers which don't want to see duplicates or miss entries should
> just take the mutex.
Well, we do have a problem with interrupted dumps. As you know once
the netlink message buffer is full, we return to userspace and
continue dumping during the next read. Expanding obviously changes
the order since we rehash from bucket N to N and 2N, so this will
indeed cause duplicate (doesn't matter) and missed entries.
^ permalink raw reply
* Stable request for "hw csum failure" fixes
From: Jay Vosburgh @ 2015-01-16 18:30 UTC (permalink / raw)
To: netdev; +Cc: David S. Miller
Please consider commits
commit 17e96834fd35997ca7cdfbf15413bcd5a36ad448
Author: Govindarajulu Varadarajan <_govind@gmx.com>
Date: Thu Dec 18 15:58:42 2014 +0530
enic: fix rx skb checksum
commit 2c26d34bbcc0b3f30385d5587aa232289e2eed8e
Author: Jay Vosburgh <jay.vosburgh@canonical.com>
Date: Fri Dec 19 15:32:00 2014 -0800
net/core: Handle csum for CHECKSUM_COMPLETE VXLAN forwarding
for -stable kernels back to 3.10.
These patches have been confirmed to fix the "hw csum failure"
issue in the Ubuntu 3.13 kernel, and code inspection suggests that the
3.12 and 3.10 kernels would also exhibit the problem.
Thanks,
-J
---
-Jay Vosburgh, jay.vosburgh@canonical.com
^ permalink raw reply
* Re: [PATCH] net: wireless: atmel: Remove open-coded and wrong strcasecmp
From: Sergei Shtylyov @ 2015-01-16 18:17 UTC (permalink / raw)
To: Rasmus Villemoes, Simon Kelley, Kalle Valo
Cc: linux-wireless-u79uwXL29TY76Z2rM5mHXA,
netdev-u79uwXL29TY76Z2rM5mHXA,
linux-kernel-u79uwXL29TY76Z2rM5mHXA
In-Reply-To: <1421414907-26764-1-git-send-email-linux-qQsb+v5E8BnlAoU/VqSP6n9LOBIZ5rWg@public.gmane.org>
Hello.
On 01/16/2015 04:28 PM, Rasmus Villemoes wrote:
> The kernel's string library does in fact have strcasecmp, at least
> since ded220bd8f08. Moreover, this open-coded version is in fact
Please also specify that commit's summary line in parens.
> wrong: If the strings only differ in their last character, a and b
> have already been incremented to point to the terminating nul bytes,
NUL.
> so they would be treated as equal.
> Signed-off-by: Rasmus Villemoes <linux-qQsb+v5E8BnlAoU/VqSP6n9LOBIZ5rWg@public.gmane.org>
[...]
WBR, Sergei
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* [PATCH v3 net-next 0/1] ip_tunnel: Create percpu gro_cell
From: Martin KaFai Lau @ 2015-01-16 18:10 UTC (permalink / raw)
To: netdev; +Cc: Eric Dumazet, kernel-team
In the ipip tunnel, the skb->queue_mapping is lost in ipip_rcv().
All skb will be queued to the same cell->napi_skbs. The
gro_cell_poll is pinned to one core under load. In production traffic,
we also see severe rx_dropped in the tunl iface and it is probably due to
this limit: skb_queue_len(&cell->napi_skbs) > netdev_max_backlog
This patch is trying to alloc_percpu(struct gro_cell) and schedule
gro_cell_poll to process the skb in the same core.
Changes from v1:
Eric Dumazet pointed out that ____cacheline_aligned_in_smp is no longer needed.
Changes from v2:
Dropped the one-item-struct cleanup patch per comment.
Setup:
VIP_PREFIX=9.9.9.9/32
REMOTE_REAL_IP=10.228.95.75
if [ "$1" = "encap" ]
then
sudo ip tunnel add mode ipip remote ${REMOTE_REAL_IP}
sudo ip link set dev ipip0 up
sudo ip route add dev ipip0 ${VIP_PREFIX}
else
# Decapsulating host
sudo ip tunnel add mode ipip
sudo ip link set dev tunl0 up
sudo ip addr add dev lo ${VIP_PREFIX}
sudo sysctl -a | grep '\.rp_filter' | awk '{print $1;}' | \
xargs -n1 -I{} sudo sysctl {}=0
fi
Before:
[root@DECAP ~]# netserver -p 8888
[root@ENCAP ~]# super_netperf 200 -t TCP_RR -H 9.9.9.9 -p 8888 \
-l 30 -- -d 0x6 -m 8k,64k -s 1M -S 1M
332215
[root@DECAP ~]# perf probe -a gro_cell_poll
[root@DECAP ~]# perf stat -I 1000 -a -A -e probe:gro_cell_poll
117.258518273 CPU0 0 probe:gro_cell_poll
117.258518273 CPU1 0 probe:gro_cell_poll
117.258518273 CPU2 0 probe:gro_cell_poll
117.258518273 CPU3 0 probe:gro_cell_poll
117.258518273 CPU4 0 probe:gro_cell_poll
117.258518273 CPU5 0 probe:gro_cell_poll
117.258518273 CPU6 0 probe:gro_cell_poll
117.258518273 CPU7 0 probe:gro_cell_poll
