* [PATCH net-next v3 10/10] MAINTAINERS: add entry for Dynamic Interrupt Moderation
From: Andy Gospodarek @ 2018-01-09 6:13 UTC (permalink / raw)
To: netdev; +Cc: mchan, talgi, ogerlitz, Andy Gospodarek
In-Reply-To: <1515478402-6048-1-git-send-email-andy@greyhouse.net>
From: Andy Gospodarek <gospo@broadcom.com>
Signed-off-by: Andy Gospodarek <gospo@broadcom.com>
Signed-off-by: Tal Gilboa <talgi@mellanox.com>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
---
MAINTAINERS | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/MAINTAINERS b/MAINTAINERS
index 753799d..178239dc 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -4944,6 +4944,11 @@ S: Maintained
F: lib/dynamic_debug.c
F: include/linux/dynamic_debug.h
+DYNAMIC INTERRUPT MODERATION
+M: Tal Gilboa <talgi@mellanox.com>
+S: Maintained
+F: include/linux/net_dim.h
+
DZ DECSTATION DZ11 SERIAL DRIVER
M: "Maciej W. Rozycki" <macro@linux-mips.org>
S: Maintained
--
2.7.4
^ permalink raw reply related
* [PATCH net-next v3 09/10] bnxt_en: add support for software dynamic interrupt moderation
From: Andy Gospodarek @ 2018-01-09 6:13 UTC (permalink / raw)
To: netdev; +Cc: mchan, talgi, ogerlitz, Andy Gospodarek
In-Reply-To: <1515478402-6048-1-git-send-email-andy@greyhouse.net>
From: Andy Gospodarek <gospo@broadcom.com>
This implements the changes needed for the bnxt_en driver to add support
for dynamic interrupt moderation per ring.
This does add additional counters in the receive path, but testing shows
that any additional instructions are offset by throughput gain when the
default configuration is for low latency.
Signed-off-by: Andy Gospodarek <gospo@broadcom.com>
Cc: Michael Chan <mchan@broadcom.com>
---
drivers/net/ethernet/broadcom/bnxt/Makefile | 2 +-
drivers/net/ethernet/broadcom/bnxt/bnxt.c | 50 +++++++++++++++++++++++
drivers/net/ethernet/broadcom/bnxt/bnxt.h | 34 ++++++++++-----
drivers/net/ethernet/broadcom/bnxt/bnxt_dim.c | 33 +++++++++++++++
drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c | 12 ++++++
5 files changed, 119 insertions(+), 12 deletions(-)
create mode 100644 drivers/net/ethernet/broadcom/bnxt/bnxt_dim.c
diff --git a/drivers/net/ethernet/broadcom/bnxt/Makefile b/drivers/net/ethernet/broadcom/bnxt/Makefile
index 59c8ec9..7c560d5 100644
--- a/drivers/net/ethernet/broadcom/bnxt/Makefile
+++ b/drivers/net/ethernet/broadcom/bnxt/Makefile
@@ -1,4 +1,4 @@
obj-$(CONFIG_BNXT) += bnxt_en.o
-bnxt_en-y := bnxt.o bnxt_sriov.o bnxt_ethtool.o bnxt_dcb.o bnxt_ulp.o bnxt_xdp.o bnxt_vfr.o bnxt_devlink.o
+bnxt_en-y := bnxt.o bnxt_sriov.o bnxt_ethtool.o bnxt_dcb.o bnxt_ulp.o bnxt_xdp.o bnxt_vfr.o bnxt_devlink.o bnxt_dim.o
bnxt_en-$(CONFIG_BNXT_FLOWER_OFFLOAD) += bnxt_tc.o
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
index 89c3c87..cf6ebf1 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
@@ -1645,6 +1645,8 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_napi *bnapi, u32 *raw_cons,
rxr->rx_next_cons = NEXT_RX(cons);
next_rx_no_prod:
+ cpr->rx_packets += 1;
+ cpr->rx_bytes += len;
*raw_cons = tmp_raw_cons;
return rc;
@@ -1802,6 +1804,7 @@ static irqreturn_t bnxt_msix(int irq, void *dev_instance)
struct bnxt_cp_ring_info *cpr = &bnapi->cp_ring;
u32 cons = RING_CMP(cpr->cp_raw_cons);
+ cpr->event_ctr++;
prefetch(&cpr->cp_desc_ring[CP_RING(cons)][CP_IDX(cons)]);
napi_schedule(&bnapi->napi);
return IRQ_HANDLED;
@@ -2025,6 +2028,15 @@ static int bnxt_poll(struct napi_struct *napi, int budget)
break;
}
}
+ if (bp->flags & BNXT_FLAG_DIM) {
+ struct net_dim_sample dim_sample;
+
+ net_dim_sample(cpr->event_ctr,
+ cpr->rx_packets,
+ cpr->rx_bytes,
+ &dim_sample);
+ net_dim(&cpr->dim, dim_sample);
+ }
mmiowb();
return work_done;
}
@@ -2617,6 +2629,8 @@ static void bnxt_init_cp_rings(struct bnxt *bp)
struct bnxt_ring_struct *ring = &cpr->cp_ring_struct;
ring->fw_ring_id = INVALID_HW_RING_ID;
+ cpr->rx_ring_coal.coal_ticks = bp->rx_coal.coal_ticks;
+ cpr->rx_ring_coal.coal_bufs = bp->rx_coal.coal_bufs;
}
}
@@ -4593,6 +4607,36 @@ static void bnxt_hwrm_set_coal_params(struct bnxt_coal *hw_coal,
req->flags = cpu_to_le16(flags);
}
+int bnxt_hwrm_set_ring_coal(struct bnxt *bp, struct bnxt_napi *bnapi)
+{
+ struct hwrm_ring_cmpl_ring_cfg_aggint_params_input req_rx = {0};
+ struct bnxt_cp_ring_info *cpr = &bnapi->cp_ring;
+ struct bnxt_coal coal;
+ unsigned int grp_idx;
+
+ /* Tick values in micro seconds.
+ * 1 coal_buf x bufs_per_record = 1 completion record.
+ */
+ memcpy(&coal, &bp->rx_coal, sizeof(struct bnxt_coal));
+
+ coal.coal_ticks = cpr->rx_ring_coal.coal_ticks;
+ coal.coal_bufs = cpr->rx_ring_coal.coal_bufs;
+
+ if (!bnapi->rx_ring)
+ return -ENODEV;
+
+ bnxt_hwrm_cmd_hdr_init(bp, &req_rx,
+ HWRM_RING_CMPL_RING_CFG_AGGINT_PARAMS, -1, -1);
+
+ bnxt_hwrm_set_coal_params(&coal, &req_rx);
+
+ grp_idx = bnapi->index;
+ req_rx.ring_id = cpu_to_le16(bp->grp_info[grp_idx].cp_fw_ring_id);
+
+ return hwrm_send_message(bp, &req_rx, sizeof(req_rx),
+ HWRM_CMD_TIMEOUT);
+}
+
int bnxt_hwrm_set_coal(struct bnxt *bp)
{
int i, rc = 0;
@@ -5715,7 +5759,13 @@ static void bnxt_enable_napi(struct bnxt *bp)
int i;
for (i = 0; i < bp->cp_nr_rings; i++) {
+ struct bnxt_cp_ring_info *cpr = &bp->bnapi[i]->cp_ring;
bp->bnapi[i]->in_reset = false;
+
+ if (bp->bnapi[i]->rx_ring) {
+ INIT_WORK(&cpr->dim.work, bnxt_dim_work);
+ cpr->dim.mode = NET_DIM_CQ_PERIOD_MODE_START_FROM_EQE;
+ }
napi_enable(&bp->bnapi[i]->napi);
}
}
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.h b/drivers/net/ethernet/broadcom/bnxt/bnxt.h
index 2d268fc..89887a8 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.h
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.h
@@ -24,6 +24,7 @@
#include <net/dst_metadata.h>
#include <net/switchdev.h>
#include <net/xdp.h>
+#include <linux/net_dim.h>
struct tx_bd {
__le32 tx_bd_len_flags_type;
@@ -608,6 +609,17 @@ struct bnxt_tx_ring_info {
struct bnxt_ring_struct tx_ring_struct;
};
+struct bnxt_coal {
+ u16 coal_ticks;
+ u16 coal_ticks_irq;
+ u16 coal_bufs;
+ u16 coal_bufs_irq;
+ /* RING_IDLE enabled when coal ticks < idle_thresh */
+ u16 idle_thresh;
+ u8 bufs_per_record;
+ u8 budget;
+};
+
struct bnxt_tpa_info {
void *data;
u8 *data_ptr;
@@ -672,6 +684,13 @@ struct bnxt_cp_ring_info {
u32 cp_raw_cons;
void __iomem *cp_doorbell;
+ struct bnxt_coal rx_ring_coal;
+ u64 rx_packets;
+ u64 rx_bytes;
+ u64 event_ctr;
+
+ struct net_dim dim;
+
struct tx_cmp *cp_desc_ring[MAX_CP_PAGES];
dma_addr_t cp_desc_mapping[MAX_CP_PAGES];
@@ -946,17 +965,6 @@ struct bnxt_test_info {
#define BNXT_CAG_REG_LEGACY_INT_STATUS 0x4014
#define BNXT_CAG_REG_BASE 0x300000
-struct bnxt_coal {
- u16 coal_ticks;
- u16 coal_ticks_irq;
- u16 coal_bufs;
- u16 coal_bufs_irq;
- /* RING_IDLE enabled when coal ticks < idle_thresh */
- u16 idle_thresh;
- u8 bufs_per_record;
- u8 budget;
-};
-
struct bnxt_tc_flow_stats {
u64 packets;
u64 bytes;
@@ -1128,6 +1136,7 @@ struct bnxt {
#define BNXT_FLAG_DOUBLE_DB 0x400000
#define BNXT_FLAG_FW_DCBX_AGENT 0x800000
#define BNXT_FLAG_CHIP_NITRO_A0 0x1000000
+ #define BNXT_FLAG_DIM 0x2000000
#define BNXT_FLAG_ALL_CONFIG_FEATS (BNXT_FLAG_TPA | \
BNXT_FLAG_RFS | \
@@ -1425,4 +1434,7 @@ int bnxt_setup_mq_tc(struct net_device *dev, u8 tc);
int bnxt_get_max_rings(struct bnxt *, int *, int *, bool);
void bnxt_restore_pf_fw_resources(struct bnxt *bp);
int bnxt_port_attr_get(struct bnxt *bp, struct switchdev_attr *attr);
+void bnxt_dim_work(struct work_struct *work);
+int bnxt_hwrm_set_ring_coal(struct bnxt *bp, struct bnxt_napi *bnapi);
+
#endif
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_dim.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_dim.c
new file mode 100644
index 0000000..3980f08
--- /dev/null
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_dim.c
@@ -0,0 +1,33 @@
+/* Broadcom NetXtreme-C/E network driver.
+ *
+ * Copyright (c) 2017-2018 Broadcom Limited
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation.
+ */
+
+#include <linux/net_dim.h>
+#include "bnxt_hsi.h"
+#include "bnxt.h"
+
+void bnxt_dim_work(struct work_struct *work)
+{
+ struct net_dim *dim = container_of(work, struct net_dim,
+ work);
+ struct bnxt_cp_ring_info *cpr = container_of(dim,
+ struct bnxt_cp_ring_info,
+ dim);
+ struct bnxt_napi *bnapi = container_of(cpr,
+ struct bnxt_napi,
+ cp_ring);
+ struct net_dim_cq_moder cur_profile = net_dim_get_profile(dim->mode,
+ dim->profile_ix);
+
+ cpr->rx_ring_coal.coal_ticks = cur_profile.usec;
+ cpr->rx_ring_coal.coal_bufs = cur_profile.pkts;
+
+ bnxt_hwrm_set_ring_coal(bnapi->bp, bnapi);
+ dim->state = NET_DIM_START_MEASURE;
+}
+
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c
index fe7599f..1801582 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c
@@ -49,6 +49,8 @@ static int bnxt_get_coalesce(struct net_device *dev,
memset(coal, 0, sizeof(*coal));
+ coal->use_adaptive_rx_coalesce = bp->flags & BNXT_FLAG_DIM;
+
hw_coal = &bp->rx_coal;
mult = hw_coal->bufs_per_record;
coal->rx_coalesce_usecs = hw_coal->coal_ticks;
@@ -77,6 +79,15 @@ static int bnxt_set_coalesce(struct net_device *dev,
int rc = 0;
u16 mult;
+ if (coal->use_adaptive_rx_coalesce) {
+ bp->flags |= BNXT_FLAG_DIM;
+ } else {
+ if (bp->flags & BNXT_FLAG_DIM) {
+ bp->flags &= ~(BNXT_FLAG_DIM);
+ goto reset_coalesce;
+ }
+ }
+
hw_coal = &bp->rx_coal;
mult = hw_coal->bufs_per_record;
hw_coal->coal_ticks = coal->rx_coalesce_usecs;
@@ -104,6 +115,7 @@ static int bnxt_set_coalesce(struct net_device *dev,
update_stats = true;
}
+reset_coalesce:
if (netif_running(dev)) {
if (update_stats) {
rc = bnxt_close_nic(bp, true, false);
--
2.7.4
^ permalink raw reply related
* [PATCH net-next v3 07/10] net/mlx5e: Move dynamic interrupt coalescing code to include/linux
From: Andy Gospodarek @ 2018-01-09 6:13 UTC (permalink / raw)
To: netdev; +Cc: mchan, talgi, ogerlitz, Andy Gospodarek
In-Reply-To: <1515478402-6048-1-git-send-email-andy@greyhouse.net>
From: Andy Gospodarek <gospo@broadcom.com>
This move allows drivers to add private structure elements to track the
number of packets, bytes, and interrupts events per ring. A driver
also defines a workqueue handler to act on this collected data once per
poll and modify the coalescing parameters per ring.
Signed-off-by: Andy Gospodarek <gospo@broadcom.com>
Acked-by: Tal Gilboa <talgi@mellanox.com>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/Makefile | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en.h | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en_dim.c | 1 +
drivers/net/ethernet/mellanox/mlx5/core/net_dim.c | 307 ------------------
drivers/net/ethernet/mellanox/mlx5/core/net_dim.h | 108 -------
include/linux/net_dim.h | 377 ++++++++++++++++++++++
6 files changed, 380 insertions(+), 417 deletions(-)
delete mode 100644 drivers/net/ethernet/mellanox/mlx5/core/net_dim.c
delete mode 100644 drivers/net/ethernet/mellanox/mlx5/core/net_dim.h
create mode 100644 include/linux/net_dim.h
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Makefile b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
index b46b6de2..c805769 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/Makefile
+++ b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
@@ -15,7 +15,7 @@ mlx5_core-$(CONFIG_MLX5_FPGA) += fpga/cmd.o fpga/core.o fpga/conn.o fpga/sdk.o \
mlx5_core-$(CONFIG_MLX5_CORE_EN) += en_main.o en_common.o en_fs.o en_ethtool.o \
en_tx.o en_rx.o en_dim.o en_txrx.o en_stats.o vxlan.o \
- en_arfs.o en_fs_ethtool.o en_selftest.o net_dim.o
+ en_arfs.o en_fs_ethtool.o en_selftest.o
mlx5_core-$(CONFIG_MLX5_MPFS) += lib/mpfs.o
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index 4d1d298..29b9675 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -47,10 +47,10 @@
#include <linux/rhashtable.h>
#include <net/switchdev.h>
#include <net/xdp.h>
+#include <linux/net_dim.h>
#include "wq.h"
#include "mlx5_core.h"
#include "en_stats.h"
-#include "net_dim.h"
#define MLX5_SET_CFG(p, f, v) MLX5_SET(create_flow_group_in, p, f, v)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.c b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.c
index f620325..2b89951 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.c
@@ -30,6 +30,7 @@
* SOFTWARE.
*/
+#include <linux/net_dim.h>
#include "en.h"
void mlx5e_rx_dim_work(struct work_struct *work)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/net_dim.c b/drivers/net/ethernet/mellanox/mlx5/core/net_dim.c
deleted file mode 100644
index decb370..0000000
--- a/drivers/net/ethernet/mellanox/mlx5/core/net_dim.c
+++ /dev/null
@@ -1,307 +0,0 @@
-/*
- * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
- * Copyright (c) 2017-2018, Broadcom Limited. All rights reserved.
- *
- * This software is available to you under a choice of one of two
- * licenses. You may choose to be licensed under the terms of the GNU
- * General Public License (GPL) Version 2, available from the file
- * COPYING in the main directory of this source tree, or the
- * OpenIB.org BSD license below:
- *
- * Redistribution and use in source and binary forms, with or
- * without modification, are permitted provided that the following
- * conditions are met:
- *
- * - Redistributions of source code must retain the above
- * copyright notice, this list of conditions and the following
- * disclaimer.
- *
- * - Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following
- * disclaimer in the documentation and/or other materials
- * provided with the distribution.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
- * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
- * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
- * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- */
-
-#include "en.h"
-
-#define NET_DIM_PARAMS_NUM_PROFILES 5
-/* Adaptive moderation profiles */
-#define NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE 256
-#define NET_DIM_DEF_PROFILE_CQE 1
-#define NET_DIM_DEF_PROFILE_EQE 1
-
-/* All profiles sizes must be NET_PARAMS_DIM_NUM_PROFILES */
-#define NET_DIM_EQE_PROFILES { \
- {1, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
- {8, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
- {64, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
- {128, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
- {256, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
-}
-
-#define NET_DIM_CQE_PROFILES { \
- {2, 256}, \
- {8, 128}, \
- {16, 64}, \
- {32, 64}, \
- {64, 64} \
-}
-
-static const struct net_dim_cq_moder
-profile[NET_DIM_CQ_PERIOD_NUM_MODES][NET_DIM_PARAMS_NUM_PROFILES] = {
- NET_DIM_EQE_PROFILES,
- NET_DIM_CQE_PROFILES,
-};
-
-struct net_dim_cq_moder net_dim_get_profile(u8 cq_period_mode, int ix)
-{
- struct net_dim_cq_moder cq_moder;
-
- cq_moder = profile[cq_period_mode][ix];
- cq_moder.cq_period_mode = cq_period_mode;
- return cq_moder;
-}
-
-struct net_dim_cq_moder net_dim_get_def_profile(u8 rx_cq_period_mode)
-{
- int default_profile_ix;
-
- if (rx_cq_period_mode == NET_DIM_CQ_PERIOD_MODE_START_FROM_CQE)
- default_profile_ix = NET_DIM_DEF_PROFILE_CQE;
- else /* NET_DIM_CQ_PERIOD_MODE_START_FROM_EQE */
- default_profile_ix = NET_DIM_DEF_PROFILE_EQE;
-
- return net_dim_get_profile(rx_cq_period_mode, default_profile_ix);
-}
-
-static bool net_dim_on_top(struct net_dim *dim)
-{
- switch (dim->tune_state) {
- case NET_DIM_PARKING_ON_TOP:
- case NET_DIM_PARKING_TIRED:
- return true;
- case NET_DIM_GOING_RIGHT:
- return (dim->steps_left > 1) && (dim->steps_right == 1);
- default: /* NET_DIM_GOING_LEFT */
- return (dim->steps_right > 1) && (dim->steps_left == 1);
- }
-}
-
-static void net_dim_turn(struct net_dim *dim)
-{
- switch (dim->tune_state) {
- case NET_DIM_PARKING_ON_TOP:
- case NET_DIM_PARKING_TIRED:
- break;
- case NET_DIM_GOING_RIGHT:
- dim->tune_state = NET_DIM_GOING_LEFT;
- dim->steps_left = 0;
- break;
- case NET_DIM_GOING_LEFT:
- dim->tune_state = NET_DIM_GOING_RIGHT;
- dim->steps_right = 0;
- break;
- }
-}
-
-static int net_dim_step(struct net_dim *dim)
-{
- if (dim->tired == (NET_DIM_PARAMS_NUM_PROFILES * 2))
- return NET_DIM_TOO_TIRED;
-
- switch (dim->tune_state) {
- case NET_DIM_PARKING_ON_TOP:
- case NET_DIM_PARKING_TIRED:
- break;
- case NET_DIM_GOING_RIGHT:
- if (dim->profile_ix == (NET_DIM_PARAMS_NUM_PROFILES - 1))
- return NET_DIM_ON_EDGE;
- dim->profile_ix++;
- dim->steps_right++;
- break;
- case NET_DIM_GOING_LEFT:
- if (dim->profile_ix == 0)
- return NET_DIM_ON_EDGE;
- dim->profile_ix--;
- dim->steps_left++;
- break;
- }
-
- dim->tired++;
- return NET_DIM_STEPPED;
-}
-
-static void net_dim_park_on_top(struct net_dim *dim)
-{
- dim->steps_right = 0;
- dim->steps_left = 0;
- dim->tired = 0;
- dim->tune_state = NET_DIM_PARKING_ON_TOP;
-}
-
-static void net_dim_park_tired(struct net_dim *dim)
-{
- dim->steps_right = 0;
- dim->steps_left = 0;
- dim->tune_state = NET_DIM_PARKING_TIRED;
-}
-
-static void net_dim_exit_parking(struct net_dim *dim)
-{
- dim->tune_state = dim->profile_ix ? NET_DIM_GOING_LEFT :
- NET_DIM_GOING_RIGHT;
- net_dim_step(dim);
-}
-
-#define IS_SIGNIFICANT_DIFF(val, ref) \
- (((100 * abs((val) - (ref))) / (ref)) > 10) /* more than 10% difference */
-
-static int net_dim_stats_compare(struct net_dim_stats *curr,
- struct net_dim_stats *prev)
-{
- if (!prev->bpms)
- return curr->bpms ? NET_DIM_STATS_BETTER :
- NET_DIM_STATS_SAME;
-
- if (IS_SIGNIFICANT_DIFF(curr->bpms, prev->bpms))
- return (curr->bpms > prev->bpms) ? NET_DIM_STATS_BETTER :
- NET_DIM_STATS_WORSE;
-
- if (IS_SIGNIFICANT_DIFF(curr->ppms, prev->ppms))
- return (curr->ppms > prev->ppms) ? NET_DIM_STATS_BETTER :
- NET_DIM_STATS_WORSE;
-
- if (IS_SIGNIFICANT_DIFF(curr->epms, prev->epms))
- return (curr->epms < prev->epms) ? NET_DIM_STATS_BETTER :
- NET_DIM_STATS_WORSE;
-
- return NET_DIM_STATS_SAME;
-}
-
-static bool net_dim_decision(struct net_dim_stats *curr_stats,
- struct net_dim *dim)
-{
- int prev_state = dim->tune_state;
- int prev_ix = dim->profile_ix;
- int stats_res;
- int step_res;
-
- switch (dim->tune_state) {
- case NET_DIM_PARKING_ON_TOP:
- stats_res = net_dim_stats_compare(curr_stats, &dim->prev_stats);
- if (stats_res != NET_DIM_STATS_SAME)
- net_dim_exit_parking(dim);
- break;
-
- case NET_DIM_PARKING_TIRED:
- dim->tired--;
- if (!dim->tired)
- net_dim_exit_parking(dim);
- break;
-
- case NET_DIM_GOING_RIGHT:
- case NET_DIM_GOING_LEFT:
- stats_res = net_dim_stats_compare(curr_stats, &dim->prev_stats);
- if (stats_res != NET_DIM_STATS_BETTER)
- net_dim_turn(dim);
-
- if (net_dim_on_top(dim)) {
- net_dim_park_on_top(dim);
- break;
- }
-
- step_res = net_dim_step(dim);
- switch (step_res) {
- case NET_DIM_ON_EDGE:
- net_dim_park_on_top(dim);
- break;
- case NET_DIM_TOO_TIRED:
- net_dim_park_tired(dim);
- break;
- }
-
- break;
- }
-
- if ((prev_state != NET_DIM_PARKING_ON_TOP) ||
- (dim->tune_state != NET_DIM_PARKING_ON_TOP))
- dim->prev_stats = *curr_stats;
-
- return dim->profile_ix != prev_ix;
-}
-
-static void net_dim_sample(u16 event_ctr,
- u64 packets,
- u64 bytes,
- struct net_dim_sample *s)
-{
- s->time = ktime_get();
- s->pkt_ctr = packets;
- s->byte_ctr = bytes;
- s->event_ctr = event_ctr;
-}
-
-#define NET_DIM_NEVENTS 64
-#define BITS_PER_TYPE(type) (sizeof(type) * BITS_PER_BYTE)
-#define BIT_GAP(bits, end, start) ((((end) - (start)) + BIT_ULL(bits)) & (BIT_ULL(bits) - 1))
-
-static void net_dim_calc_stats(struct net_dim_sample *start,
- struct net_dim_sample *end,
- struct net_dim_stats *curr_stats)
-{
- /* u32 holds up to 71 minutes, should be enough */
- u32 delta_us = ktime_us_delta(end->time, start->time);
- u32 npkts = BIT_GAP(BITS_PER_TYPE(u32), end->pkt_ctr, start->pkt_ctr);
- u32 nbytes = BIT_GAP(BITS_PER_TYPE(u32), end->byte_ctr,
- start->byte_ctr);
-
- if (!delta_us)
- return;
-
- curr_stats->ppms = DIV_ROUND_UP(npkts * USEC_PER_MSEC, delta_us);
- curr_stats->bpms = DIV_ROUND_UP(nbytes * USEC_PER_MSEC, delta_us);
- curr_stats->epms = DIV_ROUND_UP(NET_DIM_NEVENTS * USEC_PER_MSEC,
- delta_us);
-}
-
-void net_dim(struct net_dim *dim,
- u16 event_ctr,
- u64 packets,
- u64 bytes)
-{
- struct net_dim_sample end_sample;
- struct net_dim_stats curr_stats;
- u16 nevents;
-
- switch (dim->state) {
- case NET_DIM_MEASURE_IN_PROGRESS:
- nevents = BIT_GAP(BITS_PER_TYPE(u16), event_ctr,
- dim->start_sample.event_ctr);
- if (nevents < NET_DIM_NEVENTS)
- break;
- net_dim_sample(event_ctr, packets, bytes, &end_sample);
- net_dim_calc_stats(&dim->start_sample, &end_sample,
- &curr_stats);
- if (net_dim_decision(&curr_stats, dim)) {
- dim->state = NET_DIM_APPLY_NEW_PROFILE;
- schedule_work(&dim->work);
- break;
- }
- /* fall through */
- case NET_DIM_START_MEASURE:
- net_dim_sample(event_ctr, packets, bytes, &dim->start_sample);
- dim->state = NET_DIM_MEASURE_IN_PROGRESS;
- break;
- case NET_DIM_APPLY_NEW_PROFILE:
- break;
- }
-}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/net_dim.h b/drivers/net/ethernet/mellanox/mlx5/core/net_dim.h
deleted file mode 100644
index f0a70cd..0000000
--- a/drivers/net/ethernet/mellanox/mlx5/core/net_dim.h
+++ /dev/null
@@ -1,108 +0,0 @@
-/*
- * Copyright (c) 2013-2015, Mellanox Technologies, Ltd. All rights reserved.
- * Copyright (c) 2017-2018, Broadcom Limited. All rights reserved.
- *
- * This software is available to you under a choice of one of two
- * licenses. You may choose to be licensed under the terms of the GNU
- * General Public License (GPL) Version 2, available from the file
- * COPYING in the main directory of this source tree, or the
- * OpenIB.org BSD license below:
- *
- * Redistribution and use in source and binary forms, with or
- * without modification, are permitted provided that the following
- * conditions are met:
- *
- * - Redistributions of source code must retain the above
- * copyright notice, this list of conditions and the following
- * disclaimer.
- *
- * - Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following
- * disclaimer in the documentation and/or other materials
- * provided with the distribution.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
- * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
- * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
- * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- */
-
-#ifndef NET_DIM_H
-#define NET_DIM_H
-
-struct net_dim_cq_moder {
- u16 usec;
- u16 pkts;
- u8 cq_period_mode;
-};
-
-struct net_dim_sample {
- ktime_t time;
- u32 pkt_ctr;
- u32 byte_ctr;
- u16 event_ctr;
-};
-
-struct net_dim_stats {
- int ppms; /* packets per msec */
- int bpms; /* bytes per msec */
- int epms; /* events per msec */
-};
-
-struct net_dim { /* Adaptive Moderation */
- u8 state;
- struct net_dim_stats prev_stats;
- struct net_dim_sample start_sample;
- struct work_struct work;
- u8 profile_ix;
- u8 mode;
- u8 tune_state;
- u8 steps_right;
- u8 steps_left;
- u8 tired;
-};
-
-enum {
- NET_DIM_CQ_PERIOD_MODE_START_FROM_EQE = 0x0,
- NET_DIM_CQ_PERIOD_MODE_START_FROM_CQE = 0x1,
- NET_DIM_CQ_PERIOD_NUM_MODES
-};
-
-/* Adaptive moderation logic */
-enum {
- NET_DIM_START_MEASURE,
- NET_DIM_MEASURE_IN_PROGRESS,
- NET_DIM_APPLY_NEW_PROFILE,
-};
-
-enum {
- NET_DIM_PARKING_ON_TOP,
- NET_DIM_PARKING_TIRED,
- NET_DIM_GOING_RIGHT,
- NET_DIM_GOING_LEFT,
-};
-
-enum {
- NET_DIM_STATS_WORSE,
- NET_DIM_STATS_SAME,
- NET_DIM_STATS_BETTER,
-};
-
-enum {
- NET_DIM_STEPPED,
- NET_DIM_TOO_TIRED,
- NET_DIM_ON_EDGE,
-};
-
-void net_dim(struct net_dim *dim,
- u16 event_ctr,
- u64 packets,
- u64 bytes);
-struct net_dim_cq_moder net_dim_get_def_profile(u8 rx_cq_period_mode);
-struct net_dim_cq_moder net_dim_get_profile(u8 cq_period_mode, int ix);
-
-#endif /* NET_DIM_H */
diff --git a/include/linux/net_dim.h b/include/linux/net_dim.h
new file mode 100644
index 0000000..741510f
--- /dev/null
+++ b/include/linux/net_dim.h
@@ -0,0 +1,377 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2017-2018, Broadcom Limited. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef NET_DIM_H
+#define NET_DIM_H
+
+#include <linux/module.h>
+
+struct net_dim_cq_moder {
+ u16 usec;
+ u16 pkts;
+ u8 cq_period_mode;
+};
+
+struct net_dim_sample {
+ ktime_t time;
+ u32 pkt_ctr;
+ u32 byte_ctr;
+ u16 event_ctr;
+};
+
+struct net_dim_stats {
+ int ppms; /* packets per msec */
+ int bpms; /* bytes per msec */
+ int epms; /* events per msec */
+};
+
+struct net_dim { /* Adaptive Moderation */
+ u8 state;
+ struct net_dim_stats prev_stats;
+ struct net_dim_sample start_sample;
+ struct work_struct work;
+ u8 profile_ix;
+ u8 mode;
+ u8 tune_state;
+ u8 steps_right;
+ u8 steps_left;
+ u8 tired;
+};
+
+enum {
+ NET_DIM_CQ_PERIOD_MODE_START_FROM_EQE = 0x0,
+ NET_DIM_CQ_PERIOD_MODE_START_FROM_CQE = 0x1,
+ NET_DIM_CQ_PERIOD_NUM_MODES
+};
+
+/* Adaptive moderation logic */
+enum {
+ NET_DIM_START_MEASURE,
+ NET_DIM_MEASURE_IN_PROGRESS,
+ NET_DIM_APPLY_NEW_PROFILE,
+};
+
+enum {
+ NET_DIM_PARKING_ON_TOP,
+ NET_DIM_PARKING_TIRED,
+ NET_DIM_GOING_RIGHT,
+ NET_DIM_GOING_LEFT,
+};
+
+enum {
+ NET_DIM_STATS_WORSE,
+ NET_DIM_STATS_SAME,
+ NET_DIM_STATS_BETTER,
+};
+
+enum {
+ NET_DIM_STEPPED,
+ NET_DIM_TOO_TIRED,
+ NET_DIM_ON_EDGE,
+};
+
+#define NET_DIM_PARAMS_NUM_PROFILES 5
+/* Adaptive moderation profiles */
+#define NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE 256
+#define NET_DIM_DEF_PROFILE_CQE 1
+#define NET_DIM_DEF_PROFILE_EQE 1
+
+/* All profiles sizes must be NET_PARAMS_DIM_NUM_PROFILES */
+#define NET_DIM_EQE_PROFILES { \
+ {1, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+ {8, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+ {64, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+ {128, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+ {256, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+}
+
+#define NET_DIM_CQE_PROFILES { \
+ {2, 256}, \
+ {8, 128}, \
+ {16, 64}, \
+ {32, 64}, \
+ {64, 64} \
+}
+
+static const struct net_dim_cq_moder
+profile[NET_DIM_CQ_PERIOD_NUM_MODES][NET_DIM_PARAMS_NUM_PROFILES] = {
+ NET_DIM_EQE_PROFILES,
+ NET_DIM_CQE_PROFILES,
+};
+
+static inline struct net_dim_cq_moder net_dim_get_profile(u8 cq_period_mode,
+ int ix)
+{
+ struct net_dim_cq_moder cq_moder;
+
+ cq_moder = profile[cq_period_mode][ix];
+ cq_moder.cq_period_mode = cq_period_mode;
+ return cq_moder;
+}
+
+static inline struct net_dim_cq_moder net_dim_get_def_profile(u8 rx_cq_period_mode)
+{
+ int default_profile_ix;
+
+ if (rx_cq_period_mode == NET_DIM_CQ_PERIOD_MODE_START_FROM_CQE)
+ default_profile_ix = NET_DIM_DEF_PROFILE_CQE;
+ else /* NET_DIM_CQ_PERIOD_MODE_START_FROM_EQE */
+ default_profile_ix = NET_DIM_DEF_PROFILE_EQE;
+
+ return net_dim_get_profile(rx_cq_period_mode, default_profile_ix);
+}
+
+static inline bool net_dim_on_top(struct net_dim *dim)
+{
+ switch (dim->tune_state) {
+ case NET_DIM_PARKING_ON_TOP:
+ case NET_DIM_PARKING_TIRED:
+ return true;
+ case NET_DIM_GOING_RIGHT:
+ return (dim->steps_left > 1) && (dim->steps_right == 1);
+ default: /* NET_DIM_GOING_LEFT */
+ return (dim->steps_right > 1) && (dim->steps_left == 1);
+ }
+}
+
+static inline void net_dim_turn(struct net_dim *dim)
+{
+ switch (dim->tune_state) {
+ case NET_DIM_PARKING_ON_TOP:
+ case NET_DIM_PARKING_TIRED:
+ break;
+ case NET_DIM_GOING_RIGHT:
+ dim->tune_state = NET_DIM_GOING_LEFT;
+ dim->steps_left = 0;
+ break;
+ case NET_DIM_GOING_LEFT:
+ dim->tune_state = NET_DIM_GOING_RIGHT;
+ dim->steps_right = 0;
+ break;
+ }
+}
+
+static inline int net_dim_step(struct net_dim *dim)
+{
+ if (dim->tired == (NET_DIM_PARAMS_NUM_PROFILES * 2))
+ return NET_DIM_TOO_TIRED;
+
+ switch (dim->tune_state) {
+ case NET_DIM_PARKING_ON_TOP:
+ case NET_DIM_PARKING_TIRED:
+ break;
+ case NET_DIM_GOING_RIGHT:
+ if (dim->profile_ix == (NET_DIM_PARAMS_NUM_PROFILES - 1))
+ return NET_DIM_ON_EDGE;
+ dim->profile_ix++;
+ dim->steps_right++;
+ break;
+ case NET_DIM_GOING_LEFT:
+ if (dim->profile_ix == 0)
+ return NET_DIM_ON_EDGE;
+ dim->profile_ix--;
+ dim->steps_left++;
+ break;
+ }
+
+ dim->tired++;
+ return NET_DIM_STEPPED;
+}
+
+static inline void net_dim_park_on_top(struct net_dim *dim)
+{
+ dim->steps_right = 0;
+ dim->steps_left = 0;
+ dim->tired = 0;
+ dim->tune_state = NET_DIM_PARKING_ON_TOP;
+}
+
+static inline void net_dim_park_tired(struct net_dim *dim)
+{
+ dim->steps_right = 0;
+ dim->steps_left = 0;
+ dim->tune_state = NET_DIM_PARKING_TIRED;
+}
+
+static inline void net_dim_exit_parking(struct net_dim *dim)
+{
+ dim->tune_state = dim->profile_ix ? NET_DIM_GOING_LEFT :
+ NET_DIM_GOING_RIGHT;
+ net_dim_step(dim);
+}
+
+#define IS_SIGNIFICANT_DIFF(val, ref) \
+ (((100 * abs((val) - (ref))) / (ref)) > 10) /* more than 10% difference */
+
+static inline int net_dim_stats_compare(struct net_dim_stats *curr,
+ struct net_dim_stats *prev)
+{
+ if (!prev->bpms)
+ return curr->bpms ? NET_DIM_STATS_BETTER :
+ NET_DIM_STATS_SAME;
+
+ if (IS_SIGNIFICANT_DIFF(curr->bpms, prev->bpms))
+ return (curr->bpms > prev->bpms) ? NET_DIM_STATS_BETTER :
+ NET_DIM_STATS_WORSE;
+
+ if (IS_SIGNIFICANT_DIFF(curr->ppms, prev->ppms))
+ return (curr->ppms > prev->ppms) ? NET_DIM_STATS_BETTER :
+ NET_DIM_STATS_WORSE;
+
+ if (IS_SIGNIFICANT_DIFF(curr->epms, prev->epms))
+ return (curr->epms < prev->epms) ? NET_DIM_STATS_BETTER :
+ NET_DIM_STATS_WORSE;
+
+ return NET_DIM_STATS_SAME;
+}
+
+static inline bool net_dim_decision(struct net_dim_stats *curr_stats,
+ struct net_dim *dim)
+{
+ int prev_state = dim->tune_state;
+ int prev_ix = dim->profile_ix;
+ int stats_res;
+ int step_res;
+
+ switch (dim->tune_state) {
+ case NET_DIM_PARKING_ON_TOP:
+ stats_res = net_dim_stats_compare(curr_stats, &dim->prev_stats);
+ if (stats_res != NET_DIM_STATS_SAME)
+ net_dim_exit_parking(dim);
+ break;
+
+ case NET_DIM_PARKING_TIRED:
+ dim->tired--;
+ if (!dim->tired)
+ net_dim_exit_parking(dim);
+ break;
+
+ case NET_DIM_GOING_RIGHT:
+ case NET_DIM_GOING_LEFT:
+ stats_res = net_dim_stats_compare(curr_stats, &dim->prev_stats);
+ if (stats_res != NET_DIM_STATS_BETTER)
+ net_dim_turn(dim);
+
+ if (net_dim_on_top(dim)) {
+ net_dim_park_on_top(dim);
+ break;
+ }
+
+ step_res = net_dim_step(dim);
+ switch (step_res) {
+ case NET_DIM_ON_EDGE:
+ net_dim_park_on_top(dim);
+ break;
+ case NET_DIM_TOO_TIRED:
+ net_dim_park_tired(dim);
+ break;
+ }
+
+ break;
+ }
+
+ if ((prev_state != NET_DIM_PARKING_ON_TOP) ||
+ (dim->tune_state != NET_DIM_PARKING_ON_TOP))
+ dim->prev_stats = *curr_stats;
+
+ return dim->profile_ix != prev_ix;
+}
+
+static inline void net_dim_sample(u16 event_ctr,
+ u64 packets,
+ u64 bytes,
+ struct net_dim_sample *s)
+{
+ s->time = ktime_get();
+ s->pkt_ctr = packets;
+ s->byte_ctr = bytes;
+ s->event_ctr = event_ctr;
+}
+
+#define NET_DIM_NEVENTS 64
+#define BITS_PER_TYPE(type) (sizeof(type) * BITS_PER_BYTE)
+#define BIT_GAP(bits, end, start) ((((end) - (start)) + BIT_ULL(bits)) & (BIT_ULL(bits) - 1))
+
+static inline void net_dim_calc_stats(struct net_dim_sample *start,
+ struct net_dim_sample *end,
+ struct net_dim_stats *curr_stats)
+{
+ /* u32 holds up to 71 minutes, should be enough */
+ u32 delta_us = ktime_us_delta(end->time, start->time);
+ u32 npkts = BIT_GAP(BITS_PER_TYPE(u32), end->pkt_ctr, start->pkt_ctr);
+ u32 nbytes = BIT_GAP(BITS_PER_TYPE(u32), end->byte_ctr,
+ start->byte_ctr);
+
+ if (!delta_us)
+ return;
+
+ curr_stats->ppms = DIV_ROUND_UP(npkts * USEC_PER_MSEC, delta_us);
+ curr_stats->bpms = DIV_ROUND_UP(nbytes * USEC_PER_MSEC, delta_us);
+ curr_stats->epms = DIV_ROUND_UP(NET_DIM_NEVENTS * USEC_PER_MSEC,
+ delta_us);
+}
+
+static inline void net_dim(struct net_dim *dim,
+ u16 event_ctr,
+ u64 packets,
+ u64 bytes)
+{
+ struct net_dim_sample end_sample;
+ struct net_dim_stats curr_stats;
+ u16 nevents;
+
+ switch (dim->state) {
+ case NET_DIM_MEASURE_IN_PROGRESS:
+ nevents = BIT_GAP(BITS_PER_TYPE(u16), event_ctr,
+ dim->start_sample.event_ctr);
+ if (nevents < NET_DIM_NEVENTS)
+ break;
+ net_dim_sample(event_ctr, packets, bytes, &end_sample);
+ net_dim_calc_stats(&dim->start_sample, &end_sample,
+ &curr_stats);
+ if (net_dim_decision(&curr_stats, dim)) {
+ dim->state = NET_DIM_APPLY_NEW_PROFILE;
+ schedule_work(&dim->work);
+ break;
+ }
+ /* fall through */
+ case NET_DIM_START_MEASURE:
+ net_dim_sample(event_ctr, packets, bytes, &dim->start_sample);
+ dim->state = NET_DIM_MEASURE_IN_PROGRESS;
+ break;
+ case NET_DIM_APPLY_NEW_PROFILE:
+ break;
+ }
+}
+
+#endif /* NET_DIM_H */
--
2.7.4
^ permalink raw reply related
* [PATCH net-next v3 08/10] net/dim: use struct net_dim_sample as arg to net_dim
From: Andy Gospodarek @ 2018-01-09 6:13 UTC (permalink / raw)
To: netdev; +Cc: mchan, talgi, ogerlitz, Andy Gospodarek
In-Reply-To: <1515478402-6048-1-git-send-email-andy@greyhouse.net>
From: Andy Gospodarek <gospo@broadcom.com>
Simplify the arguments net_dim() by formatting them into a struct
net_dim_sample before calling the function.
Signed-off-by: Andy Gospodarek <gospo@broadcom.com>
Suggested-by: Tal Gilboa <talgi@mellanox.com>
Acked-by: Tal Gilboa <talgi@mellanox.com>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c | 13 ++++++++-----
include/linux/net_dim.h | 10 +++-------
2 files changed, 11 insertions(+), 12 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
index a1c94fd..f292bb3 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
@@ -78,11 +78,14 @@ int mlx5e_napi_poll(struct napi_struct *napi, int budget)
for (i = 0; i < c->num_tc; i++)
mlx5e_cq_arm(&c->sq[i].cq);
- if (MLX5E_TEST_BIT(c->rq.state, MLX5E_RQ_STATE_AM))
- net_dim(&c->rq.dim,
- c->rq.cq.event_ctr,
- c->rq.stats.packets,
- c->rq.stats.bytes);
+ if (MLX5E_TEST_BIT(c->rq.state, MLX5E_RQ_STATE_AM)) {
+ struct net_dim_sample dim_sample;
+ net_dim_sample(c->rq.cq.event_ctr,
+ c->rq.stats.packets,
+ c->rq.stats.bytes,
+ &dim_sample);
+ net_dim(&c->rq.dim, dim_sample);
+ }
mlx5e_cq_arm(&c->rq.cq);
mlx5e_cq_arm(&c->icosq.cq);
diff --git a/include/linux/net_dim.h b/include/linux/net_dim.h
index 741510f..1c7e450 100644
--- a/include/linux/net_dim.h
+++ b/include/linux/net_dim.h
@@ -342,21 +342,18 @@ static inline void net_dim_calc_stats(struct net_dim_sample *start,
}
static inline void net_dim(struct net_dim *dim,
- u16 event_ctr,
- u64 packets,
- u64 bytes)
+ struct net_dim_sample end_sample)
{
- struct net_dim_sample end_sample;
struct net_dim_stats curr_stats;
u16 nevents;
switch (dim->state) {
case NET_DIM_MEASURE_IN_PROGRESS:
- nevents = BIT_GAP(BITS_PER_TYPE(u16), event_ctr,
+ nevents = BIT_GAP(BITS_PER_TYPE(u16),
+ end_sample.event_ctr,
dim->start_sample.event_ctr);
if (nevents < NET_DIM_NEVENTS)
break;
- net_dim_sample(event_ctr, packets, bytes, &end_sample);
net_dim_calc_stats(&dim->start_sample, &end_sample,
&curr_stats);
if (net_dim_decision(&curr_stats, dim)) {
@@ -366,7 +363,6 @@ static inline void net_dim(struct net_dim *dim,
}
/* fall through */
case NET_DIM_START_MEASURE:
- net_dim_sample(event_ctr, packets, bytes, &dim->start_sample);
dim->state = NET_DIM_MEASURE_IN_PROGRESS;
break;
case NET_DIM_APPLY_NEW_PROFILE:
--
2.7.4
^ permalink raw reply related
* [PATCH net-next v3 06/10] net/mlx5e: Change Mellanox references in DIM code
From: Andy Gospodarek @ 2018-01-09 6:13 UTC (permalink / raw)
To: netdev; +Cc: mchan, talgi, ogerlitz, Andy Gospodarek
In-Reply-To: <1515478402-6048-1-git-send-email-andy@greyhouse.net>
From: Andy Gospodarek <gospo@broadcom.com>
Change all appropriate mlx5_am* and MLX5_AM* references to net_dim and
NET_DIM, respectively, in code that handles dynamic interrupt
moderation. Also change all references from 'am' to 'dim' when used as
local variables and add generic profile references.
Signed-off-by: Andy Gospodarek <gospo@broadcom.com>
Acked-by: Tal Gilboa <talgi@mellanox.com>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 9 +-
drivers/net/ethernet/mellanox/mlx5/core/en_dim.c | 14 +-
.../net/ethernet/mellanox/mlx5/core/en_ethtool.c | 6 +-
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 40 ++-
drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c | 8 +-
drivers/net/ethernet/mellanox/mlx5/core/net_dim.c | 286 ++++++++++-----------
drivers/net/ethernet/mellanox/mlx5/core/net_dim.h | 63 ++---
7 files changed, 225 insertions(+), 201 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index 4ee06e7..4d1d298 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -238,8 +238,8 @@ struct mlx5e_params {
u16 num_channels;
u8 num_tc;
bool rx_cqe_compress_def;
- struct mlx5e_cq_moder rx_cq_moderation;
- struct mlx5e_cq_moder tx_cq_moderation;
+ struct net_dim_cq_moder rx_cq_moderation;
+ struct net_dim_cq_moder tx_cq_moderation;
bool lro_en;
u32 lro_wqe_sz;
u16 tx_max_inline;
@@ -249,7 +249,7 @@ struct mlx5e_params {
u32 indirection_rqt[MLX5E_INDIR_RQT_SIZE];
bool vlan_strip_disable;
bool scatter_fcs_en;
- bool rx_am_enabled;
+ bool rx_dim_enabled;
u32 lro_timeout;
u32 pflags;
struct bpf_prog *xdp_prog;
@@ -528,7 +528,7 @@ struct mlx5e_rq {
unsigned long state;
int ix;
- struct mlx5e_rx_am am; /* Adaptive Moderation */
+ struct net_dim dim; /* Dynamic Interrupt Moderation */
/* XDP */
struct bpf_prog *xdp_prog;
@@ -1079,4 +1079,5 @@ void mlx5e_build_nic_params(struct mlx5_core_dev *mdev,
struct mlx5e_params *params,
u16 max_channels);
u8 mlx5e_params_calculate_tx_min_inline(struct mlx5_core_dev *mdev);
+void mlx5e_rx_dim_work(struct work_struct *work);
#endif /* __MLX5_EN_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.c b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.c
index b9b434b..f620325 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.c
@@ -32,17 +32,17 @@
#include "en.h"
-void mlx5e_rx_am_work(struct work_struct *work)
+void mlx5e_rx_dim_work(struct work_struct *work)
{
- struct mlx5e_rx_am *am = container_of(work, struct mlx5e_rx_am,
- work);
- struct mlx5e_rq *rq = container_of(am, struct mlx5e_rq, am);
- struct mlx5e_cq_moder cur_profile = mlx5e_am_get_profile(am->mode,
- am->profile_ix);
+ struct net_dim *dim = container_of(work, struct net_dim,
+ work);
+ struct mlx5e_rq *rq = container_of(dim, struct mlx5e_rq, dim);
+ struct net_dim_cq_moder cur_profile = net_dim_get_profile(dim->mode,
+ dim->profile_ix);
mlx5_core_modify_cq_moderation(rq->mdev, &rq->cq.mcq,
cur_profile.usec, cur_profile.pkts);
- am->state = MLX5E_AM_START_MEASURE;
+ dim->state = NET_DIM_START_MEASURE;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
index 8f05efa..51ae6df 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
@@ -480,7 +480,7 @@ int mlx5e_ethtool_get_coalesce(struct mlx5e_priv *priv,
coal->rx_max_coalesced_frames = priv->channels.params.rx_cq_moderation.pkts;
coal->tx_coalesce_usecs = priv->channels.params.tx_cq_moderation.usec;
coal->tx_max_coalesced_frames = priv->channels.params.tx_cq_moderation.pkts;
- coal->use_adaptive_rx_coalesce = priv->channels.params.rx_am_enabled;
+ coal->use_adaptive_rx_coalesce = priv->channels.params.rx_dim_enabled;
return 0;
}
@@ -534,7 +534,7 @@ int mlx5e_ethtool_set_coalesce(struct mlx5e_priv *priv,
new_channels.params.tx_cq_moderation.pkts = coal->tx_max_coalesced_frames;
new_channels.params.rx_cq_moderation.usec = coal->rx_coalesce_usecs;
new_channels.params.rx_cq_moderation.pkts = coal->rx_max_coalesced_frames;
- new_channels.params.rx_am_enabled = !!coal->use_adaptive_rx_coalesce;
+ new_channels.params.rx_dim_enabled = !!coal->use_adaptive_rx_coalesce;
if (!test_bit(MLX5E_STATE_OPENED, &priv->state)) {
priv->channels.params = new_channels.params;
@@ -542,7 +542,7 @@ int mlx5e_ethtool_set_coalesce(struct mlx5e_priv *priv,
}
/* we are opened */
- reset = !!coal->use_adaptive_rx_coalesce != priv->channels.params.rx_am_enabled;
+ reset = !!coal->use_adaptive_rx_coalesce != priv->channels.params.rx_dim_enabled;
if (!reset) {
mlx5e_set_priv_channels_coalesce(priv, coal);
priv->channels.params = new_channels.params;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 539bd1d..a5f82bb 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -677,8 +677,17 @@ static int mlx5e_alloc_rq(struct mlx5e_channel *c,
wqe->data.lkey = rq->mkey_be;
}
- INIT_WORK(&rq->am.work, mlx5e_rx_am_work);
- rq->am.mode = params->rx_cq_moderation.cq_period_mode;
+ INIT_WORK(&rq->dim.work, mlx5e_rx_dim_work);
+
+ switch (params->rx_cq_moderation.cq_period_mode) {
+ case MLX5_CQ_PERIOD_MODE_START_FROM_CQE:
+ rq->dim.mode = NET_DIM_CQ_PERIOD_MODE_START_FROM_CQE;
+ break;
+ case MLX5_CQ_PERIOD_MODE_START_FROM_EQE:
+ default:
+ rq->dim.mode = NET_DIM_CQ_PERIOD_MODE_START_FROM_EQE;
+ }
+
rq->page_cache.head = 0;
rq->page_cache.tail = 0;
@@ -925,7 +934,7 @@ static int mlx5e_open_rq(struct mlx5e_channel *c,
if (err)
goto err_destroy_rq;
- if (params->rx_am_enabled)
+ if (params->rx_dim_enabled)
c->rq.state |= BIT(MLX5E_RQ_STATE_AM);
return 0;
@@ -958,7 +967,7 @@ static void mlx5e_deactivate_rq(struct mlx5e_rq *rq)
static void mlx5e_close_rq(struct mlx5e_rq *rq)
{
- cancel_work_sync(&rq->am.work);
+ cancel_work_sync(&rq->dim.work);
mlx5e_destroy_rq(rq);
mlx5e_free_rx_descs(rq);
mlx5e_free_rq(rq);
@@ -1571,7 +1580,7 @@ static void mlx5e_destroy_cq(struct mlx5e_cq *cq)
}
static int mlx5e_open_cq(struct mlx5e_channel *c,
- struct mlx5e_cq_moder moder,
+ struct net_dim_cq_moder moder,
struct mlx5e_cq_param *param,
struct mlx5e_cq *cq)
{
@@ -1753,7 +1762,7 @@ static int mlx5e_open_channel(struct mlx5e_priv *priv, int ix,
struct mlx5e_channel_param *cparam,
struct mlx5e_channel **cp)
{
- struct mlx5e_cq_moder icocq_moder = {0, 0};
+ struct net_dim_cq_moder icocq_moder = {0, 0};
struct net_device *netdev = priv->netdev;
int cpu = mlx5e_get_cpu(priv, ix);
struct mlx5e_channel *c;
@@ -2005,7 +2014,7 @@ static void mlx5e_build_ico_cq_param(struct mlx5e_priv *priv,
mlx5e_build_common_cq_param(priv, param);
- param->cq_period_mode = MLX5_CQ_PERIOD_MODE_START_FROM_EQE;
+ param->cq_period_mode = NET_DIM_CQ_PERIOD_MODE_START_FROM_EQE;
}
static void mlx5e_build_icosq_param(struct mlx5e_priv *priv,
@@ -4047,9 +4056,18 @@ void mlx5e_set_rx_cq_mode_params(struct mlx5e_params *params, u8 cq_period_mode)
params->rx_cq_moderation.usec =
MLX5E_PARAMS_DEFAULT_RX_CQ_MODERATION_USEC_FROM_CQE;
- if (params->rx_am_enabled)
- params->rx_cq_moderation =
- mlx5e_am_get_def_profile(cq_period_mode);
+ if (params->rx_dim_enabled) {
+ switch (cq_period_mode) {
+ case MLX5_CQ_PERIOD_MODE_START_FROM_CQE:
+ params->rx_cq_moderation =
+ net_dim_get_def_profile(NET_DIM_CQ_PERIOD_MODE_START_FROM_CQE);
+ break;
+ case MLX5_CQ_PERIOD_MODE_START_FROM_EQE:
+ default:
+ params->rx_cq_moderation =
+ net_dim_get_def_profile(NET_DIM_CQ_PERIOD_MODE_START_FROM_EQE);
+ }
+ }
MLX5E_SET_PFLAG(params, MLX5E_PFLAG_RX_CQE_BASED_MODER,
params->rx_cq_moderation.cq_period_mode ==
@@ -4111,7 +4129,7 @@ void mlx5e_build_nic_params(struct mlx5_core_dev *mdev,
cq_period_mode = MLX5_CAP_GEN(mdev, cq_period_start_from_cqe) ?
MLX5_CQ_PERIOD_MODE_START_FROM_CQE :
MLX5_CQ_PERIOD_MODE_START_FROM_EQE;
- params->rx_am_enabled = MLX5_CAP_GEN(mdev, cq_moderation);
+ params->rx_dim_enabled = MLX5_CAP_GEN(mdev, cq_moderation);
mlx5e_set_rx_cq_mode_params(params, cq_period_mode);
mlx5e_set_tx_cq_mode_params(params, cq_period_mode);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
index 1849169..a1c94fd 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
@@ -79,10 +79,10 @@ int mlx5e_napi_poll(struct napi_struct *napi, int budget)
mlx5e_cq_arm(&c->sq[i].cq);
if (MLX5E_TEST_BIT(c->rq.state, MLX5E_RQ_STATE_AM))
- mlx5e_rx_am(&c->rq.am,
- c->rq.cq.event_ctr,
- c->rq.stats.packets,
- c->rq.stats.bytes);
+ net_dim(&c->rq.dim,
+ c->rq.cq.event_ctr,
+ c->rq.stats.packets,
+ c->rq.stats.bytes);
mlx5e_cq_arm(&c->rq.cq);
mlx5e_cq_arm(&c->icosq.cq);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/net_dim.c b/drivers/net/ethernet/mellanox/mlx5/core/net_dim.c
index 9fafe1d..decb370 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/net_dim.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/net_dim.c
@@ -33,22 +33,22 @@
#include "en.h"
-#define MLX5E_PARAMS_AM_NUM_PROFILES 5
+#define NET_DIM_PARAMS_NUM_PROFILES 5
/* Adaptive moderation profiles */
-#define MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE 256
-#define MLX5E_RX_AM_DEF_PROFILE_CQE 1
-#define MLX5E_RX_AM_DEF_PROFILE_EQE 1
-
-/* All profiles sizes must be MLX5E_PARAMS_AM_NUM_PROFILES */
-#define MLX5_AM_EQE_PROFILES { \
- {1, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
- {8, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
- {64, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
- {128, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
- {256, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+#define NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE 256
+#define NET_DIM_DEF_PROFILE_CQE 1
+#define NET_DIM_DEF_PROFILE_EQE 1
+
+/* All profiles sizes must be NET_PARAMS_DIM_NUM_PROFILES */
+#define NET_DIM_EQE_PROFILES { \
+ {1, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+ {8, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+ {64, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+ {128, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+ {256, NET_DIM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
}
-#define MLX5_AM_CQE_PROFILES { \
+#define NET_DIM_CQE_PROFILES { \
{2, 256}, \
{8, 128}, \
{16, 64}, \
@@ -56,193 +56,193 @@
{64, 64} \
}
-static const struct mlx5e_cq_moder
-profile[MLX5_CQ_PERIOD_NUM_MODES][MLX5E_PARAMS_AM_NUM_PROFILES] = {
- MLX5_AM_EQE_PROFILES,
- MLX5_AM_CQE_PROFILES,
+static const struct net_dim_cq_moder
+profile[NET_DIM_CQ_PERIOD_NUM_MODES][NET_DIM_PARAMS_NUM_PROFILES] = {
+ NET_DIM_EQE_PROFILES,
+ NET_DIM_CQE_PROFILES,
};
-struct mlx5e_cq_moder mlx5e_am_get_profile(u8 cq_period_mode, int ix)
+struct net_dim_cq_moder net_dim_get_profile(u8 cq_period_mode, int ix)
{
- struct mlx5e_cq_moder cq_moder;
+ struct net_dim_cq_moder cq_moder;
cq_moder = profile[cq_period_mode][ix];
cq_moder.cq_period_mode = cq_period_mode;
return cq_moder;
}
-struct mlx5e_cq_moder mlx5e_am_get_def_profile(u8 rx_cq_period_mode)
+struct net_dim_cq_moder net_dim_get_def_profile(u8 rx_cq_period_mode)
{
int default_profile_ix;
- if (rx_cq_period_mode == MLX5_CQ_PERIOD_MODE_START_FROM_CQE)
- default_profile_ix = MLX5E_RX_AM_DEF_PROFILE_CQE;
- else /* MLX5_CQ_PERIOD_MODE_START_FROM_EQE */
- default_profile_ix = MLX5E_RX_AM_DEF_PROFILE_EQE;
+ if (rx_cq_period_mode == NET_DIM_CQ_PERIOD_MODE_START_FROM_CQE)
+ default_profile_ix = NET_DIM_DEF_PROFILE_CQE;
+ else /* NET_DIM_CQ_PERIOD_MODE_START_FROM_EQE */
+ default_profile_ix = NET_DIM_DEF_PROFILE_EQE;
- return mlx5e_am_get_profile(rx_cq_period_mode, default_profile_ix);
+ return net_dim_get_profile(rx_cq_period_mode, default_profile_ix);
}
-static bool mlx5e_am_on_top(struct mlx5e_rx_am *am)
+static bool net_dim_on_top(struct net_dim *dim)
{
- switch (am->tune_state) {
- case MLX5E_AM_PARKING_ON_TOP:
- case MLX5E_AM_PARKING_TIRED:
+ switch (dim->tune_state) {
+ case NET_DIM_PARKING_ON_TOP:
+ case NET_DIM_PARKING_TIRED:
return true;
- case MLX5E_AM_GOING_RIGHT:
- return (am->steps_left > 1) && (am->steps_right == 1);
- default: /* MLX5E_AM_GOING_LEFT */
- return (am->steps_right > 1) && (am->steps_left == 1);
+ case NET_DIM_GOING_RIGHT:
+ return (dim->steps_left > 1) && (dim->steps_right == 1);
+ default: /* NET_DIM_GOING_LEFT */
+ return (dim->steps_right > 1) && (dim->steps_left == 1);
}
}
-static void mlx5e_am_turn(struct mlx5e_rx_am *am)
+static void net_dim_turn(struct net_dim *dim)
{
- switch (am->tune_state) {
- case MLX5E_AM_PARKING_ON_TOP:
- case MLX5E_AM_PARKING_TIRED:
+ switch (dim->tune_state) {
+ case NET_DIM_PARKING_ON_TOP:
+ case NET_DIM_PARKING_TIRED:
break;
- case MLX5E_AM_GOING_RIGHT:
- am->tune_state = MLX5E_AM_GOING_LEFT;
- am->steps_left = 0;
+ case NET_DIM_GOING_RIGHT:
+ dim->tune_state = NET_DIM_GOING_LEFT;
+ dim->steps_left = 0;
break;
- case MLX5E_AM_GOING_LEFT:
- am->tune_state = MLX5E_AM_GOING_RIGHT;
- am->steps_right = 0;
+ case NET_DIM_GOING_LEFT:
+ dim->tune_state = NET_DIM_GOING_RIGHT;
+ dim->steps_right = 0;
break;
}
}
-static int mlx5e_am_step(struct mlx5e_rx_am *am)
+static int net_dim_step(struct net_dim *dim)
{
- if (am->tired == (MLX5E_PARAMS_AM_NUM_PROFILES * 2))
- return MLX5E_AM_TOO_TIRED;
+ if (dim->tired == (NET_DIM_PARAMS_NUM_PROFILES * 2))
+ return NET_DIM_TOO_TIRED;
- switch (am->tune_state) {
- case MLX5E_AM_PARKING_ON_TOP:
- case MLX5E_AM_PARKING_TIRED:
+ switch (dim->tune_state) {
+ case NET_DIM_PARKING_ON_TOP:
+ case NET_DIM_PARKING_TIRED:
break;
- case MLX5E_AM_GOING_RIGHT:
- if (am->profile_ix == (MLX5E_PARAMS_AM_NUM_PROFILES - 1))
- return MLX5E_AM_ON_EDGE;
- am->profile_ix++;
- am->steps_right++;
+ case NET_DIM_GOING_RIGHT:
+ if (dim->profile_ix == (NET_DIM_PARAMS_NUM_PROFILES - 1))
+ return NET_DIM_ON_EDGE;
+ dim->profile_ix++;
+ dim->steps_right++;
break;
- case MLX5E_AM_GOING_LEFT:
- if (am->profile_ix == 0)
- return MLX5E_AM_ON_EDGE;
- am->profile_ix--;
- am->steps_left++;
+ case NET_DIM_GOING_LEFT:
+ if (dim->profile_ix == 0)
+ return NET_DIM_ON_EDGE;
+ dim->profile_ix--;
+ dim->steps_left++;
break;
}
- am->tired++;
- return MLX5E_AM_STEPPED;
+ dim->tired++;
+ return NET_DIM_STEPPED;
}
-static void mlx5e_am_park_on_top(struct mlx5e_rx_am *am)
+static void net_dim_park_on_top(struct net_dim *dim)
{
- am->steps_right = 0;
- am->steps_left = 0;
- am->tired = 0;
- am->tune_state = MLX5E_AM_PARKING_ON_TOP;
+ dim->steps_right = 0;
+ dim->steps_left = 0;
+ dim->tired = 0;
+ dim->tune_state = NET_DIM_PARKING_ON_TOP;
}
-static void mlx5e_am_park_tired(struct mlx5e_rx_am *am)
+static void net_dim_park_tired(struct net_dim *dim)
{
- am->steps_right = 0;
- am->steps_left = 0;
- am->tune_state = MLX5E_AM_PARKING_TIRED;
+ dim->steps_right = 0;
+ dim->steps_left = 0;
+ dim->tune_state = NET_DIM_PARKING_TIRED;
}
-static void mlx5e_am_exit_parking(struct mlx5e_rx_am *am)
+static void net_dim_exit_parking(struct net_dim *dim)
{
- am->tune_state = am->profile_ix ? MLX5E_AM_GOING_LEFT :
- MLX5E_AM_GOING_RIGHT;
- mlx5e_am_step(am);
+ dim->tune_state = dim->profile_ix ? NET_DIM_GOING_LEFT :
+ NET_DIM_GOING_RIGHT;
+ net_dim_step(dim);
}
#define IS_SIGNIFICANT_DIFF(val, ref) \
(((100 * abs((val) - (ref))) / (ref)) > 10) /* more than 10% difference */
-static int mlx5e_am_stats_compare(struct mlx5e_rx_am_stats *curr,
- struct mlx5e_rx_am_stats *prev)
+static int net_dim_stats_compare(struct net_dim_stats *curr,
+ struct net_dim_stats *prev)
{
if (!prev->bpms)
- return curr->bpms ? MLX5E_AM_STATS_BETTER :
- MLX5E_AM_STATS_SAME;
+ return curr->bpms ? NET_DIM_STATS_BETTER :
+ NET_DIM_STATS_SAME;
if (IS_SIGNIFICANT_DIFF(curr->bpms, prev->bpms))
- return (curr->bpms > prev->bpms) ? MLX5E_AM_STATS_BETTER :
- MLX5E_AM_STATS_WORSE;
+ return (curr->bpms > prev->bpms) ? NET_DIM_STATS_BETTER :
+ NET_DIM_STATS_WORSE;
if (IS_SIGNIFICANT_DIFF(curr->ppms, prev->ppms))
- return (curr->ppms > prev->ppms) ? MLX5E_AM_STATS_BETTER :
- MLX5E_AM_STATS_WORSE;
+ return (curr->ppms > prev->ppms) ? NET_DIM_STATS_BETTER :
+ NET_DIM_STATS_WORSE;
if (IS_SIGNIFICANT_DIFF(curr->epms, prev->epms))
- return (curr->epms < prev->epms) ? MLX5E_AM_STATS_BETTER :
- MLX5E_AM_STATS_WORSE;
+ return (curr->epms < prev->epms) ? NET_DIM_STATS_BETTER :
+ NET_DIM_STATS_WORSE;
- return MLX5E_AM_STATS_SAME;
+ return NET_DIM_STATS_SAME;
}
-static bool mlx5e_am_decision(struct mlx5e_rx_am_stats *curr_stats,
- struct mlx5e_rx_am *am)
+static bool net_dim_decision(struct net_dim_stats *curr_stats,
+ struct net_dim *dim)
{
- int prev_state = am->tune_state;
- int prev_ix = am->profile_ix;
+ int prev_state = dim->tune_state;
+ int prev_ix = dim->profile_ix;
int stats_res;
int step_res;
- switch (am->tune_state) {
- case MLX5E_AM_PARKING_ON_TOP:
- stats_res = mlx5e_am_stats_compare(curr_stats, &am->prev_stats);
- if (stats_res != MLX5E_AM_STATS_SAME)
- mlx5e_am_exit_parking(am);
+ switch (dim->tune_state) {
+ case NET_DIM_PARKING_ON_TOP:
+ stats_res = net_dim_stats_compare(curr_stats, &dim->prev_stats);
+ if (stats_res != NET_DIM_STATS_SAME)
+ net_dim_exit_parking(dim);
break;
- case MLX5E_AM_PARKING_TIRED:
- am->tired--;
- if (!am->tired)
- mlx5e_am_exit_parking(am);
+ case NET_DIM_PARKING_TIRED:
+ dim->tired--;
+ if (!dim->tired)
+ net_dim_exit_parking(dim);
break;
- case MLX5E_AM_GOING_RIGHT:
- case MLX5E_AM_GOING_LEFT:
- stats_res = mlx5e_am_stats_compare(curr_stats, &am->prev_stats);
- if (stats_res != MLX5E_AM_STATS_BETTER)
- mlx5e_am_turn(am);
+ case NET_DIM_GOING_RIGHT:
+ case NET_DIM_GOING_LEFT:
+ stats_res = net_dim_stats_compare(curr_stats, &dim->prev_stats);
+ if (stats_res != NET_DIM_STATS_BETTER)
+ net_dim_turn(dim);
- if (mlx5e_am_on_top(am)) {
- mlx5e_am_park_on_top(am);
+ if (net_dim_on_top(dim)) {
+ net_dim_park_on_top(dim);
break;
}
- step_res = mlx5e_am_step(am);
+ step_res = net_dim_step(dim);
switch (step_res) {
- case MLX5E_AM_ON_EDGE:
- mlx5e_am_park_on_top(am);
+ case NET_DIM_ON_EDGE:
+ net_dim_park_on_top(dim);
break;
- case MLX5E_AM_TOO_TIRED:
- mlx5e_am_park_tired(am);
+ case NET_DIM_TOO_TIRED:
+ net_dim_park_tired(dim);
break;
}
break;
}
- if ((prev_state != MLX5E_AM_PARKING_ON_TOP) ||
- (am->tune_state != MLX5E_AM_PARKING_ON_TOP))
- am->prev_stats = *curr_stats;
+ if ((prev_state != NET_DIM_PARKING_ON_TOP) ||
+ (dim->tune_state != NET_DIM_PARKING_ON_TOP))
+ dim->prev_stats = *curr_stats;
- return am->profile_ix != prev_ix;
+ return dim->profile_ix != prev_ix;
}
-static void mlx5e_am_sample(u16 event_ctr,
- u64 packets,
- u64 bytes,
- struct mlx5e_rx_am_sample *s)
+static void net_dim_sample(u16 event_ctr,
+ u64 packets,
+ u64 bytes,
+ struct net_dim_sample *s)
{
s->time = ktime_get();
s->pkt_ctr = packets;
@@ -250,13 +250,13 @@ static void mlx5e_am_sample(u16 event_ctr,
s->event_ctr = event_ctr;
}
-#define MLX5E_AM_NEVENTS 64
+#define NET_DIM_NEVENTS 64
#define BITS_PER_TYPE(type) (sizeof(type) * BITS_PER_BYTE)
#define BIT_GAP(bits, end, start) ((((end) - (start)) + BIT_ULL(bits)) & (BIT_ULL(bits) - 1))
-static void mlx5e_am_calc_stats(struct mlx5e_rx_am_sample *start,
- struct mlx5e_rx_am_sample *end,
- struct mlx5e_rx_am_stats *curr_stats)
+static void net_dim_calc_stats(struct net_dim_sample *start,
+ struct net_dim_sample *end,
+ struct net_dim_stats *curr_stats)
{
/* u32 holds up to 71 minutes, should be enough */
u32 delta_us = ktime_us_delta(end->time, start->time);
@@ -269,39 +269,39 @@ static void mlx5e_am_calc_stats(struct mlx5e_rx_am_sample *start,
curr_stats->ppms = DIV_ROUND_UP(npkts * USEC_PER_MSEC, delta_us);
curr_stats->bpms = DIV_ROUND_UP(nbytes * USEC_PER_MSEC, delta_us);
- curr_stats->epms = DIV_ROUND_UP(MLX5E_AM_NEVENTS * USEC_PER_MSEC,
+ curr_stats->epms = DIV_ROUND_UP(NET_DIM_NEVENTS * USEC_PER_MSEC,
delta_us);
}
-void mlx5e_rx_am(struct mlx5e_rx_am *am,
- u16 event_ctr,
- u64 packets,
- u64 bytes)
+void net_dim(struct net_dim *dim,
+ u16 event_ctr,
+ u64 packets,
+ u64 bytes)
{
- struct mlx5e_rx_am_sample end_sample;
- struct mlx5e_rx_am_stats curr_stats;
+ struct net_dim_sample end_sample;
+ struct net_dim_stats curr_stats;
u16 nevents;
- switch (am->state) {
- case MLX5E_AM_MEASURE_IN_PROGRESS:
+ switch (dim->state) {
+ case NET_DIM_MEASURE_IN_PROGRESS:
nevents = BIT_GAP(BITS_PER_TYPE(u16), event_ctr,
- am->start_sample.event_ctr);
- if (nevents < MLX5E_AM_NEVENTS)
+ dim->start_sample.event_ctr);
+ if (nevents < NET_DIM_NEVENTS)
break;
- mlx5e_am_sample(event_ctr, packets, bytes, &end_sample);
- mlx5e_am_calc_stats(&am->start_sample, &end_sample,
- &curr_stats);
- if (mlx5e_am_decision(&curr_stats, am)) {
- am->state = MLX5E_AM_APPLY_NEW_PROFILE;
- schedule_work(&am->work);
+ net_dim_sample(event_ctr, packets, bytes, &end_sample);
+ net_dim_calc_stats(&dim->start_sample, &end_sample,
+ &curr_stats);
+ if (net_dim_decision(&curr_stats, dim)) {
+ dim->state = NET_DIM_APPLY_NEW_PROFILE;
+ schedule_work(&dim->work);
break;
}
/* fall through */
- case MLX5E_AM_START_MEASURE:
- mlx5e_am_sample(event_ctr, packets, bytes, &am->start_sample);
- am->state = MLX5E_AM_MEASURE_IN_PROGRESS;
+ case NET_DIM_START_MEASURE:
+ net_dim_sample(event_ctr, packets, bytes, &dim->start_sample);
+ dim->state = NET_DIM_MEASURE_IN_PROGRESS;
break;
- case MLX5E_AM_APPLY_NEW_PROFILE:
+ case NET_DIM_APPLY_NEW_PROFILE:
break;
}
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/net_dim.h b/drivers/net/ethernet/mellanox/mlx5/core/net_dim.h
index b03a48f..f0a70cd 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/net_dim.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/net_dim.h
@@ -31,32 +31,32 @@
* SOFTWARE.
*/
-#ifndef MLX5_AM_H
-#define MLX5_AM_H
+#ifndef NET_DIM_H
+#define NET_DIM_H
-struct mlx5e_cq_moder {
+struct net_dim_cq_moder {
u16 usec;
u16 pkts;
u8 cq_period_mode;
};
-struct mlx5e_rx_am_sample {
+struct net_dim_sample {
ktime_t time;
u32 pkt_ctr;
u32 byte_ctr;
u16 event_ctr;
};
-struct mlx5e_rx_am_stats {
+struct net_dim_stats {
int ppms; /* packets per msec */
int bpms; /* bytes per msec */
int epms; /* events per msec */
};
-struct mlx5e_rx_am { /* Adaptive Moderation */
+struct net_dim { /* Adaptive Moderation */
u8 state;
- struct mlx5e_rx_am_stats prev_stats;
- struct mlx5e_rx_am_sample start_sample;
+ struct net_dim_stats prev_stats;
+ struct net_dim_sample start_sample;
struct work_struct work;
u8 profile_ix;
u8 mode;
@@ -66,38 +66,43 @@ struct mlx5e_rx_am { /* Adaptive Moderation */
u8 tired;
};
+enum {
+ NET_DIM_CQ_PERIOD_MODE_START_FROM_EQE = 0x0,
+ NET_DIM_CQ_PERIOD_MODE_START_FROM_CQE = 0x1,
+ NET_DIM_CQ_PERIOD_NUM_MODES
+};
+
/* Adaptive moderation logic */
enum {
- MLX5E_AM_START_MEASURE,
- MLX5E_AM_MEASURE_IN_PROGRESS,
- MLX5E_AM_APPLY_NEW_PROFILE,
+ NET_DIM_START_MEASURE,
+ NET_DIM_MEASURE_IN_PROGRESS,
+ NET_DIM_APPLY_NEW_PROFILE,
};
enum {
- MLX5E_AM_PARKING_ON_TOP,
- MLX5E_AM_PARKING_TIRED,
- MLX5E_AM_GOING_RIGHT,
- MLX5E_AM_GOING_LEFT,
+ NET_DIM_PARKING_ON_TOP,
+ NET_DIM_PARKING_TIRED,
+ NET_DIM_GOING_RIGHT,
+ NET_DIM_GOING_LEFT,
};
enum {
- MLX5E_AM_STATS_WORSE,
- MLX5E_AM_STATS_SAME,
- MLX5E_AM_STATS_BETTER,
+ NET_DIM_STATS_WORSE,
+ NET_DIM_STATS_SAME,
+ NET_DIM_STATS_BETTER,
};
enum {
- MLX5E_AM_STEPPED,
- MLX5E_AM_TOO_TIRED,
- MLX5E_AM_ON_EDGE,
+ NET_DIM_STEPPED,
+ NET_DIM_TOO_TIRED,
+ NET_DIM_ON_EDGE,
};
-void mlx5e_rx_am(struct mlx5e_rx_am *am,
- u16 event_ctr,
- u64 packets,
- u64 bytes);
-void mlx5e_rx_am_work(struct work_struct *work);
-struct mlx5e_cq_moder mlx5e_am_get_def_profile(u8 rx_cq_period_mode);
-struct mlx5e_cq_moder mlx5e_am_get_profile(u8 cq_period_mode, int ix);
+void net_dim(struct net_dim *dim,
+ u16 event_ctr,
+ u64 packets,
+ u64 bytes);
+struct net_dim_cq_moder net_dim_get_def_profile(u8 rx_cq_period_mode);
+struct net_dim_cq_moder net_dim_get_profile(u8 cq_period_mode, int ix);
-#endif /* MLX5_AM_H */
+#endif /* NET_DIM_H */
--
2.7.4
^ permalink raw reply related
* [PATCH net-next v3 05/10] net/mlx5e: Move generic functions to new file
From: Andy Gospodarek @ 2018-01-09 6:13 UTC (permalink / raw)
To: netdev; +Cc: mchan, talgi, ogerlitz, Andy Gospodarek
In-Reply-To: <1515478402-6048-1-git-send-email-andy@greyhouse.net>
From: Andy Gospodarek <gospo@broadcom.com>
These functions were identified as ones that could be made generic and
used by multiple drivers. Most of the contents of en_rx_am.c are moved
to net_dim.c.
Signed-off-by: Andy Gospodarek <gospo@broadcom.com>
Acked-by: Tal Gilboa <talgi@mellanox.com>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/Makefile | 4 +-
drivers/net/ethernet/mellanox/mlx5/core/en.h | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en_dim.c | 48 ++++
drivers/net/ethernet/mellanox/mlx5/core/en_dim.h | 102 -------
drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c | 320 ---------------------
drivers/net/ethernet/mellanox/mlx5/core/net_dim.c | 307 ++++++++++++++++++++
drivers/net/ethernet/mellanox/mlx5/core/net_dim.h | 103 +++++++
7 files changed, 461 insertions(+), 425 deletions(-)
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/en_dim.c
delete mode 100644 drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
delete mode 100644 drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/net_dim.c
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/net_dim.h
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Makefile b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
index 19b21b4..b46b6de2 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/Makefile
+++ b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
@@ -14,8 +14,8 @@ mlx5_core-$(CONFIG_MLX5_FPGA) += fpga/cmd.o fpga/core.o fpga/conn.o fpga/sdk.o \
fpga/ipsec.o
mlx5_core-$(CONFIG_MLX5_CORE_EN) += en_main.o en_common.o en_fs.o en_ethtool.o \
- en_tx.o en_rx.o en_rx_am.o en_txrx.o en_stats.o vxlan.o \
- en_arfs.o en_fs_ethtool.o en_selftest.o
+ en_tx.o en_rx.o en_dim.o en_txrx.o en_stats.o vxlan.o \
+ en_arfs.o en_fs_ethtool.o en_selftest.o net_dim.o
mlx5_core-$(CONFIG_MLX5_MPFS) += lib/mpfs.o
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index e2e35ed..4ee06e7 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -50,7 +50,7 @@
#include "wq.h"
#include "mlx5_core.h"
#include "en_stats.h"
-#include "en_dim.h"
+#include "net_dim.h"
#define MLX5_SET_CFG(p, f, v) MLX5_SET(create_flow_group_in, p, f, v)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.c b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.c
new file mode 100644
index 0000000..b9b434b
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.c
@@ -0,0 +1,48 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "en.h"
+
+void mlx5e_rx_am_work(struct work_struct *work)
+{
+ struct mlx5e_rx_am *am = container_of(work, struct mlx5e_rx_am,
+ work);
+ struct mlx5e_rq *rq = container_of(am, struct mlx5e_rq, am);
+ struct mlx5e_cq_moder cur_profile = mlx5e_am_get_profile(am->mode,
+ am->profile_ix);
+
+ mlx5_core_modify_cq_moderation(rq->mdev, &rq->cq.mcq,
+ cur_profile.usec, cur_profile.pkts);
+
+ am->state = MLX5E_AM_START_MEASURE;
+}
+
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
deleted file mode 100644
index a1497bab..0000000
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
+++ /dev/null
@@ -1,102 +0,0 @@
-/*
- * Copyright (c) 2013-2015, Mellanox Technologies, Ltd. All rights reserved.
- * Copyright (c) 2017-2018, Broadcom Limited
- *
- * This software is available to you under a choice of one of two
- * licenses. You may choose to be licensed under the terms of the GNU
- * General Public License (GPL) Version 2, available from the file
- * COPYING in the main directory of this source tree, or the
- * OpenIB.org BSD license below:
- *
- * Redistribution and use in source and binary forms, with or
- * without modification, are permitted provided that the following
- * conditions are met:
- *
- * - Redistributions of source code must retain the above
- * copyright notice, this list of conditions and the following
- * disclaimer.
- *
- * - Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following
- * disclaimer in the documentation and/or other materials
- * provided with the distribution.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
- * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
- * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
- * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- */
-
-#ifndef MLX5_AM_H
-#define MLX5_AM_H
-
-struct mlx5e_cq_moder {
- u16 usec;
- u16 pkts;
- u8 cq_period_mode;
-};
-
-struct mlx5e_rx_am_sample {
- ktime_t time;
- u32 pkt_ctr;
- u32 byte_ctr;
- u16 event_ctr;
-};
-
-struct mlx5e_rx_am_stats {
- int ppms; /* packets per msec */
- int bpms; /* bytes per msec */
- int epms; /* events per msec */
-};
-
-struct mlx5e_rx_am { /* Adaptive Moderation */
- u8 state;
- struct mlx5e_rx_am_stats prev_stats;
- struct mlx5e_rx_am_sample start_sample;
- struct work_struct work;
- u8 profile_ix;
- u8 mode;
- u8 tune_state;
- u8 steps_right;
- u8 steps_left;
- u8 tired;
-};
-
-/* Adaptive moderation logic */
-enum {
- MLX5E_AM_START_MEASURE,
- MLX5E_AM_MEASURE_IN_PROGRESS,
- MLX5E_AM_APPLY_NEW_PROFILE,
-};
-
-enum {
- MLX5E_AM_PARKING_ON_TOP,
- MLX5E_AM_PARKING_TIRED,
- MLX5E_AM_GOING_RIGHT,
- MLX5E_AM_GOING_LEFT,
-};
-
-enum {
- MLX5E_AM_STATS_WORSE,
- MLX5E_AM_STATS_SAME,
- MLX5E_AM_STATS_BETTER,
-};
-
-enum {
- MLX5E_AM_STEPPED,
- MLX5E_AM_TOO_TIRED,
- MLX5E_AM_ON_EDGE,
-};
-
-void mlx5e_rx_am(struct mlx5e_rx_am *am,
- u16 event_ctr,
- u64 packets,
- u64 bytes);
-void mlx5e_rx_am_work(struct work_struct *work);
-struct mlx5e_cq_moder mlx5e_am_get_def_profile(u8 rx_cq_period_mode);
-
-#endif /* MLX5_AM_H */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c
deleted file mode 100644
index 337dd60..0000000
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c
+++ /dev/null
@@ -1,320 +0,0 @@
-/*
- * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
- *
- * This software is available to you under a choice of one of two
- * licenses. You may choose to be licensed under the terms of the GNU
- * General Public License (GPL) Version 2, available from the file
- * COPYING in the main directory of this source tree, or the
- * OpenIB.org BSD license below:
- *
- * Redistribution and use in source and binary forms, with or
- * without modification, are permitted provided that the following
- * conditions are met:
- *
- * - Redistributions of source code must retain the above
- * copyright notice, this list of conditions and the following
- * disclaimer.
- *
- * - Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following
- * disclaimer in the documentation and/or other materials
- * provided with the distribution.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
- * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
- * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
- * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- */
-
-#include "en.h"
-
-/* Adaptive moderation profiles */
-#define MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE 256
-#define MLX5E_RX_AM_DEF_PROFILE_CQE 1
-#define MLX5E_RX_AM_DEF_PROFILE_EQE 1
-#define MLX5E_PARAMS_AM_NUM_PROFILES 5
-
-/* All profiles sizes must be MLX5E_PARAMS_AM_NUM_PROFILES */
-#define MLX5_AM_EQE_PROFILES { \
- {1, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
- {8, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
- {64, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
- {128, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
- {256, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
-}
-
-#define MLX5_AM_CQE_PROFILES { \
- {2, 256}, \
- {8, 128}, \
- {16, 64}, \
- {32, 64}, \
- {64, 64} \
-}
-
-static const struct mlx5e_cq_moder
-profile[MLX5_CQ_PERIOD_NUM_MODES][MLX5E_PARAMS_AM_NUM_PROFILES] = {
- MLX5_AM_EQE_PROFILES,
- MLX5_AM_CQE_PROFILES,
-};
-
-static inline struct mlx5e_cq_moder mlx5e_am_get_profile(u8 cq_period_mode, int ix)
-{
- struct mlx5e_cq_moder cq_moder;
-
- cq_moder = profile[cq_period_mode][ix];
- cq_moder.cq_period_mode = cq_period_mode;
- return cq_moder;
-}
-
-struct mlx5e_cq_moder mlx5e_am_get_def_profile(u8 rx_cq_period_mode)
-{
- int default_profile_ix;
-
- if (rx_cq_period_mode == MLX5_CQ_PERIOD_MODE_START_FROM_CQE)
- default_profile_ix = MLX5E_RX_AM_DEF_PROFILE_CQE;
- else /* MLX5_CQ_PERIOD_MODE_START_FROM_EQE */
- default_profile_ix = MLX5E_RX_AM_DEF_PROFILE_EQE;
-
- return mlx5e_am_get_profile(rx_cq_period_mode, default_profile_ix);
-}
-
-
-static bool mlx5e_am_on_top(struct mlx5e_rx_am *am)
-{
- switch (am->tune_state) {
- case MLX5E_AM_PARKING_ON_TOP:
- case MLX5E_AM_PARKING_TIRED:
- return true;
- case MLX5E_AM_GOING_RIGHT:
- return (am->steps_left > 1) && (am->steps_right == 1);
- default: /* MLX5E_AM_GOING_LEFT */
- return (am->steps_right > 1) && (am->steps_left == 1);
- }
-}
-
-static void mlx5e_am_turn(struct mlx5e_rx_am *am)
-{
- switch (am->tune_state) {
- case MLX5E_AM_PARKING_ON_TOP:
- case MLX5E_AM_PARKING_TIRED:
- break;
- case MLX5E_AM_GOING_RIGHT:
- am->tune_state = MLX5E_AM_GOING_LEFT;
- am->steps_left = 0;
- break;
- case MLX5E_AM_GOING_LEFT:
- am->tune_state = MLX5E_AM_GOING_RIGHT;
- am->steps_right = 0;
- break;
- }
-}
-
-static int mlx5e_am_step(struct mlx5e_rx_am *am)
-{
- if (am->tired == (MLX5E_PARAMS_AM_NUM_PROFILES * 2))
- return MLX5E_AM_TOO_TIRED;
-
- switch (am->tune_state) {
- case MLX5E_AM_PARKING_ON_TOP:
- case MLX5E_AM_PARKING_TIRED:
- break;
- case MLX5E_AM_GOING_RIGHT:
- if (am->profile_ix == (MLX5E_PARAMS_AM_NUM_PROFILES - 1))
- return MLX5E_AM_ON_EDGE;
- am->profile_ix++;
- am->steps_right++;
- break;
- case MLX5E_AM_GOING_LEFT:
- if (am->profile_ix == 0)
- return MLX5E_AM_ON_EDGE;
- am->profile_ix--;
- am->steps_left++;
- break;
- }
-
- am->tired++;
- return MLX5E_AM_STEPPED;
-}
-
-static void mlx5e_am_park_on_top(struct mlx5e_rx_am *am)
-{
- am->steps_right = 0;
- am->steps_left = 0;
- am->tired = 0;
- am->tune_state = MLX5E_AM_PARKING_ON_TOP;
-}
-
-static void mlx5e_am_park_tired(struct mlx5e_rx_am *am)
-{
- am->steps_right = 0;
- am->steps_left = 0;
- am->tune_state = MLX5E_AM_PARKING_TIRED;
-}
-
-static void mlx5e_am_exit_parking(struct mlx5e_rx_am *am)
-{
- am->tune_state = am->profile_ix ? MLX5E_AM_GOING_LEFT :
- MLX5E_AM_GOING_RIGHT;
- mlx5e_am_step(am);
-}
-
-#define IS_SIGNIFICANT_DIFF(val, ref) \
- (((100 * abs((val) - (ref))) / (ref)) > 10) /* more than 10% difference */
-
-static int mlx5e_am_stats_compare(struct mlx5e_rx_am_stats *curr,
- struct mlx5e_rx_am_stats *prev)
-{
- if (!prev->bpms)
- return curr->bpms ? MLX5E_AM_STATS_BETTER :
- MLX5E_AM_STATS_SAME;
-
- if (IS_SIGNIFICANT_DIFF(curr->bpms, prev->bpms))
- return (curr->bpms > prev->bpms) ? MLX5E_AM_STATS_BETTER :
- MLX5E_AM_STATS_WORSE;
-
- if (IS_SIGNIFICANT_DIFF(curr->ppms, prev->ppms))
- return (curr->ppms > prev->ppms) ? MLX5E_AM_STATS_BETTER :
- MLX5E_AM_STATS_WORSE;
-
- if (IS_SIGNIFICANT_DIFF(curr->epms, prev->epms))
- return (curr->epms < prev->epms) ? MLX5E_AM_STATS_BETTER :
- MLX5E_AM_STATS_WORSE;
-
- return MLX5E_AM_STATS_SAME;
-}
-
-static bool mlx5e_am_decision(struct mlx5e_rx_am_stats *curr_stats,
- struct mlx5e_rx_am *am)
-{
- int prev_state = am->tune_state;
- int prev_ix = am->profile_ix;
- int stats_res;
- int step_res;
-
- switch (am->tune_state) {
- case MLX5E_AM_PARKING_ON_TOP:
- stats_res = mlx5e_am_stats_compare(curr_stats, &am->prev_stats);
- if (stats_res != MLX5E_AM_STATS_SAME)
- mlx5e_am_exit_parking(am);
- break;
-
- case MLX5E_AM_PARKING_TIRED:
- am->tired--;
- if (!am->tired)
- mlx5e_am_exit_parking(am);
- break;
-
- case MLX5E_AM_GOING_RIGHT:
- case MLX5E_AM_GOING_LEFT:
- stats_res = mlx5e_am_stats_compare(curr_stats, &am->prev_stats);
- if (stats_res != MLX5E_AM_STATS_BETTER)
- mlx5e_am_turn(am);
-
- if (mlx5e_am_on_top(am)) {
- mlx5e_am_park_on_top(am);
- break;
- }
-
- step_res = mlx5e_am_step(am);
- switch (step_res) {
- case MLX5E_AM_ON_EDGE:
- mlx5e_am_park_on_top(am);
- break;
- case MLX5E_AM_TOO_TIRED:
- mlx5e_am_park_tired(am);
- break;
- }
-
- break;
- }
-
- if ((prev_state != MLX5E_AM_PARKING_ON_TOP) ||
- (am->tune_state != MLX5E_AM_PARKING_ON_TOP))
- am->prev_stats = *curr_stats;
-
- return am->profile_ix != prev_ix;
-}
-
-static void mlx5e_am_sample(u16 event_ctr,
- u64 packets,
- u64 bytes,
- struct mlx5e_rx_am_sample *s)
-{
- s->time = ktime_get();
- s->pkt_ctr = packets;
- s->byte_ctr = bytes;
- s->event_ctr = event_ctr;
-}
-
-#define MLX5E_AM_NEVENTS 64
-#define BITS_PER_TYPE(type) (sizeof(type) * BITS_PER_BYTE)
-#define BIT_GAP(bits, end, start) ((((end) - (start)) + BIT_ULL(bits)) & (BIT_ULL(bits) - 1))
-
-static void mlx5e_am_calc_stats(struct mlx5e_rx_am_sample *start,
- struct mlx5e_rx_am_sample *end,
- struct mlx5e_rx_am_stats *curr_stats)
-{
- /* u32 holds up to 71 minutes, should be enough */
- u32 delta_us = ktime_us_delta(end->time, start->time);
- u32 npkts = BIT_GAP(BITS_PER_TYPE(u32), end->pkt_ctr, start->pkt_ctr);
- u32 nbytes = BIT_GAP(BITS_PER_TYPE(u32), end->byte_ctr,
- start->byte_ctr);
-
- if (!delta_us)
- return;
-
- curr_stats->ppms = DIV_ROUND_UP(npkts * USEC_PER_MSEC, delta_us);
- curr_stats->bpms = DIV_ROUND_UP(nbytes * USEC_PER_MSEC, delta_us);
- curr_stats->epms = DIV_ROUND_UP(MLX5E_AM_NEVENTS * USEC_PER_MSEC,
- delta_us);
-}
-
-void mlx5e_rx_am_work(struct work_struct *work)
-{
- struct mlx5e_rx_am *am = container_of(work, struct mlx5e_rx_am,
- work);
- struct mlx5e_rq *rq = container_of(am, struct mlx5e_rq, am);
- struct mlx5e_cq_moder cur_profile = profile[am->mode][am->profile_ix];
-
- mlx5_core_modify_cq_moderation(rq->mdev, &rq->cq.mcq,
- cur_profile.usec, cur_profile.pkts);
-
- am->state = MLX5E_AM_START_MEASURE;
-}
-
-void mlx5e_rx_am(struct mlx5e_rx_am *am,
- u16 event_ctr,
- u64 packets,
- u64 bytes)
-{
- struct mlx5e_rx_am_sample end_sample;
- struct mlx5e_rx_am_stats curr_stats;
- u16 nevents;
-
- switch (am->state) {
- case MLX5E_AM_MEASURE_IN_PROGRESS:
- nevents = BIT_GAP(BITS_PER_TYPE(u16), event_ctr,
- am->start_sample.event_ctr);
- if (nevents < MLX5E_AM_NEVENTS)
- break;
- mlx5e_am_sample(event_ctr, packets, bytes, &end_sample);
- mlx5e_am_calc_stats(&am->start_sample, &end_sample,
- &curr_stats);
- if (mlx5e_am_decision(&curr_stats, am)) {
- am->state = MLX5E_AM_APPLY_NEW_PROFILE;
- schedule_work(&am->work);
- break;
- }
- /* fall through */
- case MLX5E_AM_START_MEASURE:
- mlx5e_am_sample(event_ctr, packets, bytes, &am->start_sample);
- am->state = MLX5E_AM_MEASURE_IN_PROGRESS;
- break;
- case MLX5E_AM_APPLY_NEW_PROFILE:
- break;
- }
-}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/net_dim.c b/drivers/net/ethernet/mellanox/mlx5/core/net_dim.c
new file mode 100644
index 0000000..9fafe1d
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/net_dim.c
@@ -0,0 +1,307 @@
+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2017-2018, Broadcom Limited. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "en.h"
+
+#define MLX5E_PARAMS_AM_NUM_PROFILES 5
+/* Adaptive moderation profiles */
+#define MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE 256
+#define MLX5E_RX_AM_DEF_PROFILE_CQE 1
+#define MLX5E_RX_AM_DEF_PROFILE_EQE 1
+
+/* All profiles sizes must be MLX5E_PARAMS_AM_NUM_PROFILES */
+#define MLX5_AM_EQE_PROFILES { \
+ {1, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+ {8, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+ {64, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+ {128, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+ {256, MLX5E_AM_DEFAULT_RX_CQ_MODERATION_PKTS_FROM_EQE}, \
+}
+
+#define MLX5_AM_CQE_PROFILES { \
+ {2, 256}, \
+ {8, 128}, \
+ {16, 64}, \
+ {32, 64}, \
+ {64, 64} \
+}
+
+static const struct mlx5e_cq_moder
+profile[MLX5_CQ_PERIOD_NUM_MODES][MLX5E_PARAMS_AM_NUM_PROFILES] = {
+ MLX5_AM_EQE_PROFILES,
+ MLX5_AM_CQE_PROFILES,
+};
+
+struct mlx5e_cq_moder mlx5e_am_get_profile(u8 cq_period_mode, int ix)
+{
+ struct mlx5e_cq_moder cq_moder;
+
+ cq_moder = profile[cq_period_mode][ix];
+ cq_moder.cq_period_mode = cq_period_mode;
+ return cq_moder;
+}
+
+struct mlx5e_cq_moder mlx5e_am_get_def_profile(u8 rx_cq_period_mode)
+{
+ int default_profile_ix;
+
+ if (rx_cq_period_mode == MLX5_CQ_PERIOD_MODE_START_FROM_CQE)
+ default_profile_ix = MLX5E_RX_AM_DEF_PROFILE_CQE;
+ else /* MLX5_CQ_PERIOD_MODE_START_FROM_EQE */
+ default_profile_ix = MLX5E_RX_AM_DEF_PROFILE_EQE;
+
+ return mlx5e_am_get_profile(rx_cq_period_mode, default_profile_ix);
+}
+
+static bool mlx5e_am_on_top(struct mlx5e_rx_am *am)
+{
+ switch (am->tune_state) {
+ case MLX5E_AM_PARKING_ON_TOP:
+ case MLX5E_AM_PARKING_TIRED:
+ return true;
+ case MLX5E_AM_GOING_RIGHT:
+ return (am->steps_left > 1) && (am->steps_right == 1);
+ default: /* MLX5E_AM_GOING_LEFT */
+ return (am->steps_right > 1) && (am->steps_left == 1);
+ }
+}
+
+static void mlx5e_am_turn(struct mlx5e_rx_am *am)
+{
+ switch (am->tune_state) {
+ case MLX5E_AM_PARKING_ON_TOP:
+ case MLX5E_AM_PARKING_TIRED:
+ break;
+ case MLX5E_AM_GOING_RIGHT:
+ am->tune_state = MLX5E_AM_GOING_LEFT;
+ am->steps_left = 0;
+ break;
+ case MLX5E_AM_GOING_LEFT:
+ am->tune_state = MLX5E_AM_GOING_RIGHT;
+ am->steps_right = 0;
+ break;
+ }
+}
+
+static int mlx5e_am_step(struct mlx5e_rx_am *am)
+{
+ if (am->tired == (MLX5E_PARAMS_AM_NUM_PROFILES * 2))
+ return MLX5E_AM_TOO_TIRED;
+
+ switch (am->tune_state) {
+ case MLX5E_AM_PARKING_ON_TOP:
+ case MLX5E_AM_PARKING_TIRED:
+ break;
+ case MLX5E_AM_GOING_RIGHT:
+ if (am->profile_ix == (MLX5E_PARAMS_AM_NUM_PROFILES - 1))
+ return MLX5E_AM_ON_EDGE;
+ am->profile_ix++;
+ am->steps_right++;
+ break;
+ case MLX5E_AM_GOING_LEFT:
+ if (am->profile_ix == 0)
+ return MLX5E_AM_ON_EDGE;
+ am->profile_ix--;
+ am->steps_left++;
+ break;
+ }
+
+ am->tired++;
+ return MLX5E_AM_STEPPED;
+}
+
+static void mlx5e_am_park_on_top(struct mlx5e_rx_am *am)
+{
+ am->steps_right = 0;
+ am->steps_left = 0;
+ am->tired = 0;
+ am->tune_state = MLX5E_AM_PARKING_ON_TOP;
+}
+
+static void mlx5e_am_park_tired(struct mlx5e_rx_am *am)
+{
+ am->steps_right = 0;
+ am->steps_left = 0;
+ am->tune_state = MLX5E_AM_PARKING_TIRED;
+}
+
+static void mlx5e_am_exit_parking(struct mlx5e_rx_am *am)
+{
+ am->tune_state = am->profile_ix ? MLX5E_AM_GOING_LEFT :
+ MLX5E_AM_GOING_RIGHT;
+ mlx5e_am_step(am);
+}
+
+#define IS_SIGNIFICANT_DIFF(val, ref) \
+ (((100 * abs((val) - (ref))) / (ref)) > 10) /* more than 10% difference */
+
+static int mlx5e_am_stats_compare(struct mlx5e_rx_am_stats *curr,
+ struct mlx5e_rx_am_stats *prev)
+{
+ if (!prev->bpms)
+ return curr->bpms ? MLX5E_AM_STATS_BETTER :
+ MLX5E_AM_STATS_SAME;
+
+ if (IS_SIGNIFICANT_DIFF(curr->bpms, prev->bpms))
+ return (curr->bpms > prev->bpms) ? MLX5E_AM_STATS_BETTER :
+ MLX5E_AM_STATS_WORSE;
+
+ if (IS_SIGNIFICANT_DIFF(curr->ppms, prev->ppms))
+ return (curr->ppms > prev->ppms) ? MLX5E_AM_STATS_BETTER :
+ MLX5E_AM_STATS_WORSE;
+
+ if (IS_SIGNIFICANT_DIFF(curr->epms, prev->epms))
+ return (curr->epms < prev->epms) ? MLX5E_AM_STATS_BETTER :
+ MLX5E_AM_STATS_WORSE;
+
+ return MLX5E_AM_STATS_SAME;
+}
+
+static bool mlx5e_am_decision(struct mlx5e_rx_am_stats *curr_stats,
+ struct mlx5e_rx_am *am)
+{
+ int prev_state = am->tune_state;
+ int prev_ix = am->profile_ix;
+ int stats_res;
+ int step_res;
+
+ switch (am->tune_state) {
+ case MLX5E_AM_PARKING_ON_TOP:
+ stats_res = mlx5e_am_stats_compare(curr_stats, &am->prev_stats);
+ if (stats_res != MLX5E_AM_STATS_SAME)
+ mlx5e_am_exit_parking(am);
+ break;
+
+ case MLX5E_AM_PARKING_TIRED:
+ am->tired--;
+ if (!am->tired)
+ mlx5e_am_exit_parking(am);
+ break;
+
+ case MLX5E_AM_GOING_RIGHT:
+ case MLX5E_AM_GOING_LEFT:
+ stats_res = mlx5e_am_stats_compare(curr_stats, &am->prev_stats);
+ if (stats_res != MLX5E_AM_STATS_BETTER)
+ mlx5e_am_turn(am);
+
+ if (mlx5e_am_on_top(am)) {
+ mlx5e_am_park_on_top(am);
+ break;
+ }
+
+ step_res = mlx5e_am_step(am);
+ switch (step_res) {
+ case MLX5E_AM_ON_EDGE:
+ mlx5e_am_park_on_top(am);
+ break;
+ case MLX5E_AM_TOO_TIRED:
+ mlx5e_am_park_tired(am);
+ break;
+ }
+
+ break;
+ }
+
+ if ((prev_state != MLX5E_AM_PARKING_ON_TOP) ||
+ (am->tune_state != MLX5E_AM_PARKING_ON_TOP))
+ am->prev_stats = *curr_stats;
+
+ return am->profile_ix != prev_ix;
+}
+
+static void mlx5e_am_sample(u16 event_ctr,
+ u64 packets,
+ u64 bytes,
+ struct mlx5e_rx_am_sample *s)
+{
+ s->time = ktime_get();
+ s->pkt_ctr = packets;
+ s->byte_ctr = bytes;
+ s->event_ctr = event_ctr;
+}
+
+#define MLX5E_AM_NEVENTS 64
+#define BITS_PER_TYPE(type) (sizeof(type) * BITS_PER_BYTE)
+#define BIT_GAP(bits, end, start) ((((end) - (start)) + BIT_ULL(bits)) & (BIT_ULL(bits) - 1))
+
+static void mlx5e_am_calc_stats(struct mlx5e_rx_am_sample *start,
+ struct mlx5e_rx_am_sample *end,
+ struct mlx5e_rx_am_stats *curr_stats)
+{
+ /* u32 holds up to 71 minutes, should be enough */
+ u32 delta_us = ktime_us_delta(end->time, start->time);
+ u32 npkts = BIT_GAP(BITS_PER_TYPE(u32), end->pkt_ctr, start->pkt_ctr);
+ u32 nbytes = BIT_GAP(BITS_PER_TYPE(u32), end->byte_ctr,
+ start->byte_ctr);
+
+ if (!delta_us)
+ return;
+
+ curr_stats->ppms = DIV_ROUND_UP(npkts * USEC_PER_MSEC, delta_us);
+ curr_stats->bpms = DIV_ROUND_UP(nbytes * USEC_PER_MSEC, delta_us);
+ curr_stats->epms = DIV_ROUND_UP(MLX5E_AM_NEVENTS * USEC_PER_MSEC,
+ delta_us);
+}
+
+void mlx5e_rx_am(struct mlx5e_rx_am *am,
+ u16 event_ctr,
+ u64 packets,
+ u64 bytes)
+{
+ struct mlx5e_rx_am_sample end_sample;
+ struct mlx5e_rx_am_stats curr_stats;
+ u16 nevents;
+
+ switch (am->state) {
+ case MLX5E_AM_MEASURE_IN_PROGRESS:
+ nevents = BIT_GAP(BITS_PER_TYPE(u16), event_ctr,
+ am->start_sample.event_ctr);
+ if (nevents < MLX5E_AM_NEVENTS)
+ break;
+ mlx5e_am_sample(event_ctr, packets, bytes, &end_sample);
+ mlx5e_am_calc_stats(&am->start_sample, &end_sample,
+ &curr_stats);
+ if (mlx5e_am_decision(&curr_stats, am)) {
+ am->state = MLX5E_AM_APPLY_NEW_PROFILE;
+ schedule_work(&am->work);
+ break;
+ }
+ /* fall through */
+ case MLX5E_AM_START_MEASURE:
+ mlx5e_am_sample(event_ctr, packets, bytes, &am->start_sample);
+ am->state = MLX5E_AM_MEASURE_IN_PROGRESS;
+ break;
+ case MLX5E_AM_APPLY_NEW_PROFILE:
+ break;
+ }
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/net_dim.h b/drivers/net/ethernet/mellanox/mlx5/core/net_dim.h
new file mode 100644
index 0000000..b03a48f
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/net_dim.h
@@ -0,0 +1,103 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd. All rights reserved.
+ * Copyright (c) 2017-2018, Broadcom Limited. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef MLX5_AM_H
+#define MLX5_AM_H
+
+struct mlx5e_cq_moder {
+ u16 usec;
+ u16 pkts;
+ u8 cq_period_mode;
+};
+
+struct mlx5e_rx_am_sample {
+ ktime_t time;
+ u32 pkt_ctr;
+ u32 byte_ctr;
+ u16 event_ctr;
+};
+
+struct mlx5e_rx_am_stats {
+ int ppms; /* packets per msec */
+ int bpms; /* bytes per msec */
+ int epms; /* events per msec */
+};
+
+struct mlx5e_rx_am { /* Adaptive Moderation */
+ u8 state;
+ struct mlx5e_rx_am_stats prev_stats;
+ struct mlx5e_rx_am_sample start_sample;
+ struct work_struct work;
+ u8 profile_ix;
+ u8 mode;
+ u8 tune_state;
+ u8 steps_right;
+ u8 steps_left;
+ u8 tired;
+};
+
+/* Adaptive moderation logic */
+enum {
+ MLX5E_AM_START_MEASURE,
+ MLX5E_AM_MEASURE_IN_PROGRESS,
+ MLX5E_AM_APPLY_NEW_PROFILE,
+};
+
+enum {
+ MLX5E_AM_PARKING_ON_TOP,
+ MLX5E_AM_PARKING_TIRED,
+ MLX5E_AM_GOING_RIGHT,
+ MLX5E_AM_GOING_LEFT,
+};
+
+enum {
+ MLX5E_AM_STATS_WORSE,
+ MLX5E_AM_STATS_SAME,
+ MLX5E_AM_STATS_BETTER,
+};
+
+enum {
+ MLX5E_AM_STEPPED,
+ MLX5E_AM_TOO_TIRED,
+ MLX5E_AM_ON_EDGE,
+};
+
+void mlx5e_rx_am(struct mlx5e_rx_am *am,
+ u16 event_ctr,
+ u64 packets,
+ u64 bytes);
+void mlx5e_rx_am_work(struct work_struct *work);
+struct mlx5e_cq_moder mlx5e_am_get_def_profile(u8 rx_cq_period_mode);
+struct mlx5e_cq_moder mlx5e_am_get_profile(u8 cq_period_mode, int ix);
+
+#endif /* MLX5_AM_H */
--
2.7.4
^ permalink raw reply related
* [PATCH net-next v3 04/10] net/mlx5e: Move AM logic enums
From: Andy Gospodarek @ 2018-01-09 6:13 UTC (permalink / raw)
To: netdev; +Cc: mchan, talgi, ogerlitz, Andy Gospodarek
In-Reply-To: <1515478402-6048-1-git-send-email-andy@greyhouse.net>
From: Andy Gospodarek <gospo@broadcom.com>
More movement to help make this code more generic.
Signed-off-by: Andy Gospodarek <gospo@broadcom.com>
Acked-by: Tal Gilboa <talgi@mellanox.com>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_dim.h | 26 ++++++++++++++++++++++
drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c | 25 ---------------------
2 files changed, 26 insertions(+), 25 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
index 7d5499a..a1497bab 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
@@ -66,6 +66,32 @@ struct mlx5e_rx_am { /* Adaptive Moderation */
u8 tired;
};
+/* Adaptive moderation logic */
+enum {
+ MLX5E_AM_START_MEASURE,
+ MLX5E_AM_MEASURE_IN_PROGRESS,
+ MLX5E_AM_APPLY_NEW_PROFILE,
+};
+
+enum {
+ MLX5E_AM_PARKING_ON_TOP,
+ MLX5E_AM_PARKING_TIRED,
+ MLX5E_AM_GOING_RIGHT,
+ MLX5E_AM_GOING_LEFT,
+};
+
+enum {
+ MLX5E_AM_STATS_WORSE,
+ MLX5E_AM_STATS_SAME,
+ MLX5E_AM_STATS_BETTER,
+};
+
+enum {
+ MLX5E_AM_STEPPED,
+ MLX5E_AM_TOO_TIRED,
+ MLX5E_AM_ON_EDGE,
+};
+
void mlx5e_rx_am(struct mlx5e_rx_am *am,
u16 event_ctr,
u64 packets,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c
index 1630076..337dd60 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c
@@ -82,31 +82,6 @@ struct mlx5e_cq_moder mlx5e_am_get_def_profile(u8 rx_cq_period_mode)
return mlx5e_am_get_profile(rx_cq_period_mode, default_profile_ix);
}
-/* Adaptive moderation logic */
-enum {
- MLX5E_AM_START_MEASURE,
- MLX5E_AM_MEASURE_IN_PROGRESS,
- MLX5E_AM_APPLY_NEW_PROFILE,
-};
-
-enum {
- MLX5E_AM_PARKING_ON_TOP,
- MLX5E_AM_PARKING_TIRED,
- MLX5E_AM_GOING_RIGHT,
- MLX5E_AM_GOING_LEFT,
-};
-
-enum {
- MLX5E_AM_STATS_WORSE,
- MLX5E_AM_STATS_SAME,
- MLX5E_AM_STATS_BETTER,
-};
-
-enum {
- MLX5E_AM_STEPPED,
- MLX5E_AM_TOO_TIRED,
- MLX5E_AM_ON_EDGE,
-};
static bool mlx5e_am_on_top(struct mlx5e_rx_am *am)
{
--
2.7.4
^ permalink raw reply related
* [PATCH net-next v3 03/10] net/mlx5e: Remove rq references in mlx5e_rx_am
From: Andy Gospodarek @ 2018-01-09 6:13 UTC (permalink / raw)
To: netdev; +Cc: mchan, talgi, ogerlitz, Andy Gospodarek
In-Reply-To: <1515478402-6048-1-git-send-email-andy@greyhouse.net>
From: Andy Gospodarek <gospo@broadcom.com>
This makes mlx5e_am_sample more generic so that it can be called easily
from a driver that does not use the same data structure to store these
values in a single structure.
Signed-off-by: Andy Gospodarek <gospo@broadcom.com>
Acked-by: Tal Gilboa <talgi@mellanox.com>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_dim.h | 6 ++++--
drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c | 22 +++++++++++++---------
drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c | 5 ++++-
3 files changed, 21 insertions(+), 12 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
index 2031a21..7d5499a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
@@ -66,8 +66,10 @@ struct mlx5e_rx_am { /* Adaptive Moderation */
u8 tired;
};
-struct mlx5e_rq;
-void mlx5e_rx_am(struct mlx5e_rq *rq);
+void mlx5e_rx_am(struct mlx5e_rx_am *am,
+ u16 event_ctr,
+ u64 packets,
+ u64 bytes);
void mlx5e_rx_am_work(struct work_struct *work);
struct mlx5e_cq_moder mlx5e_am_get_def_profile(u8 rx_cq_period_mode);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c
index e401d9d..1630076 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c
@@ -264,13 +264,15 @@ static bool mlx5e_am_decision(struct mlx5e_rx_am_stats *curr_stats,
return am->profile_ix != prev_ix;
}
-static void mlx5e_am_sample(struct mlx5e_rq *rq,
+static void mlx5e_am_sample(u16 event_ctr,
+ u64 packets,
+ u64 bytes,
struct mlx5e_rx_am_sample *s)
{
s->time = ktime_get();
- s->pkt_ctr = rq->stats.packets;
- s->byte_ctr = rq->stats.bytes;
- s->event_ctr = rq->cq.event_ctr;
+ s->pkt_ctr = packets;
+ s->byte_ctr = bytes;
+ s->event_ctr = event_ctr;
}
#define MLX5E_AM_NEVENTS 64
@@ -309,20 +311,22 @@ void mlx5e_rx_am_work(struct work_struct *work)
am->state = MLX5E_AM_START_MEASURE;
}
-void mlx5e_rx_am(struct mlx5e_rq *rq)
+void mlx5e_rx_am(struct mlx5e_rx_am *am,
+ u16 event_ctr,
+ u64 packets,
+ u64 bytes)
{
- struct mlx5e_rx_am *am = &rq->am;
struct mlx5e_rx_am_sample end_sample;
struct mlx5e_rx_am_stats curr_stats;
u16 nevents;
switch (am->state) {
case MLX5E_AM_MEASURE_IN_PROGRESS:
- nevents = BIT_GAP(BITS_PER_TYPE(u16), rq->cq.event_ctr,
+ nevents = BIT_GAP(BITS_PER_TYPE(u16), event_ctr,
am->start_sample.event_ctr);
if (nevents < MLX5E_AM_NEVENTS)
break;
- mlx5e_am_sample(rq, &end_sample);
+ mlx5e_am_sample(event_ctr, packets, bytes, &end_sample);
mlx5e_am_calc_stats(&am->start_sample, &end_sample,
&curr_stats);
if (mlx5e_am_decision(&curr_stats, am)) {
@@ -332,7 +336,7 @@ void mlx5e_rx_am(struct mlx5e_rq *rq)
}
/* fall through */
case MLX5E_AM_START_MEASURE:
- mlx5e_am_sample(rq, &am->start_sample);
+ mlx5e_am_sample(event_ctr, packets, bytes, &am->start_sample);
am->state = MLX5E_AM_MEASURE_IN_PROGRESS;
break;
case MLX5E_AM_APPLY_NEW_PROFILE:
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
index ab92298..1849169 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
@@ -79,7 +79,10 @@ int mlx5e_napi_poll(struct napi_struct *napi, int budget)
mlx5e_cq_arm(&c->sq[i].cq);
if (MLX5E_TEST_BIT(c->rq.state, MLX5E_RQ_STATE_AM))
- mlx5e_rx_am(&c->rq);
+ mlx5e_rx_am(&c->rq.am,
+ c->rq.cq.event_ctr,
+ c->rq.stats.packets,
+ c->rq.stats.bytes);
mlx5e_cq_arm(&c->rq.cq);
mlx5e_cq_arm(&c->icosq.cq);
--
2.7.4
^ permalink raw reply related
* [PATCH net-next v3 02/10] net/mlx5e: Move interrupt moderation forward declarations
From: Andy Gospodarek @ 2018-01-09 6:13 UTC (permalink / raw)
To: netdev; +Cc: mchan, talgi, ogerlitz, Andy Gospodarek
In-Reply-To: <1515478402-6048-1-git-send-email-andy@greyhouse.net>
From: Andy Gospodarek <gospo@broadcom.com>
Move these to newly created file to prepare to move these functions to a
library.
Signed-off-by: Andy Gospodarek <gospo@broadcom.com>
Acked-by: Tal Gilboa <talgi@mellanox.com>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 4 ----
drivers/net/ethernet/mellanox/mlx5/core/en_dim.h | 5 +++++
2 files changed, 5 insertions(+), 4 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index df9cbb3..e2e35ed 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -833,10 +833,6 @@ void mlx5e_dealloc_rx_wqe(struct mlx5e_rq *rq, u16 ix);
void mlx5e_dealloc_rx_mpwqe(struct mlx5e_rq *rq, u16 ix);
void mlx5e_free_rx_mpwqe(struct mlx5e_rq *rq, struct mlx5e_mpw_info *wi);
-void mlx5e_rx_am(struct mlx5e_rq *rq);
-void mlx5e_rx_am_work(struct work_struct *work);
-struct mlx5e_cq_moder mlx5e_am_get_def_profile(u8 rx_cq_period_mode);
-
void mlx5e_update_stats(struct mlx5e_priv *priv, bool full);
int mlx5e_create_flow_steering(struct mlx5e_priv *priv);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
index 9eeaa11..2031a21 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
@@ -66,4 +66,9 @@ struct mlx5e_rx_am { /* Adaptive Moderation */
u8 tired;
};
+struct mlx5e_rq;
+void mlx5e_rx_am(struct mlx5e_rq *rq);
+void mlx5e_rx_am_work(struct work_struct *work);
+struct mlx5e_cq_moder mlx5e_am_get_def_profile(u8 rx_cq_period_mode);
+
#endif /* MLX5_AM_H */
--
2.7.4
^ permalink raw reply related
* [PATCH net-next v3 01/10] net/mlx5e: Move interrupt moderation structs to new file
From: Andy Gospodarek @ 2018-01-09 6:13 UTC (permalink / raw)
To: netdev; +Cc: mchan, talgi, ogerlitz, Andy Gospodarek
In-Reply-To: <1515478402-6048-1-git-send-email-andy@greyhouse.net>
From: Andy Gospodarek <gospo@broadcom.com>
Create new header file to prepare to move code that handles irq
moderation to a library that lives in a header file.
Signed-off-by: Andy Gospodarek <gospo@broadcom.com>
Acked-by: Tal Gilboa <talgi@mellanox.com>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 33 +-----------
drivers/net/ethernet/mellanox/mlx5/core/en_dim.h | 69 ++++++++++++++++++++++++
2 files changed, 70 insertions(+), 32 deletions(-)
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index 5299310..df9cbb3 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -50,6 +50,7 @@
#include "wq.h"
#include "mlx5_core.h"
#include "en_stats.h"
+#include "en_dim.h"
#define MLX5_SET_CFG(p, f, v) MLX5_SET(create_flow_group_in, p, f, v)
@@ -227,12 +228,6 @@ enum mlx5e_priv_flag {
#define MLX5E_MAX_BW_ALLOC 100 /* Max percentage of BW allocation */
#endif
-struct mlx5e_cq_moder {
- u16 usec;
- u16 pkts;
- u8 cq_period_mode;
-};
-
struct mlx5e_params {
u8 log_sq_size;
u8 rq_wq_type;
@@ -473,32 +468,6 @@ struct mlx5e_mpw_info {
u16 skbs_frags[MLX5_MPWRQ_PAGES_PER_WQE];
};
-struct mlx5e_rx_am_stats {
- int ppms; /* packets per msec */
- int bpms; /* bytes per msec */
- int epms; /* events per msec */
-};
-
-struct mlx5e_rx_am_sample {
- ktime_t time;
- u32 pkt_ctr;
- u32 byte_ctr;
- u16 event_ctr;
-};
-
-struct mlx5e_rx_am { /* Adaptive Moderation */
- u8 state;
- struct mlx5e_rx_am_stats prev_stats;
- struct mlx5e_rx_am_sample start_sample;
- struct work_struct work;
- u8 profile_ix;
- u8 mode;
- u8 tune_state;
- u8 steps_right;
- u8 steps_left;
- u8 tired;
-};
-
/* a single cache unit is capable to serve one napi call (for non-striding rq)
* or a MPWQE (for striding rq).
*/
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
new file mode 100644
index 0000000..9eeaa11
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_dim.h
@@ -0,0 +1,69 @@
+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd. All rights reserved.
+ * Copyright (c) 2017-2018, Broadcom Limited
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef MLX5_AM_H
+#define MLX5_AM_H
+
+struct mlx5e_cq_moder {
+ u16 usec;
+ u16 pkts;
+ u8 cq_period_mode;
+};
+
+struct mlx5e_rx_am_sample {
+ ktime_t time;
+ u32 pkt_ctr;
+ u32 byte_ctr;
+ u16 event_ctr;
+};
+
+struct mlx5e_rx_am_stats {
+ int ppms; /* packets per msec */
+ int bpms; /* bytes per msec */
+ int epms; /* events per msec */
+};
+
+struct mlx5e_rx_am { /* Adaptive Moderation */
+ u8 state;
+ struct mlx5e_rx_am_stats prev_stats;
+ struct mlx5e_rx_am_sample start_sample;
+ struct work_struct work;
+ u8 profile_ix;
+ u8 mode;
+ u8 tune_state;
+ u8 steps_right;
+ u8 steps_left;
+ u8 tired;
+};
+
+#endif /* MLX5_AM_H */
--
2.7.4
^ permalink raw reply related
* [PATCH net-next v3 00/10] net: create dynamic software irq moderation library
From: Andy Gospodarek @ 2018-01-09 6:13 UTC (permalink / raw)
To: netdev; +Cc: mchan, talgi, ogerlitz, Andy Gospodarek
From: Andy Gospodarek <gospo@broadcom.com>
This converts the dynamic interrupt moderation library from the mlx5e
driver into a library so it can be used by any driver. The penultimate
patch in this set adds support for thiw new dynamic interrupt moderation
library in the bnxt_en driver and the last patch creates an entry in the
MAINTAINERS file for this library.
The main purpose of this code is to allow an administrator to make sure
that default coalesce settings are optimized for low latency, but
quickly adapt to handle high throughput/bulk traffic by altering how
much time passes before popping an interrupt.
For any new driver the following changes would be needed to use this
library:
- add elements in ring struct to track items needed by this library
- create function that can be called to actually set coalesce settings
for the driver
Credit to Rob Rice and Lee Reed for doing some of the initial proof of
concept and testing for this patch and Tal Gilboa and Or Gerlitz for
their comments, etc on this set.
v3: bnxt_en fix from Michael Chan, comment suggestion from Vasundhara
Volam, and small mlx5e header file fix from Tal Gilboa.
v2: Spelling fixes from Stephen Hemminger, bnxt_en suggestions from
Michael Chan, spelling and formatting fixes from Or Gerlitz, and
spelling and mlx5e changes suggested by Tal Gilboa.
Andy Gospodarek (10):
net/mlx5e: Move interrupt moderation structs to new file
net/mlx5e: Move interrupt moderation forward declarations
net/mlx5e: Remove rq references in mlx5e_rx_am
net/mlx5e: Move AM logic enums
net/mlx5e: Move generic functions to new file
net/mlx5e: Change Mellanox references in DIM code
net/mlx5e: Move dynamic interrupt coalescing code to include/linux
net/dim: use struct net_dim_sample as arg to net_dim
bnxt_en: add support for software dynamic interrupt moderation
MAINTAINERS: add entry for Dynamic Interrupt Moderation
MAINTAINERS | 5 +
drivers/net/ethernet/broadcom/bnxt/Makefile | 2 +-
drivers/net/ethernet/broadcom/bnxt/bnxt.c | 50 +++
drivers/net/ethernet/broadcom/bnxt/bnxt.h | 34 +-
drivers/net/ethernet/broadcom/bnxt/bnxt_dim.c | 33 ++
drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.c | 12 +
drivers/net/ethernet/mellanox/mlx5/core/Makefile | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en.h | 46 +--
drivers/net/ethernet/mellanox/mlx5/core/en_dim.c | 49 +++
.../net/ethernet/mellanox/mlx5/core/en_ethtool.c | 6 +-
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 40 ++-
drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c | 341 -------------------
drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c | 10 +-
include/linux/net_dim.h | 373 +++++++++++++++++++++
14 files changed, 593 insertions(+), 410 deletions(-)
create mode 100644 drivers/net/ethernet/broadcom/bnxt/bnxt_dim.c
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/en_dim.c
delete mode 100644 drivers/net/ethernet/mellanox/mlx5/core/en_rx_am.c
create mode 100644 include/linux/net_dim.h
--
2.7.4
^ permalink raw reply
* Re: WARNING: held lock freed!
From: Xin Long @ 2018-01-09 5:47 UTC (permalink / raw)
To: Marcelo Ricardo Leitner
Cc: syzbot, davem, LKML, linux-sctp, network dev, Neil Horman,
syzkaller-bugs, Vlad Yasevich
In-Reply-To: <20180108120952.GC727@localhost.localdomain>
On Mon, Jan 8, 2018 at 8:09 PM, Marcelo Ricardo Leitner
<marcelo.leitner@gmail.com> wrote:
> On Mon, Jan 08, 2018 at 06:01:14PM +0800, Xin Long wrote:
>> On Mon, Jan 8, 2018 at 6:58 AM, syzbot
>> <syzbot+ac6ea7baa4432811eb50@syzkaller.appspotmail.com> wrote:
>> > Hello,
>> >
>> > syzkaller hit the following crash on
>> > 3219e264b984ec0a13923aa66385819c2e98d582
>> > git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/master
>> > compiler: gcc (GCC) 7.1.1 20170620
>> > .config is attached
>> > Raw console output is attached.
>> > C reproducer is attached
>> > syzkaller reproducer is attached. See https://goo.gl/kgGztJ
>> > for information about syzkaller reproducers
>> >
>> >
>> > IMPORTANT: if you fix the bug, please add the following tag to the commit:
>> > Reported-by: syzbot+ac6ea7baa4432811eb50@syzkaller.appspotmail.com
>> > It will help syzbot understand when the bug is fixed. See footer for
>> > details.
>> > If you forward the report, please keep this part and the footer.
>> >
>> >
>> > =========================
>> > WARNING: held lock freed!
>> > 4.15.0-rc6+ #250 Not tainted
>> > -------------------------
>> > syzkaller065230/3505 is freeing memory 000000007fcba654-00000000499fef26,
>> > with a lock still held there!
>> > (sk_lock-AF_INET6){+.+.}, at: [<00000000abe26b8e>] lock_sock
>> > include/net/sock.h:1461 [inline]
>> > (sk_lock-AF_INET6){+.+.}, at: [<00000000abe26b8e>]
>> > sctp_wait_for_sndbuf+0x509/0x8d0 net/sctp/socket.c:8042
>> > 1 lock held by syzkaller065230/3505:
>> > #0: (sk_lock-AF_INET6){+.+.}, at: [<00000000abe26b8e>] lock_sock
>> > include/net/sock.h:1461 [inline]
>> > #0: (sk_lock-AF_INET6){+.+.}, at: [<00000000abe26b8e>]
>> > sctp_wait_for_sndbuf+0x509/0x8d0 net/sctp/socket.c:8042
>> >
>> > stack backtrace:
>> > CPU: 0 PID: 3505 Comm: syzkaller065230 Not tainted 4.15.0-rc6+ #250
>> > Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
>> > Google 01/01/2011
>> > Call Trace:
>> > __dump_stack lib/dump_stack.c:17 [inline]
>> > dump_stack+0x194/0x257 lib/dump_stack.c:53
>> > print_freed_lock_bug kernel/locking/lockdep.c:4379 [inline]
>> > debug_check_no_locks_freed+0x32f/0x3c0 kernel/locking/lockdep.c:4412
>> > kmem_cache_free+0x68/0x2a0 mm/slab.c:3743
>> > sk_prot_free net/core/sock.c:1504 [inline]
>> > __sk_destruct+0x622/0x910 net/core/sock.c:1585
>> > sk_destruct+0x47/0x80 net/core/sock.c:1593
>> > __sk_free+0x57/0x230 net/core/sock.c:1601
>> > sk_free+0x2a/0x40 net/core/sock.c:1612
>> > sock_put include/net/sock.h:1656 [inline]
>> > sctp_association_destroy net/sctp/associola.c:424 [inline]
>> > sctp_association_put+0x14c/0x2f0 net/sctp/associola.c:883
>> > sctp_wait_for_sndbuf+0x673/0x8d0 net/sctp/socket.c:8053
>> > sctp_sendmsg+0x277d/0x3360 net/sctp/socket.c:1974
>> > inet_sendmsg+0x11f/0x5e0 net/ipv4/af_inet.c:763
>> > sock_sendmsg_nosec net/socket.c:636 [inline]
>> > sock_sendmsg+0xca/0x110 net/socket.c:646
>> > SYSC_sendto+0x361/0x5c0 net/socket.c:1727
>> > SyS_sendto+0x40/0x50 net/socket.c:1695
>> > entry_SYSCALL_64_fastpath+0x23/0x9a
>> > RIP: 0033:0x445db9
>> > RSP: 002b:00007f2467415d98 EFLAGS: 00000212 ORIG_RAX: 000000000000002c
>> > RAX: ffffffffffffffda RBX: 00000000006dbc84 RCX: 0000000000445db9
>> > RDX: 0000000000000001 RSI: 000000002010bf14 RDI: 0000000000000005
>> > RBP: 0000000000000000 R08: 00000000204d9000 R09: 000000000000001c
>> > R10: 0000000000000000 R11: 0000000000000212 R12: 00000000006dbc80
>> > R13: 00000000209a9000 R14: 0100000000000000 R15: 0000000000000001
>> > ==================================================================
>> > BUG: KASAN: use-after-free in debug_spin_lock_before
>> > kernel/locking/spinlock_debug.c:83 [inline]
>> > BUG: KASAN: use-after-free in do_raw_spin_lock+0x1e0/0x220
>> > kernel/locking/spinlock_debug.c:112
>> > Read of size 4 at addr ffff8801c17ab08c by task syzkaller065230/3505
>>
>> It seems this commit is buggy:
>> commit cea0cc80a6777beb6eb643d4ad53690e1ad1d4ff
>> Author: Xin Long <lucien.xin@gmail.com>
>> Date: Wed Nov 15 16:57:26 2017 +0800
>>
>> sctp: use the right sk after waking up from wait_buf sleep
>>
>> I already had a patch to drop it and fix the old issue with a better way.
>
> Maybe you can quickly share it, so we have an idea what you are
> talking about?
Sure, in sctp_wait_for_sndbuf().
lock_sock(sk);
- if (sk != asoc->base.sk) {
- release_sock(sk);
- sk = asoc->base.sk;
- lock_sock(sk);
- }
+ if (sk != asoc->base.sk)
+ goto do_error;
I had this patch when doing cleanup in sctp_sendmsg, it will also
make that cleanup easier.
Some comments for it:
After commit cea0cc80a677 ("sctp: use the right sk after waking up from
wait_buf sleep"), it may change to lock another sk if the asoc has been
peeled off in sctp_wait_for_sndbuf.
However, the asoc's new sk could be already closed elsewhere, as it's in
the sendmsg context of the old sk that can't avoid the new sk's closing.
If the sk's last one refcnt is held by this asoc, later on after putting
this asoc, the new sk will be freed, while under it's own lock.
This patch is to revert that commit, but fix the old issue by returning
error under the old sk's lock.
Note that it's safe to do that, as after this error returns, no more use on this
asoc in sctp_sendmsg from then on, and for the err path:
out_free:
if (new_asoc)
sctp_association_free(asoc);
This would never happen. So it's safe.
^ permalink raw reply
* [PATCH v2 bpf] bpf: introduce BPF_JIT_ALWAYS_ON config
From: Alexei Starovoitov @ 2018-01-09 4:52 UTC (permalink / raw)
To: davem; +Cc: daniel, torvalds, netdev, linux-kernel, kernel-team
The BPF interpreter has been used as part of the spectre 2 attack CVE-2017-5715.
A quote from goolge project zero blog:
"At this point, it would normally be necessary to locate gadgets in
the host kernel code that can be used to actually leak data by reading
from an attacker-controlled location, shifting and masking the result
appropriately and then using the result of that as offset to an
attacker-controlled address for a load. But piecing gadgets together
and figuring out which ones work in a speculation context seems annoying.
So instead, we decided to use the eBPF interpreter, which is built into
the host kernel - while there is no legitimate way to invoke it from inside
a VM, the presence of the code in the host kernel's text section is sufficient
to make it usable for the attack, just like with ordinary ROP gadgets."
To make attacker job harder introduce BPF_JIT_ALWAYS_ON config
option that removes interpreter from the kernel in favor of JIT-only mode.
So far eBPF JIT is supported by:
x64, arm64, arm32, sparc64, s390, powerpc64, mips64
The start of JITed program is randomized and code page is marked as read-only.
In addition "constant blinding" can be turned on with net.core.bpf_jit_harden
v1->v2:
- fix init order, test_bpf and cBPF (Daniel's feedback)
- fix offloaded bpf (Jakub's feedback)
- add 'return 0' dummy in case something can invoke prog->bpf_func
- retarget bpf tree. For bpf-next the patch would need one extra hunk.
It will be sent when the trees are merged back to net-next
Considered doing:
int bpf_jit_enable __read_mostly = BPF_EBPF_JIT_DEFAULT;
but it seems better to land the patch as-is and in bpf-next remove
bpf_jit_enable global variable from all JITs, consolidate in one place
and remove this jit_init() function.
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
---
init/Kconfig | 7 +++++++
kernel/bpf/core.c | 19 +++++++++++++++++++
lib/test_bpf.c | 11 +++++++----
net/core/filter.c | 6 ++----
net/core/sysctl_net_core.c | 6 ++++++
net/socket.c | 9 +++++++++
6 files changed, 50 insertions(+), 8 deletions(-)
diff --git a/init/Kconfig b/init/Kconfig
index 2934249fba46..5e2a4a391ba9 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -1392,6 +1392,13 @@ config BPF_SYSCALL
Enable the bpf() system call that allows to manipulate eBPF
programs and maps via file descriptors.
+config BPF_JIT_ALWAYS_ON
+ bool "Permanently enable BPF JIT and remove BPF interpreter"
+ depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
+ help
+ Enables BPF JIT and removes BPF interpreter to avoid
+ speculative execution of BPF instructions by the interpreter
+
config USERFAULTFD
bool "Enable userfaultfd() system call"
select ANON_INODES
diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c
index 86b50aa26ee8..b529982c3126 100644
--- a/kernel/bpf/core.c
+++ b/kernel/bpf/core.c
@@ -767,6 +767,7 @@ noinline u64 __bpf_call_base(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5)
}
EXPORT_SYMBOL_GPL(__bpf_call_base);
+#ifndef CONFIG_BPF_JIT_ALWAYS_ON
/**
* __bpf_prog_run - run eBPF program on a given context
* @ctx: is the data we are operating on
@@ -1317,6 +1318,8 @@ EVAL6(PROG_NAME_LIST, 224, 256, 288, 320, 352, 384)
EVAL4(PROG_NAME_LIST, 416, 448, 480, 512)
};
+#endif
+
bool bpf_prog_array_compatible(struct bpf_array *array,
const struct bpf_prog *fp)
{
@@ -1354,6 +1357,12 @@ static int bpf_check_tail_call(const struct bpf_prog *fp)
return 0;
}
+static unsigned int __bpf_prog_ret0(const void *ctx,
+ const struct bpf_insn *insn)
+{
+ return 0;
+}
+
/**
* bpf_prog_select_runtime - select exec runtime for BPF program
* @fp: bpf_prog populated with internal BPF program
@@ -1364,9 +1373,13 @@ static int bpf_check_tail_call(const struct bpf_prog *fp)
*/
struct bpf_prog *bpf_prog_select_runtime(struct bpf_prog *fp, int *err)
{
+#ifndef CONFIG_BPF_JIT_ALWAYS_ON
u32 stack_depth = max_t(u32, fp->aux->stack_depth, 1);
fp->bpf_func = interpreters[(round_up(stack_depth, 32) / 32) - 1];
+#else
+ fp->bpf_func = __bpf_prog_ret0;
+#endif
/* eBPF JITs can rewrite the program in case constant
* blinding is active. However, in case of error during
@@ -1376,6 +1389,12 @@ struct bpf_prog *bpf_prog_select_runtime(struct bpf_prog *fp, int *err)
*/
if (!bpf_prog_is_dev_bound(fp->aux)) {
fp = bpf_int_jit_compile(fp);
+#ifdef CONFIG_BPF_JIT_ALWAYS_ON
+ if (!fp->jited) {
+ *err = -ENOTSUPP;
+ return fp;
+ }
+#endif
} else {
*err = bpf_prog_offload_compile(fp);
if (*err)
diff --git a/lib/test_bpf.c b/lib/test_bpf.c
index 9e9748089270..f369889e521d 100644
--- a/lib/test_bpf.c
+++ b/lib/test_bpf.c
@@ -6250,9 +6250,8 @@ static struct bpf_prog *generate_filter(int which, int *err)
return NULL;
}
}
- /* We don't expect to fail. */
if (*err) {
- pr_cont("FAIL to attach err=%d len=%d\n",
+ pr_cont("FAIL to prog_create err=%d len=%d\n",
*err, fprog.len);
return NULL;
}
@@ -6276,6 +6275,10 @@ static struct bpf_prog *generate_filter(int which, int *err)
* checks.
*/
fp = bpf_prog_select_runtime(fp, err);
+ if (*err) {
+ pr_cont("FAIL to select_runtime err=%d\n", *err);
+ return NULL;
+ }
break;
}
@@ -6461,8 +6464,8 @@ static __init int test_bpf(void)
pass_cnt++;
continue;
}
-
- return err;
+ err_cnt++;
+ continue;
}
pr_cont("jited:%u ", fp->jited);
diff --git a/net/core/filter.c b/net/core/filter.c
index 6a85e67fafce..d339ef170df6 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -1054,11 +1054,9 @@ static struct bpf_prog *bpf_migrate_filter(struct bpf_prog *fp)
*/
goto out_err_free;
- /* We are guaranteed to never error here with cBPF to eBPF
- * transitions, since there's no issue with type compatibility
- * checks on program arrays.
- */
fp = bpf_prog_select_runtime(fp, &err);
+ if (err)
+ goto out_err_free;
kfree(old_prog);
return fp;
diff --git a/net/core/sysctl_net_core.c b/net/core/sysctl_net_core.c
index cbc3dde4cfcc..a47ad6cd41c0 100644
--- a/net/core/sysctl_net_core.c
+++ b/net/core/sysctl_net_core.c
@@ -325,7 +325,13 @@ static struct ctl_table net_core_table[] = {
.data = &bpf_jit_enable,
.maxlen = sizeof(int),
.mode = 0644,
+#ifndef CONFIG_BPF_JIT_ALWAYS_ON
.proc_handler = proc_dointvec
+#else
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = &one,
+ .extra2 = &one,
+#endif
},
# ifdef CONFIG_HAVE_EBPF_JIT
{
diff --git a/net/socket.c b/net/socket.c
index 05f361faec45..78acd6ce74c7 100644
--- a/net/socket.c
+++ b/net/socket.c
@@ -2619,6 +2619,15 @@ static int __init sock_init(void)
core_initcall(sock_init); /* early initcall */
+static int __init jit_init(void)
+{
+#ifdef CONFIG_BPF_JIT_ALWAYS_ON
+ bpf_jit_enable = 1;
+#endif
+ return 0;
+}
+pure_initcall(jit_init);
+
#ifdef CONFIG_PROC_FS
void socket_seq_show(struct seq_file *seq)
{
--
2.9.5
^ permalink raw reply related
* Re: [PATCH 16/18] net: mpls: prevent bounds-check bypass via speculative execution
From: Linus Torvalds @ 2018-01-09 4:21 UTC (permalink / raw)
To: Dan Williams
Cc: Eric W. Biederman, Linux Kernel Mailing List, linux-arch,
Peter Zijlstra, Netdev, Greg KH, Thomas Gleixner, David S. Miller,
Elena Reshetova, Alan Cox
In-Reply-To: <CA+55aFwDZ_K1uzuTq95hUXUVLFsCPCqGAMHpwa4PLCRvszmqkQ@mail.gmail.com>
On Mon, Jan 8, 2018 at 8:13 PM, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> # carry will be clear if idx >= max
> cmpq %idx,%max
Bah. Other way around.
cmpq %max,%idx
I'm a moron.
> # mask will be clear if carry was clear, ~0 otherwise
> sbbq %mask,%mask
>
> to generate mask directly. I might have screwed that up. Worth perhaps trying?
More importantly, worth _testing_ and fixing my hand-waving "asm like
this" crap.
But I do think that simple two-instruction cmpq/sbbq sequence could
get it right in just two trivial ALU instructions.
Linus
^ permalink raw reply
* Re: [PATCH 16/18] net: mpls: prevent bounds-check bypass via speculative execution
From: Linus Torvalds @ 2018-01-09 4:13 UTC (permalink / raw)
To: Dan Williams
Cc: Eric W. Biederman, Linux Kernel Mailing List, linux-arch,
Peter Zijlstra, Netdev, Greg KH, Thomas Gleixner, David S. Miller,
Elena Reshetova, Alan Cox
In-Reply-To: <CAPcyv4iErvcOOSkaQbMa=9OJCmxNO7sDqi3qzg2ODvKqCApULQ@mail.gmail.com>
On Mon, Jan 8, 2018 at 7:42 PM, Dan Williams <dan.j.williams@intel.com> wrote:
>
> originally from Linus and tweaked by Alexei and I:
Sadly, that tweak - while clever - is wrong.
> unsigned long _mask = ~(long)(_m - 1 - _i) >> BITS_PER_LONG - 1;\
Why?
Because "(long)(_m-1-_i)" is not negative just because "i >= m". It
can still be positive.
Think "m = 100", "i=bignum". The subtraction will overflow and become
positive again, the shift will shift to zero, and then the mask will
become ~0.
Now, you can fix it, but you need to be a tiny bit more clever. In
particular, just make sure that you retain the high bit of "_i",
basically making the rule be that a negative index is not ever valid.
And then instead of "(_m - 1 - _i)", you use "(_i | (_m - 1 - _i))".
Now the sign bit is set if _i had it set, _or_ if the subtraction
turned negative, and you don't have to worry about the overflow
situation.
But it does require that extra step to be trustworthy. Still purely
cheap arithmetic operations, although there is possibly some
additional register pressure there.
Somebody might be able to come up with something even more minimal (or
find a fault in my fix of your tweak).
Obviously, with architecture-specific code, you may well be able to do
better, using the carry flag of the subtraction.
For example, on x86, I think you could do it with just two instructions:
# carry will be clear if idx >= max
cmpq %idx,%max
# mask will be clear if carry was clear, ~0 otherwise
sbbq %mask,%mask
to generate mask directly. I might have screwed that up. Worth perhaps trying?
Linus
^ permalink raw reply
* Re: [PATCH 16/18] net: mpls: prevent bounds-check bypass via speculative execution
From: Dan Williams @ 2018-01-09 3:42 UTC (permalink / raw)
To: Eric W. Biederman
Cc: Linux Kernel Mailing List, linux-arch, Peter Zijlstra, Netdev,
Greg KH, Thomas Gleixner, Linus Torvalds, David S. Miller,
Elena Reshetova, Alan Cox
In-Reply-To: <87lgh7n2tf.fsf@xmission.com>
On Mon, Jan 8, 2018 at 7:11 PM, Eric W. Biederman <ebiederm@xmission.com> wrote:
> Dan Williams <dan.j.williams@intel.com> writes:
>
>> Static analysis reports that 'index' may be a user controlled value that
>> is used as a data dependency reading 'rt' from the 'platform_label'
>> array. In order to avoid potential leaks of kernel memory values, block
>> speculative execution of the instruction stream that could issue further
>> reads based on an invalid 'rt' value.
>
>
> In detail.
> a) This code is fast path packet forwarding code. Introducing an
> unconditional pipeline stall is not ok.
>
> AKA either there is no speculation and so this is invulnerable
> or there is speculation and you are creating an unconditional
> pipeline stall here.
>
> My back of the napkin caluculations say that a pipeline stall
> is about 20 cycles. Which is about the same length of time
> as a modern cache miss.
>
> On a good day this code will perform with 0 cache misses. On a less
> good day 1 cache miss. Which means you are quite possibly doubling
> the runtime of mpls_forward.
>
> b) The array is dynamically allocated which should provide some
> protection, as it will be more difficult to predict the address
> of the array which is needed to craft an malicious userspace value.
>
> c) The code can be trivially modified to say:
>
> static struct mpls_route *mpls_route_input_rcu(struct net *net, unsigned index)
> {
> struct mpls_route *rt = NULL;
>
> if (index < net->mpls.platform_labels) {
> struct mpls_route __rcu **platform_label =
> rcu_dereference(net->mpls.platform_label);
> rt = rcu_dereference(platform_label[index & ((1 << 20) - 1)]);
> }
> return rt;
> }
>
> AKA a static mask will ensure that there is not a primitive that can be
> used to access all of memory. That is max a 1 cycle slowdown in the
> code, which is a much better trade off.
>
> d) If we care more it is straight forward to modify
> resize_platform_label_table() to ensure that the size of the array
> is always a power of 2.
>
> e) The fact that a pointer is returned from the array and it is treated
> like a pointer would seem to provide a defense against the
> exfiltration technique of using the value read as an index into
> a small array, that user space code can probe aliased cached
> lines of, to see which value was dereferenced.
>
>
> So to this patch in particular.
> Nacked-by: "Eric W. Biederman" <ebiederm@xmission.com>
>
> This code path will be difficult to exploit. This change messes with
> performance. There are ways to make this code path useless while
> preserving the performance of the code.
>
Thanks, Eric understood. The discussion over the weekend came to the
conclusion that using a mask will be the default approach. The
nospec_array_ptr() will be defined to something similar to the
following, originally from Linus and tweaked by Alexei and I:
#define __nospec_array_ptr(base, idx, sz) \
({ \
union { typeof(&base[0]) _ptr; unsigned long _bit; } __u; \
unsigned long _i = (idx); \
unsigned long _m = (max); \
unsigned long _mask = ~(long)(_m - 1 - _i) >> BITS_PER_LONG - 1;\
OPTIMIZER_HIDE_VAR(_mask); \
__u._ptr = &base[_i & _mask]; \
__u._bit &= _mask; \
__u._ptr; \
})
Does that address your performance concerns?
^ permalink raw reply
* Re: [v2] net: gianfar_ptp: move set_fipers() to spinlock protecting area
From: Richard Cochran @ 2018-01-09 3:39 UTC (permalink / raw)
To: Yangbo Lu; +Cc: Fabio Estevam, Claudiu Manoil, netdev, linux-kernel
In-Reply-To: <20180109030233.16013-1-yangbo.lu@nxp.com>
On Tue, Jan 09, 2018 at 11:02:33AM +0800, Yangbo Lu wrote:
> set_fipers() calling should be protected by spinlock in
> case that any interrupt breaks related registers setting
> and the function we expect. This patch is to move set_fipers()
> to spinlock protecting area in ptp_gianfar_adjtime().
>
> Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Acked-by: Richard Cochran <richardcochran@gmail.com>
^ permalink raw reply
* Re: [PATCH] net: gianfar_ptp: move set_fipers() to spinlock protecting area
From: Richard Cochran @ 2018-01-09 3:39 UTC (permalink / raw)
To: Fabio Estevam; +Cc: Yangbo Lu, Claudiu Manoil, netdev, linux-kernel
In-Reply-To: <CAOMZO5CGvXbJtuOsnv6uU5pMUw44BC0wOm+QQpkO1vnZUiJEng@mail.gmail.com>
On Mon, Jan 08, 2018 at 10:53:40AM -0200, Fabio Estevam wrote:
> On Mon, Jan 8, 2018 at 8:13 AM, Yangbo Lu <yangbo.lu@nxp.com> wrote:
> > set_fipers() calling should be protected by spinlock.
> > This patch is to move set_fipers() to spinlock protecting
> > area in ptp_gianfar_adjtime() function.
>
> It would be nice to explay why.
Maybe this is important?
/* Caller must hold etsects->lock. */
static void set_fipers(struct etsects *etsects)
{
set_alarm(etsects);
gfar_write(&etsects->regs->tmr_fiper1, etsects->tmr_fiper1);
gfar_write(&etsects->regs->tmr_fiper2, etsects->tmr_fiper2);
}
Thanks,
Richard
^ permalink raw reply
* Re: [PATCH net-next 12/20] net: hns3: Add packet statistics of netdev
From: lipeng (Y) @ 2018-01-09 3:27 UTC (permalink / raw)
To: David Miller; +Cc: kubakici, netdev, linux-kernel, linuxarm, salil.mehta
In-Reply-To: <20180108.220627.2197261866078222474.davem@davemloft.net>
On 2018/1/9 11:06, David Miller wrote:
> From: "lipeng (Y)" <lipeng321@huawei.com>
> Date: Tue, 9 Jan 2018 10:48:04 +0800
>
>> So I think it is OK if you can revert [patch 12/20 ]("net: hns3: Add
>> packet statistics of netdev").
> I think it is OK if you send the revert patch, which is what I
> am asking for :-)
>
> .
sure, i will send the revert patch.
I have tested it in my local branch.
Thanks
Peng Li
>
^ permalink raw reply
* RE: [PATCH] net: gianfar_ptp: move set_fipers() to spinlock protecting area
From: Y.b. Lu @ 2018-01-09 3:23 UTC (permalink / raw)
To: Fabio Estevam
Cc: Claudiu Manoil, Richard Cochran, netdev@vger.kernel.org,
linux-kernel
In-Reply-To: <CAOMZO5CGvXbJtuOsnv6uU5pMUw44BC0wOm+QQpkO1vnZUiJEng@mail.gmail.com>
> -----Original Message-----
> From: Fabio Estevam [mailto:festevam@gmail.com]
> Sent: 2018年1月8日 20:54
> To: Y.b. Lu <yangbo.lu@nxp.com>
> Cc: Claudiu Manoil <claudiu.manoil@nxp.com>; Richard Cochran
> <richardcochran@gmail.com>; netdev@vger.kernel.org; linux-kernel
> <linux-kernel@vger.kernel.org>
> Subject: Re: [PATCH] net: gianfar_ptp: move set_fipers() to spinlock protecting
> area
>
> On Mon, Jan 8, 2018 at 8:13 AM, Yangbo Lu <yangbo.lu@nxp.com> wrote:
> > set_fipers() calling should be protected by spinlock.
> > This patch is to move set_fipers() to spinlock protecting area in
> > ptp_gianfar_adjtime() function.
>
> It would be nice to explay why.
[Y.b. Lu] Sent out v2 patch and explained in commit message :)
Thanks.
>
> Thanks
^ permalink raw reply
* [GIT] Networking
From: David Miller @ 2018-01-09 3:13 UTC (permalink / raw)
To: torvalds; +Cc: akpm, netdev, linux-kernel
Highlights:
1) Frag and UDP handling fixes in i40e driver, from Amritha Nambiar and
Alexander Duyck.
2) Undo unintentional UAPI change in netfilter conntrack, from Florian
Westphal.
3) Revert a change to how error codes are returned from
dev_get_valid_name(), it broke some apps.
4) Cannot cache routes for ipv6 tunnels in the tunnel is ipv4/ipv6
dual-stack. From Eli Cooper.
5) Fix missed PMTU updates in geneve, from Xin Long.
6) Cure double free in macvlan, from Gao Feng.
7) Fix heap out-of-bounds write in rds_message_alloc_sgs(), from
Mohamed Ghannam.
8) FEC bug fixes from FUgang Duan (mis-accounting of dev_id, missed
deferral of probe when the regulator is not ready yet).
9) Missing DMA mapping error checks in 3c59x, from Neil Horman.
10) Turn off Broadcom tags for some b53 switches, from Florian
Fainelli.
11) Fix OOPS when get_target_net() is passed an SKB whose NETLINK_CB()
isn't initialized. From Andrei Vagin.
12) Fix crashes in fib6_add(), from Wei Wang.
13) PMTU bug fixes in SCTP from Marcelo Ricardo Leitner.
Please pull, thanks a lot!
The following changes since commit 2758b3e3e630ba304fc4aca434d591e70e528298:
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net (2017-12-28 23:20:21 -0800)
are available in the Git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/davem/net.git
for you to fetch changes up to 50f3d740d376f664f6accc7e86c9afd8f1c7e1e4:
sh_eth: fix TXALCR1 offsets (2018-01-08 14:31:38 -0500)
----------------------------------------------------------------
Alexander Duyck (1):
i40e/i40evf: Account for frags split over multiple descriptors in check linearize
Amritha Nambiar (1):
i40e: Remove UDP support for big buffer
Andrei Vagin (1):
rtnetlink: give a user socket to get_target_net()
Arjun Vynipadath (1):
cxgb4: Fix FW flash errors
Benjamin Poirier (1):
e1000e: Fix e1000_check_for_copper_link_ich8lan return value.
Christophe JAILLET (1):
mdio-sun4i: Fix a memory leak
David S. Miller (10):
Revert "net: core: dev_get_valid_name is now the same as dev_alloc_name_ns"
Merge branch '1GbE' of git://git.kernel.org/.../jkirsher/net-queue
Merge branch 'fec-clean-up-in-the-cases-of-probe-error'
Merge branch 'ena-fixes'
Merge branch '40GbE' of git://git.kernel.org/.../jkirsher/net-queue
Merge tag 'mac80211-for-davem-2018-01-04' of git://git.kernel.org/.../jberg/mac80211
Merge git://git.kernel.org/.../pablo/nf
Merge tag 'linux-can-fixes-for-4.15-20180104' of git://git.kernel.org/.../mkl/linux-can
Merge branch 'bnxt_en_fixes'
Merge branch 'SCTP-PMTU-discovery-fixes'
Eduardo Otubo (1):
xen-netfront: enable device after manual module load
Eli Cooper (1):
ip6_tunnel: disable dst caching if tunnel is dual-stack
Felix Janda (1):
uapi libc compat: add fallback for unsupported libcs
Florian Fainelli (1):
net: dsa: b53: Turn off Broadcom tags for more switches
Florian Westphal (1):
netfilter: uapi: correct UNTRACKED conntrack state bit number
Fugang Duan (3):
net: fec: restore dev_id in the cases of probe error
net: fec: defer probe if regulator is not ready
net: fec: free/restore resource in related probe error pathes
Gao Feng (1):
macvlan: Fix one possible double free
Gustavo A. R. Silva (1):
phylink: mark expected switch fall-throughs in phylink_mii_ioctl
Hangbin Liu (1):
netfilter: nf_tables: fix potential NULL-ptr deref in nf_tables_dump_obj_done()
Hao Chen (1):
nl80211: Check for the required netlink attribute presence
Hauke Mehrtens (1):
uapi/if_ether.h: prevent redefinition of struct ethhdr
Ido Schimmel (2):
mlxsw: spectrum_router: Fix NULL pointer deref
mlxsw: spectrum: Relax sanity checks during enslavement
Jacob Keller (1):
i40e: don't remove netdev->dev_addr when syncing uc list
Jerome Brunet (1):
net: stmmac: enable EEE in MII, GMII or RGMII only
Jiri Pirko (1):
i40e: flower: Fix return value for unsupported offload
Johannes Berg (1):
mac80211: mesh: drop frames appearing to be from us
Jon Maloy (1):
tipc: fix problems with multipoint-to-point flow control
Luu An Phu (1):
can: flex_can: Correct the checking for frame length in flexcan_start_xmit()
Marcelo Ricardo Leitner (3):
sctp: fix error path in sctp_stream_init
sctp: do not retransmit upon FragNeeded if PMTU discovery is disabled
sctp: fix the handling of ICMP Frag Needed for too small MTUs
Martin Lederhilger (1):
can: ems_usb: improve error reporting for error warning and error passive
Mohamed Ghannam (2):
RDS: Heap OOB write in rds_message_alloc_sgs()
RDS: null pointer dereference in rds_atomic_free_op
Neil Horman (1):
3c59x: fix missing dma_mapping_error check and bad ring refill logic
Netanel Belgazal (2):
net: ena: unmask MSI-X only after device initialization is completed
net: ena: fix error handling in ena_down() sequence
Oliver Hartkopp (1):
can: vxcan: improve handling of missing peer name attribute
Pablo Neira Ayuso (1):
netfilter: nf_tables: fix chain filter in nf_tables_dump_rules()
Pravin B Shelar (1):
MAINTAINERS: Update my email address.
Roi Dayan (1):
net/sched: Fix update of lastuse in act modules implementing stats_update
Russell King (2):
phylink: ensure we report link down when LOS asserted
sfp: fix sfp-bus oops when removing socket/upstream
SZ Lin (林上智) (1):
NET: usb: qmi_wwan: add support for YUGA CLM920-NC5 PID 0x9625
Sergei Shtylyov (3):
sh_eth: fix TSU resource handling
sh_eth: fix SH7757 GEther initialization
sh_eth: fix TXALCR1 offsets
Stephen Hemminger (1):
ethtool: do not print warning for applications using legacy API
Sunil Challa (1):
bnxt_en: Fix population of flow_type in bnxt_hwrm_cfa_flow_alloc()
Tushar Dave (1):
e1000: fix disabling already-disabled warning
Venkat Duvvuru (1):
bnxt_en: Fix the 'Invalid VF' id check in bnxt_vf_ndo_prep routine.
Wei Wang (1):
ipv6: fix general protection fault in fib6_add()
William Tu (1):
vxlan: trivial indenting fix.
Wolfgang Grandegger (1):
can: gs_usb: fix return value of the "set_bittiming" callback
Xin Long (2):
geneve: update skb dst pmtu on tx path
ip6_tunnel: allow ip6gre dev mtu to be set below 1280
MAINTAINERS | 2 +-
drivers/net/can/flexcan.c | 2 +-
drivers/net/can/usb/ems_usb.c | 1 +
drivers/net/can/usb/gs_usb.c | 2 +-
drivers/net/can/vxcan.c | 2 +-
drivers/net/dsa/b53/b53_common.c | 9 ++++---
drivers/net/ethernet/3com/3c59x.c | 90 ++++++++++++++++++++++++++++--------------------------------------
drivers/net/ethernet/amazon/ena/ena_netdev.c | 45 ++++++++++++++++++++++-----------
drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c | 2 +-
drivers/net/ethernet/broadcom/bnxt/bnxt_tc.c | 2 +-
drivers/net/ethernet/chelsio/cxgb4/cxgb4.h | 1 -
drivers/net/ethernet/chelsio/cxgb4/t4_hw.c | 17 ++++++-------
drivers/net/ethernet/freescale/fec_main.c | 7 +++++-
drivers/net/ethernet/intel/e1000/e1000.h | 3 ++-
drivers/net/ethernet/intel/e1000/e1000_main.c | 27 ++++++++++++++++----
drivers/net/ethernet/intel/e1000e/ich8lan.c | 11 +++++---
drivers/net/ethernet/intel/i40e/i40e_main.c | 37 +++++++++++++++++++--------
drivers/net/ethernet/intel/i40e/i40e_txrx.c | 26 ++++++++++++++++---
drivers/net/ethernet/intel/i40evf/i40e_txrx.c | 26 ++++++++++++++++---
drivers/net/ethernet/mellanox/mlxsw/spectrum.c | 11 ++++++--
drivers/net/ethernet/mellanox/mlxsw/spectrum.h | 2 ++
drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c | 2 +-
drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c | 6 +++++
drivers/net/ethernet/renesas/sh_eth.c | 33 ++++++++++++++++++------
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 6 +++++
drivers/net/geneve.c | 14 +++++++++++
drivers/net/macvlan.c | 7 +++++-
drivers/net/phy/mdio-sun4i.c | 6 +++--
drivers/net/phy/phylink.c | 5 ++--
drivers/net/phy/sfp-bus.c | 6 +++--
drivers/net/usb/qmi_wwan.c | 1 +
drivers/net/xen-netfront.c | 1 +
include/linux/sh_eth.h | 1 -
include/net/sctp/structs.h | 2 +-
include/net/vxlan.h | 2 +-
include/uapi/linux/if_ether.h | 3 +++
include/uapi/linux/libc-compat.h | 61 ++++++++++++++++++++++++++++++++++++++++++++-
include/uapi/linux/netfilter/nf_conntrack_common.h | 2 +-
net/core/dev.c | 14 ++++++++++-
net/core/ethtool.c | 15 ++---------
net/core/rtnetlink.c | 10 ++++----
net/ipv6/ip6_fib.c | 35 +++++++++++++++-----------
net/ipv6/ip6_tunnel.c | 15 +++++------
net/mac80211/rx.c | 2 ++
net/netfilter/nf_tables_api.c | 8 +++---
net/rds/rdma.c | 4 +++
net/sched/act_gact.c | 2 +-
net/sched/act_mirred.c | 2 +-
net/sctp/input.c | 28 ++++++++++++---------
net/sctp/stream.c | 22 ++++++++--------
net/sctp/transport.c | 29 +++++++++++++--------
net/tipc/group.c | 22 +++++++++++++---
net/wireless/nl80211.c | 3 ++-
53 files changed, 474 insertions(+), 220 deletions(-)
^ permalink raw reply
* Re: [PATCH 16/18] net: mpls: prevent bounds-check bypass via speculative execution
From: Eric W. Biederman @ 2018-01-09 3:11 UTC (permalink / raw)
To: Dan Williams
Cc: linux-kernel, linux-arch, peterz, netdev, gregkh, tglx, torvalds,
David S. Miller, Elena Reshetova, alan
In-Reply-To: <151520108080.32271.16420298348259030860.stgit@dwillia2-desk3.amr.corp.intel.com>
Dan Williams <dan.j.williams@intel.com> writes:
> Static analysis reports that 'index' may be a user controlled value that
> is used as a data dependency reading 'rt' from the 'platform_label'
> array. In order to avoid potential leaks of kernel memory values, block
> speculative execution of the instruction stream that could issue further
> reads based on an invalid 'rt' value.
In detail.
a) This code is fast path packet forwarding code. Introducing an
unconditional pipeline stall is not ok.
AKA either there is no speculation and so this is invulnerable
or there is speculation and you are creating an unconditional
pipeline stall here.
My back of the napkin caluculations say that a pipeline stall
is about 20 cycles. Which is about the same length of time
as a modern cache miss.
On a good day this code will perform with 0 cache misses. On a less
good day 1 cache miss. Which means you are quite possibly doubling
the runtime of mpls_forward.
b) The array is dynamically allocated which should provide some
protection, as it will be more difficult to predict the address
of the array which is needed to craft an malicious userspace value.
c) The code can be trivially modified to say:
static struct mpls_route *mpls_route_input_rcu(struct net *net, unsigned index)
{
struct mpls_route *rt = NULL;
if (index < net->mpls.platform_labels) {
struct mpls_route __rcu **platform_label =
rcu_dereference(net->mpls.platform_label);
rt = rcu_dereference(platform_label[index & ((1 << 20) - 1)]);
}
return rt;
}
AKA a static mask will ensure that there is not a primitive that can be
used to access all of memory. That is max a 1 cycle slowdown in the
code, which is a much better trade off.
d) If we care more it is straight forward to modify
resize_platform_label_table() to ensure that the size of the array
is always a power of 2.
e) The fact that a pointer is returned from the array and it is treated
like a pointer would seem to provide a defense against the
exfiltration technique of using the value read as an index into
a small array, that user space code can probe aliased cached
lines of, to see which value was dereferenced.
So to this patch in particular.
Nacked-by: "Eric W. Biederman" <ebiederm@xmission.com>
This code path will be difficult to exploit. This change messes with
performance. There are ways to make this code path useless while
preserving the performance of the code.
Eric
>
> Based on an original patch by Elena Reshetova.
>
> Cc: "David S. Miller" <davem@davemloft.net>
> Cc: Eric W. Biederman <ebiederm@xmission.com>
> Cc: netdev@vger.kernel.org
> Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
> ---
> net/mpls/af_mpls.c | 12 +++++++-----
> 1 file changed, 7 insertions(+), 5 deletions(-)
>
> diff --git a/net/mpls/af_mpls.c b/net/mpls/af_mpls.c
> index 8ca9915befc8..ebcf0e246cfe 100644
> --- a/net/mpls/af_mpls.c
> +++ b/net/mpls/af_mpls.c
> @@ -8,6 +8,7 @@
> #include <linux/ipv6.h>
> #include <linux/mpls.h>
> #include <linux/netconf.h>
> +#include <linux/compiler.h>
> #include <linux/vmalloc.h>
> #include <linux/percpu.h>
> #include <net/ip.h>
> @@ -77,12 +78,13 @@ static void rtmsg_lfib(int event, u32 label, struct mpls_route *rt,
> static struct mpls_route *mpls_route_input_rcu(struct net *net, unsigned index)
> {
> struct mpls_route *rt = NULL;
> + struct mpls_route __rcu **platform_label =
> + rcu_dereference(net->mpls.platform_label);
> + struct mpls_route __rcu **rtp;
>
> - if (index < net->mpls.platform_labels) {
> - struct mpls_route __rcu **platform_label =
> - rcu_dereference(net->mpls.platform_label);
> - rt = rcu_dereference(platform_label[index]);
> - }
> + if ((rtp = nospec_array_ptr(platform_label, index,
> + net->mpls.platform_labels)))
> + rt = rcu_dereference(*rtp);
> return rt;
> }
>
^ permalink raw reply
* Re: [PATCH net-next 12/20] net: hns3: Add packet statistics of netdev
From: David Miller @ 2018-01-09 3:06 UTC (permalink / raw)
To: lipeng321; +Cc: kubakici, netdev, linux-kernel, linuxarm, salil.mehta
In-Reply-To: <937b5edd-8b05-56db-5510-6d09b0004a0b@huawei.com>
From: "lipeng (Y)" <lipeng321@huawei.com>
Date: Tue, 9 Jan 2018 10:48:04 +0800
> So I think it is OK if you can revert [patch 12/20 ]("net: hns3: Add
> packet statistics of netdev").
I think it is OK if you send the revert patch, which is what I
am asking for :-)
^ permalink raw reply
* Re: [PATCH net-next v2 1/6] net: Fix netdev_WARN_ONCE macro
From: David Miller @ 2018-01-09 3:05 UTC (permalink / raw)
To: joe; +Cc: galp, netdev, tariqt, saeedm
In-Reply-To: <1515465721.9619.78.camel@perches.com>
From: Joe Perches <joe@perches.com>
Date: Mon, 08 Jan 2018 18:42:01 -0800
> On Sun, 2018-01-07 at 12:08 +0200, Gal Pressman wrote:
>> netdev_WARN_ONCE is broken (whoops..), this fix will remove the
>> unnecessary "condition" parameter, add the missing comma and change
>> "arg" to "args".
>>
>> Fixes: 375ef2b1f0d0 ("net: Introduce netdev_*_once functions")
>> Signed-off-by: Gal Pressman <galp@mellanox.com>
>> Reviewed-by: Saeed Mahameed <saeedm@mellanox.com>
>> ---
>> include/linux/netdevice.h | 4 ++--
>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
>> index 352066e..5ff1ef9 100644
>> --- a/include/linux/netdevice.h
>> +++ b/include/linux/netdevice.h
>> @@ -4407,8 +4407,8 @@ do { \
>> WARN(1, "netdevice: %s%s\n" format, netdev_name(dev), \
>> netdev_reg_state(dev), ##args)
>>
>> -#define netdev_WARN_ONCE(dev, condition, format, arg...) \
>> - WARN_ONCE(1, "netdevice: %s%s\n" format, netdev_name(dev) \
>> +#define netdev_WARN_ONCE(dev, format, args...) \
>> + WARN_ONCE(1, "netdevice: %s%s\n" format, netdev_name(dev), \
>
> You sure you want the newline before the format?
Hmmm, Gal please send me a relative fix for this.
^ permalink raw reply
* Re: [PATCH v2] openvswitch: Trim off padding before L3+ netfilter processing
From: Ed Swierk @ 2018-01-09 3:02 UTC (permalink / raw)
To: Pravin Shelar; +Cc: ovs-dev, netdev, Benjamin Warren, Keith Holleman
In-Reply-To: <CAOrHB_A1sQWforWuUve5phrxPjSOpbTEQ4F5h3beh1C9qWmUTw@mail.gmail.com>
On 1/6/18 10:57, Pravin Shelar wrote:
> On Fri, Jan 5, 2018 at 10:59 PM, Ed Swierk <eswierk@skyportsystems.com> wrote:
>>
>>
>> On Jan 5, 2018 22:17, "Pravin Shelar" <pshelar@ovn.org> wrote:
>>
>> On Fri, Jan 5, 2018 at 3:20 PM, Ed Swierk <eswierk@skyportsystems.com>
>> wrote:
>>> On Fri, Jan 5, 2018 at 10:14 AM, Ed Swierk <eswierk@skyportsystems.com>
>>> wrote:
>>>> On Thu, Jan 4, 2018 at 7:36 PM, Pravin Shelar <pshelar@ovn.org> wrote:
>>>>> OVS already pull all required headers in skb linear data, so no need
>>>>> to redo all of it. only check required is the ip-checksum validation.
>>>>> I think we could avoid it in most of cases by checking skb length to
>>>>> ipheader length before verifying the ip header-checksum.
>>>>
>>>> Shouldn't the IP header checksum be verified even earlier, like in
>>>> key_extract(), before actually using any of the fields in the IP
>>>> header?
>>>
>>> Something like this for verifying the IP header checksum (not tested):
>>>
>> AFAIU openflow does not need this verification, so it is not required
>> in flow extract.
>>
>>
>> Okay. How about my proposed trimming implementation, caching the pad length
>> in the ovs cb?
>>
> Caching the length is not that simple, OVS actions can change the
> length. Keeping it consistent with packet would be more work, so lets
> calculate it in ovs-ct function.
>
Something like this?
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index a38c80e..282325d 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -4084,6 +4084,8 @@ struct sk_buff *skb_checksum_trimmed(struct sk_buff *skb,
unsigned int transport_len,
__sum16(*skb_chkf)(struct sk_buff *skb));
+int skb_network_trim(struct sk_buff *skb);
+
/**
* skb_head_is_locked - Determine if the skb->head is locked down
* @skb: skb to check
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 08f5740..c68e927 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -4740,6 +4740,41 @@ struct sk_buff *skb_checksum_trimmed(struct sk_buff *skb,
}
EXPORT_SYMBOL(skb_checksum_trimmed);
+/**
+ * skb_network_trim - trim skb to length specified by the network header
+ * @skb: the skb to trim
+ *
+ * Trims the skb to the length specified by the network header,
+ * removing any trailing padding. Leaves the skb alone if the protocol
+ * is not IP or IPv6. Frees the skb on error.
+ *
+ * Caller needs to pull the skb to the network header.
+ */
+int skb_network_trim(struct sk_buff *skb)
+{
+ unsigned int len;
+ int err;
+
+ switch (skb->protocol) {
+ case htons(ETH_P_IP):
+ len = ntohs(ip_hdr(skb)->tot_len);
+ break;
+ case htons(ETH_P_IPV6):
+ len = sizeof(struct ipv6hdr)
+ + ntohs(ipv6_hdr(skb)->payload_len);
+ break;
+ default:
+ len = skb->len;
+ }
+
+ err = pskb_trim_rcsum(skb, len);
+ if (unlikely(err))
+ kfree_skb(skb);
+
+ return err;
+}
+EXPORT_SYMBOL(skb_network_trim);
+
void __skb_warn_lro_forwarding(const struct sk_buff *skb)
{
net_warn_ratelimited("%s: received packets cannot be forwarded while LRO is enabled\n",
diff --git a/net/openvswitch/conntrack.c b/net/openvswitch/conntrack.c
index b27c5c6..73418d3 100644
--- a/net/openvswitch/conntrack.c
+++ b/net/openvswitch/conntrack.c
@@ -1112,6 +1112,10 @@ int ovs_ct_execute(struct net *net, struct sk_buff *skb,
nh_ofs = skb_network_offset(skb);
skb_pull_rcsum(skb, nh_ofs);
+ err = skb_network_trim(skb);
+ if (err)
+ return err;
+
if (key->ip.frag != OVS_FRAG_TYPE_NONE) {
err = handle_fragments(net, key, info->zone.id, skb);
if (err)
^ permalink raw reply related
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox