* Re: [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues
2012-06-13 12:44 ` [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues Merav Sicron
@ 2012-06-13 9:52 ` Eric Dumazet
2012-06-13 15:14 ` Merav Sicron
0 siblings, 1 reply; 28+ messages in thread
From: Eric Dumazet @ 2012-06-13 9:52 UTC (permalink / raw)
To: Merav Sicron; +Cc: eilong, davem, netdev
On Wed, 2012-06-13 at 15:44 +0300, Merav Sicron wrote:
> This patch removed the limitation in the code for 16 RSS queues. The default
> (without other instruction from the user) number of queues is determined
> according to the number of MSI-X vectors supported for that function, the number
> of CPUs in the system, and a maximum of 8 queues.
Thats a very confusing changelog
You meant : " a minimum of 8 queues" ?
You should give more explanations, because its a sensible area for
performances.
^ permalink raw reply [flat|nested] 28+ messages in thread
* [net-next patch 0/12] bnx2x: ethtool and other enhancements
@ 2012-06-13 12:44 Merav Sicron
2012-06-13 12:44 ` [net-next patch 1/12] bnx2x: Add support for external LB Merav Sicron
` (11 more replies)
0 siblings, 12 replies; 28+ messages in thread
From: Merav Sicron @ 2012-06-13 12:44 UTC (permalink / raw)
To: eilong, davem, netdev; +Cc: Merav Sicron
Hi Dave,
This patch series adds few features to bnx2x:
Add support for external LB in ethtool self-test
Enable UDP RSS on 4-tupple, controlled by ethtool
Support up to 63 RSS queues (rather than 16)
Add support for setting the desired number of RSS queues via ethtool
Allow to configure dcbx admin params from all drivers on a single physical port
Add FCoE capabilities advertisement
Please consider applying this patch series to net-next.
Thanks,
Merav
^ permalink raw reply [flat|nested] 28+ messages in thread
* [net-next patch 1/12] bnx2x: Add support for external LB
2012-06-13 12:44 [net-next patch 0/12] bnx2x: ethtool and other enhancements Merav Sicron
@ 2012-06-13 12:44 ` Merav Sicron
2012-06-13 18:17 ` Ben Hutchings
2012-06-13 12:44 ` [net-next patch 2/12] bnx2x: Return only online tests for MF Merav Sicron
` (10 subsequent siblings)
11 siblings, 1 reply; 28+ messages in thread
From: Merav Sicron @ 2012-06-13 12:44 UTC (permalink / raw)
To: eilong, davem, netdev; +Cc: Merav Sicron
This change enables to do self-test with external loopback via ethtool.
Signed-off-by: Merav Sicron <meravs@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x.h | 5 +-
drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c | 4 +-
.../net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c | 76 ++++++++++++++++++--
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c | 7 +-
4 files changed, 82 insertions(+), 10 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index e30e2a2..fb6de4a 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -1832,6 +1832,7 @@ static inline u32 reg_poll(struct bnx2x *bp, u32 reg, u32 expected, int ms,
#define LOAD_NORMAL 0
#define LOAD_OPEN 1
#define LOAD_DIAG 2
+#define LOAD_LOOPBACK_EXT 3
#define UNLOAD_NORMAL 0
#define UNLOAD_CLOSE 1
#define UNLOAD_RECOVERY 2
@@ -1915,12 +1916,14 @@ static inline u32 reg_poll(struct bnx2x *bp, u32 reg, u32 expected, int ms,
#define PCICFG_LINK_SPEED_SHIFT 16
-#define BNX2X_NUM_TESTS 7
+#define BNX2X_NUM_TESTS 8
#define BNX2X_PHY_LOOPBACK 0
#define BNX2X_MAC_LOOPBACK 1
+#define BNX2X_EXT_LOOPBACK 2
#define BNX2X_PHY_LOOPBACK_FAILED 1
#define BNX2X_MAC_LOOPBACK_FAILED 2
+#define BNX2X_EXT_LOOPBACK_FAILED 3
#define BNX2X_LOOPBACK_FAILED (BNX2X_MAC_LOOPBACK_FAILED | \
BNX2X_PHY_LOOPBACK_FAILED)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
index ad0743b..09fcc55 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
@@ -2161,6 +2161,7 @@ int bnx2x_nic_load(struct bnx2x *bp, int load_mode)
break;
case LOAD_DIAG:
+ case LOAD_LOOPBACK_EXT:
bp->state = BNX2X_STATE_DIAG;
break;
@@ -2200,7 +2201,8 @@ int bnx2x_nic_load(struct bnx2x *bp, int load_mode)
return -EBUSY;
}
- bnx2x_dcbx_init(bp);
+ if (bp->state != BNX2X_STATE_DIAG)
+ bnx2x_dcbx_init(bp);
return 0;
#ifndef BNX2X_STOP_ON_ERROR
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
index bf30e28..116ac25 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
@@ -1538,7 +1538,8 @@ static const struct {
} bnx2x_tests_str_arr[BNX2X_NUM_TESTS] = {
{ "register_test (offline)" },
{ "memory_test (offline)" },
- { "loopback_test (offline)" },
+ { "int_loopback_test (offline)" },
+ { "ext_loopback_test (offline)" },
{ "nvram_test (online)" },
{ "interrupt_test (online)" },
{ "link_test (online)" },
@@ -1943,6 +1944,14 @@ static void bnx2x_wait_for_link(struct bnx2x *bp, u8 link_up, u8 is_serdes)
if (cnt <= 0 && bnx2x_link_test(bp, is_serdes))
DP(BNX2X_MSG_ETHTOOL, "Timeout waiting for link up\n");
+
+ cnt = 1400;
+ while (!bp->link_vars.link_up && cnt--)
+ msleep(20);
+
+ if (cnt <= 0 && !bp->link_vars.link_up)
+ DP(BNX2X_MSG_ETHTOOL,
+ "Timeout waiting for link init\n");
}
}
@@ -1968,13 +1977,16 @@ static int bnx2x_run_loopback(struct bnx2x *bp, int loopback_mode)
u16 len;
int rc = -ENODEV;
u8 *data;
- struct netdev_queue *txq = netdev_get_tx_queue(bp->dev, txdata->txq_index);
+ struct netdev_queue *txq = netdev_get_tx_queue(bp->dev,
+ txdata->txq_index);
/* check the loopback mode */
switch (loopback_mode) {
case BNX2X_PHY_LOOPBACK:
- if (bp->link_params.loopback_mode != LOOPBACK_XGXS)
+ if (bp->link_params.loopback_mode != LOOPBACK_XGXS) {
+ DP(BNX2X_MSG_ETHTOOL, "PHY loopback not supported\n");
return -EINVAL;
+ }
break;
case BNX2X_MAC_LOOPBACK:
if (CHIP_IS_E3(bp)) {
@@ -1991,6 +2003,13 @@ static int bnx2x_run_loopback(struct bnx2x *bp, int loopback_mode)
bnx2x_phy_init(&bp->link_params, &bp->link_vars);
break;
+ case BNX2X_EXT_LOOPBACK:
+ if (bp->link_params.loopback_mode != LOOPBACK_EXT) {
+ DP(BNX2X_MSG_ETHTOOL,
+ "Can't configure external loopback\n");
+ return -EINVAL;
+ }
+ break;
default:
DP(BNX2X_MSG_ETHTOOL, "Command parameters not supported\n");
return -EINVAL;
@@ -2162,6 +2181,38 @@ static int bnx2x_test_loopback(struct bnx2x *bp)
return rc;
}
+static int bnx2x_test_ext_loopback(struct bnx2x *bp)
+{
+ int rc;
+ u8 is_serdes =
+ (bp->link_vars.link_status & LINK_STATUS_SERDES_LINK) > 0;
+
+ if (BP_NOMCP(bp))
+ return -ENODEV;
+
+ if (!netif_running(bp->dev))
+ return BNX2X_EXT_LOOPBACK_FAILED;
+
+ bnx2x_nic_unload(bp, UNLOAD_NORMAL);
+ rc = bnx2x_nic_load(bp, LOAD_LOOPBACK_EXT);
+ if (rc != 0) {
+ DP(BNX2X_MSG_ETHTOOL,
+ "Can't perform self-test, nic_load (for external lb) failed\n");
+ return -ENODEV;
+ }
+ bnx2x_wait_for_link(bp, 1, is_serdes);
+
+ bnx2x_netif_stop(bp, 1);
+
+ rc = bnx2x_run_loopback(bp, BNX2X_EXT_LOOPBACK);
+ if (rc)
+ DP(BNX2X_MSG_ETHTOOL, "EXT loopback failed (res %d)\n", rc);
+
+ bnx2x_netif_start(bp);
+
+ return rc;
+}
+
#define CRC32_RESIDUAL 0xdebb20e3
static int bnx2x_test_nvram(struct bnx2x *bp)
@@ -2263,6 +2314,10 @@ static void bnx2x_self_test(struct net_device *dev,
etest->flags |= ETH_TEST_FL_FAILED;
return;
}
+ DP(BNX2X_MSG_ETHTOOL,
+ "Self-test command parameters: offline = %d, external_lb = %d\n",
+ (etest->flags & ETH_TEST_FL_OFFLINE),
+ (etest->flags & ETH_TEST_FL_EXTERNAL_LB)>>2);
memset(buf, 0, sizeof(u64) * BNX2X_NUM_TESTS);
@@ -2300,10 +2355,17 @@ static void bnx2x_self_test(struct net_device *dev,
etest->flags |= ETH_TEST_FL_FAILED;
}
- buf[2] = bnx2x_test_loopback(bp);
+ buf[2] = bnx2x_test_loopback(bp); /* internal LB */
if (buf[2] != 0)
etest->flags |= ETH_TEST_FL_FAILED;
+ if (etest->flags & ETH_TEST_FL_EXTERNAL_LB) {
+ buf[3] = bnx2x_test_ext_loopback(bp); /* external LB */
+ if (buf[3] != 0)
+ etest->flags |= ETH_TEST_FL_FAILED;
+ etest->flags |= ETH_TEST_FL_EXTERNAL_LB_DONE;
+ }
+
bnx2x_nic_unload(bp, UNLOAD_NORMAL);
/* restore input for TX port IF */
@@ -2314,16 +2376,16 @@ static void bnx2x_self_test(struct net_device *dev,
bnx2x_wait_for_link(bp, link_up, is_serdes);
}
if (bnx2x_test_nvram(bp) != 0) {
- buf[3] = 1;
+ buf[4] = 1;
etest->flags |= ETH_TEST_FL_FAILED;
}
if (bnx2x_test_intr(bp) != 0) {
- buf[4] = 1;
+ buf[5] = 1;
etest->flags |= ETH_TEST_FL_FAILED;
}
if (bnx2x_link_test(bp, is_serdes) != 0) {
- buf[5] = 1;
+ buf[6] = 1;
etest->flags |= ETH_TEST_FL_FAILED;
}
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index a622bb7..21930be 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -2124,7 +2124,12 @@ u8 bnx2x_initial_phy_init(struct bnx2x *bp, int load_mode)
}
}
- rc = bnx2x_phy_init(&bp->link_params, &bp->link_vars);
+ if (load_mode == LOAD_LOOPBACK_EXT) {
+ struct link_params *lp = &bp->link_params;
+ lp->loopback_mode = LOOPBACK_EXT;
+ }
+
+ rc = bnx2x_phy_init(&bp->link_params, &bp->link_vars);
bnx2x_release_phy_lock(bp);
--
1.7.10
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [net-next patch 2/12] bnx2x: Return only online tests for MF
2012-06-13 12:44 [net-next patch 0/12] bnx2x: ethtool and other enhancements Merav Sicron
2012-06-13 12:44 ` [net-next patch 1/12] bnx2x: Add support for external LB Merav Sicron
@ 2012-06-13 12:44 ` Merav Sicron
2012-06-13 12:44 ` [net-next patch 3/12] bnx2x: Add support for 4-tupple UDP RSS Merav Sicron
` (9 subsequent siblings)
11 siblings, 0 replies; 28+ messages in thread
From: Merav Sicron @ 2012-06-13 12:44 UTC (permalink / raw)
To: eilong, davem, netdev; +Cc: Merav Sicron
1. In multi-function device, show only the online tests in self-test results as
only these test are performed (offline tests cannot be performed as they may
corrupt the traffic of other functions on the same physical port). Note that
multi-function mode cannot change while the driver is up.
2. Check result code in NIC load and act accordingly.
Signed-off-by: Merav Sicron <meravs@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x.h | 6 +-
.../net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c | 85 +++++++++++++-------
2 files changed, 62 insertions(+), 29 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index fb6de4a..568c3c7 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -1915,8 +1915,10 @@ static inline u32 reg_poll(struct bnx2x *bp, u32 reg, u32 expected, int ms,
#define PCICFG_LINK_SPEED 0xf0000
#define PCICFG_LINK_SPEED_SHIFT 16
-
-#define BNX2X_NUM_TESTS 8
+#define BNX2X_NUM_TESTS_SF 7
+#define BNX2X_NUM_TESTS_MF 3
+#define BNX2X_NUM_TESTS(bp) (IS_MF(bp) ? BNX2X_NUM_TESTS_MF : \
+ BNX2X_NUM_TESTS_SF)
#define BNX2X_PHY_LOOPBACK 0
#define BNX2X_MAC_LOOPBACK 1
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
index 116ac25..d7ffc74 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
@@ -826,7 +826,7 @@ static void bnx2x_get_drvinfo(struct net_device *dev,
((phy_fw_ver[0] != '\0') ? " phy " : ""), phy_fw_ver);
strlcpy(info->bus_info, pci_name(bp->pdev), sizeof(info->bus_info));
info->n_stats = BNX2X_NUM_STATS;
- info->testinfo_len = BNX2X_NUM_TESTS;
+ info->testinfo_len = BNX2X_NUM_TESTS(bp);
info->eedump_len = bp->common.flash_size;
info->regdump_len = bnx2x_get_regs_len(dev);
}
@@ -1533,17 +1533,14 @@ static int bnx2x_set_pauseparam(struct net_device *dev,
return 0;
}
-static const struct {
- char string[ETH_GSTRING_LEN];
-} bnx2x_tests_str_arr[BNX2X_NUM_TESTS] = {
- { "register_test (offline)" },
- { "memory_test (offline)" },
- { "int_loopback_test (offline)" },
- { "ext_loopback_test (offline)" },
- { "nvram_test (online)" },
- { "interrupt_test (online)" },
- { "link_test (online)" },
- { "idle check (online)" }
+char *bnx2x_tests_str_arr[BNX2X_NUM_TESTS_SF] = {
+ "register_test (offline) ",
+ "memory_test (offline) ",
+ "int_loopback_test (offline)",
+ "ext_loopback_test (offline)",
+ "nvram_test (online) ",
+ "interrupt_test (online) ",
+ "link_test (online) "
};
static u32 bnx2x_eee_to_adv(u32 eee_adv)
@@ -2308,6 +2305,8 @@ static void bnx2x_self_test(struct net_device *dev,
{
struct bnx2x *bp = netdev_priv(dev);
u8 is_serdes;
+ int rc;
+
if (bp->recovery_state != BNX2X_RECOVERY_DONE) {
netdev_err(bp->dev,
"Handling parity error recovery. Try again later\n");
@@ -2319,17 +2318,18 @@ static void bnx2x_self_test(struct net_device *dev,
(etest->flags & ETH_TEST_FL_OFFLINE),
(etest->flags & ETH_TEST_FL_EXTERNAL_LB)>>2);
- memset(buf, 0, sizeof(u64) * BNX2X_NUM_TESTS);
+ memset(buf, 0, sizeof(u64) * BNX2X_NUM_TESTS(bp));
- if (!netif_running(dev))
+ if (!netif_running(dev)) {
+ DP(BNX2X_MSG_ETHTOOL,
+ "Can't perform self-test when interface is down\n");
return;
+ }
- /* offline tests are not supported in MF mode */
- if (IS_MF(bp))
- etest->flags &= ~ETH_TEST_FL_OFFLINE;
is_serdes = (bp->link_vars.link_status & LINK_STATUS_SERDES_LINK) > 0;
- if (etest->flags & ETH_TEST_FL_OFFLINE) {
+ /* offline tests are not supported in MF mode */
+ if ((etest->flags & ETH_TEST_FL_OFFLINE) && !IS_MF(bp)) {
int port = BP_PORT(bp);
u32 val;
u8 link_up;
@@ -2342,7 +2342,14 @@ static void bnx2x_self_test(struct net_device *dev,
link_up = bp->link_vars.link_up;
bnx2x_nic_unload(bp, UNLOAD_NORMAL);
- bnx2x_nic_load(bp, LOAD_DIAG);
+ rc = bnx2x_nic_load(bp, LOAD_DIAG);
+ if (rc != 0) {
+ etest->flags |= ETH_TEST_FL_FAILED;
+ DP(BNX2X_MSG_ETHTOOL,
+ "Can't perform self-test, nic_load (for offline) failed\n");
+ return;
+ }
+
/* wait until link state is restored */
bnx2x_wait_for_link(bp, 1, is_serdes);
@@ -2370,22 +2377,36 @@ static void bnx2x_self_test(struct net_device *dev,
/* restore input for TX port IF */
REG_WR(bp, NIG_REG_EGRESS_UMP0_IN_EN + port*4, val);
-
- bnx2x_nic_load(bp, LOAD_NORMAL);
+ rc = bnx2x_nic_load(bp, LOAD_NORMAL);
+ if (rc != 0) {
+ etest->flags |= ETH_TEST_FL_FAILED;
+ DP(BNX2X_MSG_ETHTOOL,
+ "Can't perform self-test, nic_load (for online) failed\n");
+ return;
+ }
/* wait until link state is restored */
bnx2x_wait_for_link(bp, link_up, is_serdes);
}
if (bnx2x_test_nvram(bp) != 0) {
- buf[4] = 1;
+ if (!IS_MF(bp))
+ buf[4] = 1;
+ else
+ buf[0] = 1;
etest->flags |= ETH_TEST_FL_FAILED;
}
if (bnx2x_test_intr(bp) != 0) {
- buf[5] = 1;
+ if (!IS_MF(bp))
+ buf[5] = 1;
+ else
+ buf[1] = 1;
etest->flags |= ETH_TEST_FL_FAILED;
}
if (bnx2x_link_test(bp, is_serdes) != 0) {
- buf[6] = 1;
+ if (!IS_MF(bp))
+ buf[6] = 1;
+ else
+ buf[2] = 1;
etest->flags |= ETH_TEST_FL_FAILED;
}
@@ -2430,7 +2451,7 @@ static int bnx2x_get_sset_count(struct net_device *dev, int stringset)
return num_stats;
case ETH_SS_TEST:
- return BNX2X_NUM_TESTS;
+ return BNX2X_NUM_TESTS(bp);
default:
return -EINVAL;
@@ -2440,7 +2461,7 @@ static int bnx2x_get_sset_count(struct net_device *dev, int stringset)
static void bnx2x_get_strings(struct net_device *dev, u32 stringset, u8 *buf)
{
struct bnx2x *bp = netdev_priv(dev);
- int i, j, k;
+ int i, j, k, offset, start;
char queue_name[MAX_QUEUE_NAME_LEN+1];
switch (stringset) {
@@ -2471,7 +2492,17 @@ static void bnx2x_get_strings(struct net_device *dev, u32 stringset, u8 *buf)
break;
case ETH_SS_TEST:
- memcpy(buf, bnx2x_tests_str_arr, sizeof(bnx2x_tests_str_arr));
+ /* First 4 tests cannot be done in MF mode */
+ if (!IS_MF(bp))
+ start = 0;
+ else
+ start = 4;
+ for (i = 0, j = start; j < (start + BNX2X_NUM_TESTS(bp));
+ i++, j++) {
+ offset = sprintf(buf+32*i, "%s",
+ bnx2x_tests_str_arr[j]);
+ *(buf+offset) = '\0';
+ }
break;
}
}
--
1.7.10
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [net-next patch 3/12] bnx2x: Add support for 4-tupple UDP RSS
2012-06-13 12:44 [net-next patch 0/12] bnx2x: ethtool and other enhancements Merav Sicron
2012-06-13 12:44 ` [net-next patch 1/12] bnx2x: Add support for external LB Merav Sicron
2012-06-13 12:44 ` [net-next patch 2/12] bnx2x: Return only online tests for MF Merav Sicron
@ 2012-06-13 12:44 ` Merav Sicron
2012-06-13 18:20 ` Ben Hutchings
2012-06-13 12:44 ` [net-next patch 4/12] bnx2x: Allow more than 64 L2 CIDs Merav Sicron
` (8 subsequent siblings)
11 siblings, 1 reply; 28+ messages in thread
From: Merav Sicron @ 2012-06-13 12:44 UTC (permalink / raw)
To: eilong, davem, netdev; +Cc: Merav Sicron
This change enables to control via ethtool whether to do UDP RSS on 2-tupple
(IP source / destination only) or on 4-tupple (include UDP source / destination
port). It also enables to read back the RSS configuration.
Signed-off-by: Merav Sicron <meravs@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c | 16 ++-
drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h | 8 +-
.../net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c | 147 +++++++++++++++++++-
drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c | 8 ++
drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h | 6 +
5 files changed, 170 insertions(+), 15 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
index 09fcc55..5648078 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
@@ -1651,14 +1651,13 @@ static void bnx2x_set_rx_buf_size(struct bnx2x *bp)
static int bnx2x_init_rss_pf(struct bnx2x *bp)
{
int i;
- u8 ind_table[T_ETH_INDIRECTION_TABLE_SIZE] = {0};
u8 num_eth_queues = BNX2X_NUM_ETH_QUEUES(bp);
/* Prepare the initial contents fo the indirection table if RSS is
* enabled
*/
- for (i = 0; i < sizeof(ind_table); i++)
- ind_table[i] =
+ for (i = 0; i < sizeof(bp->rss_conf_obj.ind_table); i++)
+ bp->rss_conf_obj.ind_table[i] =
bp->fp->cl_id +
ethtool_rxfh_indir_default(i, num_eth_queues);
@@ -1670,12 +1669,11 @@ static int bnx2x_init_rss_pf(struct bnx2x *bp)
* For 57712 and newer on the other hand it's a per-function
* configuration.
*/
- return bnx2x_config_rss_eth(bp, ind_table,
- bp->port.pmf || !CHIP_IS_E1x(bp));
+ return bnx2x_config_rss_eth(bp, bp->port.pmf || !CHIP_IS_E1x(bp));
}
int bnx2x_config_rss_pf(struct bnx2x *bp, struct bnx2x_rss_config_obj *rss_obj,
- u8 *ind_table, bool config_hash)
+ bool config_hash)
{
struct bnx2x_config_rss_params params = {NULL};
int i;
@@ -1698,11 +1696,15 @@ int bnx2x_config_rss_pf(struct bnx2x *bp, struct bnx2x_rss_config_obj *rss_obj,
__set_bit(BNX2X_RSS_IPV4_TCP, ¶ms.rss_flags);
__set_bit(BNX2X_RSS_IPV6, ¶ms.rss_flags);
__set_bit(BNX2X_RSS_IPV6_TCP, ¶ms.rss_flags);
+ if (rss_obj->udp_rss_v4)
+ __set_bit(BNX2X_RSS_IPV4_UDP, ¶ms.rss_flags);
+ if (rss_obj->udp_rss_v6)
+ __set_bit(BNX2X_RSS_IPV6_UDP, ¶ms.rss_flags);
/* Hash bits */
params.rss_result_mask = MULTI_MASK;
- memcpy(params.ind_table, ind_table, sizeof(params.ind_table));
+ memcpy(params.ind_table, rss_obj->ind_table, sizeof(params.ind_table));
if (config_hash) {
/* RSS keys */
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
index 7cd99b7..bb47984 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
@@ -94,7 +94,7 @@ void bnx2x_send_unload_done(struct bnx2x *bp);
* @config_hash: re-configure RSS hash keys configuration
*/
int bnx2x_config_rss_pf(struct bnx2x *bp, struct bnx2x_rss_config_obj *rss_obj,
- u8 *ind_table, bool config_hash);
+ bool config_hash);
/**
* bnx2x__init_func_obj - init function object
@@ -865,11 +865,9 @@ static inline int func_by_vn(struct bnx2x *bp, int vn)
return 2 * vn + BP_PORT(bp);
}
-static inline int bnx2x_config_rss_eth(struct bnx2x *bp, u8 *ind_table,
- bool config_hash)
+static inline int bnx2x_config_rss_eth(struct bnx2x *bp, bool config_hash)
{
- return bnx2x_config_rss_pf(bp, &bp->rss_conf_obj, ind_table,
- config_hash);
+ return bnx2x_config_rss_pf(bp, &bp->rss_conf_obj, config_hash);
}
/**
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
index d7ffc74..36c0277 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
@@ -2600,6 +2600,52 @@ static int bnx2x_set_phys_id(struct net_device *dev,
return 0;
}
+static int bnx2x_get_rss_flags(struct bnx2x *bp, struct ethtool_rxnfc *info)
+{
+
+ switch (info->flow_type) {
+ case TCP_V4_FLOW:
+ case TCP_V6_FLOW:
+ info->data = RXH_IP_SRC | RXH_IP_DST |
+ RXH_L4_B_0_1 | RXH_L4_B_2_3;
+ break;
+ case UDP_V4_FLOW:
+ if (bp->rss_conf_obj.udp_rss_v4)
+ info->data = RXH_IP_SRC | RXH_IP_DST |
+ RXH_L4_B_0_1 | RXH_L4_B_2_3;
+ else
+ info->data = RXH_IP_SRC | RXH_IP_DST;
+ break;
+ case UDP_V6_FLOW:
+ if (bp->rss_conf_obj.udp_rss_v6)
+ info->data = RXH_IP_SRC | RXH_IP_DST |
+ RXH_L4_B_0_1 | RXH_L4_B_2_3;
+ else
+ info->data = RXH_IP_SRC | RXH_IP_DST;
+ break;
+ case IPV4_FLOW:
+ case IPV6_FLOW:
+ info->data = RXH_IP_SRC | RXH_IP_DST;
+ break;
+ case SCTP_V4_FLOW:
+ case AH_ESP_V4_FLOW:
+ case AH_V4_FLOW:
+ case ESP_V4_FLOW:
+ case SCTP_V6_FLOW:
+ case AH_ESP_V6_FLOW:
+ case AH_V6_FLOW:
+ case ESP_V6_FLOW:
+ case IP_USER_FLOW:
+ case ETHER_FLOW:
+ info->data = 0;
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
static int bnx2x_get_rxnfc(struct net_device *dev, struct ethtool_rxnfc *info,
u32 *rules __always_unused)
{
@@ -2609,7 +2655,102 @@ static int bnx2x_get_rxnfc(struct net_device *dev, struct ethtool_rxnfc *info,
case ETHTOOL_GRXRINGS:
info->data = BNX2X_NUM_ETH_QUEUES(bp);
return 0;
+ case ETHTOOL_GRXFH:
+ return bnx2x_get_rss_flags(bp, info);
+ default:
+ DP(BNX2X_MSG_ETHTOOL, "Command parameters not supported\n");
+ return -EOPNOTSUPP;
+ }
+}
+static int bnx2x_set_rss_flags(struct bnx2x *bp, struct ethtool_rxnfc *info)
+{
+ int udp_rss_requested;
+
+ DP(BNX2X_MSG_ETHTOOL,
+ "Set rss flags command parameters: flow type = %d, data = %llu\n",
+ info->flow_type, info->data);
+
+ switch (info->flow_type) {
+ case TCP_V4_FLOW:
+ case TCP_V6_FLOW:
+ /* For TCP only 4-tupple hash is supported */
+ if (info->data ^ (RXH_IP_SRC | RXH_IP_DST |
+ RXH_L4_B_0_1 | RXH_L4_B_2_3)) {
+ DP(BNX2X_MSG_ETHTOOL,
+ "Command parameters not supported\n");
+ return -EINVAL;
+ } else {
+ return 0;
+ }
+
+ case UDP_V4_FLOW:
+ case UDP_V6_FLOW:
+ /* For UDP either 2-tupple hash or 4-tupple hash is supported */
+ if (info->data == (RXH_IP_SRC | RXH_IP_DST |
+ RXH_L4_B_0_1 | RXH_L4_B_2_3))
+ udp_rss_requested = 1;
+ else if (info->data == (RXH_IP_SRC | RXH_IP_DST))
+ udp_rss_requested = 0;
+ else
+ return -EINVAL;
+ if ((info->flow_type == UDP_V4_FLOW) &&
+ (bp->rss_conf_obj.udp_rss_v4 != udp_rss_requested)) {
+ bp->rss_conf_obj.udp_rss_v4 = udp_rss_requested;
+ DP(BNX2X_MSG_ETHTOOL,
+ "rss re-configured, UDP 4-tupple %s\n",
+ udp_rss_requested ? "enabled" : "disabled");
+ return bnx2x_config_rss_pf(bp, &bp->rss_conf_obj, 0);
+ } else if ((info->flow_type == UDP_V6_FLOW) &&
+ (bp->rss_conf_obj.udp_rss_v6 != udp_rss_requested)) {
+ bp->rss_conf_obj.udp_rss_v6 = udp_rss_requested;
+ return bnx2x_config_rss_pf(bp, &bp->rss_conf_obj, 0);
+ DP(BNX2X_MSG_ETHTOOL,
+ "rss re-configured, UDP 4-tupple %s\n",
+ udp_rss_requested ? "enabled" : "disabled");
+ } else {
+ return 0;
+ }
+ case IPV4_FLOW:
+ case IPV6_FLOW:
+ /* For IP only 2-tupple hash is supported */
+ if (info->data ^ (RXH_IP_SRC | RXH_IP_DST)) {
+ DP(BNX2X_MSG_ETHTOOL,
+ "Command parameters not supported\n");
+ return -EINVAL;
+ } else {
+ return 0;
+ }
+ case SCTP_V4_FLOW:
+ case AH_ESP_V4_FLOW:
+ case AH_V4_FLOW:
+ case ESP_V4_FLOW:
+ case SCTP_V6_FLOW:
+ case AH_ESP_V6_FLOW:
+ case AH_V6_FLOW:
+ case ESP_V6_FLOW:
+ case IP_USER_FLOW:
+ case ETHER_FLOW:
+ /* RSS is not supported for these protocols */
+ if (info->data) {
+ DP(BNX2X_MSG_ETHTOOL,
+ "Command parameters not supported\n");
+ return -EINVAL;
+ } else {
+ return 0;
+ }
+ default:
+ return -EINVAL;
+ }
+}
+
+static int bnx2x_set_rxnfc(struct net_device *dev, struct ethtool_rxnfc *info)
+{
+ struct bnx2x *bp = netdev_priv(dev);
+
+ switch (info->cmd) {
+ case ETHTOOL_SRXFH:
+ return bnx2x_set_rss_flags(bp, info);
default:
DP(BNX2X_MSG_ETHTOOL, "Command parameters not supported\n");
return -EOPNOTSUPP;
@@ -2649,7 +2790,6 @@ static int bnx2x_set_rxfh_indir(struct net_device *dev, const u32 *indir)
{
struct bnx2x *bp = netdev_priv(dev);
size_t i;
- u8 ind_table[T_ETH_INDIRECTION_TABLE_SIZE] = {0};
for (i = 0; i < T_ETH_INDIRECTION_TABLE_SIZE; i++) {
/*
@@ -2661,10 +2801,10 @@ static int bnx2x_set_rxfh_indir(struct net_device *dev, const u32 *indir)
* align the received table to the Client ID of the leading RSS
* queue
*/
- ind_table[i] = indir[i] + bp->fp->cl_id;
+ bp->rss_conf_obj.ind_table[i] = indir[i] + bp->fp->cl_id;
}
- return bnx2x_config_rss_eth(bp, ind_table, false);
+ return bnx2x_config_rss_eth(bp, false);
}
static const struct ethtool_ops bnx2x_ethtool_ops = {
@@ -2694,6 +2834,7 @@ static const struct ethtool_ops bnx2x_ethtool_ops = {
.set_phys_id = bnx2x_set_phys_id,
.get_ethtool_stats = bnx2x_get_ethtool_stats,
.get_rxnfc = bnx2x_get_rxnfc,
+ .set_rxnfc = bnx2x_set_rxnfc,
.get_rxfh_indir_size = bnx2x_get_rxfh_indir_size,
.get_rxfh_indir = bnx2x_get_rxfh_indir,
.set_rxfh_indir = bnx2x_set_rxfh_indir,
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c
index 6c14b4a..734fd87 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c
@@ -4107,6 +4107,10 @@ static int bnx2x_setup_rss(struct bnx2x *bp,
data->capabilities |=
ETH_RSS_UPDATE_RAMROD_DATA_IPV4_TCP_CAPABILITY;
+ if (test_bit(BNX2X_RSS_IPV4_UDP, &p->rss_flags))
+ data->capabilities |=
+ ETH_RSS_UPDATE_RAMROD_DATA_IPV4_UDP_CAPABILITY;
+
if (test_bit(BNX2X_RSS_IPV6, &p->rss_flags))
data->capabilities |=
ETH_RSS_UPDATE_RAMROD_DATA_IPV6_CAPABILITY;
@@ -4115,6 +4119,10 @@ static int bnx2x_setup_rss(struct bnx2x *bp,
data->capabilities |=
ETH_RSS_UPDATE_RAMROD_DATA_IPV6_TCP_CAPABILITY;
+ if (test_bit(BNX2X_RSS_IPV6_UDP, &p->rss_flags))
+ data->capabilities |=
+ ETH_RSS_UPDATE_RAMROD_DATA_IPV6_UDP_CAPABILITY;
+
/* Hashing mask */
data->rss_result_mask = p->rss_result_mask;
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h
index efd80bd..76818ef 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h
@@ -694,8 +694,10 @@ enum {
BNX2X_RSS_IPV4,
BNX2X_RSS_IPV4_TCP,
+ BNX2X_RSS_IPV4_UDP,
BNX2X_RSS_IPV6,
BNX2X_RSS_IPV6_TCP,
+ BNX2X_RSS_IPV6_UDP,
};
struct bnx2x_config_rss_params {
@@ -729,6 +731,10 @@ struct bnx2x_rss_config_obj {
/* Last configured indirection table */
u8 ind_table[T_ETH_INDIRECTION_TABLE_SIZE];
+ /* flags for enabling 4-tupple hash on UDP */
+ u8 udp_rss_v4;
+ u8 udp_rss_v6;
+
int (*config_rss)(struct bnx2x *bp,
struct bnx2x_config_rss_params *p);
};
--
1.7.10
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [net-next patch 4/12] bnx2x: Allow more than 64 L2 CIDs
2012-06-13 12:44 [net-next patch 0/12] bnx2x: ethtool and other enhancements Merav Sicron
` (2 preceding siblings ...)
2012-06-13 12:44 ` [net-next patch 3/12] bnx2x: Add support for 4-tupple UDP RSS Merav Sicron
@ 2012-06-13 12:44 ` Merav Sicron
2012-06-13 12:44 ` [net-next patch 5/12] bnx2x: Make the transmission queues adjacent Merav Sicron
` (7 subsequent siblings)
11 siblings, 0 replies; 28+ messages in thread
From: Merav Sicron @ 2012-06-13 12:44 UTC (permalink / raw)
To: eilong, davem, netdev; +Cc: Merav Sicron
With increased number of RSS queues, each multiplied by the number of traffic-
classes, we may have up to 64*3=192 CIDs. The current driver scheme with regard
to context allocation supports only 64 CIDs. The new scheme enables scatter-
gatehr list of pages for the context.
Signed-off-by: Merav Sicron <meravs@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x.h | 10 +++-
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c | 67 ++++++++++++++++------
2 files changed, 57 insertions(+), 20 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index 568c3c7..feec032 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -993,8 +993,8 @@ union cdu_context {
};
/* CDU host DB constants */
-#define CDU_ILT_PAGE_SZ_HW 3
-#define CDU_ILT_PAGE_SZ (8192 << CDU_ILT_PAGE_SZ_HW) /* 64K */
+#define CDU_ILT_PAGE_SZ_HW 2
+#define CDU_ILT_PAGE_SZ (8192 << CDU_ILT_PAGE_SZ_HW) /* 32K */
#define ILT_PAGE_CIDS (CDU_ILT_PAGE_SZ / sizeof(union cdu_context))
#ifdef BCM_CNIC
@@ -1435,7 +1435,11 @@ struct bnx2x {
dma_addr_t fw_stats_data_mapping;
int fw_stats_data_sz;
- struct hw_context context;
+ /* For max 196 cids (64*3 + non-eth), 32KB ILT page size and 1KB
+ * context size we need 8 ILT entries.
+ */
+#define ILT_MAX_L2_LINES 8
+ struct hw_context context[ILT_MAX_L2_LINES];
struct bnx2x_ilt *ilt;
#define BP_ILT(bp) ((bp)->ilt)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index 21930be..16e009f 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -7068,12 +7068,10 @@ static int bnx2x_init_hw_func(struct bnx2x *bp)
cdu_ilt_start = ilt->clients[ILT_CLIENT_CDU].start;
for (i = 0; i < L2_ILT_LINES(bp); i++) {
- ilt->lines[cdu_ilt_start + i].page =
- bp->context.vcxt + (ILT_PAGE_CIDS * i);
+ ilt->lines[cdu_ilt_start + i].page = bp->context[i].vcxt;
ilt->lines[cdu_ilt_start + i].page_mapping =
- bp->context.cxt_mapping + (CDU_ILT_PAGE_SZ * i);
- /* cdu ilt pages are allocated manually so there's no need to
- set the size */
+ bp->context[i].cxt_mapping;
+ ilt->lines[cdu_ilt_start + i].size = bp->context[i].size;
}
bnx2x_ilt_init_op(bp, INITOP_SET);
@@ -7340,6 +7338,8 @@ static int bnx2x_init_hw_func(struct bnx2x *bp)
void bnx2x_free_mem(struct bnx2x *bp)
{
+ int i;
+
/* fastpath */
bnx2x_free_fp_mem(bp);
/* end of fastpath */
@@ -7353,9 +7353,9 @@ void bnx2x_free_mem(struct bnx2x *bp)
BNX2X_PCI_FREE(bp->slowpath, bp->slowpath_mapping,
sizeof(struct bnx2x_slowpath));
- BNX2X_PCI_FREE(bp->context.vcxt, bp->context.cxt_mapping,
- bp->context.size);
-
+ for (i = 0; i < L2_ILT_LINES(bp); i++)
+ BNX2X_PCI_FREE(bp->context[i].vcxt, bp->context[i].cxt_mapping,
+ bp->context[i].size);
bnx2x_ilt_mem_op(bp, ILT_MEMOP_FREE);
BNX2X_FREE(bp->ilt->lines);
@@ -7441,6 +7441,8 @@ alloc_mem_err:
int bnx2x_alloc_mem(struct bnx2x *bp)
{
+ int i, allocated, context_size;
+
#ifdef BCM_CNIC
if (!CHIP_IS_E1x(bp))
/* size = the status block + ramrod buffers */
@@ -7470,11 +7472,29 @@ int bnx2x_alloc_mem(struct bnx2x *bp)
if (bnx2x_alloc_fw_stats_mem(bp))
goto alloc_mem_err;
- bp->context.size = sizeof(union cdu_context) * BNX2X_L2_CID_COUNT(bp);
-
- BNX2X_PCI_ALLOC(bp->context.vcxt, &bp->context.cxt_mapping,
- bp->context.size);
+ /* Allocate memory for CDU context:
+ * This memory is allocated separately and not in the generic ILT
+ * functions because CDU differs in few aspects:
+ * 1. There are multiple entities allocating memory for context -
+ * 'regular' driver, CNIC and SRIOV driver. Each separately controls
+ * its own ILT lines.
+ * 2. Since CDU page-size is not a single 4KB page (which is the case
+ * for the other ILT clients), to be efficient we want to support
+ * allocation of sub-page-size in the last entry.
+ * 3. Context pointers are used by the driver to pass to FW / update
+ * the context (for the other ILT clients the pointers are used just to
+ * free the memory during unload).
+ */
+ context_size = sizeof(union cdu_context) * BNX2X_L2_CID_COUNT(bp);
+ for (i = 0, allocated = 0; allocated < context_size; i++) {
+ bp->context[i].size = min(CDU_ILT_PAGE_SZ,
+ (context_size - allocated));
+ BNX2X_PCI_ALLOC(bp->context[i].vcxt,
+ &bp->context[i].cxt_mapping,
+ bp->context[i].size);
+ allocated += bp->context[i].size;
+ }
BNX2X_ALLOC(bp->ilt->lines, sizeof(struct ilt_line) * ILT_MAX_LINES);
if (bnx2x_ilt_mem_op(bp, ILT_MEMOP_ALLOC))
@@ -7748,6 +7768,8 @@ static void bnx2x_pf_q_prep_init(struct bnx2x *bp,
{
u8 cos;
+ int cxt_index, cxt_offset;
+
/* FCoE Queue uses Default SB, thus has no HC capabilities */
if (!IS_FCOE_FP(fp)) {
__set_bit(BNX2X_Q_FLG_HC, &init_params->rx.flags);
@@ -7784,9 +7806,13 @@ static void bnx2x_pf_q_prep_init(struct bnx2x *bp,
fp->index, init_params->max_cos);
/* set the context pointers queue object */
- for (cos = FIRST_TX_COS_INDEX; cos < init_params->max_cos; cos++)
+ for (cos = FIRST_TX_COS_INDEX; cos < init_params->max_cos; cos++) {
+ cxt_index = fp->txdata[cos].cid / ILT_PAGE_CIDS;
+ cxt_offset = fp->txdata[cos].cid - (cxt_index *
+ ILT_PAGE_CIDS);
init_params->cxts[cos] =
- &bp->context.vcxt[fp->txdata[cos].cid].eth;
+ &bp->context[cxt_index].vcxt[cxt_offset].eth;
+ }
}
int bnx2x_setup_tx_only(struct bnx2x *bp, struct bnx2x_fastpath *fp,
@@ -12202,6 +12228,7 @@ static int bnx2x_set_iscsi_eth_mac_addr(struct bnx2x *bp)
static void bnx2x_cnic_sp_post(struct bnx2x *bp, int count)
{
struct eth_spe *spe;
+ int cxt_index, cxt_offset;
#ifdef BNX2X_STOP_ON_ERROR
if (unlikely(bp->panic))
@@ -12224,10 +12251,16 @@ static void bnx2x_cnic_sp_post(struct bnx2x *bp, int count)
* ramrod
*/
if (type == ETH_CONNECTION_TYPE) {
- if (cmd == RAMROD_CMD_ID_ETH_CLIENT_SETUP)
- bnx2x_set_ctx_validation(bp, &bp->context.
- vcxt[BNX2X_ISCSI_ETH_CID].eth,
+ if (cmd == RAMROD_CMD_ID_ETH_CLIENT_SETUP) {
+ cxt_index = BNX2X_ISCSI_ETH_CID /
+ ILT_PAGE_CIDS;
+ cxt_offset = BNX2X_ISCSI_ETH_CID -
+ (cxt_index * ILT_PAGE_CIDS);
+ bnx2x_set_ctx_validation(bp,
+ &bp->context[cxt_index].
+ vcxt[cxt_offset].eth,
BNX2X_ISCSI_ETH_CID);
+ }
}
/*
--
1.7.10
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [net-next patch 5/12] bnx2x: Make the transmission queues adjacent
2012-06-13 12:44 [net-next patch 0/12] bnx2x: ethtool and other enhancements Merav Sicron
` (3 preceding siblings ...)
2012-06-13 12:44 ` [net-next patch 4/12] bnx2x: Allow more than 64 L2 CIDs Merav Sicron
@ 2012-06-13 12:44 ` Merav Sicron
2012-06-13 12:44 ` [net-next patch 6/12] bnx2x: Move the CNIC L2 CIDs to be right after the RSS CIDs Merav Sicron
` (6 subsequent siblings)
11 siblings, 0 replies; 28+ messages in thread
From: Merav Sicron @ 2012-06-13 12:44 UTC (permalink / raw)
To: eilong, davem, netdev; +Cc: Merav Sicron
In the current scheme the transmission queues of traffic-class 0 were 0-15, the
transmission queues of traffic-class 1 were 16-31 and so on. If the number of
RSS queues was smaller than 16, there were gaps in transmission queues
numbering, as well as in CIDs numbering. This is both a waste (especially when
16 is increased to 64), and may causes problems with flushing queues when
reducing the number of RSS queues (using ethtool -L). The new scheme eliminates
the gaps.
Signed-off-by: Merav Sicron <meravs@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x.h | 66 +++++++-----
drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c | 105 +++++++++++++-------
drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h | 14 +--
.../net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c | 2 +-
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c | 42 ++++----
drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c | 4 +-
6 files changed, 139 insertions(+), 94 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index feec032..36ebc3d 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -276,29 +276,32 @@ enum {
#define FIRST_TX_ONLY_COS_INDEX 1
#define FIRST_TX_COS_INDEX 0
-/* defines for decodeing the fastpath index and the cos index out of the
- * transmission queue index
- */
#define MAX_TXQS_PER_COS FP_SB_MAX_E1x
-#define TXQ_TO_FP(txq_index) ((txq_index) % MAX_TXQS_PER_COS)
-#define TXQ_TO_COS(txq_index) ((txq_index) / MAX_TXQS_PER_COS)
-
/* rules for calculating the cids of tx-only connections */
-#define CID_TO_FP(cid) ((cid) % MAX_TXQS_PER_COS)
-#define CID_COS_TO_TX_ONLY_CID(cid, cos) (cid + cos * MAX_TXQS_PER_COS)
+#define CID_TO_FP(cid, bp) ((cid) % BNX2X_NUM_NON_CNIC_QUEUES(bp))
+#define CID_COS_TO_TX_ONLY_CID(cid, cos, bp) \
+ (cid + cos * BNX2X_NUM_NON_CNIC_QUEUES(bp))
/* fp index inside class of service range */
-#define FP_COS_TO_TXQ(fp, cos) ((fp)->index + cos * MAX_TXQS_PER_COS)
-
-/*
- * 0..15 eth cos0
- * 16..31 eth cos1 if applicable
- * 32..47 eth cos2 If applicable
- * fcoe queue follows eth queues (16, 32, 48 depending on cos)
+#define FP_COS_TO_TXQ(fp, cos, bp) \
+ ((fp)->index + cos * BNX2X_NUM_NON_CNIC_QUEUES(bp))
+
+/* Indexes for transmission queues array:
+ * txdata for RSS i CoS j is at location i + (j * num of RSS)
+ * txdata for FCoE (if exist) is at location max cos * num of RSS
+ * txdata for FWD (if exist) is one location after FCoE
+ * txdata for OOO (if exist) is one location after FWD
*/
-#define MAX_ETH_TXQ_IDX(bp) (MAX_TXQS_PER_COS * (bp)->max_cos)
-#define FCOE_TXQ_IDX(bp) (MAX_ETH_TXQ_IDX(bp))
+enum {
+ FCOE_TXQ_IDX_OFFSET,
+ FWD_TXQ_IDX_OFFSET,
+ OOO_TXQ_IDX_OFFSET,
+};
+#define MAX_ETH_TXQ_IDX(bp) (BNX2X_NUM_NON_CNIC_QUEUES(bp) * (bp)->max_cos)
+#ifdef BCM_CNIC
+#define FCOE_TXQ_IDX(bp) (MAX_ETH_TXQ_IDX(bp) + FCOE_TXQ_IDX_OFFSET)
+#endif
/* fast path */
/*
@@ -481,6 +484,8 @@ struct bnx2x_fp_txdata {
__le16 *tx_cons_sb;
int txq_index;
+ struct bnx2x_fastpath *parent_fp;
+ int tx_ring_size;
};
enum bnx2x_tpa_mode_t {
@@ -507,7 +512,7 @@ struct bnx2x_fastpath {
enum bnx2x_tpa_mode_t mode;
u8 max_cos; /* actual number of active tx coses */
- struct bnx2x_fp_txdata txdata[BNX2X_MULTI_TX_COS];
+ struct bnx2x_fp_txdata *txdata_ptr[BNX2X_MULTI_TX_COS];
struct sw_rx_bd *rx_buf_ring; /* BDs mappings ring */
struct sw_rx_page *rx_page_ring; /* SGE pages mappings ring */
@@ -579,19 +584,22 @@ struct bnx2x_fastpath {
/* Use 2500 as a mini-jumbo MTU for FCoE */
#define BNX2X_FCOE_MINI_JUMBO_MTU 2500
-/* FCoE L2 `fastpath' entry is right after the eth entries */
-#define FCOE_IDX BNX2X_NUM_ETH_QUEUES(bp)
-#define bnx2x_fcoe_fp(bp) (&bp->fp[FCOE_IDX])
-#define bnx2x_fcoe(bp, var) (bnx2x_fcoe_fp(bp)->var)
-#define bnx2x_fcoe_tx(bp, var) (bnx2x_fcoe_fp(bp)-> \
- txdata[FIRST_TX_COS_INDEX].var)
+#define FCOE_IDX_OFFSET 0
+
+#define FCOE_IDX(bp) (BNX2X_NUM_NON_CNIC_QUEUES(bp) + \
+ FCOE_IDX_OFFSET)
+#define bnx2x_fcoe_fp(bp) (&bp->fp[FCOE_IDX(bp)])
+#define bnx2x_fcoe(bp, var) (bnx2x_fcoe_fp(bp)->var)
+#define bnx2x_fcoe_tx(bp, var) (bnx2x_fcoe_fp(bp)-> \
+ txdata_ptr[FIRST_TX_COS_INDEX] \
+ ->var)
#define IS_ETH_FP(fp) (fp->index < \
BNX2X_NUM_ETH_QUEUES(fp->bp))
#ifdef BCM_CNIC
-#define IS_FCOE_FP(fp) (fp->index == FCOE_IDX)
-#define IS_FCOE_IDX(idx) ((idx) == FCOE_IDX)
+#define IS_FCOE_FP(fp) (fp->index == FCOE_IDX(fp->bp))
+#define IS_FCOE_IDX(idx) ((idx) == FCOE_IDX(bp))
#else
#define IS_FCOE_FP(fp) false
#define IS_FCOE_IDX(idx) false
@@ -1202,6 +1210,8 @@ struct bnx2x {
* are grouped together in the beginning of the structure
*/
struct bnx2x_fastpath *fp;
+ struct bnx2x_fp_txdata *bnx2x_txq;
+ int bnx2x_txq_size;
void __iomem *regview;
void __iomem *doorbells;
u16 db_size;
@@ -1404,6 +1414,7 @@ struct bnx2x {
u8 igu_dsb_id;
u8 igu_base_sb;
u8 igu_sb_cnt;
+
dma_addr_t def_status_blk_mapping;
struct bnx2x_slowpath *slowpath;
@@ -1458,7 +1469,6 @@ struct bnx2x {
NON_ETH_CONTEXT_USE + CNIC_PRESENT)
#define L2_ILT_LINES(bp) (DIV_ROUND_UP(BNX2X_L2_CID_COUNT(bp),\
ILT_PAGE_CIDS))
-#define BNX2X_DB_SIZE(bp) (BNX2X_L2_CID_COUNT(bp) * (1 << BNX2X_DB_SHIFT))
int qm_cid_count;
@@ -1617,6 +1627,8 @@ struct bnx2x {
extern int num_queues;
#define BNX2X_NUM_QUEUES(bp) (bp->num_queues)
#define BNX2X_NUM_ETH_QUEUES(bp) (BNX2X_NUM_QUEUES(bp) - NON_ETH_CONTEXT_USE)
+#define BNX2X_NUM_NON_CNIC_QUEUES(bp) (BNX2X_NUM_QUEUES(bp) - \
+ NON_ETH_CONTEXT_USE)
#define BNX2X_NUM_RX_QUEUES(bp) BNX2X_NUM_QUEUES(bp)
#define is_multi(bp) (BNX2X_NUM_QUEUES(bp) > 1)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
index 5648078..33132b1 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
@@ -40,12 +40,15 @@
* Makes sure the contents of the bp->fp[to].napi is kept
* intact. This is done by first copying the napi struct from
* the target to the source, and then mem copying the entire
- * source onto the target
+ * source onto the target. Update txdata pointers and related
+ * content.
*/
static inline void bnx2x_move_fp(struct bnx2x *bp, int from, int to)
{
struct bnx2x_fastpath *from_fp = &bp->fp[from];
struct bnx2x_fastpath *to_fp = &bp->fp[to];
+ int old_max_eth_txqs, new_max_eth_txqs;
+ int old_txdata_index = 0, new_txdata_index = 0;
/* Copy the NAPI object as it has been already initialized */
from_fp->napi = to_fp->napi;
@@ -53,6 +56,24 @@ static inline void bnx2x_move_fp(struct bnx2x *bp, int from, int to)
/* Move bnx2x_fastpath contents */
memcpy(to_fp, from_fp, sizeof(*to_fp));
to_fp->index = to;
+
+ /* Update txdata pointers in fp and move txdata content accordingly:
+ * Each fp consumes 'max_cos' txdata structures, so the index should be
+ * decremented by max_cos x delta.
+ */
+
+ old_max_eth_txqs = BNX2X_NUM_ETH_QUEUES(bp) * (bp)->max_cos;
+ new_max_eth_txqs = (BNX2X_NUM_ETH_QUEUES(bp) - from + to) *
+ (bp)->max_cos;
+ if (from == FCOE_IDX(bp)) {
+ old_txdata_index = old_max_eth_txqs + FCOE_TXQ_IDX_OFFSET;
+ new_txdata_index = new_max_eth_txqs + FCOE_TXQ_IDX_OFFSET;
+ }
+
+ memcpy(&bp->bnx2x_txq[old_txdata_index],
+ &bp->bnx2x_txq[new_txdata_index],
+ sizeof(struct bnx2x_fp_txdata));
+ to_fp->txdata_ptr[0] = &bp->bnx2x_txq[new_txdata_index];
}
int load_count[2][3] = { {0} }; /* per-path: 0-common, 1-port0, 2-port1 */
@@ -873,7 +894,7 @@ static irqreturn_t bnx2x_msix_fp_int(int irq, void *fp_cookie)
prefetch(fp->rx_cons_sb);
for_each_cos_in_tx_queue(fp, cos)
- prefetch(fp->txdata[cos].tx_cons_sb);
+ prefetch(fp->txdata_ptr[cos]->tx_cons_sb);
prefetch(&fp->sb_running_index[SM_RX_ID]);
napi_schedule(&bnx2x_fp(bp, fp->index, napi));
@@ -1190,7 +1211,7 @@ static void bnx2x_free_tx_skbs(struct bnx2x *bp)
for_each_tx_queue(bp, i) {
struct bnx2x_fastpath *fp = &bp->fp[i];
for_each_cos_in_tx_queue(fp, cos) {
- struct bnx2x_fp_txdata *txdata = &fp->txdata[cos];
+ struct bnx2x_fp_txdata *txdata = fp->txdata_ptr[cos];
unsigned pkts_compl = 0, bytes_compl = 0;
u16 sw_prod = txdata->tx_pkt_prod;
@@ -1202,7 +1223,8 @@ static void bnx2x_free_tx_skbs(struct bnx2x *bp)
sw_cons++;
}
netdev_tx_reset_queue(
- netdev_get_tx_queue(bp->dev, txdata->txq_index));
+ netdev_get_tx_queue(bp->dev,
+ txdata->txq_index));
}
}
}
@@ -1564,6 +1586,8 @@ void bnx2x_set_num_queues(struct bnx2x *bp)
#endif
/* Add special queues */
bp->num_queues += NON_ETH_CONTEXT_USE;
+
+ BNX2X_DEV_INFO("set number of queues to %d\n", bp->num_queues);
}
/**
@@ -1592,8 +1616,8 @@ static int bnx2x_set_real_num_queues(struct bnx2x *bp)
{
int rc, tx, rx;
- tx = MAX_TXQS_PER_COS * bp->max_cos;
- rx = BNX2X_NUM_ETH_QUEUES(bp);
+ tx = BNX2X_NUM_ETH_QUEUES(bp) * bp->max_cos;
+ rx = BNX2X_NUM_QUEUES(bp) - NON_ETH_CONTEXT_USE;
/* account for fcoe queue */
#ifdef BCM_CNIC
@@ -1838,6 +1862,7 @@ bool bnx2x_test_firmware_version(struct bnx2x *bp, bool is_err)
static void bnx2x_bz_fp(struct bnx2x *bp, int index)
{
struct bnx2x_fastpath *fp = &bp->fp[index];
+ int cos;
struct napi_struct orig_napi = fp->napi;
/* bzero bnx2x_fastpath contents */
if (bp->stats_init)
@@ -1887,6 +1912,16 @@ static void bnx2x_bz_fp(struct bnx2x *bp, int index)
/* Special queues support only one CoS */
fp->max_cos = 1;
+ /* Init txdata pointers */
+#ifdef BCM_CNIC
+ if (IS_FCOE_FP(fp))
+ fp->txdata_ptr[0] = &bp->bnx2x_txq[FCOE_TXQ_IDX(bp)];
+#endif
+ if (IS_ETH_FP(fp))
+ for_each_cos_in_tx_queue(fp, cos)
+ fp->txdata_ptr[cos] = &bp->bnx2x_txq[cos *
+ BNX2X_NUM_ETH_QUEUES(bp) + index];
+
/*
* set the tpa flag for each queue. The tpa flag determines the queue
* minimal size so it must be set prior to queue memory allocation
@@ -1936,11 +1971,13 @@ int bnx2x_nic_load(struct bnx2x *bp, int load_mode)
/*
* Zero fastpath structures preserving invariants like napi, which are
* allocated only once, fp index, max_cos, bp pointer.
- * Also set fp->disable_tpa.
+ * Also set fp->disable_tpa and txdata_ptr.
*/
DP(NETIF_MSG_IFUP, "num queues: %d", bp->num_queues);
for_each_queue(bp, i)
bnx2x_bz_fp(bp, i);
+ memset(bp->bnx2x_txq, 0, bp->bnx2x_txq_size *
+ sizeof(struct bnx2x_fp_txdata));
/* Set the receive queues buffer size */
@@ -2287,6 +2324,7 @@ int bnx2x_nic_unload(struct bnx2x *bp, int unload_mode)
/* Stop Tx */
bnx2x_tx_disable(bp);
+ netdev_reset_tc(bp->dev);
#ifdef BCM_CNIC
bnx2x_cnic_notify(bp, CNIC_CTL_STOP_CMD);
@@ -2445,8 +2483,8 @@ int bnx2x_poll(struct napi_struct *napi, int budget)
#endif
for_each_cos_in_tx_queue(fp, cos)
- if (bnx2x_tx_queue_has_work(&fp->txdata[cos]))
- bnx2x_tx_int(bp, &fp->txdata[cos]);
+ if (bnx2x_tx_queue_has_work(fp->txdata_ptr[cos]))
+ bnx2x_tx_int(bp, fp->txdata_ptr[cos]);
if (bnx2x_has_rx_work(fp)) {
@@ -2825,7 +2863,6 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev)
{
struct bnx2x *bp = netdev_priv(dev);
- struct bnx2x_fastpath *fp;
struct netdev_queue *txq;
struct bnx2x_fp_txdata *txdata;
struct sw_tx_bd *tx_buf;
@@ -2835,7 +2872,7 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev)
struct eth_tx_parse_bd_e2 *pbd_e2 = NULL;
u32 pbd_e2_parsing_data = 0;
u16 pkt_prod, bd_prod;
- int nbd, txq_index, fp_index, txdata_index;
+ int nbd, txq_index;
dma_addr_t mapping;
u32 xmit_type = bnx2x_xmit_type(bp, skb);
int i;
@@ -2854,31 +2891,12 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev)
BUG_ON(txq_index >= MAX_ETH_TXQ_IDX(bp) + FCOE_PRESENT);
- /* decode the fastpath index and the cos index from the txq */
- fp_index = TXQ_TO_FP(txq_index);
- txdata_index = TXQ_TO_COS(txq_index);
-
-#ifdef BCM_CNIC
- /*
- * Override the above for the FCoE queue:
- * - FCoE fp entry is right after the ETH entries.
- * - FCoE L2 queue uses bp->txdata[0] only.
- */
- if (unlikely(!NO_FCOE(bp) && (txq_index ==
- bnx2x_fcoe_tx(bp, txq_index)))) {
- fp_index = FCOE_IDX;
- txdata_index = 0;
- }
-#endif
+ txdata = &bp->bnx2x_txq[txq_index];
/* enable this debug print to view the transmission queue being used
DP(NETIF_MSG_TX_QUEUED, "indices: txq %d, fp %d, txdata %d\n",
txq_index, fp_index, txdata_index); */
- /* locate the fastpath and the txdata */
- fp = &bp->fp[fp_index];
- txdata = &fp->txdata[txdata_index];
-
/* enable this debug print to view the tranmission details
DP(NETIF_MSG_TX_QUEUED,
"transmitting packet cid %d fp index %d txdata_index %d tx_data ptr %p fp pointer %p\n",
@@ -2886,7 +2904,7 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev)
if (unlikely(bnx2x_tx_avail(bp, txdata) <
(skb_shinfo(skb)->nr_frags + 3))) {
- fp->eth_q_stats.driver_xoff++;
+ txdata->parent_fp->eth_q_stats.driver_xoff++;
netif_tx_stop_queue(txq);
BNX2X_ERR("BUG! Tx ring full when queue awake!\n");
return NETDEV_TX_BUSY;
@@ -3168,7 +3186,7 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev)
* fp->bd_tx_cons */
smp_mb();
- fp->eth_q_stats.driver_xoff++;
+ txdata->parent_fp->eth_q_stats.driver_xoff++;
if (bnx2x_tx_avail(bp, txdata) >= MAX_SKB_FRAGS + 3)
netif_tx_wake_queue(txq);
}
@@ -3234,7 +3252,7 @@ int bnx2x_setup_tc(struct net_device *dev, u8 num_tc)
/* configure traffic class to transmission queue mapping */
for (cos = 0; cos < bp->max_cos; cos++) {
count = BNX2X_NUM_ETH_QUEUES(bp);
- offset = cos * MAX_TXQS_PER_COS;
+ offset = cos * BNX2X_NUM_NON_CNIC_QUEUES(bp);
netdev_set_tc_queue(dev, cos, count, offset);
DP(BNX2X_MSG_SP | NETIF_MSG_IFUP,
"mapping tc %d to offset %d count %d\n",
@@ -3333,7 +3351,7 @@ static void bnx2x_free_fp_mem_at(struct bnx2x *bp, int fp_index)
if (!skip_tx_queue(bp, fp_index)) {
/* fastpath tx rings: tx_buf tx_desc */
for_each_cos_in_tx_queue(fp, cos) {
- struct bnx2x_fp_txdata *txdata = &fp->txdata[cos];
+ struct bnx2x_fp_txdata *txdata = fp->txdata_ptr[cos];
DP(NETIF_MSG_IFDOWN,
"freeing tx memory of fp %d cos %d cid %d\n",
@@ -3490,7 +3508,7 @@ static int bnx2x_alloc_fp_mem_at(struct bnx2x *bp, int index)
if (!skip_tx_queue(bp, index)) {
/* fastpath tx rings: tx_buf tx_desc */
for_each_cos_in_tx_queue(fp, cos) {
- struct bnx2x_fp_txdata *txdata = &fp->txdata[cos];
+ struct bnx2x_fp_txdata *txdata = fp->txdata_ptr[cos];
DP(NETIF_MSG_IFUP,
"allocating tx memory of fp %d cos %d\n",
@@ -3573,7 +3591,7 @@ int bnx2x_alloc_fp_mem(struct bnx2x *bp)
#ifdef BCM_CNIC
if (!NO_FCOE(bp))
/* FCoE */
- if (bnx2x_alloc_fp_mem_at(bp, FCOE_IDX))
+ if (bnx2x_alloc_fp_mem_at(bp, FCOE_IDX(bp)))
/* we will fail load process instead of mark
* NO_FCOE_FLAG
*/
@@ -3598,7 +3616,7 @@ int bnx2x_alloc_fp_mem(struct bnx2x *bp)
*/
/* move FCoE fp even NO_FCOE_FLAG is on */
- bnx2x_move_fp(bp, FCOE_IDX, FCOE_IDX - delta);
+ bnx2x_move_fp(bp, FCOE_IDX(bp), FCOE_IDX(bp) - delta);
#endif
bp->num_queues -= delta;
BNX2X_ERR("Adjusted num of queues from %d to %d\n",
@@ -3611,6 +3629,7 @@ int bnx2x_alloc_fp_mem(struct bnx2x *bp)
void bnx2x_free_mem_bp(struct bnx2x *bp)
{
kfree(bp->fp);
+ kfree(bp->bnx2x_txq);
kfree(bp->msix_table);
kfree(bp->ilt);
}
@@ -3635,6 +3654,16 @@ int __devinit bnx2x_alloc_mem_bp(struct bnx2x *bp)
goto alloc_err;
bp->fp = fp;
+ /* Allocate memory for the transmission queues array */
+ bp->bnx2x_txq_size = BNX2X_MAX_RSS_COUNT(bp) * BNX2X_MULTI_TX_COS;
+#ifdef BCM_CNIC
+ bp->bnx2x_txq_size++;
+#endif
+ bp->bnx2x_txq = kcalloc(bp->bnx2x_txq_size,
+ sizeof(struct bnx2x_fp_txdata), GFP_KERNEL);
+ if (!bp->bnx2x_txq)
+ goto alloc_err;
+
/* msix table */
tbl = kcalloc(msix_table_size, sizeof(*tbl), GFP_KERNEL);
if (!tbl)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
index bb47984..112ffcc 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
@@ -728,7 +728,7 @@ static inline bool bnx2x_has_tx_work(struct bnx2x_fastpath *fp)
{
u8 cos;
for_each_cos_in_tx_queue(fp, cos)
- if (bnx2x_tx_queue_has_work(&fp->txdata[cos]))
+ if (bnx2x_tx_queue_has_work(fp->txdata_ptr[cos]))
return true;
return false;
}
@@ -1066,12 +1066,14 @@ static inline u32 bnx2x_rx_ustorm_prods_offset(struct bnx2x_fastpath *fp)
}
static inline void bnx2x_init_txdata(struct bnx2x *bp,
- struct bnx2x_fp_txdata *txdata, u32 cid, int txq_index,
- __le16 *tx_cons_sb)
+ struct bnx2x_fp_txdata *txdata, u32 cid,
+ int txq_index, __le16 *tx_cons_sb,
+ struct bnx2x_fastpath *fp)
{
txdata->cid = cid;
txdata->txq_index = txq_index;
txdata->tx_cons_sb = tx_cons_sb;
+ txdata->parent_fp = fp;
DP(NETIF_MSG_IFUP, "created tx data cid %d, txq %d\n",
txdata->cid, txdata->txq_index);
@@ -1114,9 +1116,9 @@ static inline void bnx2x_init_fcoe_fp(struct bnx2x *bp)
bnx2x_fcoe(bp, fw_sb_id) = DEF_SB_ID;
bnx2x_fcoe(bp, igu_sb_id) = bp->igu_dsb_id;
bnx2x_fcoe(bp, rx_cons_sb) = BNX2X_FCOE_L2_RX_INDEX;
-
- bnx2x_init_txdata(bp, &bnx2x_fcoe(bp, txdata[0]),
- fp->cid, FCOE_TXQ_IDX(bp), BNX2X_FCOE_L2_TX_INDEX);
+ bnx2x_init_txdata(bp, bnx2x_fcoe(bp, txdata_ptr[0]),
+ fp->cid, FCOE_TXQ_IDX(bp), BNX2X_FCOE_L2_TX_INDEX,
+ fp);
DP(NETIF_MSG_IFUP, "created fcoe tx data (fp index %d)\n", fp->index);
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
index 7378b1d..037eee6 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
@@ -1959,7 +1959,7 @@ static int bnx2x_run_loopback(struct bnx2x *bp, int loopback_mode)
unsigned char *packet;
struct bnx2x_fastpath *fp_rx = &bp->fp[0];
struct bnx2x_fastpath *fp_tx = &bp->fp[0];
- struct bnx2x_fp_txdata *txdata = &fp_tx->txdata[0];
+ struct bnx2x_fp_txdata *txdata = fp_tx->txdata_ptr[0];
u16 tx_start_idx, tx_idx;
u16 rx_start_idx, rx_idx;
u16 pkt_prod, bd_prod;
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index 16e009f..f80059e 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -758,7 +758,7 @@ void bnx2x_panic_dump(struct bnx2x *bp)
/* Tx */
for_each_cos_in_tx_queue(fp, cos)
{
- txdata = fp->txdata[cos];
+ txdata = *fp->txdata_ptr[cos];
BNX2X_ERR("fp%d: tx_pkt_prod(0x%x) tx_pkt_cons(0x%x) tx_bd_prod(0x%x) tx_bd_cons(0x%x) *tx_cons_sb(0x%x)\n",
i, txdata.tx_pkt_prod,
txdata.tx_pkt_cons, txdata.tx_bd_prod,
@@ -876,7 +876,7 @@ void bnx2x_panic_dump(struct bnx2x *bp)
for_each_tx_queue(bp, i) {
struct bnx2x_fastpath *fp = &bp->fp[i];
for_each_cos_in_tx_queue(fp, cos) {
- struct bnx2x_fp_txdata *txdata = &fp->txdata[cos];
+ struct bnx2x_fp_txdata *txdata = fp->txdata_ptr[cos];
start = TX_BD(le16_to_cpu(*txdata->tx_cons_sb) - 10);
end = TX_BD(le16_to_cpu(*txdata->tx_cons_sb) + 245);
@@ -1710,7 +1710,7 @@ irqreturn_t bnx2x_interrupt(int irq, void *dev_instance)
/* Handle Rx or Tx according to SB id */
prefetch(fp->rx_cons_sb);
for_each_cos_in_tx_queue(fp, cos)
- prefetch(fp->txdata[cos].tx_cons_sb);
+ prefetch(fp->txdata_ptr[cos]->tx_cons_sb);
prefetch(&fp->sb_running_index[SM_RX_ID]);
napi_schedule(&bnx2x_fp(bp, fp->index, napi));
status &= ~mask;
@@ -2921,7 +2921,7 @@ static void bnx2x_pf_tx_q_prep(struct bnx2x *bp,
struct bnx2x_fastpath *fp, struct bnx2x_txq_setup_params *txq_init,
u8 cos)
{
- txq_init->dscr_map = fp->txdata[cos].tx_desc_mapping;
+ txq_init->dscr_map = fp->txdata_ptr[cos]->tx_desc_mapping;
txq_init->sb_cq_index = HC_INDEX_ETH_FIRST_TX_CQ_CONS + cos;
txq_init->traffic_type = LLFC_TRAFFIC_TYPE_NW;
txq_init->fw_sb_id = fp->fw_sb_id;
@@ -3068,11 +3068,11 @@ static void bnx2x_drv_info_fcoe_stat(struct bnx2x *bp)
/* insert FCoE stats from ramrod response */
if (!NO_FCOE(bp)) {
struct tstorm_per_queue_stats *fcoe_q_tstorm_stats =
- &bp->fw_stats_data->queue_stats[FCOE_IDX].