117.258518273 CPU8 0 probe:gro_cell_poll
117.258518273 CPU9 0 probe:gro_cell_poll
117.258518273 CPU10 0 probe:gro_cell_poll
117.258518273 CPU11 0 probe:gro_cell_poll
117.258518273 CPU12 0 probe:gro_cell_poll
117.258518273 CPU13 0 probe:gro_cell_poll
117.258518273 CPU14 0 probe:gro_cell_poll
117.258518273 CPU15 4,882 probe:gro_cell_poll
117.258518273 CPU16 0 probe:gro_cell_poll
117.258518273 CPU17 0 probe:gro_cell_poll
117.258518273 CPU18 0 probe:gro_cell_poll
117.258518273 CPU19 0 probe:gro_cell_poll
117.258518273 CPU20 0 probe:gro_cell_poll
117.258518273 CPU21 0 probe:gro_cell_poll
117.258518273 CPU22 0 probe:gro_cell_poll
117.258518273 CPU23 0 probe:gro_cell_poll
117.258518273 CPU24 0 probe:gro_cell_poll
117.258518273 CPU25 0 probe:gro_cell_poll
117.258518273 CPU26 0 probe:gro_cell_poll
117.258518273 CPU27 0 probe:gro_cell_poll
117.258518273 CPU28 0 probe:gro_cell_poll
117.258518273 CPU29 0 probe:gro_cell_poll
117.258518273 CPU30 0 probe:gro_cell_poll
117.258518273 CPU31 0 probe:gro_cell_poll
117.258518273 CPU32 0 probe:gro_cell_poll
117.258518273 CPU33 0 probe:gro_cell_poll
117.258518273 CPU34 0 probe:gro_cell_poll
117.258518273 CPU35 0 probe:gro_cell_poll
117.258518273 CPU36 0 probe:gro_cell_poll
117.258518273 CPU37 0 probe:gro_cell_poll
117.258518273 CPU38 0 probe:gro_cell_poll
117.258518273 CPU39 0 probe:gro_cell_poll
After:
[root@DECAP ~]# netserver -p 8888
[root@ENCAP ~]# super_netperf 200 -t TCP_RR -H 9.9.9.9 -p 8888 \
-l 30 -- -d 0x6 -m 8k,64k -s 1M -S 1M
877530
[root@DECAP ~]# perf probe -a gro_cell_poll
[root@DECAP ~]# perf stat -I 1000 -a -A -e probe:gro_cell_poll
40.085714389 CPU0 13,607 probe:gro_cell_poll
40.085714389 CPU1 13,188 probe:gro_cell_poll
40.085714389 CPU2 12,913 probe:gro_cell_poll
40.085714389 CPU3 12,790 probe:gro_cell_poll
40.085714389 CPU4 13,395 probe:gro_cell_poll
40.085714389 CPU5 13,121 probe:gro_cell_poll
40.085714389 CPU6 11,083 probe:gro_cell_poll
40.085714389 CPU7 12,945 probe:gro_cell_poll
40.085714389 CPU8 13,704 probe:gro_cell_poll
40.085714389 CPU9 13,514 probe:gro_cell_poll
40.085714389 CPU10 0 probe:gro_cell_poll
40.085714389 CPU11 0 probe:gro_cell_poll
40.085714389 CPU12 0 probe:gro_cell_poll
40.085714389 CPU13 0 probe:gro_cell_poll
40.085714389 CPU14 0 probe:gro_cell_poll
40.085714389 CPU15 0 probe:gro_cell_poll
40.085714389 CPU16 0 probe:gro_cell_poll
40.085714389 CPU17 0 probe:gro_cell_poll
40.085714389 CPU18 0 probe:gro_cell_poll
40.085714389 CPU19 0 probe:gro_cell_poll
40.085714389 CPU20 10,402 probe:gro_cell_poll
40.085714389 CPU21 12,312 probe:gro_cell_poll
40.085714389 CPU22 11,913 probe:gro_cell_poll
40.085714389 CPU23 12,964 probe:gro_cell_poll
40.085714389 CPU24 13,727 probe:gro_cell_poll
40.085714389 CPU25 12,943 probe:gro_cell_poll
40.085714389 CPU26 13,558 probe:gro_cell_poll
40.085714389 CPU27 12,676 probe:gro_cell_poll
40.085714389 CPU28 13,754 probe:gro_cell_poll
40.085714389 CPU29 13,379 probe:gro_cell_poll
40.085714389 CPU30 0 probe:gro_cell_poll
40.085714389 CPU31 0 probe:gro_cell_poll
40.085714389 CPU32 0 probe:gro_cell_poll
40.085714389 CPU33 0 probe:gro_cell_poll
40.085714389 CPU34 0 probe:gro_cell_poll
40.085714389 CPU35 0 probe:gro_cell_poll
40.085714389 CPU36 0 probe:gro_cell_poll
40.085714389 CPU37 0 probe:gro_cell_poll
40.085714389 CPU38 0 probe:gro_cell_poll
40.085714389 CPU39 0 probe:gro_cell_poll
^ permalink raw reply
* [PATCH v3 net-next 1/1] ip_tunnel: Create percpu gro_cell
From: Martin KaFai Lau @ 2015-01-16 18:11 UTC (permalink / raw)
To: netdev; +Cc: Eric Dumazet, kernel-team
In-Reply-To: <1421431860-1960597-1-git-send-email-kafai@fb.com>
In the ipip tunnel, the skb->queue_mapping is lost in ipip_rcv().
All skb will be queued to the same cell->napi_skbs. The
gro_cell_poll is pinned to one core under load. In production traffic,
we also see severe rx_dropped in the tunl iface and it is probably due to
this limit: skb_queue_len(&cell->napi_skbs) > netdev_max_backlog.
This patch is trying to alloc_percpu(struct gro_cell) and schedule
gro_cell_poll to process the skb in the same core.
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
include/net/gro_cells.h | 29 ++++++++++++-----------------
1 file changed, 12 insertions(+), 17 deletions(-)
diff --git a/include/net/gro_cells.h b/include/net/gro_cells.h
index 734d9b5..0f712c0 100644
--- a/include/net/gro_cells.h
+++ b/include/net/gro_cells.h
@@ -8,25 +8,23 @@
struct gro_cell {
struct sk_buff_head napi_skbs;
struct napi_struct napi;
-} ____cacheline_aligned_in_smp;
+};
struct gro_cells {
- unsigned int gro_cells_mask;
- struct gro_cell *cells;
+ struct gro_cell __percpu *cells;
};
static inline void gro_cells_receive(struct gro_cells *gcells, struct sk_buff *skb)
{
- struct gro_cell *cell = gcells->cells;
+ struct gro_cell *cell;
struct net_device *dev = skb->dev;
- if (!cell || skb_cloned(skb) || !(dev->features & NETIF_F_GRO)) {
+ if (!gcells->cells || skb_cloned(skb) || !(dev->features & NETIF_F_GRO)) {
netif_rx(skb);
return;
}
- if (skb_rx_queue_recorded(skb))
- cell += skb_get_rx_queue(skb) & gcells->gro_cells_mask;
+ cell = this_cpu_ptr(gcells->cells);
if (skb_queue_len(&cell->napi_skbs) > netdev_max_backlog) {
atomic_long_inc(&dev->rx_dropped);
@@ -72,15 +70,12 @@ static inline int gro_cells_init(struct gro_cells *gcells, struct net_device *de
{
int i;
- gcells->gro_cells_mask = roundup_pow_of_two(netif_get_num_default_rss_queues()) - 1;
- gcells->cells = kcalloc(gcells->gro_cells_mask + 1,
- sizeof(struct gro_cell),
- GFP_KERNEL);
+ gcells->cells = alloc_percpu(struct gro_cell);
if (!gcells->cells)
return -ENOMEM;
- for (i = 0; i <= gcells->gro_cells_mask; i++) {
- struct gro_cell *cell = gcells->cells + i;
+ for_each_possible_cpu(i) {
+ struct gro_cell *cell = per_cpu_ptr(gcells->cells, i);
skb_queue_head_init(&cell->napi_skbs);
netif_napi_add(dev, &cell->napi, gro_cell_poll, 64);
@@ -91,16 +86,16 @@ static inline int gro_cells_init(struct gro_cells *gcells, struct net_device *de
static inline void gro_cells_destroy(struct gro_cells *gcells)
{
- struct gro_cell *cell = gcells->cells;
int i;
- if (!cell)
+ if (!gcells->cells)
return;
- for (i = 0; i <= gcells->gro_cells_mask; i++,cell++) {
+ for_each_possible_cpu(i) {
+ struct gro_cell *cell = per_cpu_ptr(gcells->cells, i);
netif_napi_del(&cell->napi);
skb_queue_purge(&cell->napi_skbs);
}
- kfree(gcells->cells);
+ free_percpu(gcells->cells);
gcells->cells = NULL;
}
--
1.8.1
^ permalink raw reply related
* [PATCH net-next v2 2/2] net: bridge: reject DSA-enabled master netdevices as bridge members
From: Florian Fainelli @ 2015-01-16 17:56 UTC (permalink / raw)
To: netdev; +Cc: davem, stephen, kaber, bridge, buytenh, Florian Fainelli
In-Reply-To: <1421430962-18119-1-git-send-email-f.fainelli@gmail.com>
DSA-enabled master network devices with a switch tagging protocol should
strip the protocol specific format before handing the frame over to
higher layer.
When adding such a DSA master network device as a bridge member, we go
through the following code path when receiving a frame:
__netif_receive_skb_core
-> first ptype check against ptype_all is not returning any
handler for this skb
-> check and invoke rx_handler:
-> deliver frame to the bridge layer: br_handle_frame
DSA registers a ptype handler with the fake ETH_XDSA ethertype, which is
called *after* the bridge-layer rx_handler has run. br_handle_frame()
tries to parse the frame it received from the DSA master network device,
and will not be able to match any of its conditions and jumps straight
at the end of the end of br_handle_frame() and returns
RX_HANDLER_CONSUMED there.
Since we returned RX_HANDLER_CONSUMED, __netif_receive_skb_core() stops
RX processing for this frame and returns NET_RX_SUCCESS, so we never get
a chance to call our switch tag packet processing logic and deliver
frames to the DSA slave network devices, and so we do not get any
functional bridge members at all.
Instead of cluttering the bridge receive path with DSA-specific checks,
and rely on assumptions about how __netif_receive_skb_core() is
processing frames, we simply deny adding the DSA master network device
(conduit interface) as a bridge member, leaving only the slave DSA
network devices to be bridge members, since those will work correctly in
all circumstances.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
net/bridge/br_if.c | 10 ++++++++--
1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/net/bridge/br_if.c b/net/bridge/br_if.c
index 81e49fb73169..b087d278c679 100644
--- a/net/bridge/br_if.c
+++ b/net/bridge/br_if.c
@@ -436,10 +436,16 @@ int br_add_if(struct net_bridge *br, struct net_device *dev)
int err = 0;
bool changed_addr;
- /* Don't allow bridging non-ethernet like devices */
+ /* Don't allow bridging non-ethernet like devices, or DSA-enabled
+ * master network devices since the bridge layer rx_handler prevents
+ * the DSA fake ethertype handler to be invoked, so we do not strip off
+ * the DSA switch tag protocol header and the bridge layer just return
+ * RX_HANDLER_CONSUMED, stopping RX processing for these frames.
+ */
if ((dev->flags & IFF_LOOPBACK) ||
dev->type != ARPHRD_ETHER || dev->addr_len != ETH_ALEN ||
- !is_valid_ether_addr(dev->dev_addr))
+ !is_valid_ether_addr(dev->dev_addr) ||
+ netdev_uses_dsa(dev))
return -EINVAL;
/* No bridging of bridges */
--
2.1.0
^ permalink raw reply related
* [PATCH net-next v2 1/2] net: ipv4: handle DSA enabled master network devices
From: Florian Fainelli @ 2015-01-16 17:56 UTC (permalink / raw)
To: netdev; +Cc: davem, stephen, kaber, bridge, buytenh, Florian Fainelli
In-Reply-To: <1421430962-18119-1-git-send-email-f.fainelli@gmail.com>
The logic to configure a network interface for kernel IP
auto-configuration is very simplistic, and does not handle the case
where a device is stacked onto another such as with DSA. This causes the
kernel not to open and configure the master network device in a DSA
switch tree, and therefore slave network devices using this master
network devices as conduit device cannot be open.
This restriction comes from a check in net/dsa/slave.c, which is
basically checking the master netdev flags for IFF_UP and returns
-ENETDOWN if it is not the case.
Automatically bringing-up DSA master network devices allows DSA slave
network devices to be used as valid interfaces for e.g: NFS root booting
by allowing kernel IP autoconfiguration to succeed on these interfaces.
On the reverse path, make sure we do not attempt to close a DSA-enabled
device as this would implicitely prevent the slave DSA network device
from operating.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
Changes in v2:
- fixed typo in first comment hunk
- remove spurious newline change
- removed second comment, the code is obvious enough
net/ipv4/ipconfig.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/net/ipv4/ipconfig.c b/net/ipv4/ipconfig.c
index 7fa18bc7e47f..b26376ef87f6 100644
--- a/net/ipv4/ipconfig.c
+++ b/net/ipv4/ipconfig.c
@@ -209,9 +209,9 @@ static int __init ic_open_devs(void)
last = &ic_first_dev;
rtnl_lock();
- /* bring loopback device up first */
+ /* bring loopback and DSA master network devices up first */
for_each_netdev(&init_net, dev) {
- if (!(dev->flags & IFF_LOOPBACK))
+ if (!(dev->flags & IFF_LOOPBACK) && !netdev_uses_dsa(dev))
continue;
if (dev_change_flags(dev, dev->flags | IFF_UP) < 0)
pr_err("IP-Config: Failed to open %s\n", dev->name);
@@ -306,7 +306,7 @@ static void __init ic_close_devs(void)
while ((d = next)) {
next = d->next;
dev = d->dev;
- if (dev != ic_dev) {
+ if (dev != ic_dev && !netdev_uses_dsa(dev)) {
DBG(("IP-Config: Downing %s\n", dev->name));
dev_change_flags(dev, d->flags);
}
--
2.1.0
^ permalink raw reply related
* [PATCH net-next v2 0/2] net: DSA fixes for bridge and ip-autoconf
From: Florian Fainelli @ 2015-01-16 17:56 UTC (permalink / raw)
To: netdev; +Cc: davem, stephen, kaber, bridge, buytenh, Florian Fainelli
Hi David,
These two patches address some real world use cases of the DSA master and slave
network devices.
You have already seen patch 1 previously and you rejected it since my
explanations were not good enough to provide a justification as to why it is
useful, hopefully this time my explanation is better.
Patch 2 solves a different, yet very real problem as well at the bridge layer
when using DSA network devices.
Florian Fainelli (2):
net: ipv4: handle DSA enabled master network devices
net: bridge: reject DSA-enabled master netdevices as bridge members
net/bridge/br_if.c | 10 ++++++++--
net/ipv4/ipconfig.c | 6 +++---
2 files changed, 11 insertions(+), 5 deletions(-)
--
2.1.0
^ permalink raw reply
* [PATCH net-next] mii: Handle link state changes for forced modes in mii_check_media()
From: Ben Hutchings @ 2015-01-16 17:55 UTC (permalink / raw)
To: ct178-internal, David S. Miller; +Cc: netdev, linux-kernel
mii_check_media() does not update the link (carrier) state or log link
changes when the link mode is forced. Drivers using the mii library
must do this themselves, but most of them do not.
Instead of changing them all, provide a sensible default behaviour
similar to mii_check_link() when the mode is forced.
via-rhine depends on it being a no-op in this case, so make its call
to mii_check_media() conditional.
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
---
I've only tested this with the asix driver, which is where I found the
bug. I reviewed all the other drivers calling mii_check_media().
Ben.
drivers/net/ethernet/amd/pcnet32.c | 2 +-
drivers/net/ethernet/via/via-rhine.c | 3 ++-
drivers/net/mii.c | 12 +++++++-----
3 files changed, 10 insertions(+), 7 deletions(-)
diff --git a/drivers/net/ethernet/amd/pcnet32.c b/drivers/net/ethernet/amd/pcnet32.c
index e2e3aaf..11d6e65 100644
--- a/drivers/net/ethernet/amd/pcnet32.c
+++ b/drivers/net/ethernet/amd/pcnet32.c
@@ -2806,7 +2806,7 @@ static void pcnet32_check_media(struct net_device *dev, int verbose)
/*
* Check for loss of link and link establishment.
- * Can not use mii_check_media because it does nothing if mode is forced.
+ * Could possibly be changed to use mii_check_media instead.
*/
static void pcnet32_watchdog(struct net_device *dev)
diff --git a/drivers/net/ethernet/via/via-rhine.c b/drivers/net/ethernet/via/via-rhine.c
index 0ac7610..b3b7f5a 100644
--- a/drivers/net/ethernet/via/via-rhine.c
+++ b/drivers/net/ethernet/via/via-rhine.c
@@ -1326,7 +1326,8 @@ static void rhine_check_media(struct net_device *dev, unsigned int init_media)
struct rhine_private *rp = netdev_priv(dev);
void __iomem *ioaddr = rp->base;
- mii_check_media(&rp->mii_if, netif_msg_link(rp), init_media);
+ if (!mii->force_media)
+ mii_check_media(&rp->mii_if, netif_msg_link(rp), init_media);
if (rp->mii_if.full_duplex)
iowrite8(ioread8(ioaddr + ChipCmd1) | Cmd1FDuplex,
diff --git a/drivers/net/mii.c b/drivers/net/mii.c
index 4a99c39..993570b 100644
--- a/drivers/net/mii.c
+++ b/drivers/net/mii.c
@@ -302,7 +302,7 @@ void mii_check_link (struct mii_if_info *mii)
}
/**
- * mii_check_media - check the MII interface for a duplex change
+ * mii_check_media - check the MII interface for a carrier/speed/duplex change
* @mii: the MII interface
* @ok_to_print: OK to print link up/down messages
* @init_media: OK to save duplex mode in @mii
@@ -318,10 +318,6 @@ unsigned int mii_check_media (struct mii_if_info *mii,
int advertise, lpa, media, duplex;
int lpa2 = 0;
- /* if forced media, go no further */
- if (mii->force_media)
- return 0; /* duplex did not change */
-
/* check current and old link status */
old_carrier = netif_carrier_ok(mii->dev) ? 1 : 0;
new_carrier = (unsigned int) mii_link_ok(mii);
@@ -345,6 +341,12 @@ unsigned int mii_check_media (struct mii_if_info *mii,
*/
netif_carrier_on(mii->dev);
+ if (mii->force_media) {
+ if (ok_to_print)
+ netdev_info(mii->dev, "link up\n");
+ return 0; /* duplex did not change */
+ }
+
/* get MII advertise and LPA values */
if ((!init_media) && (mii->advertising))
advertise = mii->advertising;
--
1.7.10.4
^ permalink raw reply related
* Re: [PATCH 1/6] selftests: Introduce minimal shared logic for running tests
From: Shuah Khan @ 2015-01-16 17:53 UTC (permalink / raw)
To: Michael Ellerman, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Cc: mmarek-AlSwsSmVLrQ, gregkh-hQyY1W1yCW8ekmWlsbkhG0B+6BGkLq7r,
akpm-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b,
rostedt-nx8X9YLhiw1AfugRpC6u6w, mingo-H+wXaHxf7aLQT0dZR+AlfA,
davem-fT/PcQaiUtIeIZ0/mPfg9Q, keescook-F7+t8E8rja9g9hUCZPvPmw,
tranmanphong-Re5JQEeQqe8AvxtiuMwx3w, cov-sgV2jX0FEOL9JmXXK+q4OQ,
dh.herrmann-Re5JQEeQqe8AvxtiuMwx3w, hughd-hpIqsD4AKlfQT0dZR+AlfA,
bobby.prani-Re5JQEeQqe8AvxtiuMwx3w,
serge.hallyn-GeWIH/nMZzLQT0dZR+AlfA,
ebiederm-aS9lmoZGLiVWk0Htik3J/w, tim.bird-/MT0OVThwyLZJqsBc5GL+g,
josh-iaAMLnmF4UmaiuxdJuQwMA, koct9i-Re5JQEeQqe8AvxtiuMwx3w,
linux-kbuild-u79uwXL29TY76Z2rM5mHXA,
linux-api-u79uwXL29TY76Z2rM5mHXA, netdev-u79uwXL29TY76Z2rM5mHXA
In-Reply-To: <1420794375-31881-1-git-send-email-mpe-Gsx/Oe8HsFggBc27wqDAHg@public.gmane.org>
On 01/09/2015 02:06 AM, Michael Ellerman wrote:
> This adds a Make include file which most selftests can then include to
> get the run_tests logic.
>
> On its own this has the advantage of some reduction in repetition, and
> also means the pass/fail message is defined in fewer places.
>
> However the key advantage is it will allow us to implement install very
> simply in a subsequent patch.
>
> The default implementation just executes each program in $(TEST_PROGS).
>
> We use a variable to hold the default implementation of $(RUN_TESTS)
> because that gives us a clean way to override it if necessary, ie. using
> override. The mount, memory-hotplug and mqueue tests use that to provide
> a different implementation.
>
> Tests are not run via /bin/bash, so if they are scripts they must be
> executable, we add u+x to several.
>
> Signed-off-by: Michael Ellerman <mpe-Gsx/Oe8HsFggBc27wqDAHg@public.gmane.org>
I like the shared logic approach in general provided it leaves the
flexibility to not use the shared logic if a test have the need to
do so. This series requires some patch planning. shared logic patch
followed by individual test patches as opposed a single patch.
I would like to see the shared logic work done on top of my patch v4
series.
thanks,
-- Shuah
--
Shuah Khan
Sr. Linux Kernel Developer
Open Source Innovation Group
Samsung Research America (Silicon Valley)
shuahkh-JPH+aEBZ4P+UEJcrhfAQsw@public.gmane.org | (970) 217-8978
^ permalink raw reply
* [PATCH net 2/2] sh_eth: Fix ethtool operation crash when net device is down
From: Ben Hutchings @ 2015-01-16 17:51 UTC (permalink / raw)
To: ct178-internal, David S. Miller
Cc: netdev, linux-kernel, Nobuhiro Iwamatsu, Mitsuhiro Kimura,
Hisashi Nakamura, Yoshihiro Kaneko
In-Reply-To: <1421430592.1222.190.camel@xylophone.i.decadent.org.uk>
The driver connects and disconnects the PHY device whenever the
net device is brought up and down. The ethtool get_settings,
set_settings and nway_reset operations will dereference a null
or dangling pointer if called while it is down.
I think it would be preferable to keep the PHY connected, but there
may be good reasons not to.
As an immediate fix for this bug:
- Set the phydev pointer to NULL after disconnecting the PHY
- Change those three operations to return -ENODEV while the PHY is
not connected
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
---
drivers/net/ethernet/renesas/sh_eth.c | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/drivers/net/ethernet/renesas/sh_eth.c b/drivers/net/ethernet/renesas/sh_eth.c
index 0c4d5b5..28e3822 100644
--- a/drivers/net/ethernet/renesas/sh_eth.c
+++ b/drivers/net/ethernet/renesas/sh_eth.c
@@ -1827,6 +1827,9 @@ static int sh_eth_get_settings(struct net_device *ndev,
unsigned long flags;
int ret;
+ if (!mdp->phydev)
+ return -ENODEV;
+
spin_lock_irqsave(&mdp->lock, flags);
ret = phy_ethtool_gset(mdp->phydev, ecmd);
spin_unlock_irqrestore(&mdp->lock, flags);
@@ -1841,6 +1844,9 @@ static int sh_eth_set_settings(struct net_device *ndev,
unsigned long flags;
int ret;
+ if (!mdp->phydev)
+ return -ENODEV;
+
spin_lock_irqsave(&mdp->lock, flags);
/* disable tx and rx */
@@ -1875,6 +1881,9 @@ static int sh_eth_nway_reset(struct net_device *ndev)
unsigned long flags;
int ret;
+ if (!mdp->phydev)
+ return -ENODEV;
+
spin_lock_irqsave(&mdp->lock, flags);
ret = phy_start_aneg(mdp->phydev);
spin_unlock_irqrestore(&mdp->lock, flags);
@@ -2184,6 +2193,7 @@ static int sh_eth_close(struct net_device *ndev)
if (mdp->phydev) {
phy_stop(mdp->phydev);
phy_disconnect(mdp->phydev);
+ mdp->phydev = NULL;
}
free_irq(ndev->irq, ndev);
--
1.7.10.4
^ permalink raw reply related
* [PATCH net 1/2] sh_eth: Fix promiscuous mode on chips without TSU
From: Ben Hutchings @ 2015-01-16 17:51 UTC (permalink / raw)
To: ct178-internal, David S. Miller
Cc: netdev, linux-kernel, Nobuhiro Iwamatsu, Mitsuhiro Kimura,
Hisashi Nakamura, Yoshihiro Kaneko
In-Reply-To: <1421430592.1222.190.camel@xylophone.i.decadent.org.uk>
Currently net_device_ops::set_rx_mode is only implemented for
chips with a TSU (multiple address table). However we do need
to turn the PRM (promiscuous) flag on and off for other chips.
- Remove the unlikely() from the TSU functions that we may safely
call for chips without a TSU
- Make setting of the MCT flag conditional on the tsu capability flag
- Rename sh_eth_set_multicast_list() to sh_eth_set_rx_mode() and plumb
it into both net_device_ops structures
- Remove the previously-unreachable branch in sh_eth_rx_mode() that
would otherwise reset the flags to defaults for non-TSU chips
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
---
drivers/net/ethernet/renesas/sh_eth.c | 18 +++++++++---------
1 file changed, 9 insertions(+), 9 deletions(-)
diff --git a/drivers/net/ethernet/renesas/sh_eth.c b/drivers/net/ethernet/renesas/sh_eth.c
index 167737f..0c4d5b5 100644
--- a/drivers/net/ethernet/renesas/sh_eth.c
+++ b/drivers/net/ethernet/renesas/sh_eth.c
@@ -2417,7 +2417,7 @@ static int sh_eth_tsu_purge_all(struct net_device *ndev)
struct sh_eth_private *mdp = netdev_priv(ndev);
int i, ret;
- if (unlikely(!mdp->cd->tsu))
+ if (!mdp->cd->tsu)
return 0;
for (i = 0; i < SH_ETH_TSU_CAM_ENTRIES; i++) {
@@ -2440,7 +2440,7 @@ static void sh_eth_tsu_purge_mcast(struct net_device *ndev)
void *reg_offset = sh_eth_tsu_get_offset(mdp, TSU_ADRH0);
int i;
- if (unlikely(!mdp->cd->tsu))
+ if (!mdp->cd->tsu)
return;
for (i = 0; i < SH_ETH_TSU_CAM_ENTRIES; i++, reg_offset += 8) {
@@ -2450,8 +2450,8 @@ static void sh_eth_tsu_purge_mcast(struct net_device *ndev)
}
}
-/* Multicast reception directions set */
-static void sh_eth_set_multicast_list(struct net_device *ndev)
+/* Update promiscuous flag and multicast filter */
+static void sh_eth_set_rx_mode(struct net_device *ndev)
{
struct sh_eth_private *mdp = netdev_priv(ndev);
u32 ecmr_bits;
@@ -2462,7 +2462,9 @@ static void sh_eth_set_multicast_list(struct net_device *ndev)
/* Initial condition is MCT = 1, PRM = 0.
* Depending on ndev->flags, set PRM or clear MCT
*/
- ecmr_bits = (sh_eth_read(ndev, ECMR) & ~ECMR_PRM) | ECMR_MCT;
+ ecmr_bits = sh_eth_read(ndev, ECMR) & ~ECMR_PRM;
+ if (mdp->cd->tsu)
+ ecmr_bits |= ECMR_MCT;
if (!(ndev->flags & IFF_MULTICAST)) {
sh_eth_tsu_purge_mcast(ndev);
@@ -2491,9 +2493,6 @@ static void sh_eth_set_multicast_list(struct net_device *ndev)
}
}
}
- } else {
- /* Normal, unicast/broadcast-only mode. */
- ecmr_bits = (ecmr_bits & ~ECMR_PRM) | ECMR_MCT;
}
/* update the ethernet mode */
@@ -2701,6 +2700,7 @@ static const struct net_device_ops sh_eth_netdev_ops = {
.ndo_stop = sh_eth_close,
.ndo_start_xmit = sh_eth_start_xmit,
.ndo_get_stats = sh_eth_get_stats,
+ .ndo_set_rx_mode = sh_eth_set_rx_mode,
.ndo_tx_timeout = sh_eth_tx_timeout,
.ndo_do_ioctl = sh_eth_do_ioctl,
.ndo_validate_addr = eth_validate_addr,
@@ -2713,7 +2713,7 @@ static const struct net_device_ops sh_eth_netdev_ops_tsu = {
.ndo_stop = sh_eth_close,
.ndo_start_xmit = sh_eth_start_xmit,
.ndo_get_stats = sh_eth_get_stats,
- .ndo_set_rx_mode = sh_eth_set_multicast_list,
+ .ndo_set_rx_mode = sh_eth_set_rx_mode,
.ndo_vlan_rx_add_vid = sh_eth_vlan_rx_add_vid,
.ndo_vlan_rx_kill_vid = sh_eth_vlan_rx_kill_vid,
.ndo_tx_timeout = sh_eth_tx_timeout,
--
1.7.10.4
^ permalink raw reply related
* [PATCH net 0/2] sh_eth fixes
From: Ben Hutchings @ 2015-01-16 17:49 UTC (permalink / raw)
To: David S. Miller
Cc: netdev, linux-kernel, Nobuhiro Iwamatsu, Mitsuhiro Kimura,
Hisashi Nakamura, Yoshihiro Kaneko
I'm currently looking at Ethernet support on the R-Car H2 chip,
reviewing and testing the sh_eth driver. Here are fixes for two fairly
obvious bugs in the driver; I will probably have some more later.
These are not tested on any of the other supported chips.
Ben.
Ben Hutchings (2):
sh_eth: Fix promiscuous mode on chips without TSU
sh_eth: Fix ethtool operation crash when net device is down
drivers/net/ethernet/renesas/sh_eth.c | 28 +++++++++++++++++++---------
1 file changed, 19 insertions(+), 9 deletions(-)
--
1.7.10.4
^ permalink raw reply
* Re: [PATCH 2/6] selftests: Add install target
From: Shuah Khan @ 2015-01-16 17:46 UTC (permalink / raw)
To: Michael Ellerman, linux-kernel
Cc: mmarek, gregkh, akpm, rostedt, mingo, davem, keescook,
tranmanphong, cov, dh.herrmann, hughd, bobby.prani, serge.hallyn,
ebiederm, tim.bird, josh, koct9i, linux-kbuild, linux-api, netdev
In-Reply-To: <1420794375-31881-2-git-send-email-mpe@ellerman.id.au>
On 01/09/2015 02:06 AM, Michael Ellerman wrote:
> This adds make install support to selftests. The basic usage is:
>
> $ cd tools/testing/selftests
> $ make install
>
> That installs into tools/testing/selftests/install, which can then be
> copied where ever necessary.
>
> The install destination is also configurable using eg:
>
> $ INSTALL_PATH=/mnt/selftests make install
Please see my response to [PATCH 4/6] kbuild: add a new
kselftest_install make target to install selftests
These are addressed by the current approach to use existing
INSTALL_MOD_PATH.
>
> The implementation uses two targets in the child makefiles. The first
> "install" is expected to install all files into $(INSTALL_PATH).
>
> The second, "emit_tests", is expected to emit the test instructions (ie.
> bash script) on stdout. Separating this from install means the child
> makefiles need no knowledge of the location of the test script.
Yes. Removing the need for individual makefiles to know the install
path details would be a good improvement to make on top of my patch v4
series.
thanks,
-- Shuah
--
Shuah Khan
Sr. Linux Kernel Developer
Open Source Innovation Group
Samsung Research America (Silicon Valley)
shuahkh@osg.samsung.com | (970) 217-8978
^ permalink raw reply
* Re: [PATCH 7/9] rhashtable: Per bucket locks & deferred expansion/shrinking
From: Thomas Graf @ 2015-01-16 16:53 UTC (permalink / raw)
To: David Laight
Cc: 'Patrick McHardy', davem@davemloft.net,
netdev@vger.kernel.org, herbert@gondor.apana.org.au,
paulmck@linux.vnet.ibm.com, edumazet@google.com,
john.r.fastabend@intel.com, josh@joshtriplett.org,
netfilter-devel@vger.kernel.org
In-Reply-To: <063D6719AE5E284EB5DD2968C1650D6D1CACADAF@AcuExch.aculab.com>
On 01/16/15 at 04:43pm, David Laight wrote:
> The walker is unlikely to see items that get inserted early in the hash
> table even without a resize.
I don't follow, you have to explain this statement.
Walkers which don't want to see duplicates or miss entries should
just take the mutex.
^ permalink raw reply
* Re: [PATCH 7/9] rhashtable: Per bucket locks & deferred expansion/shrinking
From: Patrick McHardy @ 2015-01-16 16:51 UTC (permalink / raw)
To: Thomas Graf
Cc: davem, netdev, kernel, herbert, paulmck, edumazet,
john.r.fastabend, josh, netfilter-devel
In-Reply-To: <20150116163826.GD15052@casper.infradead.org>
On 16.01, Thomas Graf wrote:
> On 01/16/15 at 04:32pm, Patrick McHardy wrote:
> > If we have to block, the dumper if of course preferred. Taking the
> > mutex should do fine I guess?
>
> That will work, it will also ensure that the walker doesn't see the
> pre unzip state which would cause entries to be duplicated in the
> walker.
>
> > I suppose walking both tables without any races would be rather
> > complicated.
>
> Very interested if you can come up with something ;-)
For now I'll stick to the mutex. Thanks!
^ permalink raw reply
* RE: [PATCH 7/9] rhashtable: Per bucket locks & deferred expansion/shrinking
From: David Laight @ 2015-01-16 16:43 UTC (permalink / raw)
To: 'Patrick McHardy', Thomas Graf
Cc: davem@davemloft.net, netdev@vger.kernel.org,
kernel@vger.kernel.org, herbert@gondor.apana.org.au,
paulmck@linux.vnet.ibm.com, edumazet@google.com,
john.r.fastabend@intel.com, josh@joshtriplett.org,
netfilter-devel@vger.kernel.org
In-Reply-To: <20150116163202.GJ30132@acer.localdomain>
From: Patrick McHardy
> On 16.01, Thomas Graf wrote:
> > On 01/16/15 at 04:03pm, Patrick McHardy wrote:
> > > On 16.01, Thomas Graf wrote:
> > > > A walker may not see insertions that occur after the walker was started
> > > > if resizing is enabled. Is that a problem for nftables?
> > >
> > > No, that would be Ok. The case I'm wondering about is:
> > >
> > > - insertion
> > > - resize starts
> > > - another insertion
> > > - walker, resize not finished yet
> >
> > Correct, walker may not see "another insertion". The window for this
> > behavior to occur is not the full resize operation, just the linking
> > period, but it may occur. The length of the window is typically
> > equal to a grace period.
> >
> > We can provide a synchronization function to block the dumper or the
> > insertion/removal until the linking is complete. The latter would
> > give the old runtime behaviour back (variable runtime of insert),
> > the blocked dumper might be preferred. What do you think?
>
> If we have to block, the dumper if of course preferred. Taking the
> mutex should do fine I guess?
>
> I suppose walking both tables without any races would be rather
> complicated.
The walker is unlikely to see items that get inserted early in the hash
table even without a resize.
I'd be more worried about the walker missing big blocks of entries or
getting duplicate entries during a resize.
This might be a problem if the walker is a user-process table dump,
in which case you can't assume it will finish in any finite time.
David
^ permalink raw reply
* Re: NETDEV WATCHDOG: internal(r8152): transmit queue 0 timed out
From: sean darcy @ 2015-01-16 16:40 UTC (permalink / raw)
To: users; +Cc: netdev, linux-usb
In-Reply-To: <54B8FF7E.2010901@gmail.com>
On 01/16/2015 07:09 AM, poma wrote:
> On 16.01.2015 10:37, Hayes Wang wrote:
>> poma [mailto:pomidorabelisima@gmail.com]
>>> Sent: Friday, January 16, 2015 4:25 PM
>> [...]
>>>> This looks like a USB problem. Is there a way to get usb (or
>>>> NetworkManager) to reinitialize the driver when this happens?
>>>
>>> I would ask these people for advice, therefore.
>>
>> Our hw engineers need to analyse the behavior of the device.
>> However, I don't think you have such instrument to provide
>> the required information. If we don't know the reason, we
>> couldn't give you the proper solution. Besides, your solution
>> would work if and only if reloading the driver is helpful.
>>
>> The issue have to debug from the hardware, and I have no idea
>> about what the software could do before analysing the hw. Maybe
>> you could try the following driver first to check if it is useful.
>>
>> http://www.realtek.com.tw/downloads/downloadsView.aspx?Langid=2&PNid=13&PFid=56&Level=5&Conn=4&DownTypeID=3&GetDown=false
>>
>> Best Regards,
>> Hayes
>>
>
> Thanks for your response, Mr. Hayes.
>
> Mr. Sean, please download and check if "timeout" is still present with built RTL8153 module from REALTEK site, as Mr. Hayes proposed.
> http://www.realtek.com.tw/downloads/downloadsView.aspx?Langid=2&PNid=13&PFid=56&Level=5&Conn=4&DownTypeID=3&GetDown=false#2
> r8152.53-2.03.0.tar.bz2
>
> Procedure - should be equal for both, Fedora 21 & 20:
>
> $ uname -r
> 3.17.8-300.fc21.x86_64
>
> $ su -c 'yum install kernel-devel'
>
> $ tar xf r8152.53-2.03.0.tar.bz2
> $ cd r8152-2.03.0/
> $ make
> $ su
>
> # cp 50-usb-realtek-net.rules /etc/udev/rules.d/
> # udevadm trigger --action=add
>
> # modprobe -rv r8152
> # cp r8152.ko /lib/modules/$(uname -r)/updates/
> # depmod
> # modprobe -v r8152
>
>
> poma
>
OK, will do.
In the meantime, here's the kernel log showing the timeout, the removal
of the usb ethernet adapter, and it's reinsertion:
Jan 15 00:15:18 kernel: r8152 2-1:1.0 internal: Tx timeout
Jan 15 00:15:23 kernel: r8152 2-1:1.0 internal: Tx timeout
Jan 15 00:15:29 kernel: usb 2-1: USB disconnect, device number 4
Jan 15 00:15:29 kernel: r8152 2-1:1.0 internal: Tx status -108
.....................
Jan 15 00:15:29 kernel: r8152 2-1:1.0 internal: Tx status -108
Jan 15 00:15:29 kernel: r8152 2-1:1.0 internal: Tx status -108
Jan 15 00:15:31 kernel: usb 2-1: new high-speed USB device number 5
using ehci-pci
Jan 15 00:15:31 kernel: usb 2-1: New USB device found, idVendor=0bda,
idProduct=8152
Jan 15 00:15:31 kernel: usb 2-1: New USB device strings: Mfr=1,
Product=2, SerialNumber=3
Jan 15 00:15:31 kernel: usb 2-1: Product: USB 10/100 LAN
Jan 15 00:15:31 kernel: usb 2-1: Manufacturer: CE-LINK
Jan 15 00:15:31 kernel: usb 2-1: SerialNumber: ...........
Jan 15 00:15:31 kernel: usb 2-1: reset high-speed USB device number 5
using ehci-pci
Jan 15 00:15:31 kernel: r8152 2-1:1.0 eth0: v1.06.1 (2014/10/01)
Jan 15 00:15:32 kernel: r8152 2-1:1.0 internal: renamed from eth0
Jan 15 00:15:32 systemd-udevd[12869]: renamed network interface eth0 to
internal
Jan 15 00:15:32 kernel: IPv6: ADDRCONF(NETDEV_UP): internal: link is not
ready
Jan 15 00:15:34 kernel: IPv6: ADDRCONF(NETDEV_CHANGE): internal: link
becomes ready
--
users mailing list
users@lists.fedoraproject.org
To unsubscribe or change subscription options:
https://admin.fedoraproject.org/mailman/listinfo/users
Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
Have a question? Ask away: http://ask.fedoraproject.org
^ permalink raw reply
* Re: [PATCH 7/9] rhashtable: Per bucket locks & deferred expansion/shrinking
From: Thomas Graf @ 2015-01-16 16:38 UTC (permalink / raw)
To: Patrick McHardy
Cc: davem, netdev, kernel, herbert, paulmck, edumazet,
john.r.fastabend, josh, netfilter-devel
In-Reply-To: <20150116163202.GJ30132@acer.localdomain>
On 01/16/15 at 04:32pm, Patrick McHardy wrote:
> If we have to block, the dumper if of course preferred. Taking the
> mutex should do fine I guess?
That will work, it will also ensure that the walker doesn't see the
pre unzip state which would cause entries to be duplicated in the
walker.
> I suppose walking both tables without any races would be rather
> complicated.
Very interested if you can come up with something ;-)
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox