Skip to contentSkip to conversation

Enhance current features in ena driver

PATCH In review [email protected] 1 participant 15 messages Apr 22, 2020

Subsystems (3)
lore
Share & tools
lore
#2 lore raw

PATCH 2/13 net: ena: avoid unnecessary admin command when RSS function set fails

In reply to [email protected]

From: Arthur Kiyanovski <[email protected]>

Currently when ena_set_hash_function() fails the hash function is restored to the previous value by calling an admin command to get the hash function from the device.

In this commit we avoid the admin command, by saving the previous hash function before calling ena_set_hash_function() and using this previous value to restore the hash function in case of failure of ena_set_hash_function().

---
 drivers/net/ethernet/amazon/ena/ena_com.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/net/ethernet/amazon/ena/ena_com.c +3 -1
index 07b0f396d3c2..66edc86c41c9 100644
@@ -2286,6 +2286,7 @@ int ena_com_fill_hash_function(struct ena_com_dev *ena_dev,
 	struct ena_admin_get_feat_resp get_resp;
 	struct ena_admin_feature_rss_flow_hash_control *hash_key =
 		rss->hash_key;
+	enum ena_admin_hash_functions old_func;
 	int rc;
 
 	/* Make sure size is a mult of DWs */
@@ -2325,12 +2326,13 @@ int ena_com_fill_hash_function(struct ena_com_dev *ena_dev,
 		return -EINVAL;
 	}
 
+	old_func = rss->hash_func;
 	rss->hash_func = func;
 	rc = ena_com_set_hash_function(ena_dev);
 
 	/* Restore the old function */
 	if (unlikely(rc))
-		ena_com_get_hash_function(ena_dev, NULL, NULL);
+		rss->hash_func = old_func;
 
 	return rc;
 }
Signature · 1 line
2.24.1.AMZN
#1 lore raw

PATCH 1/13 net: ena: fix error returning in ena_com_get_hash_function()

In reply to [email protected]

From: Arthur Kiyanovski <[email protected]>

In case the "func" parameter is NULL we now return "-EINVAL". This shouldn't happen in general, but when it does happen, this is the proper way to handle it.

We also check func for NULL in the beginning of the function, as there is no reason to do all the work and realize in the end of the function it was useless.

Fixes 1738cd3ed342 ("net: ena: Add a driver for Amazon Elastic Network Adapters (ENA)")
---
 drivers/net/ethernet/amazon/ena/ena_com.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/net/ethernet/amazon/ena/ena_com.c +4 -2
index a250046b8e18..07b0f396d3c2 100644
@@ -2345,6 +2345,9 @@ int ena_com_get_hash_function(struct ena_com_dev *ena_dev,
 		rss->hash_key;
 	int rc;
 
+	if (unlikely(!func))
+		return -EINVAL;
+
 	rc = ena_com_get_feature_ex(ena_dev, &get_resp,
 				    ENA_ADMIN_RSS_HASH_FUNCTION,
 				    rss->hash_key_dma_addr,
@@ -2357,8 +2360,7 @@ int ena_com_get_hash_function(struct ena_com_dev *ena_dev,
 	if (rss->hash_func)
 		rss->hash_func--;
 
-	if (func)
-		*func = rss->hash_func;
+	*func = rss->hash_func;
 
 	if (key)
 		memcpy(key, hash_key->key, (size_t)(hash_key->keys_num) << 2);
Signature · 1 line
2.24.1.AMZN
#9 lore raw

PATCH 9/13 net: ena: implement ena_com_get_admin_polling_mode()

In reply to [email protected]

From: Arthur Kiyanovski <[email protected]>

Before this commit there was a function prototype named ena_com_get_ena_admin_polling_mode() that was never implemented.

This commit:
1. Changes the name of the function by removing the redundant double "ena_" in it.
2. Adds an implementation to the function.
3. Fixes a typo in the description of the function.

---
 drivers/net/ethernet/amazon/ena/ena_com.c | 5 +++++
 drivers/net/ethernet/amazon/ena/ena_com.h | 4 ++--
 2 files changed, 7 insertions(+), 2 deletions(-)
drivers/net/ethernet/amazon/ena/ena_com.c +5 -0
index b51bf62af11b..b1ce95717de9 100644
@@ -1659,6 +1659,11 @@ void ena_com_set_admin_polling_mode(struct ena_com_dev *ena_dev, bool polling)
 	ena_dev->admin_queue.polling = polling;
 }
 
+bool ena_com_get_admin_polling_mode(struct ena_com_dev *ena_dev)
+{
+	return ena_dev->admin_queue.polling;
+}
+
 void ena_com_set_admin_auto_polling_mode(struct ena_com_dev *ena_dev,
 					 bool polling)
 {
drivers/net/ethernet/amazon/ena/ena_com.h +2 -2
index e2e2fd1dc820..94986cdc0b1a 100644
@@ -501,7 +501,7 @@ bool ena_com_get_admin_running_state(struct ena_com_dev *ena_dev);
  */
 void ena_com_set_admin_polling_mode(struct ena_com_dev *ena_dev, bool polling);
 
-/* ena_com_set_admin_polling_mode - Get the admin completion queue polling mode
+/* ena_com_get_admin_polling_mode - Get the admin completion queue polling mode
  * @ena_dev: ENA communication layer struct
  *
  * Get the admin completion mode.
@@ -511,7 +511,7 @@ void ena_com_set_admin_polling_mode(struct ena_com_dev *ena_dev, bool polling);
  *
  * @return state
  */
-bool ena_com_get_ena_admin_polling_mode(struct ena_com_dev *ena_dev);
+bool ena_com_get_admin_polling_mode(struct ena_com_dev *ena_dev);
 
 /* ena_com_set_admin_auto_polling_mode - Enable autoswitch to polling mode
  * @ena_dev: ENA communication layer struct
Signature · 1 line
2.24.1.AMZN
#4 lore raw

PATCH 4/13 net: ena: change default RSS hash function to Toeplitz

In reply to [email protected]

From: Arthur Kiyanovski <[email protected]>

Currently in the driver we are setting the hash function to be CRC32. Starting with this commit we want to change the default behaviour so that we set the hash function to be Toeplitz instead.

---
 drivers/net/ethernet/amazon/ena/ena_netdev.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/amazon/ena/ena_netdev.c +1 -1
index 2cc765df8da3..6baafc3aebea 100644
@@ -3991,7 +3991,7 @@ static int ena_rss_init_default(struct ena_adapter *adapter)
 		}
 	}
 
-	rc = ena_com_fill_hash_function(ena_dev, ENA_ADMIN_CRC32, NULL,
+	rc = ena_com_fill_hash_function(ena_dev, ENA_ADMIN_TOEPLITZ, NULL,
 					ENA_HASH_KEY_SIZE, 0xFFFFFFFF);
 	if (unlikely(rc && (rc != -EOPNOTSUPP))) {
 		dev_err(dev, "Cannot fill hash function\n");
Signature · 1 line
2.24.1.AMZN
#3 lore raw

PATCH 3/13 net: ena: allow setting the hash function without changing the key

In reply to [email protected]

From: Sameeh Jubran <[email protected]>

Current code does not allow setting the hash function without changing the key. This commit enables it.

To achieve this we separate ena_com_get_hash_function() to 2 functions:
ena_com_get_hash_function() - which gets only the hash function, and ena_com_get_hash_key() - which gets only the hash key.

Also return 0 instead of rc at the end of ena_get_rxfh() since all previous operations succeeded.

---
 drivers/net/ethernet/amazon/ena/ena_com.c     | 13 ++++++++----
 drivers/net/ethernet/amazon/ena/ena_com.h     | 21 +++++++++++++------
 drivers/net/ethernet/amazon/ena/ena_ethtool.c | 12 +++++++----
 3 files changed, 32 insertions(+), 14 deletions(-)
drivers/net/ethernet/amazon/ena/ena_com.c +9 -4
index 66edc86c41c9..d428d0606166 100644
@@ -2338,13 +2338,10 @@ int ena_com_fill_hash_function(struct ena_com_dev *ena_dev,
 }
 
 int ena_com_get_hash_function(struct ena_com_dev *ena_dev,
-			      enum ena_admin_hash_functions *func,
-			      u8 *key)
+			      enum ena_admin_hash_functions *func)
 {
 	struct ena_rss *rss = &ena_dev->rss;
 	struct ena_admin_get_feat_resp get_resp;
-	struct ena_admin_feature_rss_flow_hash_control *hash_key =
-		rss->hash_key;
 	int rc;
 
 	if (unlikely(!func))
@@ -2364,6 +2361,14 @@ int ena_com_get_hash_function(struct ena_com_dev *ena_dev,
 
 	*func = rss->hash_func;
 
+	return 0;
+}
+
+int ena_com_get_hash_key(struct ena_com_dev *ena_dev, u8 *key)
+{
+	struct ena_admin_feature_rss_flow_hash_control *hash_key =
+		ena_dev->rss.hash_key;
+
 	if (key)
 		memcpy(key, hash_key->key, (size_t)(hash_key->keys_num) << 2);
 
drivers/net/ethernet/amazon/ena/ena_com.h +15 -6
index 469f298199a7..e2e2fd1dc820 100644
@@ -695,13 +695,11 @@ int ena_com_fill_hash_function(struct ena_com_dev *ena_dev,
  */
 int ena_com_set_hash_function(struct ena_com_dev *ena_dev);
 
-/* ena_com_get_hash_function - Retrieve the hash function and the hash key
- * from the device.
+/* ena_com_get_hash_function - Retrieve the hash function from the device.
  * @ena_dev: ENA communication layer struct
  * @func: hash function
- * @key: hash key
  *
- * Retrieve the hash function and the hash key from the device.
+ * Retrieve the hash function from the device.
  *
  * @note: If the caller called ena_com_fill_hash_function but didn't flash
  * it to the device, the new configuration will be lost.
@@ -709,9 +707,20 @@ int ena_com_set_hash_function(struct ena_com_dev *ena_dev);
  * @return: 0 on Success and negative value otherwise.
  */
 int ena_com_get_hash_function(struct ena_com_dev *ena_dev,
-			      enum ena_admin_hash_functions *func,
-			      u8 *key);
+			      enum ena_admin_hash_functions *func);
 
+/* ena_com_get_hash_key - Retrieve the hash key
+ * @ena_dev: ENA communication layer struct
+ * @key: hash key
+ *
+ * Retrieve the hash key.
+ *
+ * @note: If the caller called ena_com_fill_hash_key but didn't flash
+ * it to the device, the new configuration will be lost.
+ *
+ * @return: 0 on Success and negative value otherwise.
+ */
+int ena_com_get_hash_key(struct ena_com_dev *ena_dev, u8 *key);
 /* ena_com_fill_hash_ctrl - Fill RSS hash control
  * @ena_dev: ENA communication layer struct.
  * @proto: The protocol to configure.
drivers/net/ethernet/amazon/ena/ena_ethtool.c +8 -4
index 9cc28b4b2627..0c3a2f14387e 100644
@@ -672,7 +672,7 @@ static int ena_get_rxfh(struct net_device *netdev, u32 *indir, u8 *key,
 	/* We call this function in order to check if the device
 	 * supports getting/setting the hash function.
 	 */
-	rc = ena_com_get_hash_function(adapter->ena_dev, &ena_func, key);
+	rc = ena_com_get_hash_function(adapter->ena_dev, &ena_func);
 	if (rc) {
 		if (rc == -EOPNOTSUPP) {
 			key = NULL;
@@ -683,6 +683,10 @@ static int ena_get_rxfh(struct net_device *netdev, u32 *indir, u8 *key,
 		return rc;
 	}
 
+	rc = ena_com_get_hash_key(adapter->ena_dev, key);
+	if (rc)
+		return rc;
+
 	switch (ena_func) {
 	case ENA_ADMIN_TOEPLITZ:
 		func = ETH_RSS_HASH_TOP;
@@ -699,7 +703,7 @@ static int ena_get_rxfh(struct net_device *netdev, u32 *indir, u8 *key,
 	if (hfunc)
 		*hfunc = func;
 
-	return rc;
+	return 0;
 }
 
 static int ena_set_rxfh(struct net_device *netdev, const u32 *indir,
@@ -707,7 +711,7 @@ static int ena_set_rxfh(struct net_device *netdev, const u32 *indir,
 {
 	struct ena_adapter *adapter = netdev_priv(netdev);
 	struct ena_com_dev *ena_dev = adapter->ena_dev;
-	enum ena_admin_hash_functions func;
+	enum ena_admin_hash_functions func = 0;
 	int rc, i;
 
 	if (indir) {
@@ -746,7 +750,7 @@ static int ena_set_rxfh(struct net_device *netdev, const u32 *indir,
 		return -EOPNOTSUPP;
 	}
 
-	if (key) {
+	if (key || func) {
 		rc = ena_com_fill_hash_function(ena_dev, func, key,
 						ENA_HASH_KEY_SIZE,
 						0xFFFFFFFF);
Signature · 1 line
2.24.1.AMZN
#7 lore raw

PATCH 7/13 net: ena: add unmask interrupts statistics to ethtool

In reply to [email protected]

From: Sameeh Jubran <[email protected]>

Add unmask interrupts statistics to ethtool.

---
 drivers/net/ethernet/amazon/ena/ena_ethtool.c | 1 +
 drivers/net/ethernet/amazon/ena/ena_netdev.c  | 3 +++
 drivers/net/ethernet/amazon/ena/ena_netdev.h  | 1 +
 3 files changed, 5 insertions(+)
drivers/net/ethernet/amazon/ena/ena_ethtool.c +1 -0
index c7df25f92dbd..74725d606964 100644
@@ -83,6 +83,7 @@ static const struct ena_stats ena_stats_tx_strings[] = {
 	ENA_STAT_TX_ENTRY(bad_req_id),
 	ENA_STAT_TX_ENTRY(llq_buffer_copy),
 	ENA_STAT_TX_ENTRY(missed_tx),
+	ENA_STAT_TX_ENTRY(unmask_interrupt),
 };
 
 static const struct ena_stats ena_stats_rx_strings[] = {
drivers/net/ethernet/amazon/ena/ena_netdev.c +3 -0
index 6baafc3aebea..3cea4c9090c2 100644
@@ -1762,6 +1762,9 @@ static void ena_unmask_interrupt(struct ena_ring *tx_ring,
 				tx_ring->smoothed_interval,
 				true);
 
+	u64_stats_update_begin(&tx_ring->syncp);
+	tx_ring->tx_stats.unmask_interrupt++;
+	u64_stats_update_end(&tx_ring->syncp);
 	/* It is a shared MSI-X.
 	 * Tx and Rx CQ have pointer to it.
 	 * So we use one of them to reach the intr reg
drivers/net/ethernet/amazon/ena/ena_netdev.h +1 -0
index 97dfd0c67e84..ebeb911c0efb 100644
@@ -248,6 +248,7 @@ struct ena_stats_tx {
 	u64 bad_req_id;
 	u64 llq_buffer_copy;
 	u64 missed_tx;
+	u64 unmask_interrupt;
 };
 
 struct ena_stats_rx {
Signature · 1 line
2.24.1.AMZN
#6 lore raw

PATCH 6/13 net: ena: remove code that does nothing

In reply to [email protected]

From: Sameeh Jubran <[email protected]>

Both key and func parameters are pointers on the stack. Setting them to NULL does nothing.
The original intent was to leave the key and func unset in this case, but for this to happen nothing needs to be done as the calling function ethtool_get_rxfh() already clears key and func.

This commit removes the above described useless code.

---
 drivers/net/ethernet/amazon/ena/ena_ethtool.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
drivers/net/ethernet/amazon/ena/ena_ethtool.c +1 -4
index 0c3a2f14387e..c7df25f92dbd 100644
@@ -674,11 +674,8 @@ static int ena_get_rxfh(struct net_device *netdev, u32 *indir, u8 *key,
 	 */
 	rc = ena_com_get_hash_function(adapter->ena_dev, &ena_func);
 	if (rc) {
-		if (rc == -EOPNOTSUPP) {
-			key = NULL;
-			hfunc = NULL;
+		if (rc == -EOPNOTSUPP)
 			rc = 0;
-		}
 
 		return rc;
 	}
Signature · 1 line
2.24.1.AMZN
#10 lore raw

PATCH 10/13 net: ena: use SHUTDOWN as reset reason when closing interface

In reply to [email protected]

From: Sameeh Jubran <[email protected]>

The 'ENA_REGS_RESET_SHUTDOWN' enum indicates a normal driver shutdown / removal procedure.

Also, a comment is added to one of the reset reason assignments for code clarity.

---
 drivers/net/ethernet/amazon/ena/ena_netdev.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/net/ethernet/amazon/ena/ena_netdev.c +2 -0
index 517681319a57..2818965427e9 100644
@@ -3439,6 +3439,7 @@ static void ena_destroy_device(struct ena_adapter *adapter, bool graceful)
 
 	ena_com_mmio_reg_read_request_destroy(ena_dev);
 
+	/* return reset reason to default value */
 	adapter->reset_reason = ENA_REGS_RESET_NORMAL;
 
 	clear_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags);
@@ -4362,6 +4363,7 @@ static void __ena_shutoff(struct pci_dev *pdev, bool shutdown)
 	cancel_work_sync(&adapter->reset_task);
 
 	rtnl_lock(); /* lock released inside the below if-else block */
+	adapter->reset_reason = ENA_REGS_RESET_SHUTDOWN;
 	ena_destroy_device(adapter, true);
 	if (shutdown) {
 		netif_device_detach(netdev);
Signature · 1 line
2.24.1.AMZN
#0 lore raw

cover Enhance current features in ena driver

From: Sameeh Jubran <[email protected]>

This patchset introduces the following:
* minor changes to RSS feature
* add total rx and tx drop counter
* add unmask_interrupt counter for ethtool statistics
* add missing implementation for ena_com_get_admin_polling_mode()
* some minor code clean-up and cosmetics
* use SHUTDOWN as reset reason when closing interface

Arthur Kiyanovski (6):
net: ena: fix error returning in ena_com_get_hash_function()
net: ena: avoid unnecessary admin command when RSS function set fails
net: ena: change default RSS hash function to Toeplitz
net: ena: implement ena_com_get_admin_polling_mode()
net: ena: move llq configuration from ena_probe to ena_device_init()
net: ena: cosmetic: extract code to ena_indirection_table_set()

Sameeh Jubran (7):
net: ena: allow setting the hash function without changing the key
net: ena: changes to RSS hash key allocation
net: ena: remove code that does nothing
net: ena: add unmask interrupts statistics to ethtool
net: ena: add support for reporting of packet drops
net: ena: use SHUTDOWN as reset reason when closing interface
net: ena: cosmetic: remove unnecessary spaces and tabs in ena_com.h
macros

 .../net/ethernet/amazon/ena/ena_admin_defs.h  |   8 +
 drivers/net/ethernet/amazon/ena/ena_com.c     |  44 +++---
 drivers/net/ethernet/amazon/ena/ena_com.h     |  39 +++--
 drivers/net/ethernet/amazon/ena/ena_ethtool.c |  68 ++++----
 drivers/net/ethernet/amazon/ena/ena_netdev.c  | 146 ++++++++++--------
 drivers/net/ethernet/amazon/ena/ena_netdev.h  |   2 +
 6 files changed, 181 insertions(+), 126 deletions(-)
Signature · 1 line
2.24.1.AMZN
#5 lore raw

PATCH 5/13 net: ena: changes to RSS hash key allocation

From: Sameeh Jubran <[email protected]>

This commit contains 2 cosmetic changes:

1. Use ena_com_check_supported_feature_id() in ena_com_hash_key_fill_default_key() instead of rewriting its implementation. This also saves us a superfluous admin command by using the cached value.

2. Change if conditions in ena_com_rss_init() to be clearer.

---
 drivers/net/ethernet/amazon/ena/ena_com.c | 16 +++++-----------
 1 file changed, 5 insertions(+), 11 deletions(-)
drivers/net/ethernet/amazon/ena/ena_com.c +5 -11
index d428d0606166..b51bf62af11b 100644
@@ -1067,16 +1067,10 @@ static void ena_com_hash_key_fill_default_key(struct ena_com_dev *ena_dev)
 static int ena_com_hash_key_allocate(struct ena_com_dev *ena_dev)
 {
 	struct ena_rss *rss = &ena_dev->rss;
-	struct ena_admin_get_feat_resp get_resp;
-	int rc;
 
-	rc = ena_com_get_feature_ex(ena_dev, &get_resp,
-				    ENA_ADMIN_RSS_HASH_FUNCTION,
-				    ena_dev->rss.hash_key_dma_addr,
-				    sizeof(ena_dev->rss.hash_key), 0);
-	if (unlikely(rc)) {
+	if (!ena_com_check_supported_feature_id(ena_dev,
+						ENA_ADMIN_RSS_HASH_FUNCTION))
 		return -EOPNOTSUPP;
-	}
 
 	rss->hash_key =
 		dma_alloc_coherent(ena_dev->dmadev, sizeof(*rss->hash_key),
@@ -2650,10 +2644,10 @@ int ena_com_rss_init(struct ena_com_dev *ena_dev, u16 indr_tbl_log_size)
 	 * ignore this error and have indirection table support only.
 	 */
 	rc = ena_com_hash_key_allocate(ena_dev);
-	if (unlikely(rc) && rc != -EOPNOTSUPP)
-		goto err_hash_key;
-	else if (rc != -EOPNOTSUPP)
+	if (likely(!rc))
 		ena_com_hash_key_fill_default_key(ena_dev);
+	else if (rc != -EOPNOTSUPP)
+		goto err_hash_key;
 
 	rc = ena_com_hash_ctrl_init(ena_dev);
 	if (unlikely(rc))
Signature · 1 line
2.24.1.AMZN
#12 lore raw

PATCH 12/13 net: ena: cosmetic: remove unnecessary spaces and tabs in ena_com.h macros

In reply to [email protected]

From: Sameeh Jubran <[email protected]>

The macros in ena_com.h have inconsistent spaces between the macro name and it's value.

This commit sets all the macros to have a single space between the name and value.

---
 drivers/net/ethernet/amazon/ena/ena_com.h | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)
drivers/net/ethernet/amazon/ena/ena_com.h +7 -7
index 94986cdc0b1a..b3a4b53b6a86 100644
@@ -54,9 +54,9 @@
 #undef pr_fmt
 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
 
-#define ENA_MAX_NUM_IO_QUEUES		128U
+#define ENA_MAX_NUM_IO_QUEUES 128U
 /* We need to queues for each IO (on for Tx and one for Rx) */
-#define ENA_TOTAL_NUM_QUEUES		(2 * (ENA_MAX_NUM_IO_QUEUES))
+#define ENA_TOTAL_NUM_QUEUES (2 * (ENA_MAX_NUM_IO_QUEUES))
 
 #define ENA_MAX_HANDLERS 256
 
@@ -73,13 +73,13 @@
 /*****************************************************************************/
 /* ENA adaptive interrupt moderation settings */
 
-#define ENA_INTR_INITIAL_TX_INTERVAL_USECS		64
-#define ENA_INTR_INITIAL_RX_INTERVAL_USECS		0
-#define ENA_DEFAULT_INTR_DELAY_RESOLUTION		1
+#define ENA_INTR_INITIAL_TX_INTERVAL_USECS 64
+#define ENA_INTR_INITIAL_RX_INTERVAL_USECS 0
+#define ENA_DEFAULT_INTR_DELAY_RESOLUTION 1
 
-#define ENA_HW_HINTS_NO_TIMEOUT				0xFFFF
+#define ENA_HW_HINTS_NO_TIMEOUT	0xFFFF
 
-#define ENA_FEATURE_MAX_QUEUE_EXT_VER	1
+#define ENA_FEATURE_MAX_QUEUE_EXT_VER 1
 
 struct ena_llq_configurations {
 	enum ena_admin_llq_header_location llq_header_location;
Signature · 1 line
2.24.1.AMZN
#8 lore raw

PATCH 8/13 net: ena: add support for reporting of packet drops

In reply to [email protected]

From: Sameeh Jubran <[email protected]>

1. Add support for getting tx drops from the device and saving them in the driver.
2. Report tx and rx drops via ethtool.
3. Report tx via netdev stats.

---
 drivers/net/ethernet/amazon/ena/ena_admin_defs.h | 8 ++++++++
 drivers/net/ethernet/amazon/ena/ena_ethtool.c    | 2 ++
 drivers/net/ethernet/amazon/ena/ena_netdev.c     | 6 ++++++
 drivers/net/ethernet/amazon/ena/ena_netdev.h     | 1 +
 4 files changed, 17 insertions(+)
drivers/net/ethernet/amazon/ena/ena_admin_defs.h +8 -0
index 8baf847e8622..7be3dcbf3d16 100644
@@ -404,6 +404,10 @@ struct ena_admin_basic_stats {
 	u32 rx_drops_low;
 
 	u32 rx_drops_high;
+
+	u32 tx_drops_low;
+
+	u32 tx_drops_high;
 };
 
 struct ena_admin_acq_get_stats_resp {
@@ -1017,6 +1021,10 @@ struct ena_admin_aenq_keep_alive_desc {
 	u32 rx_drops_low;
 
 	u32 rx_drops_high;
+
+	u32 tx_drops_low;
+
+	u32 tx_drops_high;
 };
 
 struct ena_admin_ena_mmio_req_read_less_resp {
drivers/net/ethernet/amazon/ena/ena_ethtool.c +2 -0
index 74725d606964..a736ad62af78 100644
@@ -66,6 +66,8 @@ static const struct ena_stats ena_stats_global_strings[] = {
 	ENA_STAT_GLOBAL_ENTRY(interface_up),
 	ENA_STAT_GLOBAL_ENTRY(interface_down),
 	ENA_STAT_GLOBAL_ENTRY(admin_q_pause),
+	ENA_STAT_GLOBAL_ENTRY(rx_drops),
+	ENA_STAT_GLOBAL_ENTRY(tx_drops),
 };
 
 static const struct ena_stats ena_stats_tx_strings[] = {
drivers/net/ethernet/amazon/ena/ena_netdev.c +6 -0
index 3cea4c9090c2..517681319a57 100644
@@ -3172,6 +3172,7 @@ static void ena_get_stats64(struct net_device *netdev,
 	struct ena_ring *rx_ring, *tx_ring;
 	unsigned int start;
 	u64 rx_drops;
+	u64 tx_drops;
 	int i;
 
 	if (!test_bit(ENA_FLAG_DEV_UP, &adapter->flags))
@@ -3206,9 +3207,11 @@ static void ena_get_stats64(struct net_device *netdev,
 	do {
 		start = u64_stats_fetch_begin_irq(&adapter->syncp);
 		rx_drops = adapter->dev_stats.rx_drops;
+		tx_drops = adapter->dev_stats.tx_drops;
 	} while (u64_stats_fetch_retry_irq(&adapter->syncp, start));
 
 	stats->rx_dropped = rx_drops;
+	stats->tx_dropped = tx_drops;
 
 	stats->multicast = 0;
 	stats->collisions = 0;
@@ -4517,14 +4520,17 @@ static void ena_keep_alive_wd(void *adapter_data,
 	struct ena_adapter *adapter = (struct ena_adapter *)adapter_data;
 	struct ena_admin_aenq_keep_alive_desc *desc;
 	u64 rx_drops;
+	u64 tx_drops;
 
 	desc = (struct ena_admin_aenq_keep_alive_desc *)aenq_e;
 	adapter->last_keep_alive_jiffies = jiffies;
 
 	rx_drops = ((u64)desc->rx_drops_high << 32) | desc->rx_drops_low;
+	tx_drops = ((u64)desc->tx_drops_high << 32) | desc->tx_drops_low;
 
 	u64_stats_update_begin(&adapter->syncp);
 	adapter->dev_stats.rx_drops = rx_drops;
+	adapter->dev_stats.tx_drops = tx_drops;
 	u64_stats_update_end(&adapter->syncp);
 }
 
drivers/net/ethernet/amazon/ena/ena_netdev.h +1 -0
index ebeb911c0efb..bd278c4721c6 100644
@@ -334,6 +334,7 @@ struct ena_stats_dev {
 	u64 interface_down;
 	u64 admin_q_pause;
 	u64 rx_drops;
+	u64 tx_drops;
 };
 
 enum ena_flags_t {
Signature · 1 line
2.24.1.AMZN
#11 lore raw

PATCH 11/13 net: ena: move llq configuration from ena_probe to ena_device_init()

In reply to [email protected]

From: Arthur Kiyanovski <[email protected]>

This refactor is done as preparation for an upcoming feature of allowing to increase the header size when LLQ is on.

Such a change needs to survive a device reset (for instance due to some recoverable error). In order to do that, llq initialization needs to happen both at device initialization and at device restore. We therefore move it to ena_device_init() which is called from both ena_probe() and ena_restore_device().

Also, functions are moved around in order for them to be defined before they are used.

---
 drivers/net/ethernet/amazon/ena/ena_netdev.c | 133 ++++++++++---------
 1 file changed, 67 insertions(+), 66 deletions(-)
drivers/net/ethernet/amazon/ena/ena_netdev.c +67 -66
index 2818965427e9..37c5ae52de7d 100644
@@ -3263,10 +3263,68 @@ static int ena_device_validate_params(struct ena_adapter *adapter,
 	return 0;
 }
 
+static void set_default_llq_configurations(struct ena_llq_configurations *llq_config)
+{
+	llq_config->llq_header_location = ENA_ADMIN_INLINE_HEADER;
+	llq_config->llq_stride_ctrl = ENA_ADMIN_MULTIPLE_DESCS_PER_ENTRY;
+	llq_config->llq_num_decs_before_header = ENA_ADMIN_LLQ_NUM_DESCS_BEFORE_HEADER_2;
+	llq_config->llq_ring_entry_size = ENA_ADMIN_LIST_ENTRY_SIZE_128B;
+	llq_config->llq_ring_entry_size_value = 128;
+}
+
+static int ena_set_queues_placement_policy(struct pci_dev *pdev,
+					   struct ena_com_dev *ena_dev,
+					   struct ena_admin_feature_llq_desc *llq,
+					   struct ena_llq_configurations *llq_default_configurations)
+{
+	bool has_mem_bar;
+	int rc;
+	u32 llq_feature_mask;
+
+	llq_feature_mask = 1 << ENA_ADMIN_LLQ;
+	if (!(ena_dev->supported_features & llq_feature_mask)) {
+		dev_err(&pdev->dev,
+			"LLQ is not supported Fallback to host mode policy.\n");
+		ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
+		return 0;
+	}
+
+	has_mem_bar = pci_select_bars(pdev, IORESOURCE_MEM) & BIT(ENA_MEM_BAR);
+
+	rc = ena_com_config_dev_mode(ena_dev, llq, llq_default_configurations);
+	if (unlikely(rc)) {
+		dev_err(&pdev->dev,
+			"Failed to configure the device mode.  Fallback to host mode policy.\n");
+		ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
+		return 0;
+	}
+
+	/* Nothing to config, exit */
+	if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_HOST)
+		return 0;
+
+	if (!has_mem_bar) {
+		dev_err(&pdev->dev,
+			"ENA device does not expose LLQ bar. Fallback to host mode policy.\n");
+		ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
+		return 0;
+	}
+
+	ena_dev->mem_bar = devm_ioremap_wc(&pdev->dev,
+					   pci_resource_start(pdev, ENA_MEM_BAR),
+					   pci_resource_len(pdev, ENA_MEM_BAR));
+
+	if (!ena_dev->mem_bar)
+		return -EFAULT;
+
+	return 0;
+}
+
 static int ena_device_init(struct ena_com_dev *ena_dev, struct pci_dev *pdev,
 			   struct ena_com_dev_get_features_ctx *get_feat_ctx,
 			   bool *wd_state)
 {
+	struct ena_llq_configurations llq_config;
 	struct device *dev = &pdev->dev;
 	bool readless_supported;
 	u32 aenq_groups;
@@ -3357,6 +3415,15 @@ static int ena_device_init(struct ena_com_dev *ena_dev, struct pci_dev *pdev,
 
 	*wd_state = !!(aenq_groups & BIT(ENA_ADMIN_KEEP_ALIVE));
 
+	set_default_llq_configurations(&llq_config);
+
+	rc = ena_set_queues_placement_policy(pdev, ena_dev, &get_feat_ctx->llq,
+					     &llq_config);
+	if (rc) {
+		dev_err(&pdev->dev, "ena device init failed\n");
+		goto err_admin_init;
+	}
+
 	return 0;
 
 err_admin_init:
@@ -3864,54 +3931,6 @@ static int ena_calc_max_io_queue_num(struct pci_dev *pdev,
 	return max_num_io_queues;
 }
 
-static int ena_set_queues_placement_policy(struct pci_dev *pdev,
-					   struct ena_com_dev *ena_dev,
-					   struct ena_admin_feature_llq_desc *llq,
-					   struct ena_llq_configurations *llq_default_configurations)
-{
-	bool has_mem_bar;
-	int rc;
-	u32 llq_feature_mask;
-
-	llq_feature_mask = 1 << ENA_ADMIN_LLQ;
-	if (!(ena_dev->supported_features & llq_feature_mask)) {
-		dev_err(&pdev->dev,
-			"LLQ is not supported Fallback to host mode policy.\n");
-		ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
-		return 0;
-	}
-
-	has_mem_bar = pci_select_bars(pdev, IORESOURCE_MEM) & BIT(ENA_MEM_BAR);
-
-	rc = ena_com_config_dev_mode(ena_dev, llq, llq_default_configurations);
-	if (unlikely(rc)) {
-		dev_err(&pdev->dev,
-			"Failed to configure the device mode.  Fallback to host mode policy.\n");
-		ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
-		return 0;
-	}
-
-	/* Nothing to config, exit */
-	if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_HOST)
-		return 0;
-
-	if (!has_mem_bar) {
-		dev_err(&pdev->dev,
-			"ENA device does not expose LLQ bar. Fallback to host mode policy.\n");
-		ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
-		return 0;
-	}
-
-	ena_dev->mem_bar = devm_ioremap_wc(&pdev->dev,
-					   pci_resource_start(pdev, ENA_MEM_BAR),
-					   pci_resource_len(pdev, ENA_MEM_BAR));
-
-	if (!ena_dev->mem_bar)
-		return -EFAULT;
-
-	return 0;
-}
-
 static void ena_set_dev_offloads(struct ena_com_dev_get_features_ctx *feat,
 				 struct net_device *netdev)
 {
@@ -4027,14 +4046,6 @@ static void ena_release_bars(struct ena_com_dev *ena_dev, struct pci_dev *pdev)
 	pci_release_selected_regions(pdev, release_bars);
 }
 
-static void set_default_llq_configurations(struct ena_llq_configurations *llq_config)
-{
-	llq_config->llq_header_location = ENA_ADMIN_INLINE_HEADER;
-	llq_config->llq_ring_entry_size = ENA_ADMIN_LIST_ENTRY_SIZE_128B;
-	llq_config->llq_stride_ctrl = ENA_ADMIN_MULTIPLE_DESCS_PER_ENTRY;
-	llq_config->llq_num_decs_before_header = ENA_ADMIN_LLQ_NUM_DESCS_BEFORE_HEADER_2;
-	llq_config->llq_ring_entry_size_value = 128;
-}
 
 static int ena_calc_io_queue_size(struct ena_calc_queue_size_ctx *ctx)
 {
@@ -4116,7 +4127,6 @@ static int ena_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 {
 	struct ena_com_dev_get_features_ctx get_feat_ctx;
 	struct ena_calc_queue_size_ctx calc_queue_ctx = { 0 };
-	struct ena_llq_configurations llq_config;
 	struct ena_com_dev *ena_dev = NULL;
 	struct ena_adapter *adapter;
 	struct net_device *netdev;
@@ -4169,15 +4179,6 @@ static int ena_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 		goto err_free_region;
 	}
 
-	set_default_llq_configurations(&llq_config);
-
-	rc = ena_set_queues_placement_policy(pdev, ena_dev, &get_feat_ctx.llq,
-					     &llq_config);
-	if (rc) {
-		dev_err(&pdev->dev, "ena device init failed\n");
-		goto err_device_destroy;
-	}
-
 	calc_queue_ctx.ena_dev = ena_dev;
 	calc_queue_ctx.get_feat_ctx = &get_feat_ctx;
 	calc_queue_ctx.pdev = pdev;
Signature · 1 line
2.24.1.AMZN
#13 lore raw

PATCH 13/13 net: ena: cosmetic: extract code to ena_indirection_table_set()

In reply to [email protected]

From: Arthur Kiyanovski <[email protected]>

Extract code to ena_indirection_table_set() to make the code cleaner.

---
 drivers/net/ethernet/amazon/ena/ena_ethtool.c | 48 ++++++++++++-------
 1 file changed, 30 insertions(+), 18 deletions(-)
drivers/net/ethernet/amazon/ena/ena_ethtool.c +30 -18
index a736ad62af78..51efb9463b5d 100644
@@ -638,6 +638,32 @@ static u32 ena_get_rxfh_key_size(struct net_device *netdev)
 	return ENA_HASH_KEY_SIZE;
 }
 
+static int ena_indirection_table_set(struct ena_adapter *adapter,
+				     const u32 *indir)
+{
+	struct ena_com_dev *ena_dev = adapter->ena_dev;
+	int i, rc;
+
+	for (i = 0; i < ENA_RX_RSS_TABLE_SIZE; i++) {
+		rc = ena_com_indirect_table_fill_entry(ena_dev,
+						       i,
+						       ENA_IO_RXQ_IDX(indir[i]));
+		if (unlikely(rc)) {
+			netif_err(adapter, drv, adapter->netdev,
+				  "Cannot fill indirect table (index is too large)\n");
+			return rc;
+		}
+	}
+
+	rc = ena_com_indirect_table_set(ena_dev);
+	if (rc) {
+		netif_err(adapter, drv, adapter->netdev,
+			  "Cannot set indirect table\n");
+		return rc == -EPERM ? -EOPNOTSUPP : rc;
+	}
+	return rc;
+}
+
 static int ena_indirection_table_get(struct ena_adapter *adapter, u32 *indir)
 {
 	struct ena_com_dev *ena_dev = adapter->ena_dev;
@@ -712,26 +738,12 @@ static int ena_set_rxfh(struct net_device *netdev, const u32 *indir,
 	struct ena_adapter *adapter = netdev_priv(netdev);
 	struct ena_com_dev *ena_dev = adapter->ena_dev;
 	enum ena_admin_hash_functions func = 0;
-	int rc, i;
+	int rc;
 
 	if (indir) {
-		for (i = 0; i < ENA_RX_RSS_TABLE_SIZE; i++) {
-			rc = ena_com_indirect_table_fill_entry(ena_dev,
-							       i,
-							       ENA_IO_RXQ_IDX(indir[i]));
-			if (unlikely(rc)) {
-				netif_err(adapter, drv, netdev,
-					  "Cannot fill indirect table (index is too large)\n");
-				return rc;
-			}
-		}
-
-		rc = ena_com_indirect_table_set(ena_dev);
-		if (rc) {
-			netif_err(adapter, drv, netdev,
-				  "Cannot set indirect table\n");
-			return rc == -EPERM ? -EOPNOTSUPP : rc;
-		}
+		rc = ena_indirection_table_set(adapter, indir);
+		if (rc)
+			return rc;
 	}
 
 	switch (hfunc) {
Signature · 1 line
2.24.1.AMZN
#14 lore raw
In reply to [email protected]

Please ignore this patch series, I'll resend to net-next

Quoted 53 lines

-----Original Message-----
From: [email protected] <[email protected]>
Sent: Wednesday, April 22, 2020 11:09 AM
To: [email protected]; [email protected]
Cc: Jubran, Samih <[email protected]>; Woodhouse, David <[email protected]>; Machulsky, Zorik <[email protected]>; Matushevsky, Alexander <[email protected]>; Bshara, Saeed <[email protected]>; Wilson, Matt <[email protected]>; Liguori, Anthony <[email protected]>; Bshara, Nafea <[email protected]>; Tzalik, Guy <[email protected]>; Belgazal, Netanel <[email protected]>; Saidi, Ali <[email protected]>; Herrenschmidt, Benjamin <[email protected]>; Kiyanovski, Arthur <[email protected]>; Dagan, Noam <[email protected]>
Subject: [PATCH V1 net 00/13] Enhance current features in ena driver

From: Sameeh Jubran <[email protected]>

This patchset introduces the following:
* minor changes to RSS feature
* add total rx and tx drop counter
* add unmask_interrupt counter for ethtool statistics
* add missing implementation for ena_com_get_admin_polling_mode()
* some minor code clean-up and cosmetics
* use SHUTDOWN as reset reason when closing interface

Arthur Kiyanovski (6):
net: ena: fix error returning in ena_com_get_hash_function()
net: ena: avoid unnecessary admin command when RSS function set fails
net: ena: change default RSS hash function to Toeplitz
net: ena: implement ena_com_get_admin_polling_mode()
net: ena: move llq configuration from ena_probe to ena_device_init()
net: ena: cosmetic: extract code to ena_indirection_table_set()

Sameeh Jubran (7):
net: ena: allow setting the hash function without changing the key
net: ena: changes to RSS hash key allocation
net: ena: remove code that does nothing
net: ena: add unmask interrupts statistics to ethtool
net: ena: add support for reporting of packet drops
net: ena: use SHUTDOWN as reset reason when closing interface
net: ena: cosmetic: remove unnecessary spaces and tabs in ena_com.h
macros

 .../net/ethernet/amazon/ena/ena_admin_defs.h  |   8 +
 drivers/net/ethernet/amazon/ena/ena_com.c     |  44 +++---
 drivers/net/ethernet/amazon/ena/ena_com.h     |  39 +++--
 drivers/net/ethernet/amazon/ena/ena_ethtool.c |  68 ++++----
drivers/net/ethernet/amazon/ena/ena_netdev.c  | 146 ++++++++++--------
 drivers/net/ethernet/amazon/ena/ena_netdev.h  |   2 +
 6 files changed, 181 insertions(+), 126 deletions(-)

--
2.24.1.AMZN