+ &bp->fw_stats_data->queue_stats[FCOE_IDX(bp)].
tstorm_queue_statistics;
struct xstorm_per_queue_stats *fcoe_q_xstorm_stats =
- &bp->fw_stats_data->queue_stats[FCOE_IDX].
+ &bp->fw_stats_data->queue_stats[FCOE_IDX(bp)].
xstorm_queue_statistics;
struct fcoe_statistics_params *fw_fcoe_stat =
@@ -4741,7 +4741,7 @@ static void bnx2x_after_function_update(struct bnx2x *bp)
#ifdef BCM_CNIC
if (!NO_FCOE(bp)) {
- fp = &bp->fp[FCOE_IDX];
+ fp = &bp->fp[FCOE_IDX(bp)];
queue_params.q_obj = &fp->q_obj;
/* clear pending completion bit */
@@ -4778,7 +4778,7 @@ static struct bnx2x_queue_sp_obj *bnx2x_cid_to_q_obj(
return &bnx2x_fcoe(bp, q_obj);
else
#endif
- return &bnx2x_fp(bp, CID_TO_FP(cid), q_obj);
+ return &bnx2x_fp(bp, CID_TO_FP(cid, bp), q_obj);
}
static void bnx2x_eq_int(struct bnx2x *bp)
@@ -5660,11 +5660,11 @@ static void bnx2x_init_eth_fp(struct bnx2x *bp, int fp_idx)
/* init tx data */
for_each_cos_in_tx_queue(fp, cos) {
- bnx2x_init_txdata(bp, &fp->txdata[cos],
- CID_COS_TO_TX_ONLY_CID(fp->cid, cos),
- FP_COS_TO_TXQ(fp, cos),
- BNX2X_TX_SB_INDEX_BASE + cos);
- cids[cos] = fp->txdata[cos].cid;
+ bnx2x_init_txdata(bp, fp->txdata_ptr[cos],
+ CID_COS_TO_TX_ONLY_CID(fp->cid, cos, bp),
+ FP_COS_TO_TXQ(fp, cos, bp),
+ BNX2X_TX_SB_INDEX_BASE + cos, fp);
+ cids[cos] = fp->txdata_ptr[cos]->cid;
}
bnx2x_init_queue_obj(bp, &fp->q_obj, fp->cl_id, cids, fp->max_cos,
@@ -5719,7 +5719,7 @@ static void bnx2x_init_tx_rings(struct bnx2x *bp)
for_each_tx_queue(bp, i)
for_each_cos_in_tx_queue(&bp->fp[i], cos)
- bnx2x_init_tx_ring_one(&bp->fp[i].txdata[cos]);
+ bnx2x_init_tx_ring_one(bp->fp[i].txdata_ptr[cos]);
}
void bnx2x_nic_init(struct bnx2x *bp, u32 load_code)
@@ -7807,8 +7807,8 @@ static void bnx2x_pf_q_prep_init(struct bnx2x *bp,
/* set the context pointers queue object */
for (cos = FIRST_TX_COS_INDEX; cos < init_params->max_cos; cos++) {
- cxt_index = fp->txdata[cos].cid / ILT_PAGE_CIDS;
- cxt_offset = fp->txdata[cos].cid - (cxt_index *
+ cxt_index = fp->txdata_ptr[cos]->cid / ILT_PAGE_CIDS;
+ cxt_offset = fp->txdata_ptr[cos]->cid - (cxt_index *
ILT_PAGE_CIDS);
init_params->cxts[cos] =
&bp->context[cxt_index].vcxt[cxt_offset].eth;
@@ -7961,7 +7961,7 @@ static int bnx2x_stop_queue(struct bnx2x *bp, int index)
tx_index++){
/* ascertain this is a normal queue*/
- txdata = &fp->txdata[tx_index];
+ txdata = fp->txdata_ptr[tx_index];
DP(NETIF_MSG_IFDOWN, "stopping tx-only queue %d\n",
txdata->txq_index);
@@ -8328,7 +8328,7 @@ void bnx2x_chip_cleanup(struct bnx2x *bp, int unload_mode)
struct bnx2x_fastpath *fp = &bp->fp[i];
for_each_cos_in_tx_queue(fp, cos)
- rc = bnx2x_clean_tx_queue(bp, &fp->txdata[cos]);
+ rc = bnx2x_clean_tx_queue(bp, fp->txdata_ptr[cos]);
#ifdef BNX2X_STOP_ON_ERROR
if (rc)
return;
@@ -11769,7 +11769,7 @@ static int __devinit bnx2x_init_one(struct pci_dev *pdev,
struct bnx2x *bp;
int pcie_width, pcie_speed;
int rc, max_non_def_sbs;
- int rx_count, tx_count, rss_count;
+ int rx_count, tx_count, rss_count, doorbell_size;
/*
* An estimated maximum supported CoS number according to the chip
* version.
@@ -11863,8 +11863,10 @@ static int __devinit bnx2x_init_one(struct pci_dev *pdev,
* Map doorbels here as we need the real value of bp->max_cos which
* is initialized in bnx2x_init_bp().
*/
+ doorbell_size = (rss_count * max_cos_est + NON_ETH_CONTEXT_USE +
+ CNIC_PRESENT) * (1 << BNX2X_DB_SHIFT);
bp->doorbells = ioremap_nocache(pci_resource_start(pdev, 2),
- min_t(u64, BNX2X_DB_SIZE(bp),
+ min_t(u64, doorbell_size,
pci_resource_len(pdev, 2)));
if (!bp->doorbells) {
dev_err(&bp->pdev->dev,
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
index 0e8bdcb..776b521 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
@@ -1432,7 +1432,7 @@ static void bnx2x_prep_fw_stats_req(struct bnx2x *bp)
query[first_queue_query_index + i];
cur_query_entry->kind = STATS_TYPE_QUEUE;
- cur_query_entry->index = bnx2x_stats_id(&bp->fp[FCOE_IDX]);
+ cur_query_entry->index = bnx2x_stats_id(&bp->fp[FCOE_IDX(bp)]);
cur_query_entry->funcID = cpu_to_le16(BP_FUNC(bp));
cur_query_entry->address.hi =
cpu_to_le32(U64_HI(cur_data_offset));
@@ -1573,7 +1573,7 @@ void bnx2x_afex_collect_stats(struct bnx2x *bp, void *void_afex_stats,
struct afex_stats *afex_stats = (struct afex_stats *)void_afex_stats;
struct bnx2x_eth_stats *estats = &bp->eth_stats;
struct per_queue_stats *fcoe_q_stats =
- &bp->fw_stats_data->queue_stats[FCOE_IDX];
+ &bp->fw_stats_data->queue_stats[FCOE_IDX(bp)];
struct tstorm_per_queue_stats *fcoe_q_tstorm_stats =
&fcoe_q_stats->tstorm_queue_statistics;
--
1.7.10
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [net-next patch 6/12] bnx2x: Move the CNIC L2 CIDs to be right after the RSS CIDs
2012-06-13 12:44 [net-next patch 0/12] bnx2x: ethtool and other enhancements Merav Sicron
` (4 preceding siblings ...)
2012-06-13 12:44 ` [net-next patch 5/12] bnx2x: Make the transmission queues adjacent Merav Sicron
@ 2012-06-13 12:44 ` Merav Sicron
2012-06-13 12:44 ` [net-next patch 7/12] bnx2x: Split the FP structure Merav Sicron
` (5 subsequent siblings)
11 siblings, 0 replies; 28+ messages in thread
From: Merav Sicron @ 2012-06-13 12:44 UTC (permalink / raw)
To: eilong, davem, netdev; +Cc: Merav Sicron
Currently the CNIC-related L2 CIDs (for sending control FCoE / iSCSI packets)
were at fixed position, according to the maximal number of RSS queues multiplied
by the number of traffic-classes. This change makes the CIDs dynamic, as they
are defined to be right after the highest RSS CID. This decreases the memory
allocated for the context.
Signed-off-by: Merav Sicron <meravs@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x.h | 19 +++++----
drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c | 1 +
drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h | 15 ++++---
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c | 47 +++++++++++++++-------
4 files changed, 52 insertions(+), 30 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index 36ebc3d..68d0d2c 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -248,13 +248,12 @@ enum {
BNX2X_MAX_CNIC_ETH_CL_ID_IDX,
};
-#define BNX2X_CNIC_START_ETH_CID 48
-enum {
+#define BNX2X_CNIC_START_ETH_CID(bp) (BNX2X_NUM_NON_CNIC_QUEUES(bp) *\
+ (bp)->max_cos)
/* iSCSI L2 */
- BNX2X_ISCSI_ETH_CID = BNX2X_CNIC_START_ETH_CID,
+#define BNX2X_ISCSI_ETH_CID(bp) (BNX2X_CNIC_START_ETH_CID(bp))
/* FCoE L2 */
- BNX2X_FCOE_ETH_CID,
-};
+#define BNX2X_FCOE_ETH_CID(bp) (BNX2X_CNIC_START_ETH_CID(bp) + 1)
/** Additional rings budgeting */
#ifdef BCM_CNIC
@@ -276,8 +275,6 @@ enum {
#define FIRST_TX_ONLY_COS_INDEX 1
#define FIRST_TX_COS_INDEX 0
-#define MAX_TXQS_PER_COS FP_SB_MAX_E1x
-
/* rules for calculating the cids of tx-only connections */
#define CID_TO_FP(cid, bp) ((cid) % BNX2X_NUM_NON_CNIC_QUEUES(bp))
#define CID_COS_TO_TX_ONLY_CID(cid, cos, bp) \
@@ -1463,10 +1460,12 @@ struct bnx2x {
/*
* Maximum CID count that might be required by the bnx2x:
- * Max Tss * Max_Tx_Multi_Cos + CNIC L2 Clients (FCoE and iSCSI related)
+ * Max RSS * Max_Tx_Multi_Cos + FCoE + iSCSI
*/
-#define BNX2X_L2_CID_COUNT(bp) (MAX_TXQS_PER_COS * BNX2X_MULTI_TX_COS +\
- NON_ETH_CONTEXT_USE + CNIC_PRESENT)
+#define BNX2X_L2_CID_COUNT(bp) (BNX2X_NUM_ETH_QUEUES(bp) * BNX2X_MULTI_TX_COS \
+ + NON_ETH_CONTEXT_USE + CNIC_PRESENT)
+#define BNX2X_L2_MAX_CID(bp) (BNX2X_MAX_RSS_COUNT(bp) * BNX2X_MULTI_TX_COS \
+ + NON_ETH_CONTEXT_USE + CNIC_PRESENT)
#define L2_ILT_LINES(bp) (DIV_ROUND_UP(BNX2X_L2_CID_COUNT(bp),\
ILT_PAGE_CIDS))
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
index 33132b1..d76df33 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
@@ -2220,6 +2220,7 @@ int bnx2x_nic_load(struct bnx2x *bp, int load_mode)
/* re-read iscsi info */
bnx2x_get_iscsi_info(bp);
bnx2x_setup_cnic_irq_info(bp);
+ bnx2x_setup_cnic_info(bp);
if (bp->state == BNX2X_STATE_OPEN)
bnx2x_cnic_notify(bp, CNIC_CTL_START_CMD);
#endif
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
index 112ffcc..de79cdf 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
@@ -244,6 +244,14 @@ int bnx2x_cnic_notify(struct bnx2x *bp, int cmd);
* @bp: driver handle
*/
void bnx2x_setup_cnic_irq_info(struct bnx2x *bp);
+
+/**
+ * bnx2x_setup_cnic_info - provides cnic with updated info
+ *
+ * @bp: driver handle
+ */
+void bnx2x_setup_cnic_info(struct bnx2x *bp);
+
#endif
/**
@@ -1107,12 +1115,7 @@ static inline void bnx2x_init_fcoe_fp(struct bnx2x *bp)
bnx2x_fcoe(bp, rx_queue) = BNX2X_NUM_ETH_QUEUES(bp);
bnx2x_fcoe(bp, cl_id) = bnx2x_cnic_eth_cl_id(bp,
BNX2X_FCOE_ETH_CL_ID_IDX);
- /** Current BNX2X_FCOE_ETH_CID deffinition implies not more than
- * 16 ETH clients per function when CNIC is enabled!
- *
- * Fix it ASAP!!!
- */
- bnx2x_fcoe(bp, cid) = BNX2X_FCOE_ETH_CID;
+ bnx2x_fcoe(bp, cid) = BNX2X_FCOE_ETH_CID(bp);
bnx2x_fcoe(bp, fw_sb_id) = DEF_SB_ID;
bnx2x_fcoe(bp, igu_sb_id) = bp->igu_dsb_id;
bnx2x_fcoe(bp, rx_cons_sb) = BNX2X_FCOE_L2_RX_INDEX;
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index f80059e..1895de5 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -4628,7 +4628,7 @@ static void bnx2x_handle_classification_eqe(struct bnx2x *bp,
case BNX2X_FILTER_MAC_PENDING:
DP(BNX2X_MSG_SP, "Got SETUP_MAC completions\n");
#ifdef BCM_CNIC
- if (cid == BNX2X_ISCSI_ETH_CID)
+ if (cid == BNX2X_ISCSI_ETH_CID(bp))
vlan_mac_obj = &bp->iscsi_l2_mac_obj;
else
#endif
@@ -4774,7 +4774,7 @@ static struct bnx2x_queue_sp_obj *bnx2x_cid_to_q_obj(
{
DP(BNX2X_MSG_SP, "retrieving fp from cid %d\n", cid);
#ifdef BCM_CNIC
- if (cid == BNX2X_FCOE_ETH_CID)
+ if (cid == BNX2X_FCOE_ETH_CID(bp))
return &bnx2x_fcoe(bp, q_obj);
else
#endif
@@ -11724,7 +11724,7 @@ void bnx2x__init_func_obj(struct bnx2x *bp)
/* must be called after sriov-enable */
static int bnx2x_set_qm_cid_count(struct bnx2x *bp)
{
- int cid_count = BNX2X_L2_CID_COUNT(bp);
+ int cid_count = BNX2X_L2_MAX_CID(bp);
#ifdef BCM_CNIC
cid_count += CNIC_CID_MAX;
@@ -11829,9 +11829,9 @@ static int __devinit bnx2x_init_one(struct pci_dev *pdev,
/*
* Maximum number of netdev Tx queues:
- * Maximum TSS queues * Maximum supported number of CoS + FCoE L2
+ * Maximum TSS queues * Maximum supported number of CoS + FCoE L2
*/
- tx_count = MAX_TXQS_PER_COS * max_cos_est + FCOE_PRESENT;
+ tx_count = rss_count * max_cos_est + FCOE_PRESENT;
/* dev zeroed in init_etherdev */
dev = alloc_etherdev_mqs(sizeof(*bp), tx_count, rx_count);
@@ -11863,11 +11863,15 @@ static int __devinit bnx2x_init_one(struct pci_dev *pdev,
* Map doorbels here as we need the real value of bp->max_cos which
* is initialized in bnx2x_init_bp().
*/
- doorbell_size = (rss_count * max_cos_est + NON_ETH_CONTEXT_USE +
- CNIC_PRESENT) * (1 << BNX2X_DB_SHIFT);
+ doorbell_size = BNX2X_L2_MAX_CID(bp) * (1 << BNX2X_DB_SHIFT);
+ if (doorbell_size > pci_resource_len(pdev, 2)) {
+ dev_err(&bp->pdev->dev,
+ "Cannot map doorbells, bar size too small, aborting\n");
+ rc = -ENOMEM;
+ goto init_one_exit;
+ }
bp->doorbells = ioremap_nocache(pci_resource_start(pdev, 2),
- min_t(u64, doorbell_size,
- pci_resource_len(pdev, 2)));
+ doorbell_size);
if (!bp->doorbells) {
dev_err(&bp->pdev->dev,
"Cannot map doorbell space, aborting\n");
@@ -12254,14 +12258,14 @@ static void bnx2x_cnic_sp_post(struct bnx2x *bp, int count)
*/
if (type == ETH_CONNECTION_TYPE) {
if (cmd == RAMROD_CMD_ID_ETH_CLIENT_SETUP) {
- cxt_index = BNX2X_ISCSI_ETH_CID /
+ cxt_index = BNX2X_ISCSI_ETH_CID(bp) /
ILT_PAGE_CIDS;
- cxt_offset = BNX2X_ISCSI_ETH_CID -
+ cxt_offset = BNX2X_ISCSI_ETH_CID(bp) -
(cxt_index * ILT_PAGE_CIDS);
bnx2x_set_ctx_validation(bp,
&bp->context[cxt_index].
vcxt[cxt_offset].eth,
- BNX2X_ISCSI_ETH_CID);
+ BNX2X_ISCSI_ETH_CID(bp));
}
}
@@ -12615,6 +12619,21 @@ void bnx2x_setup_cnic_irq_info(struct bnx2x *bp)
cp->num_irq = 2;
}
+void bnx2x_setup_cnic_info(struct bnx2x *bp)
+{
+ struct cnic_eth_dev *cp = &bp->cnic_eth_dev;
+
+
+ cp->ctx_tbl_offset = FUNC_ILT_BASE(BP_FUNC(bp)) +
+ bnx2x_cid_ilt_lines(bp);
+ cp->starting_cid = bnx2x_cid_ilt_lines(bp) * ILT_PAGE_CIDS;
+ cp->fcoe_init_cid = BNX2X_FCOE_ETH_CID(bp);
+ cp->iscsi_l2_cid = BNX2X_ISCSI_ETH_CID(bp);
+
+ if (NO_ISCSI_OOO(bp))
+ cp->drv_state |= CNIC_DRV_STATE_NO_ISCSI_OOO;
+}
+
static int bnx2x_register_cnic(struct net_device *dev, struct cnic_ops *ops,
void *data)
{
@@ -12693,10 +12712,10 @@ struct cnic_eth_dev *bnx2x_cnic_probe(struct net_device *dev)
cp->drv_ctl = bnx2x_drv_ctl;
cp->drv_register_cnic = bnx2x_register_cnic;
cp->drv_unregister_cnic = bnx2x_unregister_cnic;
- cp->fcoe_init_cid = BNX2X_FCOE_ETH_CID;
+ cp->fcoe_init_cid = BNX2X_FCOE_ETH_CID(bp);
cp->iscsi_l2_client_id =
bnx2x_cnic_eth_cl_id(bp, BNX2X_ISCSI_ETH_CL_ID_IDX);
- cp->iscsi_l2_cid = BNX2X_ISCSI_ETH_CID;
+ cp->iscsi_l2_cid = BNX2X_ISCSI_ETH_CID(bp);
if (NO_ISCSI_OOO(bp))
cp->drv_state |= CNIC_DRV_STATE_NO_ISCSI_OOO;
--
1.7.10
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [net-next patch 7/12] bnx2x: Split the FP structure
2012-06-13 12:44 [net-next patch 0/12] bnx2x: ethtool and other enhancements Merav Sicron
` (5 preceding siblings ...)
2012-06-13 12:44 ` [net-next patch 6/12] bnx2x: Move the CNIC L2 CIDs to be right after the RSS CIDs Merav Sicron
@ 2012-06-13 12:44 ` Merav Sicron
2012-06-13 12:44 ` [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues Merav Sicron
` (4 subsequent siblings)
11 siblings, 0 replies; 28+ messages in thread
From: Merav Sicron @ 2012-06-13 12:44 UTC (permalink / raw)
To: eilong, davem, netdev; +Cc: Merav Sicron, Barak Witkowski
From: Barak Witkowski <barak@broadcom.com>
This patch moves some fields out of the FP structure to different structures, in
order to minimize size of contigiuous memory allocated.
Signed-off-by: Barak Witkowski <barak@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x.h | 41 +++++----
drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c | 87 +++++++++++++++-----
drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h | 8 +-
.../net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c | 4 +-
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c | 37 +++++----
drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c | 55 ++++++++-----
6 files changed, 149 insertions(+), 83 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index 68d0d2c..60a72f5 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -549,34 +549,23 @@ struct bnx2x_fastpath {
rx_calls;
/* TPA related */
- struct bnx2x_agg_info tpa_info[ETH_MAX_AGGREGATION_QUEUES_E1H_E2];
+ struct bnx2x_agg_info *tpa_info;
u8 disable_tpa;
#ifdef BNX2X_STOP_ON_ERROR
u64 tpa_queue_used;
#endif
-
- struct tstorm_per_queue_stats old_tclient;
- struct ustorm_per_queue_stats old_uclient;
- struct xstorm_per_queue_stats old_xclient;
- struct bnx2x_eth_q_stats eth_q_stats;
- struct bnx2x_eth_q_stats_old eth_q_stats_old;
-
/* The size is calculated using the following:
sizeof name field from netdev structure +
4 ('-Xx-' string) +
4 (for the digits and to make it DWORD aligned) */
#define FP_NAME_SIZE (sizeof(((struct net_device *)0)->name) + 8)
char name[FP_NAME_SIZE];
-
- /* MACs object */
- struct bnx2x_vlan_mac_obj mac_obj;
-
- /* Queue State object */
- struct bnx2x_queue_sp_obj q_obj;
-
};
-#define bnx2x_fp(bp, nr, var) (bp->fp[nr].var)
+#define bnx2x_fp(bp, nr, var) ((bp)->fp[(nr)].var)
+#define bnx2x_sp_obj(bp, fp) ((bp)->sp_objs[(fp)->index])
+#define bnx2x_fp_stats(bp, fp) (&((bp)->fp_stats[(fp)->index]))
+#define bnx2x_fp_qstats(bp, fp) (&((bp)->fp_stats[(fp)->index].eth_q_stats))
/* Use 2500 as a mini-jumbo MTU for FCoE */
#define BNX2X_FCOE_MINI_JUMBO_MTU 2500
@@ -587,6 +576,8 @@ struct bnx2x_fastpath {
FCOE_IDX_OFFSET)
#define bnx2x_fcoe_fp(bp) (&bp->fp[FCOE_IDX(bp)])
#define bnx2x_fcoe(bp, var) (bnx2x_fcoe_fp(bp)->var)
+#define bnx2x_fcoe_inner_sp_obj(bp) (&bp->sp_objs[FCOE_IDX(bp)])
+#define bnx2x_fcoe_sp_obj(bp, var) (bnx2x_fcoe_inner_sp_obj(bp)->var)
#define bnx2x_fcoe_tx(bp, var) (bnx2x_fcoe_fp(bp)-> \
txdata_ptr[FIRST_TX_COS_INDEX] \
->var)
@@ -1202,11 +1193,29 @@ struct bnx2x_prev_path_list {
struct list_head list;
};
+struct bnx2x_sp_objs {
+ /* MACs object */
+ struct bnx2x_vlan_mac_obj mac_obj;
+
+ /* Queue State object */
+ struct bnx2x_queue_sp_obj q_obj;
+};
+
+struct bnx2x_fp_stats {
+ struct tstorm_per_queue_stats old_tclient;
+ struct ustorm_per_queue_stats old_uclient;
+ struct xstorm_per_queue_stats old_xclient;
+ struct bnx2x_eth_q_stats eth_q_stats;
+ struct bnx2x_eth_q_stats_old eth_q_stats_old;
+};
+
struct bnx2x {
/* Fields used in the tx and intr/napi performance paths
* are grouped together in the beginning of the structure
*/
struct bnx2x_fastpath *fp;
+ struct bnx2x_sp_objs *sp_objs;
+ struct bnx2x_fp_stats *fp_stats;
struct bnx2x_fp_txdata *bnx2x_txq;
int bnx2x_txq_size;
void __iomem *regview;
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
index d76df33..915e680 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
@@ -47,6 +47,10 @@ static inline void bnx2x_move_fp(struct bnx2x *bp, int from, int to)
{
struct bnx2x_fastpath *from_fp = &bp->fp[from];
struct bnx2x_fastpath *to_fp = &bp->fp[to];
+ struct bnx2x_sp_objs *from_sp_objs = &bp->sp_objs[from];
+ struct bnx2x_sp_objs *to_sp_objs = &bp->sp_objs[to];
+ struct bnx2x_fp_stats *from_fp_stats = &bp->fp_stats[from];
+ struct bnx2x_fp_stats *to_fp_stats = &bp->fp_stats[to];
int old_max_eth_txqs, new_max_eth_txqs;
int old_txdata_index = 0, new_txdata_index = 0;
@@ -57,6 +61,12 @@ static inline void bnx2x_move_fp(struct bnx2x *bp, int from, int to)
memcpy(to_fp, from_fp, sizeof(*to_fp));
to_fp->index = to;
+ /* move sp_objs contents as well, as their indices match fp ones */
+ memcpy(to_sp_objs, from_sp_objs, sizeof(*to_sp_objs));
+
+ /* move fp_stats contents as well, as their indices match fp ones */
+ memcpy(to_fp_stats, from_fp_stats, sizeof(*to_fp_stats));
+
/* Update txdata pointers in fp and move txdata content accordingly:
* Each fp consumes 'max_cos' txdata structures, so the index should be
* decremented by max_cos x delta.
@@ -500,7 +510,7 @@ static int bnx2x_fill_frag_skb(struct bnx2x *bp, struct bnx2x_fastpath *fp,
where we are and drop the whole packet */
err = bnx2x_alloc_rx_sge(bp, fp, sge_idx);
if (unlikely(err)) {
- fp->eth_q_stats.rx_skb_alloc_failed++;
+ bnx2x_fp_qstats(bp, fp)->rx_skb_alloc_failed++;
return err;
}
@@ -605,7 +615,7 @@ drop:
/* drop the packet and keep the buffer in the bin */
DP(NETIF_MSG_RX_STATUS,
"Failed to allocate or map a new skb - dropping packet!\n");
- fp->eth_q_stats.rx_skb_alloc_failed++;
+ bnx2x_fp_stats(bp, fp)->eth_q_stats.rx_skb_alloc_failed++;
}
static int bnx2x_alloc_rx_data(struct bnx2x *bp,
@@ -778,7 +788,7 @@ int bnx2x_rx_int(struct bnx2x_fastpath *fp, int budget)
DP(NETIF_MSG_RX_ERR | NETIF_MSG_RX_STATUS,
"ERROR flags %x rx packet %u\n",
cqe_fp_flags, sw_comp_cons);
- fp->eth_q_stats.rx_err_discard_pkt++;
+ bnx2x_fp_qstats(bp, fp)->rx_err_discard_pkt++;
goto reuse_rx;
}
@@ -791,7 +801,7 @@ int bnx2x_rx_int(struct bnx2x_fastpath *fp, int budget)
if (skb == NULL) {
DP(NETIF_MSG_RX_ERR | NETIF_MSG_RX_STATUS,
"ERROR packet dropped because of alloc failure\n");
- fp->eth_q_stats.rx_skb_alloc_failed++;
+ bnx2x_fp_qstats(bp, fp)->rx_skb_alloc_failed++;
goto reuse_rx;
}
memcpy(skb->data, data + pad, len);
@@ -805,14 +815,15 @@ int bnx2x_rx_int(struct bnx2x_fastpath *fp, int budget)
skb = build_skb(data, 0);
if (unlikely(!skb)) {
kfree(data);
- fp->eth_q_stats.rx_skb_alloc_failed++;
+ bnx2x_fp_qstats(bp, fp)->
+ rx_skb_alloc_failed++;
goto next_rx;
}
skb_reserve(skb, pad);
} else {
DP(NETIF_MSG_RX_ERR | NETIF_MSG_RX_STATUS,
"ERROR packet dropped because of alloc failure\n");
- fp->eth_q_stats.rx_skb_alloc_failed++;
+ bnx2x_fp_qstats(bp, fp)->rx_skb_alloc_failed++;
reuse_rx:
bnx2x_reuse_rx_data(fp, bd_cons, bd_prod);
goto next_rx;
@@ -832,7 +843,7 @@ reuse_rx:
if (likely(BNX2X_RX_CSUM_OK(cqe)))
skb->ip_summed = CHECKSUM_UNNECESSARY;
else
- fp->eth_q_stats.hw_csum_err++;
+ bnx2x_fp_qstats(bp, fp)->hw_csum_err++;
}
skb_record_rx_queue(skb, fp->rx_queue);
@@ -1765,7 +1776,7 @@ static void bnx2x_squeeze_objects(struct bnx2x *bp)
int rc;
unsigned long ramrod_flags = 0, vlan_mac_flags = 0;
struct bnx2x_mcast_ramrod_params rparam = {NULL};
- struct bnx2x_vlan_mac_obj *mac_obj = &bp->fp->mac_obj;
+ struct bnx2x_vlan_mac_obj *mac_obj = &bp->sp_objs->mac_obj;
/***************** Cleanup MACs' object first *************************/
@@ -1776,7 +1787,7 @@ static void bnx2x_squeeze_objects(struct bnx2x *bp)
/* Clean ETH primary MAC */
__set_bit(BNX2X_ETH_MAC, &vlan_mac_flags);
- rc = mac_obj->delete_all(bp, &bp->fp->mac_obj, &vlan_mac_flags,
+ rc = mac_obj->delete_all(bp, &bp->sp_objs->mac_obj, &vlan_mac_flags,
&ramrod_flags);
if (rc != 0)
BNX2X_ERR("Failed to clean ETH MACs: %d\n", rc);
@@ -1862,12 +1873,16 @@ bool bnx2x_test_firmware_version(struct bnx2x *bp, bool is_err)
static void bnx2x_bz_fp(struct bnx2x *bp, int index)
{
struct bnx2x_fastpath *fp = &bp->fp[index];
+ struct bnx2x_fp_stats *fp_stats = &bp->fp_stats[index];
+
int cos;
struct napi_struct orig_napi = fp->napi;
+ struct bnx2x_agg_info *orig_tpa_info = fp->tpa_info;
/* bzero bnx2x_fastpath contents */
- if (bp->stats_init)
+ if (bp->stats_init) {
+ memset(fp->tpa_info, 0, sizeof(*fp->tpa_info));
memset(fp, 0, sizeof(*fp));
- else {
+ } else {
/* Keep Queue statistics */
struct bnx2x_eth_q_stats *tmp_eth_q_stats;
struct bnx2x_eth_q_stats_old *tmp_eth_q_stats_old;
@@ -1875,26 +1890,27 @@ static void bnx2x_bz_fp(struct bnx2x *bp, int index)
tmp_eth_q_stats = kzalloc(sizeof(struct bnx2x_eth_q_stats),
GFP_KERNEL);
if (tmp_eth_q_stats)
- memcpy(tmp_eth_q_stats, &fp->eth_q_stats,
+ memcpy(tmp_eth_q_stats, &fp_stats->eth_q_stats,
sizeof(struct bnx2x_eth_q_stats));
tmp_eth_q_stats_old =
kzalloc(sizeof(struct bnx2x_eth_q_stats_old),
GFP_KERNEL);
if (tmp_eth_q_stats_old)
- memcpy(tmp_eth_q_stats_old, &fp->eth_q_stats_old,
+ memcpy(tmp_eth_q_stats_old, &fp_stats->eth_q_stats_old,
sizeof(struct bnx2x_eth_q_stats_old));
+ memset(fp->tpa_info, 0, sizeof(*fp->tpa_info));
memset(fp, 0, sizeof(*fp));
if (tmp_eth_q_stats) {
- memcpy(&fp->eth_q_stats, tmp_eth_q_stats,
- sizeof(struct bnx2x_eth_q_stats));
+ memcpy(&fp_stats->eth_q_stats, tmp_eth_q_stats,
+ sizeof(struct bnx2x_eth_q_stats));
kfree(tmp_eth_q_stats);
}
if (tmp_eth_q_stats_old) {
- memcpy(&fp->eth_q_stats_old, tmp_eth_q_stats_old,
+ memcpy(&fp_stats->eth_q_stats_old, tmp_eth_q_stats_old,
sizeof(struct bnx2x_eth_q_stats_old));
kfree(tmp_eth_q_stats_old);
}
@@ -1903,7 +1919,7 @@ static void bnx2x_bz_fp(struct bnx2x *bp, int index)
/* Restore the NAPI object as it has been already initialized */
fp->napi = orig_napi;
-
+ fp->tpa_info = orig_tpa_info;
fp->bp = bp;
fp->index = index;
if (IS_ETH_FP(fp))
@@ -2905,7 +2921,7 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev)
if (unlikely(bnx2x_tx_avail(bp, txdata) <
(skb_shinfo(skb)->nr_frags + 3))) {
- txdata->parent_fp->eth_q_stats.driver_xoff++;
+ bnx2x_fp_qstats(bp, txdata->parent_fp)->driver_xoff++;
netif_tx_stop_queue(txq);
BNX2X_ERR("BUG! Tx ring full when queue awake!\n");
return NETDEV_TX_BUSY;
@@ -3187,7 +3203,7 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev)
* fp->bd_tx_cons */
smp_mb();
- txdata->parent_fp->eth_q_stats.driver_xoff++;
+ bnx2x_fp_qstats(bp, txdata->parent_fp)->driver_xoff++;
if (bnx2x_tx_avail(bp, txdata) >= MAX_SKB_FRAGS + 3)
netif_tx_wake_queue(txq);
}
@@ -3424,7 +3440,7 @@ static int bnx2x_alloc_rx_bds(struct bnx2x_fastpath *fp,
cqe_ring_prod);
fp->rx_pkt = fp->rx_calls = 0;
- fp->eth_q_stats.rx_skb_alloc_failed += failure_cnt;
+ bnx2x_fp_stats(bp, fp)->eth_q_stats.rx_skb_alloc_failed += failure_cnt;
return i - failure_cnt;
}
@@ -3629,7 +3645,10 @@ int bnx2x_alloc_fp_mem(struct bnx2x *bp)
void bnx2x_free_mem_bp(struct bnx2x *bp)
{
+ kfree(bp->fp->tpa_info);
kfree(bp->fp);
+ kfree(bp->sp_objs);
+ kfree(bp->fp_stats);
kfree(bp->bnx2x_txq);
kfree(bp->msix_table);
kfree(bp->ilt);
@@ -3641,6 +3660,8 @@ int __devinit bnx2x_alloc_mem_bp(struct bnx2x *bp)
struct msix_entry *tbl;
struct bnx2x_ilt *ilt;
int msix_table_size = 0;
+ int fp_array_size;
+ int i;
/*
* The biggest MSI-X table we might need is as a maximum number of fast
@@ -3649,12 +3670,34 @@ int __devinit bnx2x_alloc_mem_bp(struct bnx2x *bp)
msix_table_size = bp->igu_sb_cnt + 1;
/* fp array: RSS plus CNIC related L2 queues */
- fp = kcalloc(BNX2X_MAX_RSS_COUNT(bp) + NON_ETH_CONTEXT_USE,
- sizeof(*fp), GFP_KERNEL);
+ fp_array_size = BNX2X_MAX_RSS_COUNT(bp) + NON_ETH_CONTEXT_USE;
+ BNX2X_DEV_INFO("fp_array_size %d", fp_array_size);
+
+ fp = kcalloc(fp_array_size, sizeof(*fp), GFP_KERNEL);
if (!fp)
goto alloc_err;
+ for (i = 0; i < fp_array_size; i++) {
+ fp[i].tpa_info =
+ kcalloc(ETH_MAX_AGGREGATION_QUEUES_E1H_E2,
+ sizeof(struct bnx2x_agg_info), GFP_KERNEL);
+ if (!(fp[i].tpa_info))
+ goto alloc_err;
+ }
+
bp->fp = fp;
+ /* allocate sp objs */
+ bp->sp_objs = kcalloc(fp_array_size, sizeof(struct bnx2x_sp_objs),
+ GFP_KERNEL);
+ if (!bp->sp_objs)
+ goto alloc_err;
+
+ /* allocate fp_stats */
+ bp->fp_stats = kcalloc(fp_array_size, sizeof(struct bnx2x_fp_stats),
+ GFP_KERNEL);
+ if (!bp->fp_stats)
+ goto alloc_err;
+
/* Allocate memory for the transmission queues array */
bp->bnx2x_txq_size = BNX2X_MAX_RSS_COUNT(bp) * BNX2X_MULTI_TX_COS;
#ifdef BCM_CNIC
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
index de79cdf..11afe5d 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
@@ -981,8 +981,8 @@ static inline void bnx2x_init_vlan_mac_fp_objs(struct bnx2x_fastpath *fp,
struct bnx2x *bp = fp->bp;
/* Configure classification DBs */
- bnx2x_init_mac_obj(bp, &fp->mac_obj, fp->cl_id, fp->cid,
- BP_FUNC(bp), bnx2x_sp(bp, mac_rdata),
+ bnx2x_init_mac_obj(bp, &bnx2x_sp_obj(bp, fp).mac_obj, fp->cl_id,
+ fp->cid, BP_FUNC(bp), bnx2x_sp(bp, mac_rdata),
bnx2x_sp_mapping(bp, mac_rdata),
BNX2X_FILTER_MAC_PENDING,
&bp->sp_state, obj_type,
@@ -1138,8 +1138,8 @@ static inline void bnx2x_init_fcoe_fp(struct bnx2x *bp)
/* No multi-CoS for FCoE L2 client */
BUG_ON(fp->max_cos != 1);
- bnx2x_init_queue_obj(bp, &fp->q_obj, fp->cl_id, &fp->cid, 1,
- BP_FUNC(bp), bnx2x_sp(bp, q_rdata),
+ bnx2x_init_queue_obj(bp, &bnx2x_sp_obj(bp, fp).q_obj, fp->cl_id,
+ &fp->cid, 1, BP_FUNC(bp), bnx2x_sp(bp, q_rdata),
bnx2x_sp_mapping(bp, q_rdata), q_type);
DP(NETIF_MSG_IFUP,
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
index 037eee6..13a51b0 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
@@ -2292,7 +2292,7 @@ static int bnx2x_test_intr(struct bnx2x *bp)
return -ENODEV;
}
- params.q_obj = &bp->fp->q_obj;
+ params.q_obj = &bp->sp_objs->q_obj;
params.cmd = BNX2X_Q_CMD_EMPTY;
__set_bit(RAMROD_COMP_WAIT, ¶ms.ramrod_flags);
@@ -2516,7 +2516,7 @@ static void bnx2x_get_ethtool_stats(struct net_device *dev,
if (is_multi(bp)) {
for_each_eth_queue(bp, i) {
- hw_stats = (u32 *)&bp->fp[i].eth_q_stats;
+ hw_stats = (u32 *)&bp->fp_stats[i].eth_q_stats;
for (j = 0; j < BNX2X_NUM_Q_STATS; j++) {
if (bnx2x_q_stats_arr[j].size == 0) {
/* skip this counter */
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index 1895de5..fbade5c 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -1583,7 +1583,7 @@ void bnx2x_sp_event(struct bnx2x_fastpath *fp, union eth_rx_cqe *rr_cqe)
int cid = SW_CID(rr_cqe->ramrod_cqe.conn_and_cmd_data);
int command = CQE_CMD(rr_cqe->ramrod_cqe.conn_and_cmd_data);
enum bnx2x_queue_cmd drv_cmd = BNX2X_Q_CMD_MAX;
- struct bnx2x_queue_sp_obj *q_obj = &fp->q_obj;
+ struct bnx2x_queue_sp_obj *q_obj = &bnx2x_sp_obj(bp, fp).q_obj;
DP(BNX2X_MSG_SP,
"fp %d cid %d got ramrod #%d state is %x type is %d\n",
@@ -3035,9 +3035,9 @@ static void bnx2x_drv_info_ether_stat(struct bnx2x *bp)
memcpy(ether_stat->version, DRV_MODULE_VERSION,
ETH_STAT_INFO_VERSION_LEN - 1);
- bp->fp[0].mac_obj.get_n_elements(bp, &bp->fp[0].mac_obj,
- DRV_INFO_ETH_STAT_NUM_MACS_REQUIRED,
- ether_stat->mac_local);
+ bp->sp_objs[0].mac_obj.get_n_elements(bp, &bp->sp_objs[0].mac_obj,
+ DRV_INFO_ETH_STAT_NUM_MACS_REQUIRED,
+ ether_stat->mac_local);
ether_stat->mtu_size = bp->dev->mtu;
@@ -4632,7 +4632,7 @@ static void bnx2x_handle_classification_eqe(struct bnx2x *bp,
vlan_mac_obj = &bp->iscsi_l2_mac_obj;
else
#endif
- vlan_mac_obj = &bp->fp[cid].mac_obj;
+ vlan_mac_obj = &bp->sp_objs[cid].mac_obj;
break;
case BNX2X_FILTER_MCAST_PENDING:
@@ -4730,7 +4730,7 @@ static void bnx2x_after_function_update(struct bnx2x *bp)
for_each_eth_queue(bp, q) {
/* Set the appropriate Queue object */
fp = &bp->fp[q];
- queue_params.q_obj = &fp->q_obj;
+ queue_params.q_obj = &bnx2x_sp_obj(bp, fp).q_obj;
/* send the ramrod */
rc = bnx2x_queue_state_change(bp, &queue_params);
@@ -4742,7 +4742,7 @@ static void bnx2x_after_function_update(struct bnx2x *bp)
#ifdef BCM_CNIC
if (!NO_FCOE(bp)) {
fp = &bp->fp[FCOE_IDX(bp)];
- queue_params.q_obj = &fp->q_obj;
+ queue_params.q_obj = &bnx2x_sp_obj(bp, fp).q_obj;
/* clear pending completion bit */
__clear_bit(RAMROD_COMP_WAIT, &queue_params.ramrod_flags);
@@ -4775,10 +4775,10 @@ static struct bnx2x_queue_sp_obj *bnx2x_cid_to_q_obj(
DP(BNX2X_MSG_SP, "retrieving fp from cid %d\n", cid);
#ifdef BCM_CNIC
if (cid == BNX2X_FCOE_ETH_CID(bp))
- return &bnx2x_fcoe(bp, q_obj);
+ return &bnx2x_fcoe_sp_obj(bp, q_obj);
else
#endif
- return &bnx2x_fp(bp, CID_TO_FP(cid, bp), q_obj);
+ return &bp->sp_objs[CID_TO_FP(cid, bp)].q_obj;
}
static void bnx2x_eq_int(struct bnx2x *bp)
@@ -5667,8 +5667,8 @@ static void bnx2x_init_eth_fp(struct bnx2x *bp, int fp_idx)
cids[cos] = fp->txdata_ptr[cos]->cid;
}
- bnx2x_init_queue_obj(bp, &fp->q_obj, fp->cl_id, cids, fp->max_cos,
- BP_FUNC(bp), bnx2x_sp(bp, q_rdata),
+ bnx2x_init_queue_obj(bp, &bnx2x_sp_obj(bp, fp).q_obj, fp->cl_id, cids,
+ fp->max_cos, BP_FUNC(bp), bnx2x_sp(bp, q_rdata),
bnx2x_sp_mapping(bp, q_rdata), q_type);
/**
@@ -7596,8 +7596,8 @@ int bnx2x_set_eth_mac(struct bnx2x *bp, bool set)
__set_bit(RAMROD_COMP_WAIT, &ramrod_flags);
/* Eth MAC is set on RSS leading client (fp[0]) */
- return bnx2x_set_mac_one(bp, bp->dev->dev_addr, &bp->fp->mac_obj, set,
- BNX2X_ETH_MAC, &ramrod_flags);
+ return bnx2x_set_mac_one(bp, bp->dev->dev_addr, &bp->sp_objs->mac_obj,
+ set, BNX2X_ETH_MAC, &ramrod_flags);
}
int bnx2x_setup_leading(struct bnx2x *bp)
@@ -7877,7 +7877,7 @@ int bnx2x_setup_queue(struct bnx2x *bp, struct bnx2x_fastpath *fp,
bnx2x_ack_sb(bp, fp->igu_sb_id, USTORM_ID, 0,
IGU_INT_ENABLE, 0);
- q_params.q_obj = &fp->q_obj;
+ q_params.q_obj = &bnx2x_sp_obj(bp, fp).q_obj;
/* We want to wait for completion in this context */
__set_bit(RAMROD_COMP_WAIT, &q_params.ramrod_flags);
@@ -7950,7 +7950,7 @@ static int bnx2x_stop_queue(struct bnx2x *bp, int index)
DP(NETIF_MSG_IFDOWN, "stopping queue %d cid %d\n", index, fp->cid);
- q_params.q_obj = &fp->q_obj;
+ q_params.q_obj = &bnx2x_sp_obj(bp, fp).q_obj;
/* We want to wait for completion in this context */
__set_bit(RAMROD_COMP_WAIT, &q_params.ramrod_flags);
@@ -8339,12 +8339,13 @@ void bnx2x_chip_cleanup(struct bnx2x *bp, int unload_mode)
usleep_range(1000, 1000);
/* Clean all ETH MACs */
- rc = bnx2x_del_all_macs(bp, &bp->fp[0].mac_obj, BNX2X_ETH_MAC, false);
+ rc = bnx2x_del_all_macs(bp, &bp->sp_objs[0].mac_obj, BNX2X_ETH_MAC,
+ false);
if (rc < 0)
BNX2X_ERR("Failed to delete all ETH macs: %d\n", rc);
/* Clean up UC list */
- rc = bnx2x_del_all_macs(bp, &bp->fp[0].mac_obj, BNX2X_UC_LIST_MAC,
+ rc = bnx2x_del_all_macs(bp, &bp->sp_objs[0].mac_obj, BNX2X_UC_LIST_MAC,
true);
if (rc < 0)
BNX2X_ERR("Failed to schedule DEL commands for UC MACs list: %d\n",
@@ -11049,7 +11050,7 @@ static int bnx2x_set_uc_list(struct bnx2x *bp)
int rc;
struct net_device *dev = bp->dev;
struct netdev_hw_addr *ha;
- struct bnx2x_vlan_mac_obj *mac_obj = &bp->fp->mac_obj;
+ struct bnx2x_vlan_mac_obj *mac_obj = &bp->sp_objs->mac_obj;
unsigned long ramrod_flags = 0;
/* First schedule a cleanup up of old configuration */
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
index 776b521..514a528 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
@@ -859,17 +859,22 @@ static int bnx2x_storm_stats_update(struct bnx2x *bp)
struct tstorm_per_queue_stats *tclient =
&bp->fw_stats_data->queue_stats[i].
tstorm_queue_statistics;
- struct tstorm_per_queue_stats *old_tclient = &fp->old_tclient;
+ struct tstorm_per_queue_stats *old_tclient =
+ &bnx2x_fp_stats(bp, fp)->old_tclient;
struct ustorm_per_queue_stats *uclient =
&bp->fw_stats_data->queue_stats[i].
ustorm_queue_statistics;
- struct ustorm_per_queue_stats *old_uclient = &fp->old_uclient;
+ struct ustorm_per_queue_stats *old_uclient =
+ &bnx2x_fp_stats(bp, fp)->old_uclient;
struct xstorm_per_queue_stats *xclient =
&bp->fw_stats_data->queue_stats[i].
xstorm_queue_statistics;
- struct xstorm_per_queue_stats *old_xclient = &fp->old_xclient;
- struct bnx2x_eth_q_stats *qstats = &fp->eth_q_stats;
- struct bnx2x_eth_q_stats_old *qstats_old = &fp->eth_q_stats_old;
+ struct xstorm_per_queue_stats *old_xclient =
+ &bnx2x_fp_stats(bp, fp)->old_xclient;
+ struct bnx2x_eth_q_stats *qstats =
+ &bnx2x_fp_stats(bp, fp)->eth_q_stats;
+ struct bnx2x_eth_q_stats_old *qstats_old =
+ &bnx2x_fp_stats(bp, fp)->eth_q_stats_old;
u32 diff;
@@ -1052,8 +1057,11 @@ static void bnx2x_net_stats_update(struct bnx2x *bp)
nstats->tx_bytes = bnx2x_hilo(&estats->total_bytes_transmitted_hi);
tmp = estats->mac_discard;
- for_each_rx_queue(bp, i)
- tmp += le32_to_cpu(bp->fp[i].old_tclient.checksum_discard);
+ for_each_rx_queue(bp, i) {
+ struct tstorm_per_queue_stats *old_tclient =
+ &bp->fp_stats[i].old_tclient;
+ tmp += le32_to_cpu(old_tclient->checksum_discard);
+ }
nstats->rx_dropped = tmp + bp->net_stats_old.rx_dropped;
nstats->tx_dropped = 0;
@@ -1103,9 +1111,9 @@ static void bnx2x_drv_stats_update(struct bnx2x *bp)
int i;
for_each_queue(bp, i) {
- struct bnx2x_eth_q_stats *qstats = &bp->fp[i].eth_q_stats;
+ struct bnx2x_eth_q_stats *qstats = &bp->fp_stats[i].eth_q_stats;
struct bnx2x_eth_q_stats_old *qstats_old =
- &bp->fp[i].eth_q_stats_old;
+ &bp->fp_stats[i].eth_q_stats_old;
UPDATE_ESTAT_QSTAT(driver_xoff);
UPDATE_ESTAT_QSTAT(rx_err_discard_pkt);
@@ -1483,15 +1491,19 @@ void bnx2x_stats_init(struct bnx2x *bp)
/* function stats */
for_each_queue(bp, i) {
- struct bnx2x_fastpath *fp = &bp->fp[i];
-
- memset(&fp->old_tclient, 0, sizeof(fp->old_tclient));
- memset(&fp->old_uclient, 0, sizeof(fp->old_uclient));
- memset(&fp->old_xclient, 0, sizeof(fp->old_xclient));
+ struct bnx2x_fp_stats *fp_stats = &bp->fp_stats[i];
+
+ memset(&fp_stats->old_tclient, 0,
+ sizeof(fp_stats->old_tclient));
+ memset(&fp_stats->old_uclient, 0,
+ sizeof(fp_stats->old_uclient));
+ memset(&fp_stats->old_xclient, 0,
+ sizeof(fp_stats->old_xclient));
if (bp->stats_init) {
- memset(&fp->eth_q_stats, 0, sizeof(fp->eth_q_stats));
- memset(&fp->eth_q_stats_old, 0,
- sizeof(fp->eth_q_stats_old));
+ memset(&fp_stats->eth_q_stats, 0,
+ sizeof(fp_stats->eth_q_stats));
+ memset(&fp_stats->eth_q_stats_old, 0,
+ sizeof(fp_stats->eth_q_stats_old));
}
}
@@ -1533,8 +1545,10 @@ void bnx2x_save_statistics(struct bnx2x *bp)
/* save queue statistics */
for_each_eth_queue(bp, i) {
struct bnx2x_fastpath *fp = &bp->fp[i];
- struct bnx2x_eth_q_stats *qstats = &fp->eth_q_stats;
- struct bnx2x_eth_q_stats_old *qstats_old = &fp->eth_q_stats_old;
+ struct bnx2x_eth_q_stats *qstats =
+ &bnx2x_fp_stats(bp, fp)->eth_q_stats;
+ struct bnx2x_eth_q_stats_old *qstats_old =
+ &bnx2x_fp_stats(bp, fp)->eth_q_stats_old;
UPDATE_QSTAT_OLD(total_unicast_bytes_received_hi);
UPDATE_QSTAT_OLD(total_unicast_bytes_received_lo);
@@ -1590,8 +1604,7 @@ void bnx2x_afex_collect_stats(struct bnx2x *bp, void *void_afex_stats,
memset(afex_stats, 0, sizeof(struct afex_stats));
for_each_eth_queue(bp, i) {
- struct bnx2x_fastpath *fp = &bp->fp[i];
- struct bnx2x_eth_q_stats *qstats = &fp->eth_q_stats;
+ struct bnx2x_eth_q_stats *qstats = &bp->fp_stats[i].eth_q_stats;
ADD_64(afex_stats->rx_unicast_bytes_hi,
qstats->total_unicast_bytes_received_hi,
--
1.7.10
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues
2012-06-13 12:44 [net-next patch 0/12] bnx2x: ethtool and other enhancements Merav Sicron
` (6 preceding siblings ...)
2012-06-13 12:44 ` [net-next patch 7/12] bnx2x: Split the FP structure Merav Sicron
@ 2012-06-13 12:44 ` Merav Sicron
2012-06-13 9:52 ` Eric Dumazet
2012-06-13 12:44 ` [net-next patch 9/12] bnx2x: Add support for ethtool -L Merav Sicron
` (3 subsequent siblings)
11 siblings, 1 reply; 28+ messages in thread
From: Merav Sicron @ 2012-06-13 12:44 UTC (permalink / raw)
To: eilong, davem, netdev; +Cc: Merav Sicron
This patch removed the limitation in the code for 16 RSS queues. The default
(without other instruction from the user) number of queues is determined
according to the number of MSI-X vectors supported for that function, the number
of CPUs in the system, and a maximum of 8 queues.
Signed-off-by: Merav Sicron <meravs@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x.h | 2 +-
drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h | 4 +++-
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c | 13 +++----------
3 files changed, 7 insertions(+), 12 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index 60a72f5..4dde45a 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -1643,7 +1643,7 @@ extern int num_queues;
#define BNX2X_MAX_QUEUES(bp) BNX2X_MAX_RSS_COUNT(bp)
/* #define is_eth_multi(bp) (BNX2X_NUM_ETH_QUEUES(bp) > 1) */
-
+#define BNX2X_DEFAULT_NUM_QUEUES 8
#define RSS_IPV4_CAP_MASK \
TSTORM_ETH_FUNCTION_COMMON_CONFIG_RSS_IPV4_CAPABILITY
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
index 11afe5d..2865575 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
@@ -815,9 +815,11 @@ static inline void bnx2x_disable_msi(struct bnx2x *bp)
static inline int bnx2x_calc_num_queues(struct bnx2x *bp)
{
+ int num = min_t(int, num_online_cpus(), BNX2X_DEFAULT_NUM_QUEUES);
+
return num_queues ?
min_t(int, num_queues, BNX2X_MAX_QUEUES(bp)) :
- min_t(int, num_online_cpus(), BNX2X_MAX_QUEUES(bp));
+ min_t(int, num, BNX2X_MAX_QUEUES(bp));
}
static inline void bnx2x_clear_sge_mask_next_elems(struct bnx2x_fastpath *fp)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index fbade5c..11bd0b6 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -11813,13 +11813,6 @@ static int __devinit bnx2x_init_one(struct pci_dev *pdev,
max_non_def_sbs = bnx2x_get_num_non_def_sbs(pdev);
- /* !!! FIXME !!!
- * Do not allow the maximum SB count to grow above 16
- * since Special CIDs starts from 16*BNX2X_MULTI_TX_COS=48.
- * We will use the FP_SB_MAX_E1x macro for this matter.
- */
- max_non_def_sbs = min_t(int, FP_SB_MAX_E1x, max_non_def_sbs);
-
WARN_ON(!max_non_def_sbs);
/* Maximum number of RSS queues: one IGU SB goes to CNIC */
@@ -11841,9 +11834,6 @@ static int __devinit bnx2x_init_one(struct pci_dev *pdev,
bp = netdev_priv(dev);
- BNX2X_DEV_INFO("Allocated netdev with %d tx and %d rx queues\n",
- tx_count, rx_count);
-
bp->igu_sb_cnt = max_non_def_sbs;
bp->msg_enable = debug;
pci_set_drvdata(pdev, dev);
@@ -11856,6 +11846,9 @@ static int __devinit bnx2x_init_one(struct pci_dev *pdev,
BNX2X_DEV_INFO("max_non_def_sbs %d\n", max_non_def_sbs);
+ BNX2X_DEV_INFO("Allocated netdev with %d tx and %d rx queues\n",
+ tx_count, rx_count);
+
rc = bnx2x_init_bp(bp);
if (rc)
goto init_one_exit;
--
1.7.10
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [net-next patch 9/12] bnx2x: Add support for ethtool -L
2012-06-13 12:44 [net-next patch 0/12] bnx2x: ethtool and other enhancements Merav Sicron
` (7 preceding siblings ...)
2012-06-13 12:44 ` [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues Merav Sicron
@ 2012-06-13 12:44 ` Merav Sicron
2012-06-15 17:03 ` Ben Hutchings
2012-06-13 12:44 ` [net-next patch 10/12] bnx2x: Support DCBX for all functions Merav Sicron
` (2 subsequent siblings)
11 siblings, 1 reply; 28+ messages in thread
From: Merav Sicron @ 2012-06-13 12:44 UTC (permalink / raw)
To: eilong, davem, netdev; +Cc: Merav Sicron
Add support for ethtool -L/-l for setting and getting the number of RSS queues.
The 'combined' field is used as we don't support separate IRQ for Rx and Tx.
Signed-off-by: Merav Sicron <meravs@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x.h | 4 +
drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c | 2 +-
drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h | 11 ++-
.../net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c | 81 ++++++++++++++++++++
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c | 15 ++--
5 files changed, 101 insertions(+), 12 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index 4dde45a..e0a3cc9 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -1408,6 +1408,7 @@ struct bnx2x {
#define BNX2X_MAX_COS 3
#define BNX2X_MAX_TX_COS 2
int num_queues;
+ int num_napi_queues;
int disable_tpa;
u32 rx_mode;
@@ -1695,6 +1696,9 @@ struct bnx2x_func_init_params {
continue; \
else
+#define for_each_napi_rx_queue(bp, var) \
+ for ((var) = 0; (var) < bp->num_napi_queues; (var)++)
+
/* Skip OOO FP */
#define for_each_tx_queue(bp, var) \
for ((var) = 0; (var) < BNX2X_NUM_QUEUES(bp); (var)++) \
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
index 915e680..c8a83a3 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
@@ -1343,7 +1343,7 @@ void bnx2x_free_irq(struct bnx2x *bp)
free_irq(bp->dev->irq, bp->dev);
}
-int __devinit bnx2x_enable_msix(struct bnx2x *bp)
+int bnx2x_enable_msix(struct bnx2x *bp)
{
int msix_vec = 0, i, rc, req_cnt;
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
index 2865575..b6ae665 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
@@ -29,6 +29,7 @@
extern int load_count[2][3]; /* per-path: 0-common, 1-port0, 2-port1 */
extern int num_queues;
+extern int int_mode;
/************************ Macros ********************************/
#define BNX2X_PCI_FREE(x, y, size) \
@@ -495,7 +496,7 @@ void bnx2x_netif_start(struct bnx2x *bp);
* fills msix_table, requests vectors, updates num_queues
* according to number of available vectors.
*/
-int __devinit bnx2x_enable_msix(struct bnx2x *bp);
+int bnx2x_enable_msix(struct bnx2x *bp);
/**
* bnx2x_enable_msi - request msi mode from OS, updated internals accordingly
@@ -788,8 +789,10 @@ static inline void bnx2x_add_all_napi(struct bnx2x *bp)
{
int i;
+ bp->num_napi_queues = bp->num_queues;
+
/* Add NAPI objects */
- for_each_rx_queue(bp, i)
+ for_each_napi_rx_queue(bp, i)
netif_napi_add(bp->dev, &bnx2x_fp(bp, i, napi),
bnx2x_poll, BNX2X_NAPI_WEIGHT);
}
@@ -798,10 +801,12 @@ static inline void bnx2x_del_all_napi(struct bnx2x *bp)
{
int i;
- for_each_rx_queue(bp, i)
+ for_each_napi_rx_queue(bp, i)
netif_napi_del(&bnx2x_fp(bp, i, napi));
}
+void bnx2x_set_int_mode(struct bnx2x *bp);
+
static inline void bnx2x_disable_msi(struct bnx2x *bp)
{
if (bp->flags & USING_MSIX_FLAG) {
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
index 13a51b0..43ee337 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
@@ -2809,6 +2809,85 @@ static int bnx2x_set_rxfh_indir(struct net_device *dev, const u32 *indir)
return bnx2x_config_rss_eth(bp, false);
}
+/**
+ * bnx2x_get_channels - gets the number of RSS queues.
+ *
+ * @dev: net device
+ * @channels: returns the number of max / current queues
+ */
+static void bnx2x_get_channels(struct net_device *dev,
+ struct ethtool_channels *channels)
+{
+ struct bnx2x *bp = netdev_priv(dev);
+
+ channels->max_rx = channels->max_tx = channels->max_other = 0;
+ channels->max_combined = BNX2X_MAX_RSS_COUNT(bp);
+ channels->rx_count = channels->tx_count = channels->other_count = 0;
+ channels->combined_count = BNX2X_NUM_ETH_QUEUES(bp);
+}
+
+/**
+ * bnx2x_change_num_queues - change the number of RSS queues.
+ *
+ * @bp: bnx2x private structure
+ *
+ * Re-configure interrupt mode to get the new number of MSI-X
+ * vectors and re-add NAPI objects.
+ */
+static void bnx2x_change_num_queues(struct bnx2x *bp, int num_rss)
+{
+ bnx2x_del_all_napi(bp);
+ bnx2x_disable_msi(bp);
+ BNX2X_NUM_QUEUES(bp) = num_rss + NON_ETH_CONTEXT_USE;
+ bnx2x_set_int_mode(bp);
+ bnx2x_add_all_napi(bp);
+}
+
+/**
+ * bnx2x_set_channels - sets the number of RSS queues.
+ *
+ * @dev: net device
+ * @channels: includes the number of queues requested
+ */
+static int bnx2x_set_channels(struct net_device *dev,
+ struct ethtool_channels *channels)
+{
+ struct bnx2x *bp = netdev_priv(dev);
+
+
+ DP(BNX2X_MSG_ETHTOOL,
+ "set-channels command parameters: rx = %d, tx = %d, other = %d, combined = %d\n",
+ channels->rx_count, channels->tx_count, channels->other_count,
+ channels->combined_count);
+
+ /* We don't support separate rx / tx channels.
+ * We don't allow setting 'other' channels.
+ */
+ if (channels->rx_count || channels->tx_count || channels->other_count
+ || (channels->combined_count > BNX2X_MAX_RSS_COUNT(bp))) {
+ DP(BNX2X_MSG_ETHTOOL, "command parameters not supported\n");
+ return -EINVAL;
+ }
+
+ /* Check if there was a change in the active parameters */
+ if (channels->combined_count == BNX2X_NUM_ETH_QUEUES(bp)) {
+ DP(BNX2X_MSG_ETHTOOL, "No change in active parameters\n");
+ return 0;
+ }
+
+ /* Set the requested number of queues in bp context.
+ * Note that the actual number of queues created during load may be
+ * less than requested if memory is low.
+ */
+ if (unlikely(!netif_running(dev))) {
+ bnx2x_change_num_queues(bp, channels->combined_count);
+ return 0;
+ }
+ bnx2x_nic_unload(bp, UNLOAD_NORMAL);
+ bnx2x_change_num_queues(bp, channels->combined_count);
+ return bnx2x_nic_load(bp, LOAD_NORMAL);
+}
+
static const struct ethtool_ops bnx2x_ethtool_ops = {
.get_settings = bnx2x_get_settings,
.set_settings = bnx2x_set_settings,
@@ -2840,6 +2919,8 @@ static const struct ethtool_ops bnx2x_ethtool_ops = {
.get_rxfh_indir_size = bnx2x_get_rxfh_indir_size,
.get_rxfh_indir = bnx2x_get_rxfh_indir,
.set_rxfh_indir = bnx2x_set_rxfh_indir,
+ .get_channels = bnx2x_get_channels,
+ .set_channels = bnx2x_set_channels,
.get_eee = bnx2x_get_eee,
.set_eee = bnx2x_set_eee,
};
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index 11bd0b6..3a4fefd 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -104,7 +104,7 @@ MODULE_PARM_DESC(disable_tpa, " Disable the TPA (LRO) feature");
#define INT_MODE_INTx 1
#define INT_MODE_MSI 2
-static int int_mode;
+int int_mode;
module_param(int_mode, int, 0);
MODULE_PARM_DESC(int_mode, " Force interrupt mode other than MSI-X "
"(1 INT#x; 2 MSI)");
@@ -7612,7 +7612,7 @@ int bnx2x_setup_leading(struct bnx2x *bp)
*
* In case of MSI-X it will also try to enable MSI-X.
*/
-static void __devinit bnx2x_set_int_mode(struct bnx2x *bp)
+void bnx2x_set_int_mode(struct bnx2x *bp)
{
switch (int_mode) {
case INT_MODE_MSI:
@@ -7623,11 +7623,6 @@ static void __devinit bnx2x_set_int_mode(struct bnx2x *bp)
BNX2X_DEV_INFO("set number of queues to 1\n");
break;
default:
- /* Set number of queues for MSI-X mode */
- bnx2x_set_num_queues(bp);
-
- BNX2X_DEV_INFO("set number of queues to %d\n", bp->num_queues);
-
/* if we can't use MSI-X we only need one fp,
* so try to enable MSI-X with the requested number of fp's
* and fallback to MSI or legacy INTx with one fp
@@ -11883,8 +11878,12 @@ static int __devinit bnx2x_init_one(struct pci_dev *pdev,
#endif
+
+ /* Set bp->num_queues for MSI-X mode*/
+ bnx2x_set_num_queues(bp);
+
/* Configure interrupt mode: try to enable MSI-X/MSI if
- * needed, set bp->num_queues appropriately.
+ * needed.
*/
bnx2x_set_int_mode(bp);
--
1.7.10
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [net-next patch 10/12] bnx2x: Support DCBX for all functions
2012-06-13 12:44 [net-next patch 0/12] bnx2x: ethtool and other enhancements Merav Sicron
` (8 preceding siblings ...)
2012-06-13 12:44 ` [net-next patch 9/12] bnx2x: Add support for ethtool -L Merav Sicron
@ 2012-06-13 12:44 ` Merav Sicron
2012-06-13 12:44 ` [net-next patch 11/12] bnx2x: Add FCoE capabilities advertisement Merav Sicron
2012-06-13 12:44 ` [net-next patch 12/12] bnx2x: Change date and version to 1.72.51-0 Merav Sicron
11 siblings, 0 replies; 28+ messages in thread
From: Merav Sicron @ 2012-06-13 12:44 UTC (permalink / raw)
To: eilong, davem, netdev; +Cc: Merav Sicron, Barak Witkowski
From: Barak Witkowski <barak@broadcom.com>
In multi-function device, allow configuring dcbx admin params from all drivers
on a single physical port.
Signed-off-by: Barak Witkowski <barak@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x.h | 1 +
drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c | 6 +++--
drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h | 2 +-
drivers/net/ethernet/broadcom/bnx2x/bnx2x_dcb.c | 30 +++++++++++++++-------
drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h | 1 +
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c | 2 ++
drivers/net/ethernet/broadcom/bnx2x/bnx2x_reg.h | 1 +
7 files changed, 31 insertions(+), 12 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index e0a3cc9..1f2e03a 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -1333,6 +1333,7 @@ struct bnx2x {
#define NO_FCOE_FLAG (1 << 15)
#define BC_SUPPORTS_PFC_STATS (1 << 17)
#define USING_SINGLE_MSIX_FLAG (1 << 20)
+#define BC_SUPPORTS_DCBX_MSG_NON_PMF (1 << 21)
#define NO_ISCSI(bp) ((bp)->flags & NO_ISCSI_FLAG)
#define NO_ISCSI_OOO(bp) ((bp)->flags & NO_ISCSI_OOO_FLAG)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
index c8a83a3..f774acf 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
@@ -2257,8 +2257,10 @@ int bnx2x_nic_load(struct bnx2x *bp, int load_mode)
return -EBUSY;
}
- if (bp->state != BNX2X_STATE_DIAG)
- bnx2x_dcbx_init(bp);
+ /* If PMF - send ADMIN DCBX msg to MFW to initiate DCBX FSM */
+ if (bp->port.pmf && (bp->state != BNX2X_STATE_DIAG))
+ bnx2x_dcbx_init(bp, false);
+
return 0;
#ifndef BNX2X_STOP_ON_ERROR
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
index b6ae665..cd98a3a 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
@@ -418,7 +418,7 @@ void bnx2x_ilt_set_info(struct bnx2x *bp);
*
* @bp: driver handle
*/
-void bnx2x_dcbx_init(struct bnx2x *bp);
+void bnx2x_dcbx_init(struct bnx2x *bp, bool update_shmem);
/**
* bnx2x_set_power_state - set power state to the requested value.
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_dcb.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_dcb.c
index 4f9244b..8a73374 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_dcb.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_dcb.c
@@ -972,23 +972,26 @@ void bnx2x_dcbx_init_params(struct bnx2x *bp)
bp->dcbx_config_params.admin_default_priority = 0;
}
-void bnx2x_dcbx_init(struct bnx2x *bp)
+void bnx2x_dcbx_init(struct bnx2x *bp, bool update_shmem)
{
u32 dcbx_lldp_params_offset = SHMEM_LLDP_DCBX_PARAMS_NONE;
+ /* only PMF can send ADMIN msg to MFW in old MFW versions */
+ if ((!bp->port.pmf) && (!(bp->flags & BC_SUPPORTS_DCBX_MSG_NON_PMF)))
+ return;
+
if (bp->dcbx_enabled <= 0)
return;
/* validate:
* chip of good for dcbx version,
* dcb is wanted
- * the function is pmf
* shmem2 contains DCBX support fields
*/
DP(BNX2X_MSG_DCB, "dcb_state %d bp->port.pmf %d\n",
bp->dcb_state, bp->port.pmf);
- if (bp->dcb_state == BNX2X_DCB_STATE_ON && bp->port.pmf &&
+ if (bp->dcb_state == BNX2X_DCB_STATE_ON &&
SHMEM2_HAS(bp, dcbx_lldp_params_offset)) {
dcbx_lldp_params_offset =
SHMEM2_RD(bp, dcbx_lldp_params_offset);
@@ -999,12 +1002,23 @@ void bnx2x_dcbx_init(struct bnx2x *bp)
bnx2x_update_drv_flags(bp, 1 << DRV_FLAGS_DCB_CONFIGURED, 0);
if (SHMEM_LLDP_DCBX_PARAMS_NONE != dcbx_lldp_params_offset) {
- bnx2x_dcbx_admin_mib_updated_params(bp,
- dcbx_lldp_params_offset);
+ /* need HW lock to avoid scenario of two drivers
+ * writing in parallel to shmem
+ */
+ bnx2x_acquire_hw_lock(bp,
+ HW_LOCK_RESOURCE_DCBX_ADMIN_MIB);
+ if (update_shmem)
+ bnx2x_dcbx_admin_mib_updated_params(bp,
+ dcbx_lldp_params_offset);
/* Let HW start negotiation */
bnx2x_fw_command(bp,
DRV_MSG_CODE_DCBX_ADMIN_PMF_MSG, 0);
+ /* release HW lock only after MFW acks that it finished
+ * reading values from shmem
+ */
+ bnx2x_release_hw_lock(bp,
+ HW_LOCK_RESOURCE_DCBX_ADMIN_MIB);
}
}
}
@@ -2063,10 +2077,8 @@ static u8 bnx2x_dcbnl_set_all(struct net_device *netdev)
"Handling parity error recovery. Try again later\n");
return 1;
}
- if (netif_running(bp->dev)) {
- bnx2x_nic_unload(bp, UNLOAD_NORMAL);
- rc = bnx2x_nic_load(bp, LOAD_NORMAL);
- }
+ if (netif_running(bp->dev))
+ bnx2x_dcbx_init(bp, true);
DP(BNX2X_MSG_DCB, "set_dcbx_params done (%d)\n", rc);
if (rc)
return 1;
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h
index c61aa37..6b77630 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h
@@ -1253,6 +1253,7 @@ struct drv_func_mb {
#define DRV_MSG_CODE_DCBX_ADMIN_PMF_MSG 0xb0000000
#define DRV_MSG_CODE_DCBX_PMF_DRV_OK 0xb2000000
+ #define REQ_BC_VER_4_DCBX_ADMIN_MSG_NON_PMF 0x00070401
#define DRV_MSG_CODE_VF_DISABLED_DONE 0xc0000000
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index 3a4fefd..26a8296 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -9732,6 +9732,8 @@ static void __devinit bnx2x_get_common_hwinfo(struct bnx2x *bp)
bp->flags |= (val >= REQ_BC_VER_4_PFC_STATS_SUPPORTED) ?
BC_SUPPORTS_PFC_STATS : 0;
+ bp->flags |= (val >= REQ_BC_VER_4_DCBX_ADMIN_MSG_NON_PMF) ?
+ BC_SUPPORTS_DCBX_MSG_NON_PMF : 0;
boot_mode = SHMEM_RD(bp,
dev_info.port_feature_config[BP_PORT(bp)].mba_config) &
PORT_FEATURE_MBA_BOOT_AGENT_TYPE_MASK;
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_reg.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_reg.h
index bfef98f..a78e356 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_reg.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_reg.h
@@ -5913,6 +5913,7 @@
#define MISC_REGISTERS_SPIO_OUTPUT_LOW 0
#define MISC_REGISTERS_SPIO_SET_POS 8
#define HW_LOCK_MAX_RESOURCE_VALUE 31
+#define HW_LOCK_RESOURCE_DCBX_ADMIN_MIB 13
#define HW_LOCK_RESOURCE_DRV_FLAGS 10
#define HW_LOCK_RESOURCE_GPIO 1
#define HW_LOCK_RESOURCE_MDIO 0
--
1.7.10
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [net-next patch 11/12] bnx2x: Add FCoE capabilities advertisement
2012-06-13 12:44 [net-next patch 0/12] bnx2x: ethtool and other enhancements Merav Sicron
` (9 preceding siblings ...)
2012-06-13 12:44 ` [net-next patch 10/12] bnx2x: Support DCBX for all functions Merav Sicron
@ 2012-06-13 12:44 ` Merav Sicron
2012-06-13 12:44 ` [net-next patch 12/12] bnx2x: Change date and version to 1.72.51-0 Merav Sicron
11 siblings, 0 replies; 28+ messages in thread
From: Merav Sicron @ 2012-06-13 12:44 UTC (permalink / raw)
To: eilong, davem, netdev; +Cc: Merav Sicron, Barak Witkowski
From: Barak Witkowski <barak@broadcom.com>
1. When FCoE offload driver is registered, copy its capabilities to the chip
scratchpad.
2. Copy FCoE/iSCSI MAC addresses in aligned manner to chip scratchpad.
Signed-off-by: Barak Witkowski <barak@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x.h | 1 +
drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h | 41 +++++++++++++++++++++-
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c | 41 +++++++++++++++++++---
drivers/net/ethernet/broadcom/cnic_if.h | 6 ++++
4 files changed, 83 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index 1f2e03a..47ec384 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -1332,6 +1332,7 @@ struct bnx2x {
#define NO_ISCSI_FLAG (1 << 14)
#define NO_FCOE_FLAG (1 << 15)
#define BC_SUPPORTS_PFC_STATS (1 << 17)
+#define BC_SUPPORTS_FCOE_FEATURES (1 << 19)
#define USING_SINGLE_MSIX_FLAG (1 << 20)
#define BC_SUPPORTS_DCBX_MSG_NON_PMF (1 << 21)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h
index 6b77630..6bad5c4 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h
@@ -1250,6 +1250,7 @@ struct drv_func_mb {
#define REQ_BC_VER_4_VRFY_AFEX_SUPPORTED 0x00070002
#define REQ_BC_VER_4_SFP_TX_DISABLE_SUPPORTED 0x00070014
#define REQ_BC_VER_4_PFC_STATS_SUPPORTED 0x00070201
+ #define REQ_BC_VER_4_FCOE_FEATURES 0x00070209
#define DRV_MSG_CODE_DCBX_ADMIN_PMF_MSG 0xb0000000
#define DRV_MSG_CODE_DCBX_PMF_DRV_OK 0xb2000000
@@ -2698,8 +2699,46 @@ struct host_func_stats {
/* VIC definitions */
#define VICSTATST_UIF_INDEX 2
+/* FCoE capabilities required from the driver */
+struct fcoe_capabilities {
+ u32 capability1;
+ /* Maximum number of I/Os per connection */
+ #define FCOE_IOS_PER_CONNECTION_MASK 0x0000ffff
+ #define FCOE_IOS_PER_CONNECTION_SHIFT 0
+ /* Maximum number of Logins per port */
+ #define FCOE_LOGINS_PER_PORT_MASK 0xffff0000
+ #define FCOE_LOGINS_PER_PORT_SHIFT 16
+
+ u32 capability2;
+ /* Maximum number of exchanges */
+ #define FCOE_NUMBER_OF_EXCHANGES_MASK 0x0000ffff
+ #define FCOE_NUMBER_OF_EXCHANGES_SHIFT 0
+ /* Maximum NPIV WWN per port */
+ #define FCOE_NPIV_WWN_PER_PORT_MASK 0xffff0000
+ #define FCOE_NPIV_WWN_PER_PORT_SHIFT 16
+
+ u32 capability3;
+ /* Maximum number of targets supported */
+ #define FCOE_TARGETS_SUPPORTED_MASK 0x0000ffff
+ #define FCOE_TARGETS_SUPPORTED_SHIFT 0
+ /* Maximum number of outstanding commands across all connections */
+ #define FCOE_OUTSTANDING_COMMANDS_MASK 0xffff0000
+ #define FCOE_OUTSTANDING_COMMANDS_SHIFT 16
+
+ u32 capability4;
+ #define FCOE_CAPABILITY4_STATEFUL 0x00000001
+ #define FCOE_CAPABILITY4_STATELESS 0x00000002
+ #define FCOE_CAPABILITY4_CAPABILITIES_REPORTED_VALID 0x00000004
+};
+
+struct glob_ncsi_oem_data {
+ u32 driver_version;
+ u32 unused[3];
+ struct fcoe_capabilities fcoe_features[NVM_PATH_MAX][PORT_MAX];
+};
+
/* current drv_info version */
-#define DRV_INFO_CUR_VER 1
+#define DRV_INFO_CUR_VER 2
/* drv_info op codes supported */
enum drv_info_opcode {
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index 26a8296..eb3e55a 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -74,6 +74,8 @@
#define FW_FILE_NAME_E1H "bnx2x/bnx2x-e1h-" FW_FILE_VERSION ".fw"
#define FW_FILE_NAME_E2 "bnx2x/bnx2x-e2-" FW_FILE_VERSION ".fw"
+#define MAC_LEADING_ZERO_CNT (ALIGN(ETH_ALEN, sizeof(u32)) - ETH_ALEN)
+
/* Time in jiffies before concluding the transmitter is hung */
#define TX_TIMEOUT (5*HZ)
@@ -3060,7 +3062,8 @@ static void bnx2x_drv_info_fcoe_stat(struct bnx2x *bp)
struct fcoe_stats_info *fcoe_stat =
&bp->slowpath->drv_info_to_mcp.fcoe_stat;
- memcpy(fcoe_stat->mac_local, bp->fip_mac, ETH_ALEN);
+ memcpy(fcoe_stat->mac_local + MAC_LEADING_ZERO_CNT,
+ bp->fip_mac, ETH_ALEN);
fcoe_stat->qos_priority =
app->traffic_type_priority[LLFC_TRAFFIC_TYPE_FCOE];
@@ -3151,7 +3154,8 @@ static void bnx2x_drv_info_iscsi_stat(struct bnx2x *bp)
struct iscsi_stats_info *iscsi_stat =
&bp->slowpath->drv_info_to_mcp.iscsi_stat;
- memcpy(iscsi_stat->mac_local, bp->cnic_eth_dev.iscsi_mac, ETH_ALEN);
+ memcpy(iscsi_stat->mac_local + MAC_LEADING_ZERO_CNT,
+ bp->cnic_eth_dev.iscsi_mac, ETH_ALEN);
iscsi_stat->qos_priority =
app->traffic_type_priority[LLFC_TRAFFIC_TYPE_ISCSI];
@@ -9732,6 +9736,9 @@ static void __devinit bnx2x_get_common_hwinfo(struct bnx2x *bp)
bp->flags |= (val >= REQ_BC_VER_4_PFC_STATS_SUPPORTED) ?
BC_SUPPORTS_PFC_STATS : 0;
+ bp->flags |= (val >= REQ_BC_VER_4_FCOE_FEATURES) ?
+ BC_SUPPORTS_FCOE_FEATURES : 0;
+
bp->flags |= (val >= REQ_BC_VER_4_DCBX_ADMIN_MSG_NON_PMF) ?
BC_SUPPORTS_DCBX_MSG_NON_PMF : 0;
boot_mode = SHMEM_RD(bp,
@@ -12548,21 +12555,45 @@ static int bnx2x_drv_ctl(struct net_device *dev, struct drv_ctl_info *ctl)
break;
}
case DRV_CTL_ULP_REGISTER_CMD: {
- int ulp_type = ctl->data.ulp_type;
+ int ulp_type = ctl->data.register_data.ulp_type;
if (CHIP_IS_E3(bp)) {
int idx = BP_FW_MB_IDX(bp);
- u32 cap;
+ u32 cap = SHMEM2_RD(bp, drv_capabilities_flag[idx]);
+ int path = BP_PATH(bp);
+ int port = BP_PORT(bp);
+ int i;
+ u32 scratch_offset;
+ u32 *host_addr;
- cap = SHMEM2_RD(bp, drv_capabilities_flag[idx]);
+ /* first write capability to shmem2 */
if (ulp_type == CNIC_ULP_ISCSI)
cap |= DRV_FLAGS_CAPABILITIES_LOADED_ISCSI;
else if (ulp_type == CNIC_ULP_FCOE)
cap |= DRV_FLAGS_CAPABILITIES_LOADED_FCOE;
SHMEM2_WR(bp, drv_capabilities_flag[idx], cap);
+
+ if ((ulp_type != CNIC_ULP_FCOE) ||
+ (!SHMEM2_HAS(bp, ncsi_oem_data_addr)) ||
+ (!(bp->flags & BC_SUPPORTS_FCOE_FEATURES)))
+ break;
+
+ /* if reached here - should write fcoe capabilities */
+ scratch_offset = SHMEM2_RD(bp, ncsi_oem_data_addr);
+ if (!scratch_offset)
+ break;
+ scratch_offset += offsetof(struct glob_ncsi_oem_data,
+ fcoe_features[path][port]);
+ host_addr = (u32 *) &(ctl->data.register_data.
+ fcoe_features);
+ for (i = 0; i < sizeof(struct fcoe_capabilities);
+ i += 4)
+ REG_WR(bp, scratch_offset + i,
+ *(host_addr + i/4));
}
break;
}
+
case DRV_CTL_ULP_UNREGISTER_CMD: {
int ulp_type = ctl->data.ulp_type;
diff --git a/drivers/net/ethernet/broadcom/cnic_if.h b/drivers/net/ethernet/broadcom/cnic_if.h
index 289274e..82872e6 100644
--- a/drivers/net/ethernet/broadcom/cnic_if.h
+++ b/drivers/net/ethernet/broadcom/cnic_if.h
@@ -131,6 +131,11 @@ struct drv_ctl_l2_ring {
u32 cid;
};
+struct drv_ctl_register_data {
+ int ulp_type;
+ struct fcoe_capabilities fcoe_features;
+};
+
struct drv_ctl_info {
int cmd;
union {
@@ -138,6 +143,7 @@ struct drv_ctl_info {
struct drv_ctl_io io;
struct drv_ctl_l2_ring ring;
int ulp_type;
+ struct drv_ctl_register_data register_data;
char bytes[MAX_DRV_CTL_DATA];
} data;
};
--
1.7.10
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [net-next patch 12/12] bnx2x: Change date and version to 1.72.51-0
2012-06-13 12:44 [net-next patch 0/12] bnx2x: ethtool and other enhancements Merav Sicron
` (10 preceding siblings ...)
2012-06-13 12:44 ` [net-next patch 11/12] bnx2x: Add FCoE capabilities advertisement Merav Sicron
@ 2012-06-13 12:44 ` Merav Sicron
11 siblings, 0 replies; 28+ messages in thread
From: Merav Sicron @ 2012-06-13 12:44 UTC (permalink / raw)
To: eilong, davem, netdev; +Cc: Merav Sicron
This change updates the date and version of the bnx2x driver.
Signed-off-by: Merav Sicron <meravs@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index e30e2a2..85a3b84 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -23,8 +23,8 @@
* (you will need to reboot afterwards) */
/* #define BNX2X_STOP_ON_ERROR */
-#define DRV_MODULE_VERSION "1.72.50-0"
-#define DRV_MODULE_RELDATE "2012/04/23"
+#define DRV_MODULE_VERSION "1.72.51-0"
+#define DRV_MODULE_RELDATE "2012/06/12"
#define BNX2X_BC_VER 0x040200
#if defined(CONFIG_DCB)
--
1.7.10
^ permalink raw reply related [flat|nested] 28+ messages in thread
* Re: [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues
2012-06-13 15:14 ` Merav Sicron
@ 2012-06-13 13:53 ` Eilon Greenstein
2012-06-13 13:59 ` Eric Dumazet
2012-06-13 22:35 ` David Miller
0 siblings, 2 replies; 28+ messages in thread
From: Eilon Greenstein @ 2012-06-13 13:53 UTC (permalink / raw)
To: Eric Dumazet; +Cc: Merav Sicron, davem, netdev
On Wed, 2012-06-13 at 18:14 +0300, Merav Sicron wrote:
> Hi Eric,
>
> On Wed, 2012-06-13 at 11:52 +0200, Eric Dumazet wrote:
> > On Wed, 2012-06-13 at 15:44 +0300, Merav Sicron wrote:
> > > This patch removed the limitation in the code for 16 RSS queues. The default
> > > (without other instruction from the user) number of queues is determined
> > > according to the number of MSI-X vectors supported for that function, the number
> > > of CPUs in the system, and a maximum of 8 queues.
> >
> > Thats a very confusing changelog
> >
> > You meant : " a minimum of 8 queues" ?
> >
> No, I meant maximum...for example in a system with 16 CPUs we will
> allocate 8 RSS queues. We saw that in most scenarios we tested there was
> no need for more than 8 queues to get the maximal throughput, and by
> limiting to 8 by default we reduce the allocated memory. We do provide
> ethtool -L support so that the user could request more RSS queues if
> desired.
>
> > You should give more explanations, because its a sensible area for
> > performances.
> >
Just to emphasis, since this is the patch series that enable the users
to control the number of queues, we can reduce the default number and
allow the user to increase it if he has a setup that needs more than 8
parallel CPUs to receive the traffic. When using a new FW on the board,
the number can be increased up to 64, so using the maximal number can be
an overkill (even if the machine has 64 CPUs, it does not mean that the
user would like us to consume 64 MSI-X vectors and all the memory to set
up 64 queues) - so a lower default value can be used to satisfy most
users while allowing them to increase the number if they wish.
Thanks,
Eilon
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues
2012-06-13 13:53 ` Eilon Greenstein
@ 2012-06-13 13:59 ` Eric Dumazet
2012-06-13 22:35 ` David Miller
1 sibling, 0 replies; 28+ messages in thread
From: Eric Dumazet @ 2012-06-13 13:59 UTC (permalink / raw)
To: eilong; +Cc: Merav Sicron, davem, netdev
On Wed, 2012-06-13 at 16:53 +0300, Eilon Greenstein wrote:
> On Wed, 2012-06-13 at 18:14 +0300, Merav Sicron wrote:
> > Hi Eric,
> >
> > On Wed, 2012-06-13 at 11:52 +0200, Eric Dumazet wrote:
> > > On Wed, 2012-06-13 at 15:44 +0300, Merav Sicron wrote:
> > > > This patch removed the limitation in the code for 16 RSS queues. The default
> > > > (without other instruction from the user) number of queues is determined
> > > > according to the number of MSI-X vectors supported for that function, the number
> > > > of CPUs in the system, and a maximum of 8 queues.
> > >
> > > Thats a very confusing changelog
> > >
> > > You meant : " a minimum of 8 queues" ?
> > >
> > No, I meant maximum...for example in a system with 16 CPUs we will
> > allocate 8 RSS queues. We saw that in most scenarios we tested there was
> > no need for more than 8 queues to get the maximal throughput, and by
> > limiting to 8 by default we reduce the allocated memory. We do provide
> > ethtool -L support so that the user could request more RSS queues if
> > desired.
> >
> > > You should give more explanations, because its a sensible area for
> > > performances.
> > >
>
> Just to emphasis, since this is the patch series that enable the users
> to control the number of queues, we can reduce the default number and
> allow the user to increase it if he has a setup that needs more than 8
> parallel CPUs to receive the traffic. When using a new FW on the board,
> the number can be increased up to 64, so using the maximal number can be
> an overkill (even if the machine has 64 CPUs, it does not mean that the
> user would like us to consume 64 MSI-X vectors and all the memory to set
> up 64 queues) - so a lower default value can be used to satisfy most
> users while allowing them to increase the number if they wish.
I am all for a reduction of default number of queues.
Even a laptop has now 8 cpus, so on servers, using one queue per cpu is
overkill for most uses.
Thanks
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues
2012-06-13 9:52 ` Eric Dumazet
@ 2012-06-13 15:14 ` Merav Sicron
2012-06-13 13:53 ` Eilon Greenstein
0 siblings, 1 reply; 28+ messages in thread
From: Merav Sicron @ 2012-06-13 15:14 UTC (permalink / raw)
To: Eric Dumazet; +Cc: eilong, davem, netdev
Hi Eric,
On Wed, 2012-06-13 at 11:52 +0200, Eric Dumazet wrote:
> On Wed, 2012-06-13 at 15:44 +0300, Merav Sicron wrote:
> > This patch removed the limitation in the code for 16 RSS queues. The default
> > (without other instruction from the user) number of queues is determined
> > according to the number of MSI-X vectors supported for that function, the number
> > of CPUs in the system, and a maximum of 8 queues.
>
> Thats a very confusing changelog
>
> You meant : " a minimum of 8 queues" ?
>
No, I meant maximum...for example in a system with 16 CPUs we will
allocate 8 RSS queues. We saw that in most scenarios we tested there was
no need for more than 8 queues to get the maximal throughput, and by
limiting to 8 by default we reduce the allocated memory. We do provide
ethtool -L support so that the user could request more RSS queues if
desired.
> You should give more explanations, because its a sensible area for
> performances.
>
Thanks,
Merav
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [net-next patch 1/12] bnx2x: Add support for external LB
2012-06-13 12:44 ` [net-next patch 1/12] bnx2x: Add support for external LB Merav Sicron
@ 2012-06-13 18:17 ` Ben Hutchings
0 siblings, 0 replies; 28+ messages in thread
From: Ben Hutchings @ 2012-06-13 18:17 UTC (permalink / raw)
To: Merav Sicron; +Cc: eilong, davem, netdev
On Wed, 2012-06-13 at 15:44 +0300, Merav Sicron wrote:
> This change enables to do self-test with external loopback via ethtool.
>
> Signed-off-by: Merav Sicron <meravs@broadcom.com>
> Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Acked-by: Ben Hutchings <bhutchings@solarflare.com>
The ethtool bits look right.
[...]
> --- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
> +++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
> @@ -2124,7 +2124,12 @@ u8 bnx2x_initial_phy_init(struct bnx2x *bp, int load_mode)
> }
> }
>
> - rc = bnx2x_phy_init(&bp->link_params, &bp->link_vars);
> + if (load_mode == LOAD_LOOPBACK_EXT) {
> + struct link_params *lp = &bp->link_params;
> + lp->loopback_mode = LOOPBACK_EXT;
> + }
> +
> + rc = bnx2x_phy_init(&bp->link_params, &bp->link_vars);
>
> bnx2x_release_phy_lock(bp);
Indentation is wrong here, though.
Ben.
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [net-next patch 3/12] bnx2x: Add support for 4-tupple UDP RSS
2012-06-13 12:44 ` [net-next patch 3/12] bnx2x: Add support for 4-tupple UDP RSS Merav Sicron
@ 2012-06-13 18:20 ` Ben Hutchings
0 siblings, 0 replies; 28+ messages in thread
From: Ben Hutchings @ 2012-06-13 18:20 UTC (permalink / raw)
To: Merav Sicron; +Cc: eilong, davem, netdev
On Wed, 2012-06-13 at 15:44 +0300, Merav Sicron wrote:
> This change enables to control via ethtool whether to do UDP RSS on 2-tupple
> (IP source / destination only) or on 4-tupple (include UDP source / destination
> port). It also enables to read back the RSS configuration.
[...]
> --- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
> +++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
> @@ -2600,6 +2600,52 @@ static int bnx2x_set_phys_id(struct net_device *dev,
> return 0;
> }
>
> +static int bnx2x_get_rss_flags(struct bnx2x *bp, struct ethtool_rxnfc *info)
> +{
> +
> + switch (info->flow_type) {
> + case TCP_V4_FLOW:
> + case TCP_V6_FLOW:
> + info->data = RXH_IP_SRC | RXH_IP_DST |
> + RXH_L4_B_0_1 | RXH_L4_B_2_3;
> + break;
> + case UDP_V4_FLOW:
> + if (bp->rss_conf_obj.udp_rss_v4)
> + info->data = RXH_IP_SRC | RXH_IP_DST |
> + RXH_L4_B_0_1 | RXH_L4_B_2_3;
> + else
> + info->data = RXH_IP_SRC | RXH_IP_DST;
> + break;
> + case UDP_V6_FLOW:
> + if (bp->rss_conf_obj.udp_rss_v6)
> + info->data = RXH_IP_SRC | RXH_IP_DST |
> + RXH_L4_B_0_1 | RXH_L4_B_2_3;
> + else
> + info->data = RXH_IP_SRC | RXH_IP_DST;
> + break;
> + case IPV4_FLOW:
> + case IPV6_FLOW:
> + info->data = RXH_IP_SRC | RXH_IP_DST;
> + break;
> + case SCTP_V4_FLOW:
> + case AH_ESP_V4_FLOW:
> + case AH_V4_FLOW:
> + case ESP_V4_FLOW:
> + case SCTP_V6_FLOW:
> + case AH_ESP_V6_FLOW:
> + case AH_V6_FLOW:
> + case ESP_V6_FLOW:
> + case IP_USER_FLOW:
> + case ETHER_FLOW:
> + info->data = 0;
> + break;
> + default:
> + return -EINVAL;
> + }
[...]
Don't try to enumerate all flow types; we may add others in future. You
should set info->data = 0 for any type for which RSS is not supported.
Otherwise I think the ethtool bits are fine.
Ben.
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues
2012-06-13 13:53 ` Eilon Greenstein
2012-06-13 13:59 ` Eric Dumazet
@ 2012-06-13 22:35 ` David Miller
2012-06-14 15:34 ` Merav Sicron
1 sibling, 1 reply; 28+ messages in thread
From: David Miller @ 2012-06-13 22:35 UTC (permalink / raw)
To: eilong; +Cc: eric.dumazet, meravs, netdev
From: "Eilon Greenstein" <eilong@broadcom.com>
Date: Wed, 13 Jun 2012 16:53:29 +0300
> Just to emphasis, since this is the patch series that enable the users
> to control the number of queues, we can reduce the default number and
> allow the user to increase it if he has a setup that needs more than 8
> parallel CPUs to receive the traffic. When using a new FW on the board,
> the number can be increased up to 64, so using the maximal number can be
> an overkill (even if the machine has 64 CPUs, it does not mean that the
> user would like us to consume 64 MSI-X vectors and all the memory to set
> up 64 queues) - so a lower default value can be used to satisfy most
> users while allowing them to increase the number if they wish.
I think you should look at other drivers for guidance in this area.
There is zero value in each and every driver author deciding what
is a good default strategy, because this means the user gets a
very inconsistent experience based purely upon the driver author's
whims.
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues
2012-06-13 22:35 ` David Miller
@ 2012-06-14 15:34 ` Merav Sicron
2012-06-14 20:36 ` David Miller
0 siblings, 1 reply; 28+ messages in thread
From: Merav Sicron @ 2012-06-14 15:34 UTC (permalink / raw)
To: David Miller; +Cc: eilong, eric.dumazet, netdev
On Wed, 2012-06-13 at 15:35 -0700, David Miller wrote:
> From: "Eilon Greenstein" <eilong@broadcom.com>
> Date: Wed, 13 Jun 2012 16:53:29 +0300
>
> > Just to emphasis, since this is the patch series that enable the users
> > to control the number of queues, we can reduce the default number and
> > allow the user to increase it if he has a setup that needs more than 8
> > parallel CPUs to receive the traffic. When using a new FW on the board,
> > the number can be increased up to 64, so using the maximal number can be
> > an overkill (even if the machine has 64 CPUs, it does not mean that the
> > user would like us to consume 64 MSI-X vectors and all the memory to set
> > up 64 queues) - so a lower default value can be used to satisfy most
> > users while allowing them to increase the number if they wish.
>
> I think you should look at other drivers for guidance in this area.
>
> There is zero value in each and every driver author deciding what
> is a good default strategy, because this means the user gets a
> very inconsistent experience based purely upon the driver author's
> whims.
>
We looked at few other drivers - their current behavior is similar to
what bnx2x had before this change: Minimum between the number of CPU and
a defined maximum (probably the HW limit).
bnx2x HW limit is 64 (in most other drivers it is smaller, but this can
change). The number of CPUs in new systems becomes bigger and bigger,
and allocaintg so many RSS queues seems like a waste. With the
relatively new ethtool -L feature the user can change the number of
queues. That's why we think (and so does Eric Dumazet) that it is better
to have a smaller default number which is good for most cases.
Do you agree with that?
Thanks,
Merav
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues
2012-06-14 15:34 ` Merav Sicron
@ 2012-06-14 20:36 ` David Miller
2012-06-17 6:53 ` Eilon Greenstein
0 siblings, 1 reply; 28+ messages in thread
From: David Miller @ 2012-06-14 20:36 UTC (permalink / raw)
To: meravs; +Cc: eilong, eric.dumazet, netdev
From: "Merav Sicron" <meravs@broadcom.com>
Date: Thu, 14 Jun 2012 18:34:06 +0300
> That's why we think (and so does Eric Dumazet) that it is better
> to have a smaller default number which is good for most cases.
> Do you agree with that?
What I think is that the thing which is more important than the
default we choose, is that it is consistently followed by all
multiqueue drivers.
By blazing your own unique path here, that is nearly guaranteed not to
happen.
I'd much rather have a bad default that every driver adheres to.
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [net-next patch 9/12] bnx2x: Add support for ethtool -L
2012-06-13 12:44 ` [net-next patch 9/12] bnx2x: Add support for ethtool -L Merav Sicron
@ 2012-06-15 17:03 ` Ben Hutchings
0 siblings, 0 replies; 28+ messages in thread
From: Ben Hutchings @ 2012-06-15 17:03 UTC (permalink / raw)
To: Merav Sicron; +Cc: eilong, davem, netdev
On Wed, 2012-06-13 at 15:44 +0300, Merav Sicron wrote:
> Add support for ethtool -L/-l for setting and getting the number of RSS queues.
> The 'combined' field is used as we don't support separate IRQ for Rx and Tx.
[...]
> --- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
> +++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
> @@ -2809,6 +2809,85 @@ static int bnx2x_set_rxfh_indir(struct net_device *dev, const u32 *indir)
> return bnx2x_config_rss_eth(bp, false);
> }
>
> +/**
> + * bnx2x_get_channels - gets the number of RSS queues.
> + *
> + * @dev: net device
> + * @channels: returns the number of max / current queues
> + */
> +static void bnx2x_get_channels(struct net_device *dev,
> + struct ethtool_channels *channels)
> +{
> + struct bnx2x *bp = netdev_priv(dev);
> +
> + channels->max_rx = channels->max_tx = channels->max_other = 0;
> + channels->max_combined = BNX2X_MAX_RSS_COUNT(bp);
> + channels->rx_count = channels->tx_count = channels->other_count = 0;
> + channels->combined_count = BNX2X_NUM_ETH_QUEUES(bp);
Don't bother setting fields to 0; you can assume the ethtool core does
that for you.
> +}
> +
> +/**
> + * bnx2x_change_num_queues - change the number of RSS queues.
> + *
> + * @bp: bnx2x private structure
> + *
> + * Re-configure interrupt mode to get the new number of MSI-X
> + * vectors and re-add NAPI objects.
> + */
> +static void bnx2x_change_num_queues(struct bnx2x *bp, int num_rss)
> +{
> + bnx2x_del_all_napi(bp);
> + bnx2x_disable_msi(bp);
> + BNX2X_NUM_QUEUES(bp) = num_rss + NON_ETH_CONTEXT_USE;
The NON_ETH_CONTEXT_USE queues could possibly be counted as 'other'.
Depends on how closely they are associated with the net device.
> + bnx2x_set_int_mode(bp);
> + bnx2x_add_all_napi(bp);
> +}
> +
> +/**
> + * bnx2x_set_channels - sets the number of RSS queues.
> + *
> + * @dev: net device
> + * @channels: includes the number of queues requested
> + */
> +static int bnx2x_set_channels(struct net_device *dev,
> + struct ethtool_channels *channels)
> +{
> + struct bnx2x *bp = netdev_priv(dev);
> +
> +
> + DP(BNX2X_MSG_ETHTOOL,
> + "set-channels command parameters: rx = %d, tx = %d, other = %d, combined = %d\n",
> + channels->rx_count, channels->tx_count, channels->other_count,
> + channels->combined_count);
> +
> + /* We don't support separate rx / tx channels.
> + * We don't allow setting 'other' channels.
> + */
> + if (channels->rx_count || channels->tx_count || channels->other_count
> + || (channels->combined_count > BNX2X_MAX_RSS_COUNT(bp))) {
Missing a check for combined_count == 0.
Ben.
> + DP(BNX2X_MSG_ETHTOOL, "command parameters not supported\n");
> + return -EINVAL;
> + }
> +
> + /* Check if there was a change in the active parameters */
> + if (channels->combined_count == BNX2X_NUM_ETH_QUEUES(bp)) {
> + DP(BNX2X_MSG_ETHTOOL, "No change in active parameters\n");
> + return 0;
> + }
> +
> + /* Set the requested number of queues in bp context.
> + * Note that the actual number of queues created during load may be
> + * less than requested if memory is low.
> + */
> + if (unlikely(!netif_running(dev))) {
> + bnx2x_change_num_queues(bp, channels->combined_count);
> + return 0;
> + }
> + bnx2x_nic_unload(bp, UNLOAD_NORMAL);
> + bnx2x_change_num_queues(bp, channels->combined_count);
> + return bnx2x_nic_load(bp, LOAD_NORMAL);
> +}
[...]
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues
2012-06-14 20:36 ` David Miller
@ 2012-06-17 6:53 ` Eilon Greenstein
2012-06-17 8:08 ` David Miller
0 siblings, 1 reply; 28+ messages in thread
From: Eilon Greenstein @ 2012-06-17 6:53 UTC (permalink / raw)
To: David Miller; +Cc: meravs, eric.dumazet, netdev
On Thu, 2012-06-14 at 13:36 -0700, David Miller wrote:
> From: "Merav Sicron" <meravs@broadcom.com>
> Date: Thu, 14 Jun 2012 18:34:06 +0300
>
> > That's why we think (and so does Eric Dumazet) that it is better
> > to have a smaller default number which is good for most cases.
> > Do you agree with that?
>
> What I think is that the thing which is more important than the
> default we choose, is that it is consistently followed by all
> multiqueue drivers.
>
> By blazing your own unique path here, that is nearly guaranteed not to
> happen.
>
> I'd much rather have a bad default that every driver adheres to.
>
The increasing number of CPUs together with the increasing number of
supported MSI-X vectors per device, is causing a lot of memory to be
allocated for the networking channels. This is a problem on low memory
platforms such as 32bits installations. 64 queues per device (and in
most cases, there are few devices) is a becoming a real problem.
If the device does not implement the ethtool .set_channels operation,
then the default is practically the only way to go, and there is no such
thing as a good default - it really depends on the use case. This is the
reason why we are only setting the default in the series that adds
support for this operation.
So what do you say about setting a new default only for devices that
supports .set_channels? There are only 2 of those right now (qlcnic,
bnx2 and hopefully bnx2x soon). Is it acceptable to keep the default
hard coded value or do you want it to be configurable? If the latter, is
a kernel parameter a valid option?
Please note that today, on a system with enough CPUs, you cannot really
tell how many channels will be created per device - each vendor is
allocating up to its HW limit - usually 8 or 16, but there is no easy
way to determine it beside reading the code. So adding a kernel
parameter to limit the number can actually increase the consistency.
Thanks,
Eilon
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues
2012-06-17 6:53 ` Eilon Greenstein
@ 2012-06-17 8:08 ` David Miller
2012-06-17 8:33 ` Eilon Greenstein
0 siblings, 1 reply; 28+ messages in thread
From: David Miller @ 2012-06-17 8:08 UTC (permalink / raw)
To: eilong; +Cc: meravs, eric.dumazet, netdev
If you're not going to read what I wrote, don't pretend that you did.
That really irritates me and makes me want to not review or apply your
patches.
I didn't say that there wasn't a problem.
I said that I don't want you to just blaze your own trail without
getting agreement and coordinating to adjust the other drivers in
unison as well. I refuse to allow you to just change the default of
your driver and then well... maybe other drivers follow and maybe
others do not. That hodge-podge inconsistent situation sucks for
users.
So now are you going to try again to explain the problem to me or
are you going to understand what my actual concern is?
Thanks.
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues
2012-06-17 8:08 ` David Miller
@ 2012-06-17 8:33 ` Eilon Greenstein
2012-06-17 9:00 ` David Miller
0 siblings, 1 reply; 28+ messages in thread
From: Eilon Greenstein @ 2012-06-17 8:33 UTC (permalink / raw)
To: David Miller; +Cc: meravs, eric.dumazet, netdev
On Sun, 2012-06-17 at 01:08 -0700, David Miller wrote:
> If you're not going to read what I wrote, don't pretend that you did.
> That really irritates me and makes me want to not review or apply your
> patches.
>
> I didn't say that there wasn't a problem.
>
> I said that I don't want you to just blaze your own trail without
> getting agreement and coordinating to adjust the other drivers in
> unison as well. I refuse to allow you to just change the default of
> your driver and then well... maybe other drivers follow and maybe
> others do not. That hodge-podge inconsistent situation sucks for
> users.
>
> So now are you going to try again to explain the problem to me or
> are you going to understand what my actual concern is?
>
I understood what you are saying, and if it looks like I was still
pushing for this patch to be accepted then I apologize, it is not what
I'm trying to do. What I am trying to do now, is to find a way to set a
default for all drivers. Since set_channels is applicable to only two
drivers now, it does not sound like a good path - so I was asking if you
can consider a kernel parameter as acceptable path. To be clear, I'm
asking if adding the default for all multi-queue drivers in the form of
a kernel parameter is a good path to explore.
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues
2012-06-17 8:33 ` Eilon Greenstein
@ 2012-06-17 9:00 ` David Miller
2012-06-17 9:31 ` Eilon Greenstein
0 siblings, 1 reply; 28+ messages in thread
From: David Miller @ 2012-06-17 9:00 UTC (permalink / raw)
To: eilong; +Cc: meravs, eric.dumazet, netdev
From: "Eilon Greenstein" <eilong@broadcom.com>
Date: Sun, 17 Jun 2012 11:33:56 +0300
> drivers now, it does not sound like a good path - so I was asking if you
> can consider a kernel parameter as acceptable path. To be clear, I'm
> asking if adding the default for all multi-queue drivers in the form of
> a kernel parameter is a good path to explore.
If it's the default, there's no need for a kernel command line
parameter. Those suck, just pick a good default.
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues
2012-06-17 9:00 ` David Miller
@ 2012-06-17 9:31 ` Eilon Greenstein
0 siblings, 0 replies; 28+ messages in thread
From: Eilon Greenstein @ 2012-06-17 9:31 UTC (permalink / raw)
To: David Miller; +Cc: meravs, eric.dumazet, netdev
On Sun, 2012-06-17 at 02:00 -0700, David Miller wrote:
> From: "Eilon Greenstein" <eilong@broadcom.com>
> Date: Sun, 17 Jun 2012 11:33:56 +0300
>
> > drivers now, it does not sound like a good path - so I was asking if you
> > can consider a kernel parameter as acceptable path. To be clear, I'm
> > asking if adding the default for all multi-queue drivers in the form of
> > a kernel parameter is a good path to explore.
>
> If it's the default, there's no need for a kernel command line
> parameter. Those suck, just pick a good default.
>
OK. So we will send an RFC series setting the default (hard coded
define) to 8 for all multi-queue drivers and that will start a
discussion if this is a good default number or not. The side-effect of
this series will be that drivers that did not implement set_channels
will not be able to change this default back to the HW limit like they
run today (assuming the HW limit is higher than the chosen default).
On top of that, we will send this series for the bnx2x again without any
default -allowing up to 64 queues unless set_channels is used.
Eilon
^ permalink raw reply [flat|nested] 28+ messages in thread
end of thread, other threads:[~2012-06-17 9:31 UTC | newest]
Thread overview: 28+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-06-13 12:44 [net-next patch 0/12] bnx2x: ethtool and other enhancements Merav Sicron
2012-06-13 12:44 ` [net-next patch 1/12] bnx2x: Add support for external LB Merav Sicron
2012-06-13 18:17 ` Ben Hutchings
2012-06-13 12:44 ` [net-next patch 2/12] bnx2x: Return only online tests for MF Merav Sicron
2012-06-13 12:44 ` [net-next patch 3/12] bnx2x: Add support for 4-tupple UDP RSS Merav Sicron
2012-06-13 18:20 ` Ben Hutchings
2012-06-13 12:44 ` [net-next patch 4/12] bnx2x: Allow more than 64 L2 CIDs Merav Sicron
2012-06-13 12:44 ` [net-next patch 5/12] bnx2x: Make the transmission queues adjacent Merav Sicron
2012-06-13 12:44 ` [net-next patch 6/12] bnx2x: Move the CNIC L2 CIDs to be right after the RSS CIDs Merav Sicron
2012-06-13 12:44 ` [net-next patch 7/12] bnx2x: Split the FP structure Merav Sicron
2012-06-13 12:44 ` [net-next patch 8/12] bnx2x: Allow up to 63 RSS queues default 8 queues Merav Sicron
2012-06-13 9:52 ` Eric Dumazet
2012-06-13 15:14 ` Merav Sicron
2012-06-13 13:53 ` Eilon Greenstein
2012-06-13 13:59 ` Eric Dumazet
2012-06-13 22:35 ` David Miller
2012-06-14 15:34 ` Merav Sicron
2012-06-14 20:36 ` David Miller
2012-06-17 6:53 ` Eilon Greenstein
2012-06-17 8:08 ` David Miller
2012-06-17 8:33 ` Eilon Greenstein
2012-06-17 9:00 ` David Miller
2012-06-17 9:31 ` Eilon Greenstein
2012-06-13 12:44 ` [net-next patch 9/12] bnx2x: Add support for ethtool -L Merav Sicron
2012-06-15 17:03 ` Ben Hutchings
2012-06-13 12:44 ` [net-next patch 10/12] bnx2x: Support DCBX for all functions Merav Sicron
2012-06-13 12:44 ` [net-next patch 11/12] bnx2x: Add FCoE capabilities advertisement Merav Sicron
2012-06-13 12:44 ` [net-next patch 12/12] bnx2x: Change date and version to 1.72.51-0 Merav Sicron
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.