* [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23
@ 2017-05-23 11:43 Saeed Mahameed
2017-05-23 11:44 ` [for-next 2/6] net/mlx5: Update the list of the PCI supported devices Saeed Mahameed
` (4 more replies)
0 siblings, 5 replies; 72+ messages in thread
From: Saeed Mahameed @ 2017-05-23 11:43 UTC (permalink / raw)
To: David S. Miller, Doug Ledford
Cc: netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
Ilan Tayari, Saeed Mahameed
Hi Dave and Doug,
This series introduces some small updates and FPGA support to the mlx5
core/ethernet and IB drivers.
For more details please see below.
Please pull and let me know if there's any problem.
Thanks,
Saeed.
---
The following changes since commit 2ea659a9ef488125eb46da6eb571de5eae5c43f6:
Linux 4.12-rc1 (2017-05-13 13:19:49 -0700)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux.git tags/mlx5-update-2017-05-23
for you to fetch changes up to b359911d6608bd16888466184e8e8faeb63bd9eb:
IB/mlx5: Bump driver version (2017-05-14 14:24:19 +0300)
----------------------------------------------------------------
mlx5-update-2017-05-23
First patch from Leon, came to remove the redundant usage of mlx5_vzalloc,
and directly use kvzalloc across all mlx5 drivers.
2nd patch from Noa, adds new device IDs into the supported devices list.
3rd and 4th patches from Ilan are adding the basic infrastructure and
support for Mellanox's mlx5 FPGA.
Last two patches from Tariq came to modify the outdated driver version
reported in ethtool and in mlx5_ib to more reflect the current driver state
and remove the redundant date string reported in the version.
Thanks,
Saeed.
----------------------------------------------------------------
Ilan Tayari (2):
net/mlx5: Introduce trigger_health_work function
net/mlx5: FPGA, Add basic support for Innova
Leon Romanovsky (1):
{net, IB}/mlx5: Replace mlx5_vzalloc with kvzalloc
Noa Osherovich (1):
net/mlx5: Update the list of the PCI supported devices
Tariq Toukan (2):
net/mlx5: Bump driver version
IB/mlx5: Bump driver version
MAINTAINERS | 10 +
drivers/infiniband/hw/mlx5/cq.c | 6 +-
drivers/infiniband/hw/mlx5/mad.c | 4 +-
drivers/infiniband/hw/mlx5/main.c | 11 +-
drivers/infiniband/hw/mlx5/mr.c | 2 +-
drivers/infiniband/hw/mlx5/qp.c | 32 ++--
drivers/infiniband/hw/mlx5/srq.c | 4 +-
drivers/net/ethernet/mellanox/mlx5/core/Kconfig | 10 +
drivers/net/ethernet/mellanox/mlx5/core/Makefile | 3 +
drivers/net/ethernet/mellanox/mlx5/core/debugfs.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c | 8 +-
.../net/ethernet/mellanox/mlx5/core/en_common.c | 4 +-
.../net/ethernet/mellanox/mlx5/core/en_ethtool.c | 4 +-
drivers/net/ethernet/mellanox/mlx5/core/en_fs.c | 24 +--
.../ethernet/mellanox/mlx5/core/en_fs_ethtool.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 33 ++--
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/eq.c | 13 +-
drivers/net/ethernet/mellanox/mlx5/core/eswitch.c | 24 +--
.../ethernet/mellanox/mlx5/core/eswitch_offloads.c | 13 +-
drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.c | 64 +++++++
drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.h | 59 ++++++
.../net/ethernet/mellanox/mlx5/core/fpga/core.c | 202 +++++++++++++++++++++
.../net/ethernet/mellanox/mlx5/core/fpga/core.h | 99 ++++++++++
drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c | 6 +-
drivers/net/ethernet/mellanox/mlx5/core/fs_core.c | 10 +-
drivers/net/ethernet/mellanox/mlx5/core/health.c | 32 ++--
drivers/net/ethernet/mellanox/mlx5/core/ipoib.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/main.c | 21 ++-
.../net/ethernet/mellanox/mlx5/core/mlx5_core.h | 3 +-
.../net/ethernet/mellanox/mlx5/core/pagealloc.c | 4 +-
drivers/net/ethernet/mellanox/mlx5/core/port.c | 6 +-
drivers/net/ethernet/mellanox/mlx5/core/qp.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/srq.c | 14 +-
drivers/net/ethernet/mellanox/mlx5/core/transobj.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/vport.c | 34 ++--
include/linux/mlx5/device.h | 6 +
include/linux/mlx5/driver.h | 11 +-
include/linux/mlx5/mlx5_ifc.h | 11 +-
include/linux/mlx5/mlx5_ifc_fpga.h | 144 +++++++++++++++
40 files changed, 779 insertions(+), 164 deletions(-)
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.c
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.h
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/fpga/core.c
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/fpga/core.h
create mode 100644 include/linux/mlx5/mlx5_ifc_fpga.h
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* [for-next 1/6] {net, IB}/mlx5: Replace mlx5_vzalloc with kvzalloc
[not found] ` <20170523114404.20387-1-saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
@ 2017-05-23 11:43 ` Saeed Mahameed
2017-05-23 11:44 ` [for-next 6/6] IB/mlx5: Bump driver version Saeed Mahameed
2017-05-25 16:02 ` [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23 David Miller
2 siblings, 0 replies; 72+ messages in thread
From: Saeed Mahameed @ 2017-05-23 11:43 UTC (permalink / raw)
To: David S. Miller, Doug Ledford
Cc: netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
Ilan Tayari, Leon Romanovsky, Saeed Mahameed
From: Leon Romanovsky <leonro-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Commit a7c3e901a46f ("mm: introduce kv[mz]alloc helpers") added
proper implementation of mlx5_vzalloc function to the MM core.
This made the mlx5_vzalloc function useless, so let's remove it.
Signed-off-by: Leon Romanovsky <leonro-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Signed-off-by: Saeed Mahameed <saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---
drivers/infiniband/hw/mlx5/cq.c | 6 ++--
drivers/infiniband/hw/mlx5/mad.c | 4 +--
drivers/infiniband/hw/mlx5/main.c | 6 ++--
drivers/infiniband/hw/mlx5/mr.c | 2 +-
drivers/infiniband/hw/mlx5/qp.c | 32 ++++++++++----------
drivers/infiniband/hw/mlx5/srq.c | 4 +--
drivers/net/ethernet/mellanox/mlx5/core/debugfs.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c | 8 ++---
.../net/ethernet/mellanox/mlx5/core/en_common.c | 4 +--
.../net/ethernet/mellanox/mlx5/core/en_ethtool.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en_fs.c | 24 ++++++---------
.../ethernet/mellanox/mlx5/core/en_fs_ethtool.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 33 ++++++++++-----------
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/eq.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/eswitch.c | 24 ++++++---------
.../ethernet/mellanox/mlx5/core/eswitch_offloads.c | 13 ++++-----
drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c | 6 ++--
drivers/net/ethernet/mellanox/mlx5/core/fs_core.c | 10 +++----
drivers/net/ethernet/mellanox/mlx5/core/ipoib.c | 2 +-
.../net/ethernet/mellanox/mlx5/core/pagealloc.c | 4 +--
drivers/net/ethernet/mellanox/mlx5/core/port.c | 6 ++--
drivers/net/ethernet/mellanox/mlx5/core/qp.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/srq.c | 14 ++++-----
drivers/net/ethernet/mellanox/mlx5/core/transobj.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/vport.c | 34 +++++++++-------------
include/linux/mlx5/driver.h | 5 ----
27 files changed, 111 insertions(+), 144 deletions(-)
diff --git a/drivers/infiniband/hw/mlx5/cq.c b/drivers/infiniband/hw/mlx5/cq.c
index 94c049b62c2f..a384d72ea3cd 100644
--- a/drivers/infiniband/hw/mlx5/cq.c
+++ b/drivers/infiniband/hw/mlx5/cq.c
@@ -788,7 +788,7 @@ static int create_cq_user(struct mlx5_ib_dev *dev, struct ib_udata *udata,
*inlen = MLX5_ST_SZ_BYTES(create_cq_in) +
MLX5_FLD_SZ_BYTES(create_cq_in, pas[0]) * ncont;
- *cqb = mlx5_vzalloc(*inlen);
+ *cqb = kvzalloc(*inlen, GFP_KERNEL);
if (!*cqb) {
err = -ENOMEM;
goto err_db;
@@ -884,7 +884,7 @@ static int create_cq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_cq *cq,
*inlen = MLX5_ST_SZ_BYTES(create_cq_in) +
MLX5_FLD_SZ_BYTES(create_cq_in, pas[0]) * cq->buf.buf.npages;
- *cqb = mlx5_vzalloc(*inlen);
+ *cqb = kvzalloc(*inlen, GFP_KERNEL);
if (!*cqb) {
err = -ENOMEM;
goto err_buf;
@@ -1314,7 +1314,7 @@ int mlx5_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata)
inlen = MLX5_ST_SZ_BYTES(modify_cq_in) +
MLX5_FLD_SZ_BYTES(modify_cq_in, pas[0]) * npas;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in) {
err = -ENOMEM;
goto ex_resize;
diff --git a/drivers/infiniband/hw/mlx5/mad.c b/drivers/infiniband/hw/mlx5/mad.c
index f1b56de64871..95db929bdc34 100644
--- a/drivers/infiniband/hw/mlx5/mad.c
+++ b/drivers/infiniband/hw/mlx5/mad.c
@@ -218,7 +218,7 @@ static int process_pma_cmd(struct ib_device *ibdev, u8 port_num,
(struct ib_pma_portcounters_ext *)(out_mad->data + 40);
int sz = MLX5_ST_SZ_BYTES(query_vport_counter_out);
- out_cnt = mlx5_vzalloc(sz);
+ out_cnt = kvzalloc(sz, GFP_KERNEL);
if (!out_cnt)
return IB_MAD_RESULT_FAILURE;
@@ -231,7 +231,7 @@ static int process_pma_cmd(struct ib_device *ibdev, u8 port_num,
(struct ib_pma_portcounters *)(out_mad->data + 40);
int sz = MLX5_ST_SZ_BYTES(ppcnt_reg);
- out_cnt = mlx5_vzalloc(sz);
+ out_cnt = kvzalloc(sz, GFP_KERNEL);
if (!out_cnt)
return IB_MAD_RESULT_FAILURE;
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index d45772da0963..b6991204e5df 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -2263,7 +2263,7 @@ static struct mlx5_ib_flow_handler *create_flow_rule(struct mlx5_ib_dev *dev,
if (!is_valid_attr(dev->mdev, flow_attr))
return ERR_PTR(-EINVAL);
- spec = mlx5_vzalloc(sizeof(*spec));
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
handler = kzalloc(sizeof(*handler), GFP_KERNEL);
if (!handler || !spec) {
err = -ENOMEM;
@@ -3456,7 +3456,7 @@ static int mlx5_ib_query_q_counters(struct mlx5_ib_dev *dev,
__be32 val;
int ret, i;
- out = mlx5_vzalloc(outlen);
+ out = kvzalloc(outlen, GFP_KERNEL);
if (!out)
return -ENOMEM;
@@ -3485,7 +3485,7 @@ static int mlx5_ib_query_cong_counters(struct mlx5_ib_dev *dev,
int ret, i;
int offset = port->cnts.num_q_counters;
- out = mlx5_vzalloc(outlen);
+ out = kvzalloc(outlen, GFP_KERNEL);
if (!out)
return -ENOMEM;
diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index 366433f71b58..763bb5b36144 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -1110,7 +1110,7 @@ static struct mlx5_ib_mr *reg_create(struct ib_mr *ibmr, struct ib_pd *pd,
inlen = MLX5_ST_SZ_BYTES(create_mkey_in) +
sizeof(*pas) * ((npages + 1) / 2) * 2;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in) {
err = -ENOMEM;
goto err_1;
diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c
index 93959e1e43a3..d17aad0f54c0 100644
--- a/drivers/infiniband/hw/mlx5/qp.c
+++ b/drivers/infiniband/hw/mlx5/qp.c
@@ -823,7 +823,7 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd,
*inlen = MLX5_ST_SZ_BYTES(create_qp_in) +
MLX5_FLD_SZ_BYTES(create_qp_in, pas[0]) * ncont;
- *in = mlx5_vzalloc(*inlen);
+ *in = kvzalloc(*inlen, GFP_KERNEL);
if (!*in) {
err = -ENOMEM;
goto err_umem;
@@ -931,7 +931,7 @@ static int create_kernel_qp(struct mlx5_ib_dev *dev,
qp->sq.qend = mlx5_get_send_wqe(qp, qp->sq.wqe_cnt);
*inlen = MLX5_ST_SZ_BYTES(create_qp_in) +
MLX5_FLD_SZ_BYTES(create_qp_in, pas[0]) * qp->buf.npages;
- *in = mlx5_vzalloc(*inlen);
+ *in = kvzalloc(*inlen, GFP_KERNEL);
if (!*in) {
err = -ENOMEM;
goto err_buf;
@@ -1060,7 +1060,7 @@ static int create_raw_packet_qp_sq(struct mlx5_ib_dev *dev,
return err;
inlen = MLX5_ST_SZ_BYTES(create_sq_in) + sizeof(u64) * ncont;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in) {
err = -ENOMEM;
goto err_umem;
@@ -1140,7 +1140,7 @@ static int create_raw_packet_qp_rq(struct mlx5_ib_dev *dev,
u32 rq_pas_size = get_rq_pas_size(qpc);
inlen = MLX5_ST_SZ_BYTES(create_rq_in) + rq_pas_size;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -1193,7 +1193,7 @@ static int create_raw_packet_qp_tir(struct mlx5_ib_dev *dev,
int err;
inlen = MLX5_ST_SZ_BYTES(create_tir_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -1372,7 +1372,7 @@ static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp,
}
inlen = MLX5_ST_SZ_BYTES(create_tir_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -1633,7 +1633,7 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd,
if (err)
return err;
} else {
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -2164,7 +2164,7 @@ static int modify_raw_packet_eth_prio(struct mlx5_core_dev *dev,
int err;
inlen = MLX5_ST_SZ_BYTES(modify_tis_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -2189,7 +2189,7 @@ static int modify_raw_packet_tx_affinity(struct mlx5_core_dev *dev,
int err;
inlen = MLX5_ST_SZ_BYTES(modify_tis_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -2434,7 +2434,7 @@ static int modify_raw_packet_qp_rq(struct mlx5_ib_dev *dev,
int err;
inlen = MLX5_ST_SZ_BYTES(modify_rq_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -2479,7 +2479,7 @@ static int modify_raw_packet_qp_sq(struct mlx5_core_dev *dev,
int err;
inlen = MLX5_ST_SZ_BYTES(modify_sq_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -4294,7 +4294,7 @@ static int query_raw_packet_qp_sq_state(struct mlx5_ib_dev *dev,
int err;
inlen = MLX5_ST_SZ_BYTES(query_sq_out);
- out = mlx5_vzalloc(inlen);
+ out = kvzalloc(inlen, GFP_KERNEL);
if (!out)
return -ENOMEM;
@@ -4321,7 +4321,7 @@ static int query_raw_packet_qp_rq_state(struct mlx5_ib_dev *dev,
int err;
inlen = MLX5_ST_SZ_BYTES(query_rq_out);
- out = mlx5_vzalloc(inlen);
+ out = kvzalloc(inlen, GFP_KERNEL);
if (!out)
return -ENOMEM;
@@ -4625,7 +4625,7 @@ static int create_rq(struct mlx5_ib_rwq *rwq, struct ib_pd *pd,
dev = to_mdev(pd->device);
inlen = MLX5_ST_SZ_BYTES(create_rq_in) + sizeof(u64) * rwq->rq_num_pas;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -4855,7 +4855,7 @@ struct ib_rwq_ind_table *mlx5_ib_create_rwq_ind_table(struct ib_device *device,
return ERR_PTR(-ENOMEM);
inlen = MLX5_ST_SZ_BYTES(create_rqt_in) + sizeof(u32) * sz;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in) {
err = -ENOMEM;
goto err;
@@ -4934,7 +4934,7 @@ int mlx5_ib_modify_wq(struct ib_wq *wq, struct ib_wq_attr *wq_attr,
return -EOPNOTSUPP;
inlen = MLX5_ST_SZ_BYTES(modify_rq_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
diff --git a/drivers/infiniband/hw/mlx5/srq.c b/drivers/infiniband/hw/mlx5/srq.c
index 7cb145f9a6db..43707b101f47 100644
--- a/drivers/infiniband/hw/mlx5/srq.c
+++ b/drivers/infiniband/hw/mlx5/srq.c
@@ -127,7 +127,7 @@ static int create_srq_user(struct ib_pd *pd, struct mlx5_ib_srq *srq,
goto err_umem;
}
- in->pas = mlx5_vzalloc(sizeof(*in->pas) * ncont);
+ in->pas = kvzalloc(sizeof(*in->pas) * ncont, GFP_KERNEL);
if (!in->pas) {
err = -ENOMEM;
goto err_umem;
@@ -189,7 +189,7 @@ static int create_srq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_srq *srq,
}
mlx5_ib_dbg(dev, "srq->buf.page_shift = %d\n", srq->buf.page_shift);
- in->pas = mlx5_vzalloc(sizeof(*in->pas) * srq->buf.npages);
+ in->pas = kvzalloc(sizeof(*in->pas) * srq->buf.npages, GFP_KERNEL);
if (!in->pas) {
err = -ENOMEM;
goto err_buf;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c b/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c
index e94a9532e218..de40b6cfee95 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c
@@ -405,7 +405,7 @@ static u64 cq_read_field(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq,
u32 *out;
int err;
- out = mlx5_vzalloc(outlen);
+ out = kvzalloc(outlen, GFP_KERNEL);
if (!out)
return param;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c b/drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c
index c8a005326e30..f4017c06ddd2 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c
@@ -180,9 +180,8 @@ static int arfs_add_default_rule(struct mlx5e_priv *priv,
struct mlx5_flow_spec *spec;
int err = 0;
- spec = mlx5_vzalloc(sizeof(*spec));
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
if (!spec) {
- netdev_err(priv->netdev, "%s: alloc failed\n", __func__);
err = -ENOMEM;
goto out;
}
@@ -237,7 +236,7 @@ static int arfs_create_groups(struct mlx5e_flow_table *ft,
ft->g = kcalloc(MLX5E_ARFS_NUM_GROUPS,
sizeof(*ft->g), GFP_KERNEL);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in || !ft->g) {
kvfree(ft->g);
kvfree(in);
@@ -481,9 +480,8 @@ static struct mlx5_flow_handle *arfs_add_rule(struct mlx5e_priv *priv,
struct mlx5_flow_table *ft;
int err = 0;
- spec = mlx5_vzalloc(sizeof(*spec));
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
if (!spec) {
- netdev_err(priv->netdev, "%s: alloc failed\n", __func__);
err = -ENOMEM;
goto out;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_common.c b/drivers/net/ethernet/mellanox/mlx5/core/en_common.c
index f1f17f7a3cd0..46e56ec4c26f 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_common.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_common.c
@@ -65,7 +65,7 @@ static int mlx5e_create_mkey(struct mlx5_core_dev *mdev, u32 pdn,
u32 *in;
int err;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -147,7 +147,7 @@ int mlx5e_refresh_tirs(struct mlx5e_priv *priv, bool enable_uc_lb)
inlen = MLX5_ST_SZ_BYTES(modify_tir_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
goto out;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
index ce7b09d72ff6..e0dd1048c966 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
@@ -1045,7 +1045,7 @@ static int mlx5e_set_rxfh(struct net_device *dev, const u32 *indir,
(hfunc != ETH_RSS_HASH_TOP))
return -EINVAL;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_fs.c b/drivers/net/ethernet/mellanox/mlx5/core/en_fs.c
index 576d6787b484..936fc6d96c18 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_fs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_fs.c
@@ -218,11 +218,9 @@ static int mlx5e_add_vlan_rule(struct mlx5e_priv *priv,
struct mlx5_flow_spec *spec;
int err = 0;
- spec = mlx5_vzalloc(sizeof(*spec));
- if (!spec) {
- netdev_err(priv->netdev, "%s: alloc failed\n", __func__);
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+ if (!spec)
return -ENOMEM;
- }
if (rule_type == MLX5E_VLAN_RULE_TYPE_MATCH_VID)
mlx5e_vport_context_update_vlans(priv);
@@ -660,11 +658,9 @@ mlx5e_generate_ttc_rule(struct mlx5e_priv *priv,
struct mlx5_flow_spec *spec;
int err = 0;
- spec = mlx5_vzalloc(sizeof(*spec));
- if (!spec) {
- netdev_err(priv->netdev, "%s: alloc failed\n", __func__);
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+ if (!spec)
return ERR_PTR(-ENOMEM);
- }
if (proto) {
spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
@@ -742,7 +738,7 @@ static int mlx5e_create_ttc_table_groups(struct mlx5e_ttc_table *ttc)
sizeof(*ft->g), GFP_KERNEL);
if (!ft->g)
return -ENOMEM;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in) {
kfree(ft->g);
return -ENOMEM;
@@ -853,11 +849,9 @@ static int mlx5e_add_l2_flow_rule(struct mlx5e_priv *priv,
u8 *mc_dmac;
u8 *mv_dmac;
- spec = mlx5_vzalloc(sizeof(*spec));
- if (!spec) {
- netdev_err(priv->netdev, "%s: alloc failed\n", __func__);
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+ if (!spec)
return -ENOMEM;
- }
mc_dmac = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
outer_headers.dmac_47_16);
@@ -917,7 +911,7 @@ static int mlx5e_create_l2_table_groups(struct mlx5e_l2_table *l2_table)
ft->g = kcalloc(MLX5E_NUM_L2_GROUPS, sizeof(*ft->g), GFP_KERNEL);
if (!ft->g)
return -ENOMEM;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in) {
kfree(ft->g);
return -ENOMEM;
@@ -1072,7 +1066,7 @@ static int mlx5e_create_vlan_table_groups(struct mlx5e_flow_table *ft)
int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
int err;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_fs_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_fs_ethtool.c
index 85bf4a389295..bdd82c9b3992 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_fs_ethtool.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_fs_ethtool.c
@@ -296,7 +296,7 @@ add_ethtool_flow_rule(struct mlx5e_priv *priv,
struct mlx5_flow_handle *rule;
int err = 0;
- spec = mlx5_vzalloc(sizeof(*spec));
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
if (!spec)
return ERR_PTR(-ENOMEM);
err = set_flow_attrs(spec->match_criteria, spec->match_value,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index a61b71b6fff3..edc485e489cc 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -252,9 +252,9 @@ static void mlx5e_update_pport_counters(struct mlx5e_priv *priv)
void *out;
u32 *in;
- in = mlx5_vzalloc(sz);
+ in = kvzalloc(sz, GFP_KERNEL);
if (!in)
- goto free_out;
+ return;
MLX5_SET(ppcnt_reg, in, local_port, 1);
@@ -288,7 +288,6 @@ static void mlx5e_update_pport_counters(struct mlx5e_priv *priv)
MLX5_REG_PPCNT, 0, 0);
}
-free_out:
kvfree(in);
}
@@ -314,7 +313,7 @@ static void mlx5e_update_pcie_counters(struct mlx5e_priv *priv)
if (!MLX5_CAP_MCAM_FEATURE(mdev, pcie_performance_group))
return;
- in = mlx5_vzalloc(sz);
+ in = kvzalloc(sz, GFP_KERNEL);
if (!in)
return;
@@ -503,7 +502,7 @@ static int mlx5e_create_umr_mkey(struct mlx5_core_dev *mdev,
if (!MLX5E_VALID_NUM_MTTS(npages))
return -EINVAL;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -711,7 +710,7 @@ static int mlx5e_create_rq(struct mlx5e_rq *rq,
inlen = MLX5_ST_SZ_BYTES(create_rq_in) +
sizeof(u64) * rq->wq_ctrl.buf.npages;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -748,7 +747,7 @@ static int mlx5e_modify_rq_state(struct mlx5e_rq *rq, int curr_state,
int err;
inlen = MLX5_ST_SZ_BYTES(modify_rq_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -776,7 +775,7 @@ static int mlx5e_modify_rq_scatter_fcs(struct mlx5e_rq *rq, bool enable)
int err;
inlen = MLX5_ST_SZ_BYTES(modify_rq_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -805,7 +804,7 @@ static int mlx5e_modify_rq_vsd(struct mlx5e_rq *rq, bool vsd)
int err;
inlen = MLX5_ST_SZ_BYTES(modify_rq_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -1134,7 +1133,7 @@ static int mlx5e_create_sq(struct mlx5_core_dev *mdev,
inlen = MLX5_ST_SZ_BYTES(create_sq_in) +
sizeof(u64) * csp->wq_ctrl->buf.npages;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -1182,7 +1181,7 @@ static int mlx5e_modify_sq(struct mlx5_core_dev *mdev, u32 sqn,
int err;
inlen = MLX5_ST_SZ_BYTES(modify_sq_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -1496,7 +1495,7 @@ static int mlx5e_create_cq(struct mlx5e_cq *cq, struct mlx5e_cq_param *param)
inlen = MLX5_ST_SZ_BYTES(create_cq_in) +
sizeof(u64) * cq->wq_ctrl.frag_buf.npages;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -2091,7 +2090,7 @@ mlx5e_create_rqt(struct mlx5e_priv *priv, int sz, struct mlx5e_rqt *rqt)
int i;
inlen = MLX5_ST_SZ_BYTES(create_rqt_in) + sizeof(u32) * sz;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -2210,7 +2209,7 @@ int mlx5e_redirect_rqt(struct mlx5e_priv *priv, u32 rqtn, int sz,
int err;
inlen = MLX5_ST_SZ_BYTES(modify_rqt_in) + sizeof(u32) * sz;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -2433,7 +2432,7 @@ static int mlx5e_modify_tirs_lro(struct mlx5e_priv *priv)
int ix;
inlen = MLX5_ST_SZ_BYTES(modify_tir_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -2850,7 +2849,7 @@ int mlx5e_create_indirect_tirs(struct mlx5e_priv *priv)
int tt;
inlen = MLX5_ST_SZ_BYTES(create_tir_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -2889,7 +2888,7 @@ int mlx5e_create_direct_tirs(struct mlx5e_priv *priv)
int ix;
inlen = MLX5_ST_SZ_BYTES(create_tir_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 11c27e4fadf6..66a9bd635176 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -1738,7 +1738,7 @@ int mlx5e_configure_flower(struct mlx5e_priv *priv, __be16 protocol,
}
flow = kzalloc(sizeof(*flow) + attr_size, GFP_KERNEL);
- parse_attr = mlx5_vzalloc(sizeof(*parse_attr));
+ parse_attr = kvzalloc(sizeof(*parse_attr), GFP_KERNEL);
if (!parse_attr || !flow) {
err = -ENOMEM;
goto err_free;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eq.c b/drivers/net/ethernet/mellanox/mlx5/core/eq.c
index ea5d8d37a75c..df0034d8f48c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eq.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eq.c
@@ -548,7 +548,7 @@ int mlx5_create_map_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq, u8 vecidx,
inlen = MLX5_ST_SZ_BYTES(create_eq_in) +
MLX5_FLD_SZ_BYTES(create_eq_in, pas[0]) * eq->buf.npages;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in) {
err = -ENOMEM;
goto err_buf;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
index 2e34d95ea776..81dfcd90b1f5 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
@@ -248,11 +248,10 @@ __esw_fdb_set_vport_rule(struct mlx5_eswitch *esw, u32 vport, bool rx_rule,
if (rx_rule)
match_header |= MLX5_MATCH_MISC_PARAMETERS;
- spec = mlx5_vzalloc(sizeof(*spec));
- if (!spec) {
- esw_warn(esw->dev, "FDB: Failed to alloc match parameters\n");
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+ if (!spec)
return NULL;
- }
+
dmac_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,
outer_headers.dmac_47_16);
dmac_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
@@ -350,10 +349,9 @@ static int esw_create_legacy_fdb_table(struct mlx5_eswitch *esw, int nvports)
return -EOPNOTSUPP;
}
- flow_group_in = mlx5_vzalloc(inlen);
+ flow_group_in = kvzalloc(inlen, GFP_KERNEL);
if (!flow_group_in)
return -ENOMEM;
- memset(flow_group_in, 0, inlen);
table_size = BIT(MLX5_CAP_ESW_FLOWTABLE_FDB(dev, log_max_ft_size));
@@ -961,7 +959,7 @@ static int esw_vport_enable_egress_acl(struct mlx5_eswitch *esw,
return -EOPNOTSUPP;
}
- flow_group_in = mlx5_vzalloc(inlen);
+ flow_group_in = kvzalloc(inlen, GFP_KERNEL);
if (!flow_group_in)
return -ENOMEM;
@@ -1078,7 +1076,7 @@ static int esw_vport_enable_ingress_acl(struct mlx5_eswitch *esw,
return -EOPNOTSUPP;
}
- flow_group_in = mlx5_vzalloc(inlen);
+ flow_group_in = kvzalloc(inlen, GFP_KERNEL);
if (!flow_group_in)
return -ENOMEM;
@@ -1241,11 +1239,9 @@ static int esw_vport_ingress_config(struct mlx5_eswitch *esw,
"vport[%d] configure ingress rules, vlan(%d) qos(%d)\n",
vport->vport, vport->info.vlan, vport->info.qos);
- spec = mlx5_vzalloc(sizeof(*spec));
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
if (!spec) {
err = -ENOMEM;
- esw_warn(esw->dev, "vport[%d] configure ingress rules failed, err(%d)\n",
- vport->vport, err);
goto out;
}
@@ -1322,11 +1318,9 @@ static int esw_vport_egress_config(struct mlx5_eswitch *esw,
"vport[%d] configure egress rules, vlan(%d) qos(%d)\n",
vport->vport, vport->info.vlan, vport->info.qos);
- spec = mlx5_vzalloc(sizeof(*spec));
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
if (!spec) {
err = -ENOMEM;
- esw_warn(esw->dev, "vport[%d] configure egress rules failed, err(%d)\n",
- vport->vport, err);
goto out;
}
@@ -2158,7 +2152,7 @@ int mlx5_eswitch_get_vport_stats(struct mlx5_eswitch *esw,
if (!LEGAL_VPORT(esw, vport))
return -EINVAL;
- out = mlx5_vzalloc(outlen);
+ out = kvzalloc(outlen, GFP_KERNEL);
if (!out)
return -ENOMEM;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index f991f669047e..3795943ef2d1 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -311,9 +311,8 @@ mlx5_eswitch_add_send_to_vport_rule(struct mlx5_eswitch *esw, int vport, u32 sqn
struct mlx5_flow_spec *spec;
void *misc;
- spec = mlx5_vzalloc(sizeof(*spec));
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
if (!spec) {
- esw_warn(esw->dev, "FDB: Failed to alloc match parameters\n");
flow_rule = ERR_PTR(-ENOMEM);
goto out;
}
@@ -401,9 +400,8 @@ static int esw_add_fdb_miss_rule(struct mlx5_eswitch *esw)
struct mlx5_flow_spec *spec;
int err = 0;
- spec = mlx5_vzalloc(sizeof(*spec));
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
if (!spec) {
- esw_warn(esw->dev, "FDB: Failed to alloc match parameters\n");
err = -ENOMEM;
goto out;
}
@@ -488,7 +486,7 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
u32 *flow_group_in;
esw_debug(esw->dev, "Create offloads FDB Tables\n");
- flow_group_in = mlx5_vzalloc(inlen);
+ flow_group_in = kvzalloc(inlen, GFP_KERNEL);
if (!flow_group_in)
return -ENOMEM;
@@ -631,7 +629,7 @@ static int esw_create_vport_rx_group(struct mlx5_eswitch *esw)
int err = 0;
int nvports = priv->sriov.num_vfs + 2;
- flow_group_in = mlx5_vzalloc(inlen);
+ flow_group_in = kvzalloc(inlen, GFP_KERNEL);
if (!flow_group_in)
return -ENOMEM;
@@ -675,9 +673,8 @@ mlx5_eswitch_create_vport_rx_rule(struct mlx5_eswitch *esw, int vport, u32 tirn)
struct mlx5_flow_spec *spec;
void *misc;
- spec = mlx5_vzalloc(sizeof(*spec));
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
if (!spec) {
- esw_warn(esw->dev, "Failed to alloc match parameters\n");
flow_rule = ERR_PTR(-ENOMEM);
goto out;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
index 19e3d2fc2099..b27c59af9640 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
@@ -235,11 +235,9 @@ static int mlx5_cmd_set_fte(struct mlx5_core_dev *dev,
u32 *in;
int err;
- in = mlx5_vzalloc(inlen);
- if (!in) {
- mlx5_core_warn(dev, "failed to allocate inbox\n");
+ in = kvzalloc(inlen, GFP_KERNEL);
+ if (!in)
return -ENOMEM;
- }
MLX5_SET(set_fte_in, in, opcode, MLX5_CMD_OP_SET_FLOW_TABLE_ENTRY);
MLX5_SET(set_fte_in, in, op_mod, opmod);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
index b8a176503d38..20a50f23fb1b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
@@ -376,11 +376,9 @@ static void del_rule(struct fs_node *node)
int err;
bool update_fte = false;
- match_value = mlx5_vzalloc(match_len);
- if (!match_value) {
- mlx5_core_warn(dev, "failed to allocate inbox\n");
+ match_value = kvzalloc(match_len, GFP_KERNEL);
+ if (!match_value)
return;
- }
fs_get_obj(rule, node);
fs_get_obj(fte, rule->node.parent);
@@ -1159,7 +1157,7 @@ static struct mlx5_flow_group *create_autogroup(struct mlx5_flow_table *ft,
if (!ft->autogroup.active)
return ERR_PTR(-ENOENT);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return ERR_PTR(-ENOMEM);
@@ -1778,7 +1776,7 @@ static struct mlx5_flow_root_namespace *create_root_ns(struct mlx5_flow_steering
struct mlx5_flow_namespace *ns;
/* Create the root namespace */
- root_ns = mlx5_vzalloc(sizeof(*root_ns));
+ root_ns = kvzalloc(sizeof(*root_ns), GFP_KERNEL);
if (!root_ns)
return NULL;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/ipoib.c b/drivers/net/ethernet/mellanox/mlx5/core/ipoib.c
index 019c230da498..9b397fe3f159 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/ipoib.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/ipoib.c
@@ -98,7 +98,7 @@ static int mlx5i_create_underlay_qp(struct mlx5_core_dev *mdev, struct mlx5_core
void *qpc;
inlen = MLX5_ST_SZ_BYTES(create_qp_in);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c b/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c
index a57d5a81eb05..efcded7ca27a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c
@@ -279,7 +279,7 @@ static int give_pages(struct mlx5_core_dev *dev, u16 func_id, int npages,
int i;
inlen += npages * MLX5_FLD_SZ_BYTES(manage_pages_in, pas[0]);
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in) {
err = -ENOMEM;
mlx5_core_warn(dev, "vzalloc failed %d\n", inlen);
@@ -376,7 +376,7 @@ static int reclaim_pages(struct mlx5_core_dev *dev, u32 func_id, int npages,
*nclaimed = 0;
outlen += npages * MLX5_FLD_SZ_BYTES(manage_pages_out, pas[0]);
- out = mlx5_vzalloc(outlen);
+ out = kvzalloc(outlen, GFP_KERNEL);
if (!out)
return -ENOMEM;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/port.c b/drivers/net/ethernet/mellanox/mlx5/core/port.c
index 141583daf5a2..1975d4388d4f 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/port.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/port.c
@@ -47,8 +47,8 @@ int mlx5_core_access_reg(struct mlx5_core_dev *dev, void *data_in,
u32 *in = NULL;
void *data;
- in = mlx5_vzalloc(inlen);
- out = mlx5_vzalloc(outlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
+ out = kvzalloc(outlen, GFP_KERNEL);
if (!in || !out)
goto out;
@@ -454,7 +454,7 @@ int mlx5_core_query_ib_ppcnt(struct mlx5_core_dev *dev,
u32 *in;
int err;
- in = mlx5_vzalloc(sz);
+ in = kvzalloc(sz, GFP_KERNEL);
if (!in) {
err = -ENOMEM;
return err;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/qp.c b/drivers/net/ethernet/mellanox/mlx5/core/qp.c
index cbbcef2884be..573a6b27fed8 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/qp.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/qp.c
@@ -527,7 +527,7 @@ int mlx5_core_query_out_of_buffer(struct mlx5_core_dev *dev, u16 counter_id,
void *out;
int err;
- out = mlx5_vzalloc(outlen);
+ out = kvzalloc(outlen, GFP_KERNEL);
if (!out)
return -ENOMEM;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/srq.c b/drivers/net/ethernet/mellanox/mlx5/core/srq.c
index 3099630015d7..f774de6f5fcb 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/srq.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/srq.c
@@ -162,7 +162,7 @@ static int create_srq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq,
pas_size = get_pas_size(in);
inlen = MLX5_ST_SZ_BYTES(create_srq_in) + pas_size;
- create_in = mlx5_vzalloc(inlen);
+ create_in = kvzalloc(inlen, GFP_KERNEL);
if (!create_in)
return -ENOMEM;
@@ -221,7 +221,7 @@ static int query_srq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq,
void *srqc;
int err;
- srq_out = mlx5_vzalloc(MLX5_ST_SZ_BYTES(query_srq_out));
+ srq_out = kvzalloc(MLX5_ST_SZ_BYTES(query_srq_out), GFP_KERNEL);
if (!srq_out)
return -ENOMEM;
@@ -256,7 +256,7 @@ static int create_xrc_srq_cmd(struct mlx5_core_dev *dev,
pas_size = get_pas_size(in);
inlen = MLX5_ST_SZ_BYTES(create_xrc_srq_in) + pas_size;
- create_in = mlx5_vzalloc(inlen);
+ create_in = kvzalloc(inlen, GFP_KERNEL);
if (!create_in)
return -ENOMEM;
@@ -320,7 +320,7 @@ static int query_xrc_srq_cmd(struct mlx5_core_dev *dev,
void *xrc_srqc;
int err;
- xrcsrq_out = mlx5_vzalloc(MLX5_ST_SZ_BYTES(query_xrc_srq_out));
+ xrcsrq_out = kvzalloc(MLX5_ST_SZ_BYTES(query_xrc_srq_out), GFP_KERNEL);
if (!xrcsrq_out)
return -ENOMEM;
memset(xrcsrq_in, 0, sizeof(xrcsrq_in));
@@ -357,7 +357,7 @@ static int create_rmp_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq,
pas_size = get_pas_size(in);
inlen = MLX5_ST_SZ_BYTES(create_rmp_in) + pas_size;
- create_in = mlx5_vzalloc(inlen);
+ create_in = kvzalloc(inlen, GFP_KERNEL);
if (!create_in)
return -ENOMEM;
@@ -390,7 +390,7 @@ static int arm_rmp_cmd(struct mlx5_core_dev *dev,
void *bitmask;
int err;
- in = mlx5_vzalloc(MLX5_ST_SZ_BYTES(modify_rmp_in));
+ in = kvzalloc(MLX5_ST_SZ_BYTES(modify_rmp_in), GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -417,7 +417,7 @@ static int query_rmp_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq,
void *rmpc;
int err;
- rmp_out = mlx5_vzalloc(MLX5_ST_SZ_BYTES(query_rmp_out));
+ rmp_out = kvzalloc(MLX5_ST_SZ_BYTES(query_rmp_out), GFP_KERNEL);
if (!rmp_out)
return -ENOMEM;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/transobj.c b/drivers/net/ethernet/mellanox/mlx5/core/transobj.c
index a00ff49eec18..5e128d7a9ffd 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/transobj.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/transobj.c
@@ -284,7 +284,7 @@ int mlx5_core_arm_rmp(struct mlx5_core_dev *dev, u32 rmpn, u16 lwm)
void *bitmask;
int err;
- in = mlx5_vzalloc(MLX5_ST_SZ_BYTES(modify_rmp_in));
+ in = kvzalloc(MLX5_ST_SZ_BYTES(modify_rmp_in), GFP_KERNEL);
if (!in)
return -ENOMEM;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/vport.c b/drivers/net/ethernet/mellanox/mlx5/core/vport.c
index 15c2294dd2b4..06019d00ab7b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/vport.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/vport.c
@@ -172,7 +172,7 @@ int mlx5_query_nic_vport_mac_address(struct mlx5_core_dev *mdev,
u8 *out_addr;
int err;
- out = mlx5_vzalloc(outlen);
+ out = kvzalloc(outlen, GFP_KERNEL);
if (!out)
return -ENOMEM;
@@ -197,11 +197,9 @@ int mlx5_modify_nic_vport_mac_address(struct mlx5_core_dev *mdev,
void *nic_vport_ctx;
u8 *perm_mac;
- in = mlx5_vzalloc(inlen);
- if (!in) {
- mlx5_core_warn(mdev, "failed to allocate inbox\n");
+ in = kvzalloc(inlen, GFP_KERNEL);
+ if (!in)
return -ENOMEM;
- }
MLX5_SET(modify_nic_vport_context_in, in,
field_select.permanent_address, 1);
@@ -231,7 +229,7 @@ int mlx5_query_nic_vport_mtu(struct mlx5_core_dev *mdev, u16 *mtu)
u32 *out;
int err;
- out = mlx5_vzalloc(outlen);
+ out = kvzalloc(outlen, GFP_KERNEL);
if (!out)
return -ENOMEM;
@@ -251,7 +249,7 @@ int mlx5_modify_nic_vport_mtu(struct mlx5_core_dev *mdev, u16 mtu)
void *in;
int err;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -501,7 +499,7 @@ int mlx5_query_nic_vport_system_image_guid(struct mlx5_core_dev *mdev,
u32 *out;
int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out);
- out = mlx5_vzalloc(outlen);
+ out = kvzalloc(outlen, GFP_KERNEL);
if (!out)
return -ENOMEM;
@@ -521,7 +519,7 @@ int mlx5_query_nic_vport_node_guid(struct mlx5_core_dev *mdev, u64 *node_guid)
u32 *out;
int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out);
- out = mlx5_vzalloc(outlen);
+ out = kvzalloc(outlen, GFP_KERNEL);
if (!out)
return -ENOMEM;
@@ -551,7 +549,7 @@ int mlx5_modify_nic_vport_node_guid(struct mlx5_core_dev *mdev,
if (!MLX5_CAP_ESW(mdev, nic_vport_node_guid_modify))
return -EOPNOTSUPP;
- in = mlx5_vzalloc(inlen);
+ in = kvzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@@ -577,7 +575,7 @@ int mlx5_query_nic_vport_qkey_viol_cntr(struct mlx5_core_dev *mdev,
u32 *out;
int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out);
- out = mlx5_vzalloc(outlen);
+ out = kvzalloc(outlen, GFP_KERNEL);
if (!out)
return -ENOMEM;
@@ -879,11 +877,9 @@ int mlx5_modify_nic_vport_promisc(struct mlx5_core_dev *mdev,
int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in);
int err;
- in = mlx5_vzalloc(inlen);
- if (!in) {
- mlx5_core_err(mdev, "failed to allocate inbox\n");
+ in = kvzalloc(inlen, GFP_KERNEL);
+ if (!in)
return -ENOMEM;
- }
MLX5_SET(modify_nic_vport_context_in, in, field_select.promisc, 1);
MLX5_SET(modify_nic_vport_context_in, in,
@@ -913,11 +909,9 @@ static int mlx5_nic_vport_update_roce_state(struct mlx5_core_dev *mdev,
int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in);
int err;
- in = mlx5_vzalloc(inlen);
- if (!in) {
- mlx5_core_warn(mdev, "failed to allocate inbox\n");
+ in = kvzalloc(inlen, GFP_KERNEL);
+ if (!in)
return -ENOMEM;
- }
MLX5_SET(modify_nic_vport_context_in, in, field_select.roce_en, 1);
MLX5_SET(modify_nic_vport_context_in, in, nic_vport_context.roce_en,
@@ -952,7 +946,7 @@ int mlx5_core_query_vport_counter(struct mlx5_core_dev *dev, u8 other_vport,
int err;
is_group_manager = MLX5_CAP_GEN(dev, vport_group_manager);
- in = mlx5_vzalloc(in_sz);
+ in = kvzalloc(in_sz, GFP_KERNEL);
if (!in) {
err = -ENOMEM;
return err;
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index bcdf739ee41a..c2740688d679 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -890,11 +890,6 @@ static inline u16 cmdif_rev(struct mlx5_core_dev *dev)
return ioread32be(&dev->iseg->cmdif_rev_fw_sub) >> 16;
}
-static inline void *mlx5_vzalloc(unsigned long size)
-{
- return kvzalloc(size, GFP_KERNEL);
-}
-
static inline u32 mlx5_base_mkey(const u32 key)
{
return key & 0xffffff00u;
--
2.11.0
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 72+ messages in thread
* [for-next 2/6] net/mlx5: Update the list of the PCI supported devices
2017-05-23 11:43 [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23 Saeed Mahameed
@ 2017-05-23 11:44 ` Saeed Mahameed
2017-05-23 11:44 ` [for-next 3/6] net/mlx5: Introduce trigger_health_work function Saeed Mahameed
` (3 subsequent siblings)
4 siblings, 0 replies; 72+ messages in thread
From: Saeed Mahameed @ 2017-05-23 11:44 UTC (permalink / raw)
To: David S. Miller, Doug Ledford
Cc: netdev, linux-rdma, Ilan Tayari, Noa Osherovich, Saeed Mahameed
From: Noa Osherovich <noaos@mellanox.com>
Add the BlueField device and VF IDs to the supported devices list.
Signed-off-by: Noa Osherovich <noaos@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/main.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c
index 0c123d571b4c..f933922d5cca 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c
@@ -1520,6 +1520,8 @@ static const struct pci_device_id mlx5_core_pci_table[] = {
{ PCI_VDEVICE(MELLANOX, 0x101a), MLX5_PCI_DEV_IS_VF}, /* ConnectX-5 Ex VF */
{ PCI_VDEVICE(MELLANOX, 0x101b) }, /* ConnectX-6 */
{ PCI_VDEVICE(MELLANOX, 0x101c), MLX5_PCI_DEV_IS_VF}, /* ConnectX-6 VF */
+ { PCI_VDEVICE(MELLANOX, 0xa2d2) }, /* BlueField integrated ConnectX-5 network controller */
+ { PCI_VDEVICE(MELLANOX, 0xa2d3), MLX5_PCI_DEV_IS_VF}, /* BlueField integrated ConnectX-5 network controller VF */
{ 0, }
};
--
2.11.0
^ permalink raw reply related [flat|nested] 72+ messages in thread
* [for-next 3/6] net/mlx5: Introduce trigger_health_work function
2017-05-23 11:43 [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23 Saeed Mahameed
2017-05-23 11:44 ` [for-next 2/6] net/mlx5: Update the list of the PCI supported devices Saeed Mahameed
@ 2017-05-23 11:44 ` Saeed Mahameed
2017-05-23 11:44 ` [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova Saeed Mahameed
` (2 subsequent siblings)
4 siblings, 0 replies; 72+ messages in thread
From: Saeed Mahameed @ 2017-05-23 11:44 UTC (permalink / raw)
To: David S. Miller, Doug Ledford
Cc: netdev, linux-rdma, Ilan Tayari, Saeed Mahameed
From: Ilan Tayari <ilant@mellanox.com>
Introduce new function for entering bad-health state.
This function will be called from FPGA-related logic in a later patch from
asynchronous event (IRQ) context, for that we change the spin lock to an
IRQ-safe one.
Signed-off-by: Ilan Tayari <ilant@mellanox.com>
Reviewed-by: Boris Pismenny <borisp@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/health.c | 32 ++++++++++++++++--------
include/linux/mlx5/driver.h | 1 +
2 files changed, 22 insertions(+), 11 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/health.c b/drivers/net/ethernet/mellanox/mlx5/core/health.c
index d0515391d33b..c3cedb6cec3f 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/health.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/health.c
@@ -185,6 +185,7 @@ static void health_care(struct work_struct *work)
struct mlx5_core_health *health;
struct mlx5_core_dev *dev;
struct mlx5_priv *priv;
+ unsigned long flags;
health = container_of(work, struct mlx5_core_health, work);
priv = container_of(health, struct mlx5_priv, health);
@@ -192,13 +193,13 @@ static void health_care(struct work_struct *work)
mlx5_core_warn(dev, "handling bad device here\n");
mlx5_handle_bad_state(dev);
- spin_lock(&health->wq_lock);
+ spin_lock_irqsave(&health->wq_lock, flags);
if (!test_bit(MLX5_DROP_NEW_HEALTH_WORK, &health->flags))
schedule_delayed_work(&health->recover_work, recover_delay);
else
dev_err(&dev->pdev->dev,
"new health works are not permitted at this stage\n");
- spin_unlock(&health->wq_lock);
+ spin_unlock_irqrestore(&health->wq_lock, flags);
}
static const char *hsynd_str(u8 synd)
@@ -269,6 +270,20 @@ static unsigned long get_next_poll_jiffies(void)
return next;
}
+void mlx5_trigger_health_work(struct mlx5_core_dev *dev)
+{
+ struct mlx5_core_health *health = &dev->priv.health;
+ unsigned long flags;
+
+ spin_lock_irqsave(&health->wq_lock, flags);
+ if (!test_bit(MLX5_DROP_NEW_HEALTH_WORK, &health->flags))
+ queue_work(health->wq, &health->work);
+ else
+ dev_err(&dev->pdev->dev,
+ "new health works are not permitted at this stage\n");
+ spin_unlock_irqrestore(&health->wq_lock, flags);
+}
+
static void poll_health(unsigned long data)
{
struct mlx5_core_dev *dev = (struct mlx5_core_dev *)data;
@@ -297,13 +312,7 @@ static void poll_health(unsigned long data)
if (in_fatal(dev) && !health->sick) {
health->sick = true;
print_health_info(dev);
- spin_lock(&health->wq_lock);
- if (!test_bit(MLX5_DROP_NEW_HEALTH_WORK, &health->flags))
- queue_work(health->wq, &health->work);
- else
- dev_err(&dev->pdev->dev,
- "new health works are not permitted at this stage\n");
- spin_unlock(&health->wq_lock);
+ mlx5_trigger_health_work(dev);
}
}
@@ -333,10 +342,11 @@ void mlx5_stop_health_poll(struct mlx5_core_dev *dev)
void mlx5_drain_health_wq(struct mlx5_core_dev *dev)
{
struct mlx5_core_health *health = &dev->priv.health;
+ unsigned long flags;
- spin_lock(&health->wq_lock);
+ spin_lock_irqsave(&health->wq_lock, flags);
set_bit(MLX5_DROP_NEW_HEALTH_WORK, &health->flags);
- spin_unlock(&health->wq_lock);
+ spin_unlock_irqrestore(&health->wq_lock, flags);
cancel_delayed_work_sync(&health->recover_work);
cancel_work_sync(&health->work);
}
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index c2740688d679..a277bb36c21f 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -915,6 +915,7 @@ int mlx5_health_init(struct mlx5_core_dev *dev);
void mlx5_start_health_poll(struct mlx5_core_dev *dev);
void mlx5_stop_health_poll(struct mlx5_core_dev *dev);
void mlx5_drain_health_wq(struct mlx5_core_dev *dev);
+void mlx5_trigger_health_work(struct mlx5_core_dev *dev);
int mlx5_buf_alloc_node(struct mlx5_core_dev *dev, int size,
struct mlx5_buf *buf, int node);
int mlx5_buf_alloc(struct mlx5_core_dev *dev, int size, struct mlx5_buf *buf);
--
2.11.0
^ permalink raw reply related [flat|nested] 72+ messages in thread
* [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-05-23 11:43 [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23 Saeed Mahameed
2017-05-23 11:44 ` [for-next 2/6] net/mlx5: Update the list of the PCI supported devices Saeed Mahameed
2017-05-23 11:44 ` [for-next 3/6] net/mlx5: Introduce trigger_health_work function Saeed Mahameed
@ 2017-05-23 11:44 ` Saeed Mahameed
[not found] ` <20170523114404.20387-5-saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
2017-05-26 3:58 ` please revert. Was: " Alexei Starovoitov
2017-05-23 11:44 ` [for-next 5/6] net/mlx5: Bump driver version Saeed Mahameed
[not found] ` <20170523114404.20387-1-saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
4 siblings, 2 replies; 72+ messages in thread
From: Saeed Mahameed @ 2017-05-23 11:44 UTC (permalink / raw)
To: David S. Miller, Doug Ledford
Cc: netdev, linux-rdma, Ilan Tayari, Saeed Mahameed
From: Ilan Tayari <ilant@mellanox.com>
Mellanox Innova is a NIC with ConnectX and an FPGA on the same
board. The FPGA is a bump-on-the-wire and thus affects operation of
the mlx5_core driver on the ConnectX ASIC.
Add basic support for Innova in mlx5_core.
This allows using the Innova card as a regular NIC, by detecting
the FPGA capability bit, and verifying its load state before
initializing ConnectX interfaces.
Also detect FPGA fatal runtime failures and enter error state if
they ever happen.
All new FPGA-related logic is placed in its own subdirectory 'fpga',
which may be built by selecting CONFIG_MLX5_FPGA.
This prepares for further support of various Innova features in later
patchsets.
Additional details about hardware architecture will be provided as
more features get submitted.
Signed-off-by: Ilan Tayari <ilant@mellanox.com>
Reviewed-by: Boris Pismenny <borisp@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
MAINTAINERS | 10 +
drivers/net/ethernet/mellanox/mlx5/core/Kconfig | 10 +
drivers/net/ethernet/mellanox/mlx5/core/Makefile | 3 +
drivers/net/ethernet/mellanox/mlx5/core/eq.c | 11 ++
drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.c | 64 +++++++
drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.h | 59 ++++++
.../net/ethernet/mellanox/mlx5/core/fpga/core.c | 202 +++++++++++++++++++++
.../net/ethernet/mellanox/mlx5/core/fpga/core.h | 99 ++++++++++
drivers/net/ethernet/mellanox/mlx5/core/main.c | 19 +-
include/linux/mlx5/device.h | 6 +
include/linux/mlx5/driver.h | 5 +
include/linux/mlx5/mlx5_ifc.h | 11 +-
include/linux/mlx5/mlx5_ifc_fpga.h | 144 +++++++++++++++
13 files changed, 640 insertions(+), 3 deletions(-)
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.c
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.h
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/fpga/core.c
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/fpga/core.h
create mode 100644 include/linux/mlx5/mlx5_ifc_fpga.h
diff --git a/MAINTAINERS b/MAINTAINERS
index f7d568b8f133..374ebf1b5d5d 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -8304,6 +8304,16 @@ W: http://www.mellanox.com
Q: http://patchwork.ozlabs.org/project/netdev/list/
F: drivers/net/ethernet/mellanox/mlx5/core/en_*
+MELLANOX ETHERNET INNOVA DRIVER
+M: Ilan Tayari <ilant@mellanox.com>
+R: Boris Pismenny <borisp@mellanox.com>
+L: netdev@vger.kernel.org
+S: Supported
+W: http://www.mellanox.com
+Q: http://patchwork.ozlabs.org/project/netdev/list/
+F: drivers/net/ethernet/mellanox/mlx5/core/fpga/*
+F: include/linux/mlx5/mlx5_ifc_fpga.h
+
MELLANOX ETHERNET SWITCH DRIVERS
M: Jiri Pirko <jiri@mellanox.com>
M: Ido Schimmel <idosch@mellanox.com>
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Kconfig b/drivers/net/ethernet/mellanox/mlx5/core/Kconfig
index fc52d742b7f7..28cf88483ca4 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/Kconfig
+++ b/drivers/net/ethernet/mellanox/mlx5/core/Kconfig
@@ -11,6 +11,16 @@ config MLX5_CORE
Core driver for low level functionality of the ConnectX-4 and
Connect-IB cards by Mellanox Technologies.
+config MLX5_FPGA
+ bool "Mellanox Technologies Innova support"
+ depends on MLX5_CORE
+ ---help---
+ Build support for the Innova family of network cards by Mellanox
+ Technologies. Innova network cards are comprised of a ConnectX chip
+ and an FPGA chip on one board. If you select this option, the
+ mlx5_core driver will include the Innova FPGA core and allow building
+ sandbox-specific client drivers.
+
config MLX5_CORE_EN
bool "Mellanox Technologies ConnectX-4 Ethernet support"
depends on NETDEVICES && ETHERNET && PCI && MLX5_CORE
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Makefile b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
index 9e644615f07a..12556c03eec4 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/Makefile
+++ b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
@@ -1,10 +1,13 @@
obj-$(CONFIG_MLX5_CORE) += mlx5_core.o
+subdir-ccflags-y += -I$(src)
mlx5_core-y := main.o cmd.o debugfs.o fw.o eq.o uar.o pagealloc.o \
health.o mcg.o cq.o srq.o alloc.o qp.o port.o mr.o pd.o \
mad.o transobj.o vport.o sriov.o fs_cmd.o fs_core.o \
fs_counters.o rl.o lag.o dev.o
+mlx5_core-$(CONFIG_MLX5_FPGA) += fpga/cmd.o fpga/core.o
+
mlx5_core-$(CONFIG_MLX5_CORE_EN) += wq.o eswitch.o eswitch_offloads.o \
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_clock.o vxlan.o \
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eq.c b/drivers/net/ethernet/mellanox/mlx5/core/eq.c
index df0034d8f48c..01d2cd7e4746 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eq.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eq.c
@@ -35,6 +35,7 @@
#include <linux/mlx5/driver.h>
#include <linux/mlx5/cmd.h>
#include "mlx5_core.h"
+#include "fpga/core.h"
#ifdef CONFIG_MLX5_CORE_EN
#include "eswitch.h"
#endif
@@ -156,6 +157,8 @@ static const char *eqe_type_str(u8 type)
return "MLX5_EVENT_TYPE_PAGE_FAULT";
case MLX5_EVENT_TYPE_PPS_EVENT:
return "MLX5_EVENT_TYPE_PPS_EVENT";
+ case MLX5_EVENT_TYPE_FPGA_ERROR:
+ return "MLX5_EVENT_TYPE_FPGA_ERROR";
default:
return "Unrecognized event";
}
@@ -476,6 +479,11 @@ static irqreturn_t mlx5_eq_int(int irq, void *eq_ptr)
if (dev->event)
dev->event(dev, MLX5_DEV_EVENT_PPS, (unsigned long)eqe);
break;
+
+ case MLX5_EVENT_TYPE_FPGA_ERROR:
+ mlx5_fpga_event(dev, eqe->type, &eqe->data.raw);
+ break;
+
default:
mlx5_core_warn(dev, "Unhandled event 0x%x on EQ 0x%x\n",
eqe->type, eq->eqn);
@@ -693,6 +701,9 @@ int mlx5_start_eqs(struct mlx5_core_dev *dev)
if (MLX5_CAP_GEN(dev, pps))
async_event_mask |= (1ull << MLX5_EVENT_TYPE_PPS_EVENT);
+ if (MLX5_CAP_GEN(dev, fpga))
+ async_event_mask |= (1ull << MLX5_EVENT_TYPE_FPGA_ERROR);
+
err = mlx5_create_map_eq(dev, &table->cmd_eq, MLX5_EQ_VEC_CMD,
MLX5_NUM_CMD_EQE, 1ull << MLX5_EVENT_TYPE_CMD,
"mlx5_cmd_eq", MLX5_EQ_TYPE_ASYNC);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.c
new file mode 100644
index 000000000000..99cba644b4fc
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.c
@@ -0,0 +1,64 @@
+/*
+ * Copyright (c) 2017, 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 <linux/etherdevice.h>
+#include <linux/mlx5/cmd.h>
+#include <linux/mlx5/driver.h>
+
+#include "mlx5_core.h"
+#include "fpga/cmd.h"
+
+int mlx5_fpga_caps(struct mlx5_core_dev *dev, u32 *caps)
+{
+ u32 in[MLX5_ST_SZ_DW(fpga_cap)] = {0};
+
+ return mlx5_core_access_reg(dev, in, sizeof(in), caps,
+ MLX5_ST_SZ_BYTES(fpga_cap),
+ MLX5_REG_FPGA_CAP, 0, 0);
+}
+
+int mlx5_fpga_query(struct mlx5_core_dev *dev, struct mlx5_fpga_query *query)
+{
+ u32 in[MLX5_ST_SZ_DW(fpga_ctrl)] = {0};
+ u32 out[MLX5_ST_SZ_DW(fpga_ctrl)];
+ int err;
+
+ err = mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out),
+ MLX5_REG_FPGA_CTRL, 0, false);
+ if (err)
+ return err;
+
+ query->status = MLX5_GET(fpga_ctrl, out, status);
+ query->admin_image = MLX5_GET(fpga_ctrl, out, flash_select_admin);
+ query->oper_image = MLX5_GET(fpga_ctrl, out, flash_select_oper);
+ return 0;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.h b/drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.h
new file mode 100644
index 000000000000..a74396a61bc3
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.h
@@ -0,0 +1,59 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies, Ltd. 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_FPGA_H__
+#define __MLX5_FPGA_H__
+
+#include <linux/mlx5/driver.h>
+
+enum mlx5_fpga_image {
+ MLX5_FPGA_IMAGE_USER = 0,
+ MLX5_FPGA_IMAGE_FACTORY,
+};
+
+enum mlx5_fpga_status {
+ MLX5_FPGA_STATUS_SUCCESS = 0,
+ MLX5_FPGA_STATUS_FAILURE = 1,
+ MLX5_FPGA_STATUS_IN_PROGRESS = 2,
+ MLX5_FPGA_STATUS_NONE = 0xFFFF,
+};
+
+struct mlx5_fpga_query {
+ enum mlx5_fpga_image admin_image;
+ enum mlx5_fpga_image oper_image;
+ enum mlx5_fpga_status status;
+};
+
+int mlx5_fpga_caps(struct mlx5_core_dev *dev, u32 *caps);
+int mlx5_fpga_query(struct mlx5_core_dev *dev, struct mlx5_fpga_query *query);
+
+#endif /* __MLX5_FPGA_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.c b/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.c
new file mode 100644
index 000000000000..d88b332e9669
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.c
@@ -0,0 +1,202 @@
+/*
+ * Copyright (c) 2017, 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 <linux/module.h>
+#include <linux/etherdevice.h>
+#include <linux/mlx5/driver.h>
+
+#include "mlx5_core.h"
+#include "fpga/core.h"
+
+static const char *const mlx5_fpga_error_strings[] = {
+ "Null Syndrome",
+ "Corrupted DDR",
+ "Flash Timeout",
+ "Internal Link Error",
+ "Watchdog HW Failure",
+ "I2C Failure",
+ "Image Changed",
+ "Temperature Critical",
+};
+
+static struct mlx5_fpga_device *mlx5_fpga_device_alloc(void)
+{
+ struct mlx5_fpga_device *fdev = NULL;
+
+ fdev = kzalloc(sizeof(*fdev), GFP_KERNEL);
+ if (!fdev)
+ return NULL;
+
+ spin_lock_init(&fdev->state_lock);
+ fdev->state = MLX5_FPGA_STATUS_NONE;
+ return fdev;
+}
+
+static const char *mlx5_fpga_image_name(enum mlx5_fpga_image image)
+{
+ switch (image) {
+ case MLX5_FPGA_IMAGE_USER:
+ return "user";
+ case MLX5_FPGA_IMAGE_FACTORY:
+ return "factory";
+ default:
+ return "unknown";
+ }
+}
+
+static int mlx5_fpga_device_load_check(struct mlx5_fpga_device *fdev)
+{
+ struct mlx5_fpga_query query;
+ int err;
+
+ err = mlx5_fpga_query(fdev->mdev, &query);
+ if (err) {
+ mlx5_fpga_err(fdev, "Failed to query status: %d\n", err);
+ return err;
+ }
+
+ fdev->last_admin_image = query.admin_image;
+ fdev->last_oper_image = query.oper_image;
+
+ mlx5_fpga_dbg(fdev, "Status %u; Admin image %u; Oper image %u\n",
+ query.status, query.admin_image, query.oper_image);
+
+ if (query.status != MLX5_FPGA_STATUS_SUCCESS) {
+ mlx5_fpga_err(fdev, "%s image failed to load; status %u\n",
+ mlx5_fpga_image_name(fdev->last_oper_image),
+ query.status);
+ return -EIO;
+ }
+
+ return 0;
+}
+
+int mlx5_fpga_device_start(struct mlx5_core_dev *mdev)
+{
+ struct mlx5_fpga_device *fdev = mdev->fpga;
+ unsigned long flags;
+ int err;
+
+ if (!fdev)
+ return 0;
+
+ err = mlx5_fpga_device_load_check(fdev);
+ if (err)
+ goto out;
+
+ err = mlx5_fpga_caps(fdev->mdev,
+ fdev->mdev->caps.hca_cur[MLX5_CAP_FPGA]);
+ if (err)
+ goto out;
+
+ mlx5_fpga_info(fdev, "device %u; %s image, version %u\n",
+ MLX5_CAP_FPGA(fdev->mdev, fpga_device),
+ mlx5_fpga_image_name(fdev->last_oper_image),
+ MLX5_CAP_FPGA(fdev->mdev, image_version));
+
+out:
+ spin_lock_irqsave(&fdev->state_lock, flags);
+ fdev->state = err ? MLX5_FPGA_STATUS_FAILURE : MLX5_FPGA_STATUS_SUCCESS;
+ spin_unlock_irqrestore(&fdev->state_lock, flags);
+ return err;
+}
+
+int mlx5_fpga_device_init(struct mlx5_core_dev *mdev)
+{
+ struct mlx5_fpga_device *fdev = NULL;
+
+ if (!MLX5_CAP_GEN(mdev, fpga)) {
+ mlx5_core_dbg(mdev, "FPGA capability not present\n");
+ return 0;
+ }
+
+ mlx5_core_dbg(mdev, "Initializing FPGA\n");
+
+ fdev = mlx5_fpga_device_alloc();
+ if (!fdev)
+ return -ENOMEM;
+
+ fdev->mdev = mdev;
+ mdev->fpga = fdev;
+
+ return 0;
+}
+
+void mlx5_fpga_device_cleanup(struct mlx5_core_dev *mdev)
+{
+ kfree(mdev->fpga);
+ mdev->fpga = NULL;
+}
+
+static const char *mlx5_fpga_syndrome_to_string(u8 syndrome)
+{
+ if (syndrome < ARRAY_SIZE(mlx5_fpga_error_strings))
+ return mlx5_fpga_error_strings[syndrome];
+ return "Unknown";
+}
+
+void mlx5_fpga_event(struct mlx5_core_dev *mdev, u8 event, void *data)
+{
+ struct mlx5_fpga_device *fdev = mdev->fpga;
+ const char *event_name;
+ bool teardown = false;
+ unsigned long flags;
+ u8 syndrome;
+
+ if (event != MLX5_EVENT_TYPE_FPGA_ERROR) {
+ mlx5_fpga_warn_ratelimited(fdev, "Unexpected event %u\n",
+ event);
+ return;
+ }
+
+ syndrome = MLX5_GET(fpga_error_event, data, syndrome);
+ event_name = mlx5_fpga_syndrome_to_string(syndrome);
+
+ spin_lock_irqsave(&fdev->state_lock, flags);
+ switch (fdev->state) {
+ case MLX5_FPGA_STATUS_SUCCESS:
+ mlx5_fpga_warn(fdev, "Error %u: %s\n", syndrome, event_name);
+ teardown = true;
+ break;
+ default:
+ mlx5_fpga_warn_ratelimited(fdev, "Unexpected error event %u: %s\n",
+ syndrome, event_name);
+ }
+ spin_unlock_irqrestore(&fdev->state_lock, flags);
+ /* We tear-down the card's interfaces and functionality because
+ * the FPGA bump-on-the-wire is misbehaving and we lose ability
+ * to communicate with the network. User may still be able to
+ * recover by re-programming or debugging the FPGA
+ */
+ if (teardown)
+ mlx5_trigger_health_work(fdev->mdev);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.h b/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.h
new file mode 100644
index 000000000000..c55044d66778
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.h
@@ -0,0 +1,99 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies, Ltd. 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_FPGA_CORE_H__
+#define __MLX5_FPGA_CORE_H__
+
+#ifdef CONFIG_MLX5_FPGA
+
+#include "fpga/cmd.h"
+
+/* Represents an Innova device */
+struct mlx5_fpga_device {
+ struct mlx5_core_dev *mdev;
+ spinlock_t state_lock; /* Protects state transitions */
+ enum mlx5_fpga_status state;
+ enum mlx5_fpga_image last_admin_image;
+ enum mlx5_fpga_image last_oper_image;
+};
+
+#define mlx5_fpga_dbg(__adev, format, ...) \
+ dev_dbg(&(__adev)->mdev->pdev->dev, "FPGA: %s:%d:(pid %d): " format, \
+ __func__, __LINE__, current->pid, ##__VA_ARGS__)
+
+#define mlx5_fpga_err(__adev, format, ...) \
+ dev_err(&(__adev)->mdev->pdev->dev, "FPGA: %s:%d:(pid %d): " format, \
+ __func__, __LINE__, current->pid, ##__VA_ARGS__)
+
+#define mlx5_fpga_warn(__adev, format, ...) \
+ dev_warn(&(__adev)->mdev->pdev->dev, "FPGA: %s:%d:(pid %d): " format, \
+ __func__, __LINE__, current->pid, ##__VA_ARGS__)
+
+#define mlx5_fpga_warn_ratelimited(__adev, format, ...) \
+ dev_warn_ratelimited(&(__adev)->mdev->pdev->dev, "FPGA: %s:%d: " \
+ format, __func__, __LINE__, ##__VA_ARGS__)
+
+#define mlx5_fpga_notice(__adev, format, ...) \
+ dev_notice(&(__adev)->mdev->pdev->dev, "FPGA: " format, ##__VA_ARGS__)
+
+#define mlx5_fpga_info(__adev, format, ...) \
+ dev_info(&(__adev)->mdev->pdev->dev, "FPGA: " format, ##__VA_ARGS__)
+
+int mlx5_fpga_device_init(struct mlx5_core_dev *mdev);
+void mlx5_fpga_device_cleanup(struct mlx5_core_dev *mdev);
+int mlx5_fpga_device_start(struct mlx5_core_dev *mdev);
+void mlx5_fpga_event(struct mlx5_core_dev *mdev, u8 event, void *data);
+
+#else
+
+static inline int mlx5_fpga_device_init(struct mlx5_core_dev *mdev)
+{
+ return 0;
+}
+
+static inline void mlx5_fpga_device_cleanup(struct mlx5_core_dev *mdev)
+{
+}
+
+static inline int mlx5_fpga_device_start(struct mlx5_core_dev *mdev)
+{
+ return 0;
+}
+
+static inline void mlx5_fpga_event(struct mlx5_core_dev *mdev, u8 event,
+ void *data)
+{
+}
+
+#endif
+
+#endif /* __MLX5_FPGA_CORE_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c
index f933922d5cca..ad0202cef203 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c
@@ -56,6 +56,7 @@
#ifdef CONFIG_MLX5_CORE_EN
#include "eswitch.h"
#endif
+#include "fpga/core.h"
MODULE_AUTHOR("Eli Cohen <eli@mellanox.com>");
MODULE_DESCRIPTION("Mellanox Connect-IB, ConnectX-4 core driver");
@@ -1113,10 +1114,16 @@ static int mlx5_load_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv,
goto err_disable_msix;
}
+ err = mlx5_fpga_device_init(dev);
+ if (err) {
+ dev_err(&pdev->dev, "fpga device init failed %d\n", err);
+ goto err_put_uars;
+ }
+
err = mlx5_start_eqs(dev);
if (err) {
dev_err(&pdev->dev, "Failed to start pages and async EQs\n");
- goto err_put_uars;
+ goto err_fpga_init;
}
err = alloc_comp_eqs(dev);
@@ -1147,6 +1154,12 @@ static int mlx5_load_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv,
goto err_sriov;
}
+ err = mlx5_fpga_device_start(dev);
+ if (err) {
+ dev_err(&pdev->dev, "fpga device start failed %d\n", err);
+ goto err_reg_dev;
+ }
+
if (mlx5_device_registered(dev)) {
mlx5_attach_device(dev);
} else {
@@ -1182,6 +1195,9 @@ static int mlx5_load_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv,
err_stop_eqs:
mlx5_stop_eqs(dev);
+err_fpga_init:
+ mlx5_fpga_device_cleanup(dev);
+
err_put_uars:
mlx5_put_uars_page(dev, priv->uar);
@@ -1246,6 +1262,7 @@ static int mlx5_unload_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv,
mlx5_irq_clear_affinity_hints(dev);
free_comp_eqs(dev);
mlx5_stop_eqs(dev);
+ mlx5_fpga_device_cleanup(dev);
mlx5_put_uars_page(dev, priv->uar);
mlx5_disable_msix(dev);
if (cleanup)
diff --git a/include/linux/mlx5/device.h b/include/linux/mlx5/device.h
index dd9a263ed368..786a43843da9 100644
--- a/include/linux/mlx5/device.h
+++ b/include/linux/mlx5/device.h
@@ -300,6 +300,8 @@ enum mlx5_event {
MLX5_EVENT_TYPE_PAGE_FAULT = 0xc,
MLX5_EVENT_TYPE_NIC_VPORT_CHANGE = 0xd,
+
+ MLX5_EVENT_TYPE_FPGA_ERROR = 0x20,
};
enum {
@@ -967,6 +969,7 @@ enum mlx5_cap_type {
MLX5_CAP_RESERVED,
MLX5_CAP_VECTOR_CALC,
MLX5_CAP_QOS,
+ MLX5_CAP_FPGA,
/* NUM OF CAP Types */
MLX5_CAP_NUM
};
@@ -1088,6 +1091,9 @@ enum mlx5_mcam_feature_groups {
#define MLX5_CAP_MCAM_FEATURE(mdev, fld) \
MLX5_GET(mcam_reg, (mdev)->caps.mcam, mng_feature_cap_mask.enhanced_features.fld)
+#define MLX5_CAP_FPGA(mdev, cap) \
+ MLX5_GET(fpga_cap, (mdev)->caps.hca_cur[MLX5_CAP_FPGA], cap)
+
enum {
MLX5_CMD_STAT_OK = 0x0,
MLX5_CMD_STAT_INT_ERR = 0x1,
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index a277bb36c21f..55bb712643cb 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -108,6 +108,8 @@ enum {
MLX5_REG_QTCT = 0x400a,
MLX5_REG_DCBX_PARAM = 0x4020,
MLX5_REG_DCBX_APP = 0x4021,
+ MLX5_REG_FPGA_CAP = 0x4022,
+ MLX5_REG_FPGA_CTRL = 0x4023,
MLX5_REG_PCAP = 0x5001,
MLX5_REG_PMTU = 0x5003,
MLX5_REG_PTYS = 0x5004,
@@ -761,6 +763,9 @@ struct mlx5_core_dev {
atomic_t num_qps;
u32 issi;
struct mlx5e_resources mlx5e_res;
+#ifdef CONFIG_MLX5_FPGA
+ struct mlx5_fpga_device *fpga;
+#endif
#ifdef CONFIG_RFS_ACCEL
struct cpu_rmap *rmap;
#endif
diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h
index 32de0724b400..6fa1eb6766af 100644
--- a/include/linux/mlx5/mlx5_ifc.h
+++ b/include/linux/mlx5/mlx5_ifc.h
@@ -32,6 +32,8 @@
#ifndef MLX5_IFC_H
#define MLX5_IFC_H
+#include "mlx5_ifc_fpga.h"
+
enum {
MLX5_EVENT_TYPE_CODING_COMPLETION_EVENTS = 0x0,
MLX5_EVENT_TYPE_CODING_PATH_MIGRATED_SUCCEEDED = 0x1,
@@ -56,7 +58,8 @@ enum {
MLX5_EVENT_TYPE_CODING_STALL_VL_EVENT = 0x1b,
MLX5_EVENT_TYPE_CODING_DROPPED_PACKET_LOGGED_EVENT = 0x1f,
MLX5_EVENT_TYPE_CODING_COMMAND_INTERFACE_COMPLETION = 0xa,
- MLX5_EVENT_TYPE_CODING_PAGE_REQUEST = 0xb
+ MLX5_EVENT_TYPE_CODING_PAGE_REQUEST = 0xb,
+ MLX5_EVENT_TYPE_CODING_FPGA_ERROR = 0x20,
};
enum {
@@ -854,7 +857,8 @@ struct mlx5_ifc_cmd_hca_cap_bits {
u8 max_tc[0x4];
u8 reserved_at_1d0[0x1];
u8 dcbx[0x1];
- u8 reserved_at_1d2[0x4];
+ u8 reserved_at_1d2[0x3];
+ u8 fpga[0x1];
u8 rol_s[0x1];
u8 rol_g[0x1];
u8 reserved_at_1d8[0x1];
@@ -2186,6 +2190,7 @@ union mlx5_ifc_hca_cap_union_bits {
struct mlx5_ifc_e_switch_cap_bits e_switch_cap;
struct mlx5_ifc_vector_calc_cap_bits vector_calc_cap;
struct mlx5_ifc_qos_cap_bits qos_cap;
+ struct mlx5_ifc_fpga_cap_bits fpga_cap;
u8 reserved_at_0[0x8000];
};
@@ -8182,6 +8187,8 @@ union mlx5_ifc_ports_control_registers_document_bits {
struct mlx5_ifc_sltp_reg_bits sltp_reg;
struct mlx5_ifc_mtpps_reg_bits mtpps_reg;
struct mlx5_ifc_mtppse_reg_bits mtppse_reg;
+ struct mlx5_ifc_fpga_ctrl_bits fpga_ctrl_bits;
+ struct mlx5_ifc_fpga_cap_bits fpga_cap_bits;
u8 reserved_at_0[0x60e0];
};
diff --git a/include/linux/mlx5/mlx5_ifc_fpga.h b/include/linux/mlx5/mlx5_ifc_fpga.h
new file mode 100644
index 000000000000..0032d10ac6cf
--- /dev/null
+++ b/include/linux/mlx5/mlx5_ifc_fpga.h
@@ -0,0 +1,144 @@
+/*
+ * Copyright (c) 2017, Mellanox Technologies, Ltd. 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_IFC_FPGA_H
+#define MLX5_IFC_FPGA_H
+
+struct mlx5_ifc_fpga_shell_caps_bits {
+ u8 max_num_qps[0x10];
+ u8 reserved_at_10[0x8];
+ u8 total_rcv_credits[0x8];
+
+ u8 reserved_at_20[0xe];
+ u8 qp_type[0x2];
+ u8 reserved_at_30[0x5];
+ u8 rae[0x1];
+ u8 rwe[0x1];
+ u8 rre[0x1];
+ u8 reserved_at_38[0x4];
+ u8 dc[0x1];
+ u8 ud[0x1];
+ u8 uc[0x1];
+ u8 rc[0x1];
+
+ u8 reserved_at_40[0x1a];
+ u8 log_ddr_size[0x6];
+
+ u8 max_fpga_qp_msg_size[0x20];
+
+ u8 reserved_at_80[0x180];
+};
+
+struct mlx5_ifc_fpga_cap_bits {
+ u8 fpga_id[0x8];
+ u8 fpga_device[0x18];
+
+ u8 register_file_ver[0x20];
+
+ u8 fpga_ctrl_modify[0x1];
+ u8 reserved_at_41[0x5];
+ u8 access_reg_query_mode[0x2];
+ u8 reserved_at_48[0x6];
+ u8 access_reg_modify_mode[0x2];
+ u8 reserved_at_50[0x10];
+
+ u8 reserved_at_60[0x20];
+
+ u8 image_version[0x20];
+
+ u8 image_date[0x20];
+
+ u8 image_time[0x20];
+
+ u8 shell_version[0x20];
+
+ u8 reserved_at_100[0x80];
+
+ struct mlx5_ifc_fpga_shell_caps_bits shell_caps;
+
+ u8 reserved_at_380[0x8];
+ u8 ieee_vendor_id[0x18];
+
+ u8 sandbox_product_version[0x10];
+ u8 sandbox_product_id[0x10];
+
+ u8 sandbox_basic_caps[0x20];
+
+ u8 reserved_at_3e0[0x10];
+ u8 sandbox_extended_caps_len[0x10];
+
+ u8 sandbox_extended_caps_addr[0x40];
+
+ u8 fpga_ddr_start_addr[0x40];
+
+ u8 fpga_cr_space_start_addr[0x40];
+
+ u8 fpga_ddr_size[0x20];
+
+ u8 fpga_cr_space_size[0x20];
+
+ u8 reserved_at_500[0x300];
+};
+
+struct mlx5_ifc_fpga_ctrl_bits {
+ u8 reserved_at_0[0x8];
+ u8 operation[0x8];
+ u8 reserved_at_10[0x8];
+ u8 status[0x8];
+
+ u8 reserved_at_20[0x8];
+ u8 flash_select_admin[0x8];
+ u8 reserved_at_30[0x8];
+ u8 flash_select_oper[0x8];
+
+ u8 reserved_at_40[0x40];
+};
+
+enum {
+ MLX5_FPGA_ERROR_EVENT_SYNDROME_CORRUPTED_DDR = 0x1,
+ MLX5_FPGA_ERROR_EVENT_SYNDROME_FLASH_TIMEOUT = 0x2,
+ MLX5_FPGA_ERROR_EVENT_SYNDROME_INTERNAL_LINK_ERROR = 0x3,
+ MLX5_FPGA_ERROR_EVENT_SYNDROME_WATCHDOG_FAILURE = 0x4,
+ MLX5_FPGA_ERROR_EVENT_SYNDROME_I2C_FAILURE = 0x5,
+ MLX5_FPGA_ERROR_EVENT_SYNDROME_IMAGE_CHANGED = 0x6,
+ MLX5_FPGA_ERROR_EVENT_SYNDROME_TEMPERATURE_CRITICAL = 0x7,
+};
+
+struct mlx5_ifc_fpga_error_event_bits {
+ u8 reserved_at_0[0x40];
+
+ u8 reserved_at_40[0x18];
+ u8 syndrome[0x8];
+
+ u8 reserved_at_60[0x80];
+};
+
+#endif /* MLX5_IFC_FPGA_H */
--
2.11.0
^ permalink raw reply related [flat|nested] 72+ messages in thread
* [for-next 5/6] net/mlx5: Bump driver version
2017-05-23 11:43 [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23 Saeed Mahameed
` (2 preceding siblings ...)
2017-05-23 11:44 ` [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova Saeed Mahameed
@ 2017-05-23 11:44 ` Saeed Mahameed
[not found] ` <20170523114404.20387-6-saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
[not found] ` <20170523114404.20387-1-saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
4 siblings, 1 reply; 72+ messages in thread
From: Saeed Mahameed @ 2017-05-23 11:44 UTC (permalink / raw)
To: David S. Miller, Doug Ledford
Cc: netdev, linux-rdma, Ilan Tayari, Tariq Toukan, Saeed Mahameed
From: Tariq Toukan <tariqt@mellanox.com>
Remove date and bump version for mlx5_core driver.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h | 3 +--
2 files changed, 2 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
index e0dd1048c966..afa89dcf30a5 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
@@ -39,7 +39,7 @@ static void mlx5e_get_drvinfo(struct net_device *dev,
struct mlx5_core_dev *mdev = priv->mdev;
strlcpy(drvinfo->driver, DRIVER_NAME, sizeof(drvinfo->driver));
- strlcpy(drvinfo->version, DRIVER_VERSION " (" DRIVER_RELDATE ")",
+ strlcpy(drvinfo->version, DRIVER_VERSION,
sizeof(drvinfo->version));
snprintf(drvinfo->fw_version, sizeof(drvinfo->fw_version),
"%d.%d.%04d (%.16s)",
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
index fbc6e9e9e305..cf69b42278df 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
@@ -39,8 +39,7 @@
#include <linux/if_link.h>
#define DRIVER_NAME "mlx5_core"
-#define DRIVER_VERSION "3.0-1"
-#define DRIVER_RELDATE "January 2015"
+#define DRIVER_VERSION "5.0-0"
#define MLX5_TOTAL_VPORTS(mdev) (1 + pci_sriov_get_totalvfs(mdev->pdev))
--
2.11.0
^ permalink raw reply related [flat|nested] 72+ messages in thread
* [for-next 6/6] IB/mlx5: Bump driver version
[not found] ` <20170523114404.20387-1-saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
2017-05-23 11:43 ` [for-next 1/6] {net, IB}/mlx5: Replace mlx5_vzalloc with kvzalloc Saeed Mahameed
@ 2017-05-23 11:44 ` Saeed Mahameed
2017-05-25 16:02 ` [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23 David Miller
2 siblings, 0 replies; 72+ messages in thread
From: Saeed Mahameed @ 2017-05-23 11:44 UTC (permalink / raw)
To: David S. Miller, Doug Ledford
Cc: netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
Ilan Tayari, Tariq Toukan, Saeed Mahameed
From: Tariq Toukan <tariqt-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Remove date and bump version for mlx5_ib driver.
Signed-off-by: Tariq Toukan <tariqt-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Signed-off-by: Saeed Mahameed <saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---
drivers/infiniband/hw/mlx5/main.c | 5 ++---
1 file changed, 2 insertions(+), 3 deletions(-)
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index b6991204e5df..42defaa0d6c6 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -60,8 +60,7 @@
#include "cmd.h"
#define DRIVER_NAME "mlx5_ib"
-#define DRIVER_VERSION "2.2-1"
-#define DRIVER_RELDATE "Feb 2014"
+#define DRIVER_VERSION "5.0-0"
MODULE_AUTHOR("Eli Cohen <eli-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>");
MODULE_DESCRIPTION("Mellanox Connect-IB HCA IB driver");
@@ -70,7 +69,7 @@ MODULE_VERSION(DRIVER_VERSION);
static char mlx5_version[] =
DRIVER_NAME ": Mellanox Connect-IB Infiniband driver v"
- DRIVER_VERSION " (" DRIVER_RELDATE ")\n";
+ DRIVER_VERSION "\n";
enum {
MLX5_ATOMIC_SIZE_QP_8BYTES = 1 << 3,
--
2.11.0
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <20170523114404.20387-5-saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
@ 2017-05-24 16:51 ` Alexei Starovoitov
2017-05-25 5:20 ` Ilan Tayari
0 siblings, 1 reply; 72+ messages in thread
From: Alexei Starovoitov @ 2017-05-24 16:51 UTC (permalink / raw)
To: Saeed Mahameed
Cc: David S. Miller, Doug Ledford, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, Ilan Tayari,
jsorensen-b10kYP2dOMg
On Tue, May 23, 2017 at 02:44:02PM +0300, Saeed Mahameed wrote:
> From: Ilan Tayari <ilant-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
>
> Mellanox Innova is a NIC with ConnectX and an FPGA on the same
> board. The FPGA is a bump-on-the-wire and thus affects operation of
> the mlx5_core driver on the ConnectX ASIC.
>
> Add basic support for Innova in mlx5_core.
>
> This allows using the Innova card as a regular NIC, by detecting
> the FPGA capability bit, and verifying its load state before
> initializing ConnectX interfaces.
That doesn't sound like cx4/cx5 hw that mlx5 driver suppose to support.
> Also detect FPGA fatal runtime failures and enter error state if
> they ever happen.
>
> All new FPGA-related logic is placed in its own subdirectory 'fpga',
> which may be built by selecting CONFIG_MLX5_FPGA.
> This prepares for further support of various Innova features in later
> patchsets.
> Additional details about hardware architecture will be provided as
> more features get submitted.
>
> Signed-off-by: Ilan Tayari <ilant-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
> Reviewed-by: Boris Pismenny <borisp-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
> Signed-off-by: Saeed Mahameed <saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
> ---
> MAINTAINERS | 10 +
> drivers/net/ethernet/mellanox/mlx5/core/Kconfig | 10 +
> drivers/net/ethernet/mellanox/mlx5/core/Makefile | 3 +
> drivers/net/ethernet/mellanox/mlx5/core/eq.c | 11 ++
> drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.c | 64 +++++++
> drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.h | 59 ++++++
> .../net/ethernet/mellanox/mlx5/core/fpga/core.c | 202 +++++++++++++++++++++
> .../net/ethernet/mellanox/mlx5/core/fpga/core.h | 99 ++++++++++
> drivers/net/ethernet/mellanox/mlx5/core/main.c | 19 +-
> include/linux/mlx5/device.h | 6 +
> include/linux/mlx5/driver.h | 5 +
> include/linux/mlx5/mlx5_ifc.h | 11 +-
> include/linux/mlx5/mlx5_ifc_fpga.h | 144 +++++++++++++++
> 13 files changed, 640 insertions(+), 3 deletions(-)
Can you put it into different driver? Dumping everything into by far
the biggest nic driver already is already huge headache in terms on
maintainability, debugging and back ports.
Look at how intel splits their drivers.
ixgb, ixgbe, ixgbevf are different drivers thought they have a lot in
common. On one side it's a bit of copy paste, but on the other side
it makes drivers much easier to develop and maintain independently.
ConnectX-6 code and any future hw support doesn't belong to
mlx5 driver at all.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* RE: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-05-24 16:51 ` Alexei Starovoitov
@ 2017-05-25 5:20 ` Ilan Tayari
[not found] ` <AM4PR0501MB1940885C2F1CEF4DDE4EA8D1DBFF0-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Ilan Tayari @ 2017-05-25 5:20 UTC (permalink / raw)
To: Alexei Starovoitov, Saeed Mahameed
Cc: David S. Miller, Doug Ledford, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg
> -----Original Message-----
> From: Alexei Starovoitov [mailto:alexei.starovoitov@gmail.com]
>
> On Tue, May 23, 2017 at 02:44:02PM +0300, Saeed Mahameed wrote:
> > From: Ilan Tayari <ilant@mellanox.com>
> >
> > Mellanox Innova is a NIC with ConnectX and an FPGA on the same
> > board. The FPGA is a bump-on-the-wire and thus affects operation of
> > the mlx5_core driver on the ConnectX ASIC.
> >
> > Add basic support for Innova in mlx5_core.
> >
> > This allows using the Innova card as a regular NIC, by detecting
> > the FPGA capability bit, and verifying its load state before
> > initializing ConnectX interfaces.
>
> That doesn't sound like cx4/cx5 hw that mlx5 driver suppose to support.
Hi Alexei,
Thanks for your feedback.
Let me address your concerns.
I didn't state it, but the current iterations of Innova cards include the
ConnectX-4LX chip, which is exactly what mlx5 driver supports.
The PCI interface is *only* through the ConnectX chip with some new
Firmware commands. This patch doesn't register any new PCI device,
so it does not make sense to create a separate driver.
>
> > Also detect FPGA fatal runtime failures and enter error state if
> > they ever happen.
> >
> > All new FPGA-related logic is placed in its own subdirectory 'fpga',
> > which may be built by selecting CONFIG_MLX5_FPGA.
> > This prepares for further support of various Innova features in later
> > patchsets.
> > Additional details about hardware architecture will be provided as
> > more features get submitted.
> >
> > Signed-off-by: Ilan Tayari <ilant@mellanox.com>
> > Reviewed-by: Boris Pismenny <borisp@mellanox.com>
> > Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
> > ---
> > MAINTAINERS | 10 +
> > drivers/net/ethernet/mellanox/mlx5/core/Kconfig | 10 +
> > drivers/net/ethernet/mellanox/mlx5/core/Makefile | 3 +
> > drivers/net/ethernet/mellanox/mlx5/core/eq.c | 11 ++
> > drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.c | 64 +++++++
> > drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.h | 59 ++++++
> > .../net/ethernet/mellanox/mlx5/core/fpga/core.c | 202
> +++++++++++++++++++++
> > .../net/ethernet/mellanox/mlx5/core/fpga/core.h | 99 ++++++++++
> > drivers/net/ethernet/mellanox/mlx5/core/main.c | 19 +-
> > include/linux/mlx5/device.h | 6 +
> > include/linux/mlx5/driver.h | 5 +
> > include/linux/mlx5/mlx5_ifc.h | 11 +-
> > include/linux/mlx5/mlx5_ifc_fpga.h | 144
> +++++++++++++++
> > 13 files changed, 640 insertions(+), 3 deletions(-)
>
> Can you put it into different driver? Dumping everything into by far
> the biggest nic driver already is already huge headache in terms on
> maintainability, debugging and back ports.
> Look at how intel splits their drivers.
> ixgb, ixgbe, ixgbevf are different drivers thought they have a lot in
> common. On one side it's a bit of copy paste, but on the other side
> it makes drivers much easier to develop and maintain independently.
> ConnectX-6 code and any future hw support doesn't belong to
> mlx5 driver at all.
If you build your driver without explicitly enabling CONFIG_MLX5_FPGA=y
(default is N) then you get none of this, and your driver is practically
the same as before.
The FPGA and CX operation is very tightly integrated.
If you don't want any of this, simply don't opt-in to CONFIG_MLX5_FPGA.
If you do want this, then splitting some of the logic to a
separate kernel object will not gain anything useful (logic would stay
the same), and just pollute the exported symbol table and open up the door
for issues of inter-module compatibility and so on.
Furthermore, the next patchset will introduce IPSec offload capabilities
Which are declared in netdev->hw_features. Those cannot be modified
after the netdevice is created, so all the extra logic has to be
integrated into the mlx5 ethernet driver. This is another reason why
a separate driver is a bad idea.
We will include details about the board and how things work in the
cover letter of the IPSec offload patchset.
Ilan.
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <AM4PR0501MB1940885C2F1CEF4DDE4EA8D1DBFF0-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
@ 2017-05-25 10:40 ` Saeed Mahameed
[not found] ` <CALzJLG-B_tAmASn_SMmPNiucq-tTpywHniRTkb4N32oGF6Y3Ng-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-26 3:07 ` Alexei Starovoitov
1 sibling, 1 reply; 72+ messages in thread
From: Saeed Mahameed @ 2017-05-25 10:40 UTC (permalink / raw)
To: Ilan Tayari
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg
On Thu, May 25, 2017 at 8:20 AM, Ilan Tayari <ilant-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org> wrote:
>> -----Original Message-----
>> From: Alexei Starovoitov [mailto:alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org]
>>
>> On Tue, May 23, 2017 at 02:44:02PM +0300, Saeed Mahameed wrote:
>> > From: Ilan Tayari <ilant-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
>> >
>> > Mellanox Innova is a NIC with ConnectX and an FPGA on the same
>> > board. The FPGA is a bump-on-the-wire and thus affects operation of
>> > the mlx5_core driver on the ConnectX ASIC.
>> >
>> > Add basic support for Innova in mlx5_core.
>> >
>> > This allows using the Innova card as a regular NIC, by detecting
>> > the FPGA capability bit, and verifying its load state before
>> > initializing ConnectX interfaces.
>>
>> That doesn't sound like cx4/cx5 hw that mlx5 driver suppose to support.
>
> Hi Alexei,
> Thanks for your feedback.
> Let me address your concerns.
>
> I didn't state it, but the current iterations of Innova cards include the
> ConnectX-4LX chip, which is exactly what mlx5 driver supports.
>
> The PCI interface is *only* through the ConnectX chip with some new
> Firmware commands. This patch doesn't register any new PCI device,
> so it does not make sense to create a separate driver.
>
>>
>> > Also detect FPGA fatal runtime failures and enter error state if
>> > they ever happen.
>> >
>> > All new FPGA-related logic is placed in its own subdirectory 'fpga',
>> > which may be built by selecting CONFIG_MLX5_FPGA.
>> > This prepares for further support of various Innova features in later
>> > patchsets.
>> > Additional details about hardware architecture will be provided as
>> > more features get submitted.
>> >
>> > Signed-off-by: Ilan Tayari <ilant-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
>> > Reviewed-by: Boris Pismenny <borisp-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
>> > Signed-off-by: Saeed Mahameed <saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
>> > ---
>> > MAINTAINERS | 10 +
>> > drivers/net/ethernet/mellanox/mlx5/core/Kconfig | 10 +
>> > drivers/net/ethernet/mellanox/mlx5/core/Makefile | 3 +
>> > drivers/net/ethernet/mellanox/mlx5/core/eq.c | 11 ++
>> > drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.c | 64 +++++++
>> > drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.h | 59 ++++++
>> > .../net/ethernet/mellanox/mlx5/core/fpga/core.c | 202
>> +++++++++++++++++++++
>> > .../net/ethernet/mellanox/mlx5/core/fpga/core.h | 99 ++++++++++
>> > drivers/net/ethernet/mellanox/mlx5/core/main.c | 19 +-
>> > include/linux/mlx5/device.h | 6 +
>> > include/linux/mlx5/driver.h | 5 +
>> > include/linux/mlx5/mlx5_ifc.h | 11 +-
>> > include/linux/mlx5/mlx5_ifc_fpga.h | 144
>> +++++++++++++++
>> > 13 files changed, 640 insertions(+), 3 deletions(-)
>>
>> Can you put it into different driver? Dumping everything into by far
>> the biggest nic driver already is already huge headache in terms on
>> maintainability, debugging and back ports.
>> Look at how intel splits their drivers.
>> ixgb, ixgbe, ixgbevf are different drivers thought they have a lot in
>> common. On one side it's a bit of copy paste, but on the other side
I don't think the ixgb example is the same, simply ixgb, ixgbe,
ixgbevf have different PCI IDs
and even different SW/FW interfaces. On the other hand, same mlx5
driver can support all of
ConnetX4/5/6 device IDs with the same code flows, same interfaces.
>> it makes drivers much easier to develop and maintain independently.
>> ConnectX-6 code and any future hw support doesn't belong to
>> mlx5 driver at all.
Sorry i must disagree with you on this for the same reasons Ilan mentioned.
We can perfectly achieve the same with modular driver design all under the
same .ko module, with some kconfig flags to reduce the amount of code/features
this .ko provides.
>
> If you build your driver without explicitly enabling CONFIG_MLX5_FPGA=y
> (default is N) then you get none of this, and your driver is practically
> the same as before.
>
> The FPGA and CX operation is very tightly integrated.
> If you don't want any of this, simply don't opt-in to CONFIG_MLX5_FPGA.
>
> If you do want this, then splitting some of the logic to a
> separate kernel object will not gain anything useful (logic would stay
> the same), and just pollute the exported symbol table and open up the door
> for issues of inter-module compatibility and so on.
>
> Furthermore, the next patchset will introduce IPSec offload capabilities
> Which are declared in netdev->hw_features. Those cannot be modified
> after the netdevice is created, so all the extra logic has to be
> integrated into the mlx5 ethernet driver. This is another reason why
> a separate driver is a bad idea.
>
> We will include details about the board and how things work in the
> cover letter of the IPSec offload patchset.
>
> Ilan.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23
[not found] ` <20170523114404.20387-1-saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
2017-05-23 11:43 ` [for-next 1/6] {net, IB}/mlx5: Replace mlx5_vzalloc with kvzalloc Saeed Mahameed
2017-05-23 11:44 ` [for-next 6/6] IB/mlx5: Bump driver version Saeed Mahameed
@ 2017-05-25 16:02 ` David Miller
2017-06-01 22:57 ` Doug Ledford
2017-06-14 19:30 ` Doug Ledford
2 siblings, 2 replies; 72+ messages in thread
From: David Miller @ 2017-05-25 16:02 UTC (permalink / raw)
To: saeedm-VPRAkNaXOzVWk0Htik3J/w
Cc: dledford-H+wXaHxf7aLQT0dZR+AlfA, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, ilant-VPRAkNaXOzVWk0Htik3J/w
From: Saeed Mahameed <saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Date: Tue, 23 May 2017 14:43:58 +0300
> Hi Dave and Doug,
>
> This series introduces some small updates and FPGA support to the mlx5
> core/ethernet and IB drivers.
>
> For more details please see below.
>
> Please pull and let me know if there's any problem.
Ok, I've pulled this into net-next.
Doug let me know if there are any merge hassles we need to coordinate on.
Thanks.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <CALzJLG-B_tAmASn_SMmPNiucq-tTpywHniRTkb4N32oGF6Y3Ng-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
@ 2017-05-25 20:48 ` Jes Sorensen
2017-05-26 8:29 ` Saeed Mahameed
0 siblings, 1 reply; 72+ messages in thread
From: Jes Sorensen @ 2017-05-25 20:48 UTC (permalink / raw)
To: Saeed Mahameed, Ilan Tayari
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA
On 05/25/2017 06:40 AM, Saeed Mahameed wrote:
> On Thu, May 25, 2017 at 8:20 AM, Ilan Tayari <ilant-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org> wrote:
>>> -----Original Message-----
>>> Can you put it into different driver? Dumping everything into by far
>>> the biggest nic driver already is already huge headache in terms on
>>> maintainability, debugging and back ports.
>>> Look at how intel splits their drivers.
>>> ixgb, ixgbe, ixgbevf are different drivers thought they have a lot in
>>> common. On one side it's a bit of copy paste, but on the other side
>
> I don't think the ixgb example is the same, simply ixgb, ixgbe,
> ixgbevf have different PCI IDs
> and even different SW/FW interfaces. On the other hand, same mlx5
> driver can support all of
> ConnetX4/5/6 device IDs with the same code flows, same interfaces.
>
>>> it makes drivers much easier to develop and maintain independently.
>>> ConnectX-6 code and any future hw support doesn't belong to
>>> mlx5 driver at all.
>
> Sorry i must disagree with you on this for the same reasons Ilan mentioned.
> We can perfectly achieve the same with modular driver design all under the
> same .ko module, with some kconfig flags to reduce the amount of code/features
> this .ko provides.
If I get this right, the FPGA is independent and could in theory be used
for non network stuff. It really should have it's own driver in that
case, and you should provide accessor functionality via the mlx5 driver.
We have this with other devices in the kernel where a primary device
driver provides an interface for an additional sub-driver to access
another device behind it. Like bt-coexist in some of the wifi drivers
allowing access to a bluetooth device behind it.
Jes
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <AM4PR0501MB1940885C2F1CEF4DDE4EA8D1DBFF0-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
2017-05-25 10:40 ` Saeed Mahameed
@ 2017-05-26 3:07 ` Alexei Starovoitov
2017-05-26 8:59 ` Saeed Mahameed
1 sibling, 1 reply; 72+ messages in thread
From: Alexei Starovoitov @ 2017-05-26 3:07 UTC (permalink / raw)
To: Ilan Tayari
Cc: Saeed Mahameed, David S. Miller, Doug Ledford,
netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
jsorensen-b10kYP2dOMg
On Thu, May 25, 2017 at 05:20:04AM +0000, Ilan Tayari wrote:
>
> If you do want this, then splitting some of the logic to a
> separate kernel object will not gain anything useful (logic would stay
> the same), and just pollute the exported symbol table and open up the door
> for issues of inter-module compatibility and so on.
in other words instead of properly designing inter-module api
you just want to add everything into one giant 'driver'
because it's easier to do so. Understandable, but it leads
to unmaintainable device infrastructure long term.
> Furthermore, the next patchset will introduce IPSec offload capabilities
> Which are declared in netdev->hw_features. Those cannot be modified
> after the netdevice is created, so all the extra logic has to be
> integrated into the mlx5 ethernet driver. This is another reason why
> a separate driver is a bad idea.
that increases my concerns even more.
ipsec offload is a new feature for new hw, yet you're trying to
hide it in the mlx5 'driver'.
mlx5 already supports cx4/cx4-lx/cx5 and not even released cx6.
All of them have different feature sets.
The only common piece is driver/fw cmd interface and _some_ aspects
of rx/tx descriptors. Everything else is different.
cx4-lx doesn't have infiniband and rdma, yet there is a ton of code
in the driver that deals with it.
cx5 has fancy storage interfaces for nvme, etc I don't think they
are part of the mlx5 'driver' yet. Are you going to dump them
in there as well?
Take a look at the simplest feature-wise cx4-lx. It has
eswitch which is full l2 switch with mac table, drop policy,
counters and such. Yet kernel knows nothing about it.
Everything is hidden in mlx5 'driver' with its own special
interfaces to manage it.
Now you want to hide fpga with custom logic behind mlx5 'driver'
and manage it through mlx5 interfaces?
That's not acceptable.
mlx fpga got to have generic kernel representation that intel
and other fpga vendors can use.
mlx5 driver has to be modularized and since it's not too late
cx6 support has to be removed from it.
Thankfully only few cx6 pci ids were added to mlx5, but driver
knows nothing about cx6 features, so we can properly design it.
Since driver/fw interface is common between cx4+ this part
can be moved into mlx_core.ko or done as library
the way chelsio did it in libcxgb.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* please revert. Was: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-05-23 11:44 ` [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova Saeed Mahameed
[not found] ` <20170523114404.20387-5-saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
@ 2017-05-26 3:58 ` Alexei Starovoitov
2017-05-26 4:13 ` David Miller
1 sibling, 1 reply; 72+ messages in thread
From: Alexei Starovoitov @ 2017-05-26 3:58 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, linux-rdma
On Tue, May 23, 2017 at 02:44:02PM +0300, Saeed Mahameed wrote:
> From: Ilan Tayari <ilant@mellanox.com>
>
> Mellanox Innova is a NIC with ConnectX and an FPGA on the same
> board. The FPGA is a bump-on-the-wire and thus affects operation of
> the mlx5_core driver on the ConnectX ASIC.
>
> Add basic support for Innova in mlx5_core.
>
> This allows using the Innova card as a regular NIC, by detecting
> the FPGA capability bit, and verifying its load state before
> initializing ConnectX interfaces.
>
> Also detect FPGA fatal runtime failures and enter error state if
> they ever happen.
>
> All new FPGA-related logic is placed in its own subdirectory 'fpga',
> which may be built by selecting CONFIG_MLX5_FPGA.
> This prepares for further support of various Innova features in later
> patchsets.
> Additional details about hardware architecture will be provided as
> more features get submitted.
>
> Signed-off-by: Ilan Tayari <ilant@mellanox.com>
> Reviewed-by: Boris Pismenny <borisp@mellanox.com>
> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
> ---
> MAINTAINERS | 10 +
> drivers/net/ethernet/mellanox/mlx5/core/Kconfig | 10 +
> drivers/net/ethernet/mellanox/mlx5/core/Makefile | 3 +
> drivers/net/ethernet/mellanox/mlx5/core/eq.c | 11 ++
> drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.c | 64 +++++++
> drivers/net/ethernet/mellanox/mlx5/core/fpga/cmd.h | 59 ++++++
> .../net/ethernet/mellanox/mlx5/core/fpga/core.c | 202 +++++++++++++++++++++
> .../net/ethernet/mellanox/mlx5/core/fpga/core.h | 99 ++++++++++
> drivers/net/ethernet/mellanox/mlx5/core/main.c | 19 +-
> include/linux/mlx5/device.h | 6 +
> include/linux/mlx5/driver.h | 5 +
> include/linux/mlx5/mlx5_ifc.h | 11 +-
> include/linux/mlx5/mlx5_ifc_fpga.h | 144 +++++++++++++++
> 13 files changed, 640 insertions(+), 3 deletions(-)
Dave, please revert this Innova fpga stuff.
I think you pushed it by accident, since it was mixed with
other valid changes.
The discussion didn't conclude.
Myself and Jes are clearly against such changes.
It definitely needs more discussion and wider consensus.
Thanks
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: please revert. Was: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-05-26 3:58 ` please revert. Was: " Alexei Starovoitov
@ 2017-05-26 4:13 ` David Miller
2017-05-26 4:40 ` Alexei Starovoitov
0 siblings, 1 reply; 72+ messages in thread
From: David Miller @ 2017-05-26 4:13 UTC (permalink / raw)
To: alexei.starovoitov; +Cc: netdev, linux-rdma
From: Alexei Starovoitov <alexei.starovoitov@gmail.com>
Date: Thu, 25 May 2017 20:58:32 -0700
> Dave, please revert this Innova fpga stuff.
> I think you pushed it by accident, since it was mixed with
> other valid changes.
> The discussion didn't conclude.
> Myself and Jes are clearly against such changes.
> It definitely needs more discussion and wider consensus.
Why don't you finish your discussion, then I can revert or leave it in
there?
If someone puts an FPGA inside of a device, and it's still behind the
PCI ID of the ethernet card, how they expose that thing is largely at
the discression of the driver author.
So feel free to discuss things with them, but in the end unless I am
convinced otherwise (and I'm currently not), what they are doing now
is fine by me.
I will also state that it can't be that every time Mellanox tries to
add something significant to their driver everyone screams "oh my,
this thing is so hard to maintain as it is, you can't add this feature
_too_."
Sorry, it's an important driver, it's big, it has a lot of features,
and all those problems people talk about simply come with the
territory.
This FPGA thing is just a drop in the bucket as far as long term
maintainence and backporting is concerned, so if that truly is the
basis of why you are against these changes, there is not much weight
to that argument.
Use a less featurefull device, which uses a driver with slower
development and less changes, if these things truly bother you.
Thanks.
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: please revert. Was: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-05-26 4:13 ` David Miller
@ 2017-05-26 4:40 ` Alexei Starovoitov
2017-05-26 14:51 ` David Miller
0 siblings, 1 reply; 72+ messages in thread
From: Alexei Starovoitov @ 2017-05-26 4:40 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-rdma
On Fri, May 26, 2017 at 12:13:27AM -0400, David Miller wrote:
> From: Alexei Starovoitov <alexei.starovoitov@gmail.com>
> Date: Thu, 25 May 2017 20:58:32 -0700
>
> > Dave, please revert this Innova fpga stuff.
> > I think you pushed it by accident, since it was mixed with
> > other valid changes.
> > The discussion didn't conclude.
> > Myself and Jes are clearly against such changes.
> > It definitely needs more discussion and wider consensus.
>
> Why don't you finish your discussion, then I can revert or leave it in
> there?
Not really. What you're saying is 'shut up. mellanox can do
whatever they like as long as it's hidden behind pcie id'.
> If someone puts an FPGA inside of a device, and it's still behind the
> PCI ID of the ethernet card, how they expose that thing is largely at
> the discression of the driver author.
So you're saying it's ok to program fpga through nic whereas larger
kernel community is trying to come up with a solution to represent
fpga as a proper device for the kernel.
I suspect there will be popcorn threads during the next merge window.
> So feel free to discuss things with them, but in the end unless I am
> convinced otherwise (and I'm currently not), what they are doing now
> is fine by me.
If you didn't merge it already it would be fair course of action,
but it's clearly not the case.
> I will also state that it can't be that every time Mellanox tries to
> add something significant to their driver everyone screams "oh my,
> this thing is so hard to maintain as it is, you can't add this feature
> _too_."
Maintaining is only small part of it. The key argument against it
that all these things fpga, eswitch, etc are _hidden_ behind the nic.
It already causes headaches in production, because kernel cannot see
them.
> Sorry, it's an important driver, it's big, it has a lot of features,
> and all those problems people talk about simply come with the
> territory.
It's not a driver feature. It's not a nic.
It's a different physical device that mlx choose to program via nic.
There is also 'bluefield' nic on the horizon.
>From host point of view it looks like the same cx5/6 nic,
but it has arm cpus on board. Are you going to allow programming
it via nic as part of mlx5 driver as well?
It's no different than fpga hiding behind the nic...
> Use a less featurefull device, which uses a driver with slower
> development and less changes, if these things truly bother you.
like google did. food for thought.
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-05-25 20:48 ` Jes Sorensen
@ 2017-05-26 8:29 ` Saeed Mahameed
[not found] ` <CALzJLG9YNpagdJAcrh6O0jJhZWtsck6KigRtVxyjkArTm=82ew-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Saeed Mahameed @ 2017-05-26 8:29 UTC (permalink / raw)
To: Jes Sorensen
Cc: Ilan Tayari, Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev, linux-rdma
On Thu, May 25, 2017 at 11:48 PM, Jes Sorensen <jsorensen@fb.com> wrote:
> On 05/25/2017 06:40 AM, Saeed Mahameed wrote:
>>
>> On Thu, May 25, 2017 at 8:20 AM, Ilan Tayari <ilant@mellanox.com> wrote:
>>>>
>>>> -----Original Message-----
>>>> Can you put it into different driver? Dumping everything into by far
>>>> the biggest nic driver already is already huge headache in terms on
>>>> maintainability, debugging and back ports.
>>>> Look at how intel splits their drivers.
>>>> ixgb, ixgbe, ixgbevf are different drivers thought they have a lot in
>>>> common. On one side it's a bit of copy paste, but on the other side
>>
>>
>> I don't think the ixgb example is the same, simply ixgb, ixgbe,
>> ixgbevf have different PCI IDs
>> and even different SW/FW interfaces. On the other hand, same mlx5
>> driver can support all of
>> ConnetX4/5/6 device IDs with the same code flows, same interfaces.
>>
>>>> it makes drivers much easier to develop and maintain independently.
>>>> ConnectX-6 code and any future hw support doesn't belong to
>>>> mlx5 driver at all.
>>
>>
>> Sorry i must disagree with you on this for the same reasons Ilan
>> mentioned.
>> We can perfectly achieve the same with modular driver design all under the
>> same .ko module, with some kconfig flags to reduce the amount of
>> code/features
>> this .ko provides.
>
>
> If I get this right, the FPGA is independent and could in theory be used for
> non network stuff. It really should have it's own driver in that case, and
> you should provide accessor functionality via the mlx5 driver.
>
Hi Jes,
No, It is clearly stated in the commit message :
"The FPGA is a bump-on-the-wire and thus affects operation of
the mlx5_core driver on the ConnectX ASIC."
Which means mlx5 FPGA user can only write logic which affects only
packets going in/out
A ConnectX chip - so it is only network stuff -.
> We have this with other devices in the kernel where a primary device driver
> provides an interface for an additional sub-driver to access another device
> behind it. Like bt-coexist in some of the wifi drivers allowing access to a
> bluetooth device behind it.
>
Blutooth over wifi or vise versa is a very good example to what you
are requesting.
But, it doesn't fit to what we are trying to do here. mlx5 FGPA is a
ConnectX card feature, not a new protocol.
> Jes
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-05-26 3:07 ` Alexei Starovoitov
@ 2017-05-26 8:59 ` Saeed Mahameed
[not found] ` <CALzJLG98D=3yMJV_q4sjVNG41AERFRU+6rwqQJsxnRuVeDTPdA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Saeed Mahameed @ 2017-05-26 8:59 UTC (permalink / raw)
To: Alexei Starovoitov
Cc: Ilan Tayari, Saeed Mahameed, David S. Miller, Doug Ledford,
netdev, linux-rdma, jsorensen
On Fri, May 26, 2017 at 6:07 AM, Alexei Starovoitov
<alexei.starovoitov@gmail.com> wrote:
> On Thu, May 25, 2017 at 05:20:04AM +0000, Ilan Tayari wrote:
>>
>> If you do want this, then splitting some of the logic to a
>> separate kernel object will not gain anything useful (logic would stay
>> the same), and just pollute the exported symbol table and open up the door
>> for issues of inter-module compatibility and so on.
>
> in other words instead of properly designing inter-module api
> you just want to add everything into one giant 'driver'
> because it's easier to do so. Understandable, but it leads
> to unmaintainable device infrastructure long term.
>
What is wrong with giant 'driver' ? :-)
BTW making the fpga a separate module is as easy as adding one line to
the make file
and moving the entry points functions (fpga_init/cleanup) form mlx5
init flow into a "mlx5_interface" instance and register it with the
mlx5_core.
But again i don't see a point of doing so, the code will still look
the same and sit in the same places.
so one giant ko file or small separate modules is basically the same,
from my point of view. modular design have nothing to do with the
compilation output binary, it should be considered regardless of the
compiler output.
Which we tried to not far a go to break the driver into sub-module
directories, each with its own responsibility, but it was too late for
that.
And for FPGA support, we did it correctly this time, all the new code
is under mlx5/core/fgpa ..
if you don't like core/fpga, ok just simply "CONFIG_MLX5_FPGA = no"
and all of that code/directory will disappear from your ko file and
will not be touched in compilation time, and the core code will still
look the same.
>> Furthermore, the next patchset will introduce IPSec offload capabilities
>> Which are declared in netdev->hw_features. Those cannot be modified
>> after the netdevice is created, so all the extra logic has to be
>> integrated into the mlx5 ethernet driver. This is another reason why
>> a separate driver is a bad idea.
>
> that increases my concerns even more.
> ipsec offload is a new feature for new hw, yet you're trying to
> hide it in the mlx5 'driver'.
>
Well, this is a well known dilemma, if one has a new feature and has no sandbox
area to put it in the first phase, it is better to start from the most
common place
for that feature which is the originating place, before defining
APIs/infrastructures,
until the feature is complete and every body is happy about it.
Same was done for GRO, XDP , you name it ..
nothing is being hidden here, we are adding new HW offloads, not that
different from
csum offloads, which no one can live without these days, and i believe
in few years IPSec and TLS and security offloads will be no less
important than csum offloads.
Someone has to start somewhere and we choose to start in our driver,
if you have a better IPSec generic interface that allows modular
module separation we are happy to hear about it.
> mlx5 already supports cx4/cx4-lx/cx5 and not even released cx6.
I will take this as a complement ;-), please see more info below.
> All of them have different feature sets.
Not different feature sets, but i expect some changes through out the
evolution of the ConnectX chip , and those changes can be:
1. internal bug fixes/internal improvements.
2. new extra features from the previous generation.
But the chip is the same chip, i.e same driver works on all of the
chips with the basic/standard feature set.
> The only common piece is driver/fw cmd interface and _some_ aspects
> of rx/tx descriptors. Everything else is different.
I can add more to the list.
basic offloads, steering management, eswitch, HW objects and semantics
QPs/RQs/SQs/CQs/EQs and their manipulation semantics, internal page
memory management, NIC initialization and teardown flows are the same
and many more.
> cx4-lx doesn't have infiniband and rdma, yet there is a ton of code
although it doesn't have infinband support, cx4-lx have rdma support,
you can perfectly load mlx5_ib and work with RDMA over ethernet
(RoCE).
which is 99% of the mlx5_ib&mlx5_core code. (really !)
so the ton of code is still valid for cx4-lx.
> in the driver that deals with it.
> cx5 has fancy storage interfaces for nvme, etc I don't think they
> are part of the mlx5 'driver' yet. Are you going to dump them
> in there as well?
> Take a look at the simplest feature-wise cx4-lx. It has
> eswitch which is full l2 switch with mac table, drop policy,
> counters and such. Yet kernel knows nothing about it.
> Everything is hidden in mlx5 'driver' with its own special
> interfaces to manage it.
1. eswitch is a very integral steering feature of the ConnectX family
and is valid for all ConnectX4/lx/5/6 .. cards.
2. kenrel doesn't and can't know about it since in the early days of
SRIOV implementation (what the cool kids of switchdev mode call the
legacy sriov mode) is the responsibility of the device driver and it
is an internal arch of the HCA. so it is not our fault it is hidden in
mlx5, we just went with the flow..
3. I agree with you, device driver should be as transparent as
possible to the kernel, and should provide as much info as needed to
the stack.
4. switchdev mode came to solve this exact issue, and the kernel stack
even up to user space are perfectly aware of the nic eswitch
capabilities.
being said, even if the kernel is aware or not aware of the eswitch it
doesn't mean it should or shouldn't sit in the same mlx5 ko file.
> Now you want to hide fpga with custom logic behind mlx5 'driver'
> and manage it through mlx5 interfaces?
> That's not acceptable.
> mlx fpga got to have generic kernel representation that intel
> and other fpga vendors can use.
>
quoting Ilan from his commit message:
"The FPGA is a bump-on-the-wire and thus affects operation of
the mlx5_core driver on the ConnectX ASIC."
This is not a general purpose FPGA ! it even doesn't have a PCI Id or
it is own PCI function.
it simply allows the ConnectX Chip user to inspect/inject or run user
specif login on traffic going in/out of the chip.
But i agree with you some serious API brainstorming should be
considered, but not at this stage,
this patch only tells the ConnectX card (if you have FPGA, please enable it).
> mlx5 driver has to be modularized and since it's not too late
> cx6 support has to be removed from it.
>
I agree with you on the first part, Yes i would like to better
modularize the driver, i will even consider taking it to the extreme
and have separate module for each sub mlx5 module
e.g:
mlx5_core.ko
mlx5_eswitch.ko
mlx5_vxaln.ko
mlx5_vf_repreentor.ko
mlx5_ipoib.ko
...
even mlx5_xdp.ko ( not really ;-) )
If this what it takes to more logically and modularly break mlx5/core
into hierarchical design,
Please count me in. And i would like to hear Dave explicitly say i am
ok with this first.
But, re cx6 I am not sure i can agree on this for the simple fact:
it took 3 years to get mlx5 (Cx4/lx) driver to provide similar feature
set as mlx4 (Cx3).
and it literally took 0 lines of codes 1 day of regression to test
mlx5 driver over CX5 !
Now All we need is to add the new features of CX5 on top of the
already existing driver and i will be more than happy to hear about
exciting Ideas on how to modularly do that in the same driver,
and provide clear separation (beleive me it is is to do, if i am
allowed to split the driver into directories) but i can't do it in a
completely separate driver with a base code library, it is simply
wrong and non scalable (even if someone is not careful enough it could
make a lot of dependency mess between mlx5 modules)
> Thankfully only few cx6 pci ids were added to mlx5, but driver
> knows nothing about cx6 features, so we can properly design it.
>
> Since driver/fw interface is common between cx4+ this part
> can be moved into mlx_core.ko or done as library
> the way chelsio did it in libcxgb.
>
driver/fw interfaces are not the only common part, it is all basic
standard netdev/rdma and linux networkinng stuff are the same.
-Saeed.
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 5/6] net/mlx5: Bump driver version
[not found] ` <20170523114404.20387-6-saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
@ 2017-05-26 12:56 ` Dennis Dalessandro
2017-05-26 16:35 ` Saeed Mahameed
0 siblings, 1 reply; 72+ messages in thread
From: Dennis Dalessandro @ 2017-05-26 12:56 UTC (permalink / raw)
To: Saeed Mahameed, David S. Miller, Doug Ledford
Cc: netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
Ilan Tayari, Tariq Toukan
On 5/23/2017 7:44 AM, Saeed Mahameed wrote:
> From: Tariq Toukan <tariqt-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
>
> Remove date and bump version for mlx5_core driver.
>
> Signed-off-by: Tariq Toukan <tariqt-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
> Signed-off-by: Saeed Mahameed <saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
So I just complained about the bnxt_re doing this. I guess I need to
raise a flag here too. Now to be clear I'm not against doing this
version stuff. I'm against using it for some internal tracking that is
meaningless to the kernel community.
There is no detail in your commit message as to what this driver version
is used for. Can you please explain how your use of driver version is
valid while theirs is not?
I realize Dave has already pulled this and I'm not asking for it to be
reverted but maybe some discussion will help guide future patch
submissions which do this stuff.
-Denny
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: please revert. Was: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-05-26 4:40 ` Alexei Starovoitov
@ 2017-05-26 14:51 ` David Miller
0 siblings, 0 replies; 72+ messages in thread
From: David Miller @ 2017-05-26 14:51 UTC (permalink / raw)
To: alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w
Cc: netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA
From: Alexei Starovoitov <alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Date: Thu, 25 May 2017 21:40:59 -0700
> On Fri, May 26, 2017 at 12:13:27AM -0400, David Miller wrote:
>> From: Alexei Starovoitov <alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
>> Date: Thu, 25 May 2017 20:58:32 -0700
>>
>> > Dave, please revert this Innova fpga stuff.
>> > I think you pushed it by accident, since it was mixed with
>> > other valid changes.
>> > The discussion didn't conclude.
>> > Myself and Jes are clearly against such changes.
>> > It definitely needs more discussion and wider consensus.
>>
>> Why don't you finish your discussion, then I can revert or leave it in
>> there?
>
> Not really. What you're saying is 'shut up. mellanox can do
> whatever they like as long as it's hidden behind pcie id'.
No, what I'm saying is, discuss things.
And if in the end you agree with Mellanox, we don't have flapping of
the change in and out of the tree over and over.
My understanding is that this FPGA offloads processing the data
stream, which in a way is not much different than JIT'ing eBPF
onto Netronome cpus. :-)
Maybe the programming interface is different, maybe all the algorithms
are implemented in discrete ASICs, but effect is quite similar.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 5/6] net/mlx5: Bump driver version
2017-05-26 12:56 ` Dennis Dalessandro
@ 2017-05-26 16:35 ` Saeed Mahameed
[not found] ` <CALzJLG_ha-XiPAMnoKrUgm_EwPx2yH0T2y4EBRfrWNYSZi1cTg-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Saeed Mahameed @ 2017-05-26 16:35 UTC (permalink / raw)
To: Dennis Dalessandro
Cc: Saeed Mahameed, David S. Miller, Doug Ledford, Linux Netdev List,
linux-rdma, Ilan Tayari, Tariq Toukan
On Fri, May 26, 2017 at 3:56 PM, Dennis Dalessandro
<dennis.dalessandro@intel.com> wrote:
> On 5/23/2017 7:44 AM, Saeed Mahameed wrote:
>>
>> From: Tariq Toukan <tariqt@mellanox.com>
>>
>> Remove date and bump version for mlx5_core driver.
>>
>> Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
>> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
>
>
> So I just complained about the bnxt_re doing this. I guess I need to raise a
> flag here too. Now to be clear I'm not against doing this version stuff. I'm
> against using it for some internal tracking that is meaningless to the
> kernel community.
>
> There is no detail in your commit message as to what this driver version is
> used for. Can you please explain how your use of driver version is valid
> while theirs is not?
>
Hi Denny,
We don't use this for internal or any kind of tracking at all, if you
look at the driver change log, we hardly touch the version if at all !
but since we decided to remove the date since it is very misleading
and we got some complaints that some think they have an old driver
just because of the date, also we bumped the version so it would align
with our drivers external package (OFED).
Anyway i don't think we are going to change this frequently or even use it.
But if you are that much against touching this ethtool field, why
don't you just jump into the mud and remove it from the tree, I
imagine it is really hard for you to watch for patches doing this and
nack them on the list.
> I realize Dave has already pulled this and I'm not asking for it to be
> reverted but maybe some discussion will help guide future patch submissions
> which do this stuff.
>
Sure, although i don't think we are going to use those version fields
in the future,
please allow me to ask, how do you do your driver versioning ? how do
you track things ?
and what is your future vision regarding ethool->drv_version ?
Thanks,
Saeed.
> -Denny
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 5/6] net/mlx5: Bump driver version
[not found] ` <CALzJLG_ha-XiPAMnoKrUgm_EwPx2yH0T2y4EBRfrWNYSZi1cTg-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
@ 2017-05-26 16:55 ` Dennis Dalessandro
[not found] ` <ee23ad82-4a2e-8546-d41b-11f979b127bb-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
2017-05-26 21:53 ` Jakub Kicinski
0 siblings, 2 replies; 72+ messages in thread
From: Dennis Dalessandro @ 2017-05-26 16:55 UTC (permalink / raw)
To: Saeed Mahameed
Cc: Saeed Mahameed, David S. Miller, Doug Ledford, Linux Netdev List,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, Ilan Tayari, Tariq Toukan
On 5/26/2017 12:35 PM, Saeed Mahameed wrote:
> On Fri, May 26, 2017 at 3:56 PM, Dennis Dalessandro
> <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org> wrote:
>> On 5/23/2017 7:44 AM, Saeed Mahameed wrote:
>>>
>>> From: Tariq Toukan <tariqt-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
>>>
>>> Remove date and bump version for mlx5_core driver.
>>>
>>> Signed-off-by: Tariq Toukan <tariqt-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
>>> Signed-off-by: Saeed Mahameed <saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
>>
>>
>> So I just complained about the bnxt_re doing this. I guess I need to raise a
>> flag here too. Now to be clear I'm not against doing this version stuff. I'm
>> against using it for some internal tracking that is meaningless to the
>> kernel community.
>>
>> There is no detail in your commit message as to what this driver version is
>> used for. Can you please explain how your use of driver version is valid
>> while theirs is not?
>>
>
> Hi Denny,
>
> We don't use this for internal or any kind of tracking at all, if you
> look at the driver change log, we hardly touch the version if at all !
>
> but since we decided to remove the date since it is very misleading
> and we got some complaints that some think they have an old driver
> just because of the date, also we bumped the version so it would align
> with our drivers external package (OFED).
To me that's not a reason to include something in the kernel. At least
your external package is something that is available and not an internal
development process, but still probably no interest to anyone in the kernel.
> Anyway i don't think we are going to change this frequently or even use it.
> But if you are that much against touching this ethtool field, why
> don't you just jump into the mud and remove it from the tree, I
> imagine it is really hard for you to watch for patches doing this and
> nack them on the list.
I have no interest in policing that. I complained about the other driver
because I know patches like that were NAKed by Greg KH in the past
(ours). I'd rather not see the rdma subsystem keep making the same
mistakes is all. We want to improve our standing in the community after
all. As I said I'm not personally against it as evidenced by our driver.
>> I realize Dave has already pulled this and I'm not asking for it to be
>> reverted but maybe some discussion will help guide future patch submissions
>> which do this stuff.
>>
>
> Sure, although i don't think we are going to use those version fields
> in the future,
> please allow me to ask, how do you do your driver versioning ? how do
> you track things ?
> and what is your future vision regarding ethool->drv_version ?
That's just the thing, we don't do anything with it either really. I'm
trying to justify its existence to myself and if you folks had some whiz
bang idea for a driver version I was interested in hearing what it was.
-Denny
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <CALzJLG98D=3yMJV_q4sjVNG41AERFRU+6rwqQJsxnRuVeDTPdA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
@ 2017-05-26 17:56 ` Alexei Starovoitov
2017-05-26 18:15 ` Jason Gunthorpe
2017-05-28 12:33 ` Or Gerlitz
0 siblings, 2 replies; 72+ messages in thread
From: Alexei Starovoitov @ 2017-05-26 17:56 UTC (permalink / raw)
To: Saeed Mahameed
Cc: Ilan Tayari, David S. Miller, Doug Ledford,
netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
jsorensen-b10kYP2dOMg, Andy Shevchenko,
linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA
On Fri, May 26, 2017 at 11:59:26AM +0300, Saeed Mahameed wrote:
>
> And for FPGA support, we did it correctly this time, all the new code
> is under mlx5/core/fgpa ..
s/correctly/incorrectly/ ?
> Well, this is a well known dilemma, if one has a new feature and has no sandbox
> area to put it in the first phase, it is better to start from the most
> common place
> for that feature which is the originating place, before defining
> APIs/infrastructures,
> until the feature is complete and every body is happy about it.
There is driver/fpga to manage fpga, but mlx fpga+nic combo
will be managed via mlx5/core/fpga/
Adding fpga folks for visibility.
> 1. eswitch is a very integral steering feature of the ConnectX family
> and is valid for all ConnectX4/lx/5/6 .. cards.
and kernel has no visibility into it and mlx has no incentive to expose it,
since community needs to agree on api whereas hiding in the driver
is 'mlx territory' and much easier to push arbitrary patches to.
> "The FPGA is a bump-on-the-wire and thus affects operation of
> the mlx5_core driver on the ConnectX ASIC."
>
> This is not a general purpose FPGA ! it even doesn't have a PCI Id or
> it is own PCI function.
> it simply allows the ConnectX Chip user to inspect/inject or run user
> specif login on traffic going in/out of the chip.
... and bluefield arm cpus are not general purpose cpus and
only used to offload networking just like this fpga?
> But i agree with you some serious API brainstorming should be
> considered, but not at this stage,
> this patch only tells the ConnectX card (if you have FPGA, please enable it).
serious api discussion needs be done before things like eswitch, fpga
exposed to users otherwise we end up in the eswitch situation that
drops packets based on its own invisible logic, but with hidden fpga
it will be even more obscure.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 5/6] net/mlx5: Bump driver version
[not found] ` <ee23ad82-4a2e-8546-d41b-11f979b127bb-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
@ 2017-05-26 18:10 ` Leon Romanovsky
0 siblings, 0 replies; 72+ messages in thread
From: Leon Romanovsky @ 2017-05-26 18:10 UTC (permalink / raw)
To: Dennis Dalessandro
Cc: Saeed Mahameed, Saeed Mahameed, David S. Miller, Doug Ledford,
Linux Netdev List, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
Ilan Tayari, Tariq Toukan
[-- Attachment #1: Type: text/plain, Size: 2348 bytes --]
On Fri, May 26, 2017 at 12:55:22PM -0400, Dennis Dalessandro wrote:
> On 5/26/2017 12:35 PM, Saeed Mahameed wrote:
> > On Fri, May 26, 2017 at 3:56 PM, Dennis Dalessandro
> > <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org> wrote:
> > > On 5/23/2017 7:44 AM, Saeed Mahameed wrote:
>
> > Anyway i don't think we are going to change this frequently or even use it.
>
> > But if you are that much against touching this ethtool field, why
> > don't you just jump into the mud and remove it from the tree, I
> > imagine it is really hard for you to watch for patches doing this and
> > nack them on the list.
>
> I have no interest in policing that. I complained about the other driver
> because I know patches like that were NAKed by Greg KH in the past (ours).
> I'd rather not see the rdma subsystem keep making the same mistakes is all.
> We want to improve our standing in the community after all. As I said I'm
> not personally against it as evidenced by our driver.
I personally would like to see patch which does two things:
1. Removes all these MODULE_VERSIONS
2. Puts some hardcoded value common to kernel, so no user space
application (like modinfo) won't be broken.
>
> > > I realize Dave has already pulled this and I'm not asking for it to be
> > > reverted but maybe some discussion will help guide future patch submissions
> > > which do this stuff.
> > >
> >
> > Sure, although i don't think we are going to use those version fields
> > in the future,
> > please allow me to ask, how do you do your driver versioning ? how do
> > you track things ?
> > and what is your future vision regarding ethool->drv_version ?
>
> That's just the thing, we don't do anything with it either really. I'm
> trying to justify its existence to myself and if you folks had some whiz
> bang idea for a driver version I was interested in hearing what it was.
No, we didn't find any real usage of it too. My attempts raise attention
to that issue in netdev weren't successful, maybe you will have
more luck than me.
https://lkml.org/lkml/2016/7/15/17
https://patchwork.kernel.org/patch/9349357/
Thanks
>
> -Denny
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-05-26 17:56 ` Alexei Starovoitov
@ 2017-05-26 18:15 ` Jason Gunthorpe
[not found] ` <20170526181517.GA3860-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2017-05-28 12:33 ` Or Gerlitz
1 sibling, 1 reply; 72+ messages in thread
From: Jason Gunthorpe @ 2017-05-26 18:15 UTC (permalink / raw)
To: Alexei Starovoitov
Cc: Saeed Mahameed, Ilan Tayari, David S. Miller, Doug Ledford,
netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
jsorensen-b10kYP2dOMg, Andy Shevchenko,
linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA
On Fri, May 26, 2017 at 10:56:25AM -0700, Alexei Starovoitov wrote:
> > for that feature which is the originating place, before defining
> > APIs/infrastructures,
> > until the feature is complete and every body is happy about it.
>
> There is driver/fpga to manage fpga, but mlx fpga+nic combo
> will be managed via mlx5/core/fpga/
>
> Adding fpga folks for visibility.
It would be good to use the existing fpga loading infrastructure to
get the bitstream into the NIC, and to use the same Xilinx bitstream
format as eg Zynq does for consistency.
I'm unclear how this works - there must be more to it than just a
'bump on the wire', there must be some communication channel between
the FPGA and Linux to set operational data (eg load keys) etc.
If that is register mapped into a PCI-BAR someplace then it really
should use the FPGA layer functions to manage binding drivers to that
register window.
If it is mailbox command based then it is not as good of a fit.
Is this FPGA expected to be customer programmable? In that case you
really need the full infrastructure to bind the right driver (possibly
a customer driver) to the current FPGA, to expose the correct
operational interface to the kernel.
Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <CALzJLG9YNpagdJAcrh6O0jJhZWtsck6KigRtVxyjkArTm=82ew-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
@ 2017-05-26 18:31 ` Jes Sorensen
2017-05-28 7:24 ` Ilan Tayari
0 siblings, 1 reply; 72+ messages in thread
From: Jes Sorensen @ 2017-05-26 18:31 UTC (permalink / raw)
To: Saeed Mahameed
Cc: Ilan Tayari, Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA
On 05/26/2017 04:29 AM, Saeed Mahameed wrote:
> On Thu, May 25, 2017 at 11:48 PM, Jes Sorensen <jsorensen-b10kYP2dOMg@public.gmane.org> wrote:
>> On 05/25/2017 06:40 AM, Saeed Mahameed wrote:
> Hi Jes,
>
> No, It is clearly stated in the commit message :
>
> "The FPGA is a bump-on-the-wire and thus affects operation of
> the mlx5_core driver on the ConnectX ASIC."
>
> Which means mlx5 FPGA user can only write logic which affects only
> packets going in/out
> A ConnectX chip - so it is only network stuff -.
>
>> We have this with other devices in the kernel where a primary device driver
>> provides an interface for an additional sub-driver to access another device
>> behind it. Like bt-coexist in some of the wifi drivers allowing access to a
>> bluetooth device behind it.
>
> Blutooth over wifi or vise versa is a very good example to what you
> are requesting.
> But, it doesn't fit to what we are trying to do here. mlx5 FGPA is a
> ConnectX card feature, not a new protocol.
In that case it would need to be an optional module that can be enabled
or disabled at build time.
Cheers,
Jes
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 5/6] net/mlx5: Bump driver version
2017-05-26 16:55 ` Dennis Dalessandro
[not found] ` <ee23ad82-4a2e-8546-d41b-11f979b127bb-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
@ 2017-05-26 21:53 ` Jakub Kicinski
[not found] ` <20170526145318.7fd8c8e2-68UzVGuGftmUSpRRplVxJ1aTQe2KTcn/@public.gmane.org>
1 sibling, 1 reply; 72+ messages in thread
From: Jakub Kicinski @ 2017-05-26 21:53 UTC (permalink / raw)
To: Dennis Dalessandro
Cc: Saeed Mahameed, Saeed Mahameed, David S. Miller, Doug Ledford,
Linux Netdev List, linux-rdma, Ilan Tayari, Tariq Toukan
On Fri, 26 May 2017 12:55:22 -0400, Dennis Dalessandro wrote:
> >> I realize Dave has already pulled this and I'm not asking for it to be
> >> reverted but maybe some discussion will help guide future patch submissions
> >> which do this stuff.
> >>
> >
> > Sure, although i don't think we are going to use those version fields
> > in the future,
> > please allow me to ask, how do you do your driver versioning ? how do
> > you track things ?
> > and what is your future vision regarding ethool->drv_version ?
>
> That's just the thing, we don't do anything with it either really. I'm
> trying to justify its existence to myself and if you folks had some whiz
> bang idea for a driver version I was interested in hearing what it was.
FWIW I put VERMAGIC_STRING in drv_version:
# ethtool -i p4p1
driver: nfp
version: 4.12.0-rc2-perf-00282-gc42dfc79
...
I think I got that idea from Felix Fietkau. The backport-ified version
of the driver [1] uses git hash and "o-o-t" sting:
# ethtool -i p4p1
driver: nfp
version: 49e5c6abf5b2 (o-o-t)
...
So it's pretty easy to tell which driver the customer is using. I
could probably throw in the output of $(git rev-list --count HEAD) to
have an automatic monotonically increasing "version" when built
out-of-tree, hm...
[1] https://github.com/Netronome/nfp-drv-kmods
^ permalink raw reply [flat|nested] 72+ messages in thread
* RE: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <20170526181517.GA3860-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
@ 2017-05-28 7:22 ` Ilan Tayari
[not found] ` <AM4PR0501MB1940330F0EBAA819C87C5278DBF20-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Ilan Tayari @ 2017-05-28 7:22 UTC (permalink / raw)
To: Jason Gunthorpe, Alexei Starovoitov
Cc: Saeed Mahameed, David S. Miller, Doug Ledford,
netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
jsorensen-b10kYP2dOMg, Andy Shevchenko,
linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
> -----Original Message-----
> From: Jason Gunthorpe [mailto:jgunthorpe@obsidianresearch.com]
>
> On Fri, May 26, 2017 at 10:56:25AM -0700, Alexei Starovoitov wrote:
>
> > > for that feature which is the originating place, before defining
> > > APIs/infrastructures,
> > > until the feature is complete and every body is happy about it.
> >
> > There is driver/fpga to manage fpga, but mlx fpga+nic combo
> > will be managed via mlx5/core/fpga/
> >
> > Adding fpga folks for visibility.
>
> It would be good to use the existing fpga loading infrastructure to
> get the bitstream into the NIC, and to use the same Xilinx bitstream
> format as eg Zynq does for consistency.
>
> I'm unclear how this works - there must be more to it than just a
> 'bump on the wire', there must be some communication channel between
> the FPGA and Linux to set operational data (eg load keys) etc.
>
> If that is register mapped into a PCI-BAR someplace then it really
> should use the FPGA layer functions to manage binding drivers to that
> register window.
>
> If it is mailbox command based then it is not as good of a fit.
>
Hi Jason,
Thanks for taking the time to consider this!
This is neither PCI-bar mapped, nor mailbox command.
The FPGA is indeed a bump-on-the-wire.
(It has I2C to the CX4 chip, but that is for debug purposes, and too slow
to perform real programming)
The FPGA is programmed with RoCE packets. We are going to share a lot of
details about this in the IPSec offload patchset, so I don't want to
repeat it here.
In short, we open a RoCEv2 QP between the connect and the FPGA chips.
Over this QP, we can communicate at high speed with the FPGA.
> Is this FPGA expected to be customer programmable? In that case you
> really need the full infrastructure to bind the right driver (possibly
> a customer driver) to the current FPGA, to expose the correct
> operational interface to the kernel.
Most of the Innova family of cards are not customer-programmable.
Over time, they may require an upgrade (essentially a FW upgrade) but not
with customer logic.
One flavor of the product, the Innova Flex, allows customer logic in the
FPGA. But even then it is "wrapped" by Mellanox shell logic.
We plan to have an in-kernel API for writing client drivers for Innova
Flex.
>
> Jason
^ permalink raw reply [flat|nested] 72+ messages in thread
* RE: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-05-26 18:31 ` Jes Sorensen
@ 2017-05-28 7:24 ` Ilan Tayari
2017-06-02 20:31 ` Jes Sorensen
0 siblings, 1 reply; 72+ messages in thread
From: Ilan Tayari @ 2017-05-28 7:24 UTC (permalink / raw)
To: Jes Sorensen, Saeed Mahameed
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev, linux-rdma
> -----Original Message-----
> From: Jes Sorensen [mailto:jsorensen@fb.com]
>
> On 05/26/2017 04:29 AM, Saeed Mahameed wrote:
> > On Thu, May 25, 2017 at 11:48 PM, Jes Sorensen <jsorensen@fb.com> wrote:
> >> On 05/25/2017 06:40 AM, Saeed Mahameed wrote:
> > Hi Jes,
> >
> > No, It is clearly stated in the commit message :
> >
> > "The FPGA is a bump-on-the-wire and thus affects operation of
> > the mlx5_core driver on the ConnectX ASIC."
> >
> > Which means mlx5 FPGA user can only write logic which affects only
> > packets going in/out
> > A ConnectX chip - so it is only network stuff -.
> >
> >> We have this with other devices in the kernel where a primary device
> driver
> >> provides an interface for an additional sub-driver to access another
> device
> >> behind it. Like bt-coexist in some of the wifi drivers allowing access
> to a
> >> bluetooth device behind it.
> >
> > Blutooth over wifi or vise versa is a very good example to what you
> > are requesting.
> > But, it doesn't fit to what we are trying to do here. mlx5 FGPA is a
> > ConnectX card feature, not a new protocol.
>
> In that case it would need to be an optional module that can be enabled
> or disabled at build time.
Jes,
It is already modular like that. See CONFIG_MLX5_FPGA.
Ilan.
>
> Cheers,
> Jes
>
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-05-26 17:56 ` Alexei Starovoitov
2017-05-26 18:15 ` Jason Gunthorpe
@ 2017-05-28 12:33 ` Or Gerlitz
1 sibling, 0 replies; 72+ messages in thread
From: Or Gerlitz @ 2017-05-28 12:33 UTC (permalink / raw)
To: Alexei Starovoitov
Cc: Saeed Mahameed, Ilan Tayari, David S. Miller, Doug Ledford,
netdev, jsorensen, Andy Shevchenko, linux-fpga, Alan Tull,
yi1.li
On Fri, May 26, 2017 at 8:56 PM, Alexei Starovoitov
<alexei.starovoitov@gmail.com> wrote:
> On Fri, May 26, 2017 at 11:59:26AM +0300, Saeed Mahameed wrote:
>> But i agree with you some serious API brainstorming should be
>> considered, but not at this stage, this patch only tells the ConnectX card
>> (if you have FPGA, please enable it).
> serious api discussion needs be done before things like eswitch, fpga
> exposed to users otherwise we end up in the eswitch situation that
> drops packets based on its own invisible logic, but with hidden fpga
> it will be even more obscure.
Alexei,
ACK for your comment on e-switch, what we call the legacy mode for SRIOV in
Linux was a big mistake.
As you know, since 4.8 we're @ MLNX are working to fix it... we introduced a
model for VF port representors which has set the foundations for
e-switch data-path
to be programmed as an offload for kernel SW switching. We also
introduced the code
to offload TC flower base data-path and support that in for both
switch (Spectrum) ASIC
(mlxsw driver) and NIC ASIC (mlx5 driver)
Or.
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 5/6] net/mlx5: Bump driver version
[not found] ` <20170526145318.7fd8c8e2-68UzVGuGftmUSpRRplVxJ1aTQe2KTcn/@public.gmane.org>
@ 2017-05-29 5:47 ` Leon Romanovsky
0 siblings, 0 replies; 72+ messages in thread
From: Leon Romanovsky @ 2017-05-29 5:47 UTC (permalink / raw)
To: Jakub Kicinski
Cc: Dennis Dalessandro, Saeed Mahameed, Saeed Mahameed,
David S. Miller, Doug Ledford, Linux Netdev List,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, Ilan Tayari, Tariq Toukan
[-- Attachment #1: Type: text/plain, Size: 1496 bytes --]
On Fri, May 26, 2017 at 02:53:18PM -0700, Jakub Kicinski wrote:
> On Fri, 26 May 2017 12:55:22 -0400, Dennis Dalessandro wrote:
> > >> I realize Dave has already pulled this and I'm not asking for it to be
> > >> reverted but maybe some discussion will help guide future patch submissions
> > >> which do this stuff.
> > >>
> > >
> > > Sure, although i don't think we are going to use those version fields
> > > in the future,
> > > please allow me to ask, how do you do your driver versioning ? how do
> > > you track things ?
> > > and what is your future vision regarding ethool->drv_version ?
> >
> > That's just the thing, we don't do anything with it either really. I'm
> > trying to justify its existence to myself and if you folks had some whiz
> > bang idea for a driver version I was interested in hearing what it was.
>
> FWIW I put VERMAGIC_STRING in drv_version:
>
> # ethtool -i p4p1
> driver: nfp
> version: 4.12.0-rc2-perf-00282-gc42dfc79
> ...
>
> I think I got that idea from Felix Fietkau. The backport-ified version
> of the driver [1] uses git hash and "o-o-t" sting:
>
> # ethtool -i p4p1
> driver: nfp
> version: 49e5c6abf5b2 (o-o-t)
> ...
>
> So it's pretty easy to tell which driver the customer is using. I
> could probably throw in the output of $(git rev-list --count HEAD) to
> have an automatic monotonically increasing "version" when built
> out-of-tree, hm...
Thanks for the pointer. I'll try to evaluate it.
>
> [1] https://github.com/Netronome/nfp-drv-kmods
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <AM4PR0501MB1940330F0EBAA819C87C5278DBF20-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
@ 2017-05-29 15:31 ` Jason Gunthorpe
[not found] ` <20170529153131.GB7924-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Jason Gunthorpe @ 2017-05-29 15:31 UTC (permalink / raw)
To: Ilan Tayari
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
Andy Shevchenko, linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
On Sun, May 28, 2017 at 07:22:27AM +0000, Ilan Tayari wrote:
> This is neither PCI-bar mapped, nor mailbox command.
> The FPGA is indeed a bump-on-the-wire.
> (It has I2C to the CX4 chip, but that is for debug purposes, and too slow
> to perform real programming)
Wait.. So if it truely has nothing to do with the existing mellanox
driver, then nothing more than the fpga loader should be in the mlx5
directory?
> One flavor of the product, the Innova Flex, allows customer logic in the
> FPGA. But even then it is "wrapped" by Mellanox shell logic.
> We plan to have an in-kernel API for writing client drivers for Innova
> Flex.
The FPGA subsystem already has APIs for partial reconfiguration use
cases, you probably should not re-invent that under the mlx5 driver.
Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* RE: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <20170529153131.GB7924-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
@ 2017-05-29 15:58 ` Ilan Tayari
[not found] ` <AM4PR0501MB1940D05A19F098286B99EAD0DBF30-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Ilan Tayari @ 2017-05-29 15:58 UTC (permalink / raw)
To: Jason Gunthorpe
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
Andy Shevchenko, linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
> -----Original Message-----
> From: Jason Gunthorpe [mailto:jgunthorpe@obsidianresearch.com]
> Subject: Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
>
> On Sun, May 28, 2017 at 07:22:27AM +0000, Ilan Tayari wrote:
>
> > This is neither PCI-bar mapped, nor mailbox command.
> > The FPGA is indeed a bump-on-the-wire.
> > (It has I2C to the CX4 chip, but that is for debug purposes, and too
> slow
> > to perform real programming)
>
> Wait.. So if it truely has nothing to do with the existing mellanox
> driver, then nothing more than the fpga loader should be in the mlx5
> directory?
True, except in specific cases when the FPGA may mangle the packets in
a way that the netdevice configures, and the driver needs to adapt the
data path.
Such is the case of IPSec and TLS offloads.
Those are tied to the mlx5 Ethernet driver (isolated with a kconfig).
>
> > One flavor of the product, the Innova Flex, allows customer logic in the
> > FPGA. But even then it is "wrapped" by Mellanox shell logic.
> > We plan to have an in-kernel API for writing client drivers for Innova
> > Flex.
>
> The FPGA subsystem already has APIs for partial reconfiguration use
> cases, you probably should not re-invent that under the mlx5 driver.
When the time comes to introduce the API for that, we will investigate all
Options available. Currently this is not in the plans for Linux.
(We do have an out-of-tree 'hack' driver for this, but it's planned to be
replaced with something more appropriate later on)
>
> Jason
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <AM4PR0501MB1940D05A19F098286B99EAD0DBF30-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
@ 2017-05-29 16:02 ` Jason Gunthorpe
2017-05-29 16:05 ` Ilan Tayari
0 siblings, 1 reply; 72+ messages in thread
From: Jason Gunthorpe @ 2017-05-29 16:02 UTC (permalink / raw)
To: Ilan Tayari
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
Andy Shevchenko, linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
On Mon, May 29, 2017 at 03:58:33PM +0000, Ilan Tayari wrote:
> > From: Jason Gunthorpe [mailto:jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org]
> > Subject: Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
> >
> > On Sun, May 28, 2017 at 07:22:27AM +0000, Ilan Tayari wrote:
> >
> > > This is neither PCI-bar mapped, nor mailbox command.
> > > The FPGA is indeed a bump-on-the-wire.
> > > (It has I2C to the CX4 chip, but that is for debug purposes, and too
> > slow
> > > to perform real programming)
> >
> > Wait.. So if it truely has nothing to do with the existing mellanox
> > driver, then nothing more than the fpga loader should be in the mlx5
> > directory?
>
> True, except in specific cases when the FPGA may mangle the packets in
> a way that the netdevice configures, and the driver needs to adapt the
> data path.
> Such is the case of IPSec and TLS offloads.
> Those are tied to the mlx5 Ethernet driver (isolated with a kconfig).
But there is nothing stopping this sort of FPGA mangling logic being
downstream of any NIC, Mellanox is just the first to do this.
I think you'd be better to add something to the net stack to model
this post-nic mangling hardware, than trying to hide it in a driver.
Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* RE: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-05-29 16:02 ` Jason Gunthorpe
@ 2017-05-29 16:05 ` Ilan Tayari
[not found] ` <AM4PR0501MB194037FF8F17466BC9ECC73DDBF30-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Ilan Tayari @ 2017-05-29 16:05 UTC (permalink / raw)
To: Jason Gunthorpe
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev, linux-rdma, jsorensen, Andy Shevchenko,
linux-fpga, Alan Tull, yi1.li, Boris Pismenny
> -----Original Message-----
> From: Jason Gunthorpe [mailto:jgunthorpe@obsidianresearch.com]
> Subject: Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
>
> On Mon, May 29, 2017 at 03:58:33PM +0000, Ilan Tayari wrote:
> > > From: Jason Gunthorpe [mailto:jgunthorpe@obsidianresearch.com]
> > > Subject: Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for
> Innova
> > >
> > > On Sun, May 28, 2017 at 07:22:27AM +0000, Ilan Tayari wrote:
> > >
> > > > This is neither PCI-bar mapped, nor mailbox command.
> > > > The FPGA is indeed a bump-on-the-wire.
> > > > (It has I2C to the CX4 chip, but that is for debug purposes, and too
> > > slow
> > > > to perform real programming)
> > >
> > > Wait.. So if it truely has nothing to do with the existing mellanox
> > > driver, then nothing more than the fpga loader should be in the mlx5
> > > directory?
> >
> > True, except in specific cases when the FPGA may mangle the packets in
> > a way that the netdevice configures, and the driver needs to adapt the
> > data path.
>
> > Such is the case of IPSec and TLS offloads.
> > Those are tied to the mlx5 Ethernet driver (isolated with a kconfig).
>
> But there is nothing stopping this sort of FPGA mangling logic being
> downstream of any NIC, Mellanox is just the first to do this.
>
> I think you'd be better to add something to the net stack to model
> this post-nic mangling hardware, than trying to hide it in a driver.
Of course.
For IPSec, this is already in the kernel.
See this patchset:
http://www.mail-archive.com/netdev@vger.kernel.org/msg162876.html
For TLS, Dave Watson is working with our guys to add it.
>
> Jason
^ permalink raw reply [flat|nested] 72+ messages in thread
* RE: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <AM4PR0501MB194037FF8F17466BC9ECC73DDBF30-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
@ 2017-05-29 16:09 ` Ilan Tayari
[not found] ` <AM4PR0501MB19409139227E11A4A7F82F0FDBF30-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Ilan Tayari @ 2017-05-29 16:09 UTC (permalink / raw)
To: Ilan Tayari, Jason Gunthorpe
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
Andy Shevchenko, linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="utf-8", Size: 2158 bytes --]
> -----Original Message-----
> From: netdev-owner@vger.kernel.org [mailto:netdev-owner@vger.kernel.org]
>
> > -----Original Message-----
> > From: Jason Gunthorpe [mailto:jgunthorpe@obsidianresearch.com]
> > Subject: Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
> >
> > On Mon, May 29, 2017 at 03:58:33PM +0000, Ilan Tayari wrote:
> > > > From: Jason Gunthorpe [mailto:jgunthorpe@obsidianresearch.com]
> > > > Subject: Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for
> > Innova
> > > >
> > > > On Sun, May 28, 2017 at 07:22:27AM +0000, Ilan Tayari wrote:
> > > >
> > > > > This is neither PCI-bar mapped, nor mailbox command.
> > > > > The FPGA is indeed a bump-on-the-wire.
> > > > > (It has I2C to the CX4 chip, but that is for debug purposes, and
> too
> > > > slow
> > > > > to perform real programming)
> > > >
> > > > Wait.. So if it truely has nothing to do with the existing mellanox
> > > > driver, then nothing more than the fpga loader should be in the mlx5
> > > > directory?
> > >
> > > True, except in specific cases when the FPGA may mangle the packets in
> > > a way that the netdevice configures, and the driver needs to adapt the
> > > data path.
> >
> > > Such is the case of IPSec and TLS offloads.
> > > Those are tied to the mlx5 Ethernet driver (isolated with a kconfig).
> >
> > But there is nothing stopping this sort of FPGA mangling logic being
> > downstream of any NIC, Mellanox is just the first to do this.
> >
> > I think you'd be better to add something to the net stack to model
> > this post-nic mangling hardware, than trying to hide it in a driver.
>
> Of course.
>
> For IPSec, this is already in the kernel.
> See this patchset:
> http://www.mail-archive.com/netdev@vger.kernel.org/msg162876.html
Sorry, I pointed at the RFC by mistake.
This is the relevant pull request:
https://patchwork.ozlabs.org/patch/752707/
>
> For TLS, Dave Watson is working with our guys to add it.
>
> >
> > Jason
N‹§²æìr¸›yúèšØb²X¬¶Ç§vØ^–)Þº{.nÇ+‰·¥Š{±ÙšŠ{ayº\x1dʇڙë,j\a¢f£¢·hš‹»öì\x17/oSc¾™Ú³9˜uÀ¦æå‰È&jw¨®\x03(階ŠÝ¢j"ú\x1a¶^[m§ÿïêäz¹Þ–Šàþf£¢·hšˆ§~ˆmš
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <AM4PR0501MB19409139227E11A4A7F82F0FDBF30-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
@ 2017-06-01 15:37 ` Jason Gunthorpe
[not found] ` <20170601153704.GA1680-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Jason Gunthorpe @ 2017-06-01 15:37 UTC (permalink / raw)
To: Ilan Tayari
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
Andy Shevchenko, linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
On Mon, May 29, 2017 at 04:09:06PM +0000, Ilan Tayari wrote:
> > For IPSec, this is already in the kernel.
> > See this patchset:
> > http://www.mail-archive.com/netdev-u79uwXL29TY76Z2rM5mHXA@public.gmane.org/msg162876.html
>
> Sorry, I pointed at the RFC by mistake.
>
> This is the relevant pull request:
> https://patchwork.ozlabs.org/patch/752707/
This is connecting ipsec to a netdev, while Innova seems to be a
network connected ipsec accelerator configured using IP packets.
Those two things don't seem to be the same.
Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23
2017-05-25 16:02 ` [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23 David Miller
@ 2017-06-01 22:57 ` Doug Ledford
[not found] ` <1496357879.7171.76.camel-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2017-06-14 19:30 ` Doug Ledford
1 sibling, 1 reply; 72+ messages in thread
From: Doug Ledford @ 2017-06-01 22:57 UTC (permalink / raw)
To: David Miller, saeedm; +Cc: netdev, linux-rdma, ilant
On Thu, 2017-05-25 at 12:02 -0400, David Miller wrote:
> From: Saeed Mahameed <saeedm@mellanox.com>
> Date: Tue, 23 May 2017 14:43:58 +0300
>
> > Hi Dave and Doug,
> >
> > This series introduces some small updates and FPGA support to the
> mlx5
> > core/ethernet and IB drivers.
> >
> > For more details please see below.
> >
> > Please pull and let me know if there's any problem.
>
> Ok, I've pulled this into net-next.
>
> Doug let me know if there are any merge hassles we need to coordinate
> on.
>
> Thanks.
Will do. But is the question of a possible revert settled? Because
once I pull, a revert means I have to pull all the way up to the revert
as well...
--
Doug Ledford <dledford@redhat.com>
GPG KeyID: B826A3330E572FDD
Key fingerprint = AE6B 1BDA 122B 23B4 265B 1274 B826 A333 0E57 2FDD
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23
[not found] ` <1496357879.7171.76.camel-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
@ 2017-06-02 15:39 ` Leon Romanovsky
[not found] ` <20170602153940.GX5406-U/DQcQFIOTAAJjI8aNfphQ@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Leon Romanovsky @ 2017-06-02 15:39 UTC (permalink / raw)
To: Doug Ledford
Cc: David Miller, saeedm-VPRAkNaXOzVWk0Htik3J/w,
netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
ilant-VPRAkNaXOzVWk0Htik3J/w
[-- Attachment #1: Type: text/plain, Size: 1361 bytes --]
On Thu, Jun 01, 2017 at 06:57:59PM -0400, Doug Ledford wrote:
> On Thu, 2017-05-25 at 12:02 -0400, David Miller wrote:
> > From: Saeed Mahameed <saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
> > Date: Tue, 23 May 2017 14:43:58 +0300
> >
> > > Hi Dave and Doug,
> > >
> > > This series introduces some small updates and FPGA support to the
> > mlx5
> > > core/ethernet and IB drivers.
> > >
> > > For more details please see below.
> > >
> > > Please pull and let me know if there's any problem.
> >
> > Ok, I've pulled this into net-next.
> >
> > Doug let me know if there are any merge hassles we need to coordinate
> > on.
> >
> > Thanks.
>
> Will do. But is the question of a possible revert settled? Because
> once I pull, a revert means I have to pull all the way up to the revert
> as well...
Revert is unlikely to happen, Saeed is working with Alexey to resolve
his raised concerns.
Thanks
>
> --
> Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
> GPG KeyID: B826A3330E572FDD
>
> Key fingerprint = AE6B 1BDA 122B 23B4 265B 1274 B826 A333 0E57 2FDD
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23
[not found] ` <20170602153940.GX5406-U/DQcQFIOTAAJjI8aNfphQ@public.gmane.org>
@ 2017-06-02 16:06 ` Alexei Starovoitov
[not found] ` <20170602160641.ylowbobe5v72ui7g-+o4/htvd0TCa6kscz5V53/3mLCh9rsb+VpNB7YpNyf8@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Alexei Starovoitov @ 2017-06-02 16:06 UTC (permalink / raw)
To: Leon Romanovsky
Cc: Doug Ledford, David Miller, saeedm-VPRAkNaXOzVWk0Htik3J/w,
netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
ilant-VPRAkNaXOzVWk0Htik3J/w
On Fri, Jun 02, 2017 at 06:39:40PM +0300, Leon Romanovsky wrote:
> On Thu, Jun 01, 2017 at 06:57:59PM -0400, Doug Ledford wrote:
> > On Thu, 2017-05-25 at 12:02 -0400, David Miller wrote:
> > > From: Saeed Mahameed <saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
> > > Date: Tue, 23 May 2017 14:43:58 +0300
> > >
> > > > Hi Dave and Doug,
> > > >
> > > > This series introduces some small updates and FPGA support to the
> > > mlx5
> > > > core/ethernet and IB drivers.
> > > >
> > > > For more details please see below.
> > > >
> > > > Please pull and let me know if there's any problem.
> > >
> > > Ok, I've pulled this into net-next.
> > >
> > > Doug let me know if there are any merge hassles we need to coordinate
> > > on.
> > >
> > > Thanks.
> >
> > Will do. But is the question of a possible revert settled? Because
> > once I pull, a revert means I have to pull all the way up to the revert
> > as well...
>
> Revert is unlikely to happen, Saeed is working with Alexey to resolve
> his raised concerns.
I still think the direction is absolutely wrong.
Reinventing fpga interfaces via nic driver is no-go.
mlx needs to send a patch to revert it to show that they are willing
to listen to the community.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23
[not found] ` <20170602160641.ylowbobe5v72ui7g-+o4/htvd0TCa6kscz5V53/3mLCh9rsb+VpNB7YpNyf8@public.gmane.org>
@ 2017-06-02 16:08 ` David Miller
[not found] ` <20170602.120839.1394660754953676217.davem-fT/PcQaiUtIeIZ0/mPfg9Q@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: David Miller @ 2017-06-02 16:08 UTC (permalink / raw)
To: alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w
Cc: leon-DgEjT+Ai2ygdnm+yROfE0A, dledford-H+wXaHxf7aLQT0dZR+AlfA,
saeedm-VPRAkNaXOzVWk0Htik3J/w, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, ilant-VPRAkNaXOzVWk0Htik3J/w
From: Alexei Starovoitov <alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Date: Fri, 2 Jun 2017 09:06:43 -0700
> On Fri, Jun 02, 2017 at 06:39:40PM +0300, Leon Romanovsky wrote:
>> On Thu, Jun 01, 2017 at 06:57:59PM -0400, Doug Ledford wrote:
>> > On Thu, 2017-05-25 at 12:02 -0400, David Miller wrote:
>> > > From: Saeed Mahameed <saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
>> > > Date: Tue, 23 May 2017 14:43:58 +0300
>> > >
>> > > > Hi Dave and Doug,
>> > > >
>> > > > This series introduces some small updates and FPGA support to the
>> > > mlx5
>> > > > core/ethernet and IB drivers.
>> > > >
>> > > > For more details please see below.
>> > > >
>> > > > Please pull and let me know if there's any problem.
>> > >
>> > > Ok, I've pulled this into net-next.
>> > >
>> > > Doug let me know if there are any merge hassles we need to coordinate
>> > > on.
>> > >
>> > > Thanks.
>> >
>> > Will do. But is the question of a possible revert settled? Because
>> > once I pull, a revert means I have to pull all the way up to the revert
>> > as well...
>>
>> Revert is unlikely to happen, Saeed is working with Alexey to resolve
>> his raised concerns.
>
> I still think the direction is absolutely wrong.
> Reinventing fpga interfaces via nic driver is no-go.
> mlx needs to send a patch to revert it to show that they are willing
> to listen to the community.
"Two people from Facebook" is not the commnity Alexei. :-)
And ironically, you guys will want to be one of the main users of
these facilities for the KTLS offload bits, so the desire to separate
it out and make "backports easier" really confuses me.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23
[not found] ` <20170602.120839.1394660754953676217.davem-fT/PcQaiUtIeIZ0/mPfg9Q@public.gmane.org>
@ 2017-06-02 16:57 ` Alexei Starovoitov
[not found] ` <20170602165736.nwunidodmu6xsmuv-+o4/htvd0TCa6kscz5V53/3mLCh9rsb+VpNB7YpNyf8@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Alexei Starovoitov @ 2017-06-02 16:57 UTC (permalink / raw)
To: David Miller
Cc: leon-DgEjT+Ai2ygdnm+yROfE0A, dledford-H+wXaHxf7aLQT0dZR+AlfA,
saeedm-VPRAkNaXOzVWk0Htik3J/w, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, ilant-VPRAkNaXOzVWk0Htik3J/w
On Fri, Jun 02, 2017 at 12:08:39PM -0400, David Miller wrote:
> From: Alexei Starovoitov <alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> Date: Fri, 2 Jun 2017 09:06:43 -0700
>
> > On Fri, Jun 02, 2017 at 06:39:40PM +0300, Leon Romanovsky wrote:
> >> On Thu, Jun 01, 2017 at 06:57:59PM -0400, Doug Ledford wrote:
> >> > On Thu, 2017-05-25 at 12:02 -0400, David Miller wrote:
> >> > > From: Saeed Mahameed <saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
> >> > > Date: Tue, 23 May 2017 14:43:58 +0300
> >> > >
> >> > > > Hi Dave and Doug,
> >> > > >
> >> > > > This series introduces some small updates and FPGA support to the
> >> > > mlx5
> >> > > > core/ethernet and IB drivers.
> >> > > >
> >> > > > For more details please see below.
> >> > > >
> >> > > > Please pull and let me know if there's any problem.
> >> > >
> >> > > Ok, I've pulled this into net-next.
> >> > >
> >> > > Doug let me know if there are any merge hassles we need to coordinate
> >> > > on.
> >> > >
> >> > > Thanks.
> >> >
> >> > Will do. But is the question of a possible revert settled? Because
> >> > once I pull, a revert means I have to pull all the way up to the revert
> >> > as well...
> >>
> >> Revert is unlikely to happen, Saeed is working with Alexey to resolve
> >> his raised concerns.
> >
> > I still think the direction is absolutely wrong.
> > Reinventing fpga interfaces via nic driver is no-go.
> > mlx needs to send a patch to revert it to show that they are willing
> > to listen to the community.
>
> "Two people from Facebook" is not the commnity Alexei. :-)
I've read the thread that fpga folks are not excited about
fpga behind the nic either... but yeah i'm mainly expressing
my opinions here.
> And ironically, you guys will want to be one of the main users of
> these facilities for the KTLS offload bits, so the desire to separate
> it out and make "backports easier" really confuses me.
We will not be using ktls offload via hidden fpga which
is not properly exposed to the kernel.
We've learned our lesson with eswitch.
ktls patches are independent of hw offload. It's great that
mlx fpga can accelerate ktls tx which is extra confirmation
that ktls api is on the right track. There is still 'ktls rx'
to finish and 'ktls rx offload' is even trickier, since it's
more intrusive into the core networking. Once crypto scatter gather
is improved, we expect to see 8% perf gain out of sw ktls alone.
So ktls hw offload is complementary and not mandatory.
Back to eswitch analogy. All I hear from mlx is "at this stage
it's not approriate ...". Well, I've been asking to do
'the right thing' for eswitch for months now and eswitch was
in the driver for years. What are the chances that this fpga
will ever be exposed? Once the code is merged there is no
incentive for the vendor to do it differently.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-05-28 7:24 ` Ilan Tayari
@ 2017-06-02 20:31 ` Jes Sorensen
[not found] ` <4c164e09-0103-7daf-e9f8-9260223ada08-b10kYP2dOMg@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Jes Sorensen @ 2017-06-02 20:31 UTC (permalink / raw)
To: Ilan Tayari, Saeed Mahameed
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev, linux-rdma, Kernel Team
On 05/28/2017 03:24 AM, Ilan Tayari wrote:
>> -----Original Message-----
>> From: Jes Sorensen [mailto:jsorensen@fb.com]
>>
>> On 05/26/2017 04:29 AM, Saeed Mahameed wrote:
>>> On Thu, May 25, 2017 at 11:48 PM, Jes Sorensen <jsorensen@fb.com> wrote:
>>>> On 05/25/2017 06:40 AM, Saeed Mahameed wrote:
>>> Hi Jes,
>>>
>>> No, It is clearly stated in the commit message :
>>>
>>> "The FPGA is a bump-on-the-wire and thus affects operation of
>>> the mlx5_core driver on the ConnectX ASIC."
>>>
>>> Which means mlx5 FPGA user can only write logic which affects only
>>> packets going in/out
>>> A ConnectX chip - so it is only network stuff -.
>>>
>>>> We have this with other devices in the kernel where a primary device
>> driver
>>>> provides an interface for an additional sub-driver to access another
>> device
>>>> behind it. Like bt-coexist in some of the wifi drivers allowing access
>> to a
>>>> bluetooth device behind it.
>>>
>>> Blutooth over wifi or vise versa is a very good example to what you
>>> are requesting.
>>> But, it doesn't fit to what we are trying to do here. mlx5 FGPA is a
>>> ConnectX card feature, not a new protocol.
>>
>> In that case it would need to be an optional module that can be enabled
>> or disabled at build time.
>
> Jes,
>
> It is already modular like that. See CONFIG_MLX5_FPGA.
[jes@xpeas netdev.git]$ grep CONFIG_MLX5_FPGA
drivers/net/ethernet/mellanox/mlx5/core/*
[jes@xpeas netdev.git]$
Which git tree am I supposed to look at?
Jes
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <4c164e09-0103-7daf-e9f8-9260223ada08-b10kYP2dOMg@public.gmane.org>
@ 2017-06-02 20:33 ` Doug Ledford
0 siblings, 0 replies; 72+ messages in thread
From: Doug Ledford @ 2017-06-02 20:33 UTC (permalink / raw)
To: Jes Sorensen, Ilan Tayari, Saeed Mahameed
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
Kernel Team
On Fri, 2017-06-02 at 16:31 -0400, Jes Sorensen wrote:
>
> > It is already modular like that. See CONFIG_MLX5_FPGA.
>
> [jes@xpeas netdev.git]$ grep CONFIG_MLX5_FPGA
> drivers/net/ethernet/mellanox/mlx5/core/*
> [jes@xpeas netdev.git]$
>
> Which git tree am I supposed to look at?
net-next
--
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
GPG KeyID: B826A3330E572FDD
Key fingerprint = AE6B 1BDA 122B 23B4 265B 1274 B826 A333 0E57 2FDD
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23
[not found] ` <20170602165736.nwunidodmu6xsmuv-+o4/htvd0TCa6kscz5V53/3mLCh9rsb+VpNB7YpNyf8@public.gmane.org>
@ 2017-06-03 19:46 ` Or Gerlitz
2017-06-03 22:45 ` Saeed Mahameed
1 sibling, 0 replies; 72+ messages in thread
From: Or Gerlitz @ 2017-06-03 19:46 UTC (permalink / raw)
To: Alexei Starovoitov
Cc: David Miller, Leon Romanovsky, Doug Ledford, Saeed Mahameed,
Linux Netdev List, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
Ilan Tayari
On Fri, Jun 2, 2017 at 7:57 PM, Alexei Starovoitov
<alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
> Back to eswitch analogy. All I hear from mlx is "at this stage
> it's not approriate ...". Well, I've been asking to do
> 'the right thing' for eswitch for months now and eswitch was
> in the driver for years. What are the chances that this fpga
> will ever be exposed? Once the code is merged there is no
> incentive for the vendor to do it differently.
Alexei, can you please clarify the claims re eswitch? are you
referring to the interfaces to program SRIOV eswitch? if not, what?
Or.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23
[not found] ` <20170602165736.nwunidodmu6xsmuv-+o4/htvd0TCa6kscz5V53/3mLCh9rsb+VpNB7YpNyf8@public.gmane.org>
2017-06-03 19:46 ` Or Gerlitz
@ 2017-06-03 22:45 ` Saeed Mahameed
1 sibling, 0 replies; 72+ messages in thread
From: Saeed Mahameed @ 2017-06-03 22:45 UTC (permalink / raw)
To: Alexei Starovoitov
Cc: David Miller, Leon Romanovsky, Doug Ledford, Saeed Mahameed,
Linux Netdev List, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
Ilan Tayari
On Fri, Jun 2, 2017 at 7:57 PM, Alexei Starovoitov
<alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
> On Fri, Jun 02, 2017 at 12:08:39PM -0400, David Miller wrote:
>> From: Alexei Starovoitov <alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
>> Date: Fri, 2 Jun 2017 09:06:43 -0700
>>
>> > On Fri, Jun 02, 2017 at 06:39:40PM +0300, Leon Romanovsky wrote:
>> >> On Thu, Jun 01, 2017 at 06:57:59PM -0400, Doug Ledford wrote:
>> >> > On Thu, 2017-05-25 at 12:02 -0400, David Miller wrote:
>> >> > > From: Saeed Mahameed <saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
>> >> > > Date: Tue, 23 May 2017 14:43:58 +0300
>> >> > >
>> >> > > > Hi Dave and Doug,
>> >> > > >
>> >> > > > This series introduces some small updates and FPGA support to the
>> >> > > mlx5
>> >> > > > core/ethernet and IB drivers.
>> >> > > >
>> >> > > > For more details please see below.
>> >> > > >
>> >> > > > Please pull and let me know if there's any problem.
>> >> > >
>> >> > > Ok, I've pulled this into net-next.
>> >> > >
>> >> > > Doug let me know if there are any merge hassles we need to coordinate
>> >> > > on.
>> >> > >
>> >> > > Thanks.
>> >> >
>> >> > Will do. But is the question of a possible revert settled? Because
>> >> > once I pull, a revert means I have to pull all the way up to the revert
>> >> > as well...
>> >>
>> >> Revert is unlikely to happen, Saeed is working with Alexey to resolve
>> >> his raised concerns.
>> >
>> > I still think the direction is absolutely wrong.
>> > Reinventing fpga interfaces via nic driver is no-go.
>> > mlx needs to send a patch to revert it to show that they are willing
>> > to listen to the community.
>>
>> "Two people from Facebook" is not the commnity Alexei. :-)
>
> I've read the thread that fpga folks are not excited about
> fpga behind the nic either... but yeah i'm mainly expressing
> my opinions here.
>
>> And ironically, you guys will want to be one of the main users of
>> these facilities for the KTLS offload bits, so the desire to separate
>> it out and make "backports easier" really confuses me.
>
> We will not be using ktls offload via hidden fpga which
> is not properly exposed to the kernel.
> We've learned our lesson with eswitch.
>
KTLS/IPSec or any mlx5 FPGA/Innova applications is yet to be submitted,
and as Ilan said the kernel API is WIP, and we will not push anything
until the kernel API is submitted and accepted.
The patch in question that you are asking to revert, is adding no new
fancy functionality or FPGA capabilities to the mlx5 driver yet.
it is only basic mlx5 core stuff to add support of new device
"Mellanox Innova™ Flex 4 Lx EN Adapter Card" [1], so such device which
is already in production
can work with the mlx5_core.ko as is (ConnectX4-Lx) with no special
offloads or special Innova applications.
so this patch is mandatory for those who have this device to have
basic and standard ConnectX4-Lx networking interfaces.
It also adds the needed code separation between mlx5 basic core and
Innova stuff for future Innova applications and API development
(IPsec,KTLS, etc ..), so it will be easy for you to remove Innova
support at compilation time. And when the time comes, each new part
will be introduced with the suitable kernel API in the right place.
> ktls patches are independent of hw offload. It's great that
> mlx fpga can accelerate ktls tx which is extra confirmation
> that ktls api is on the right track. There is still 'ktls rx'
> to finish and 'ktls rx offload' is even trickier, since it's
> more intrusive into the core networking. Once crypto scatter gather
> is improved, we expect to see 8% perf gain out of sw ktls alone.
> So ktls hw offload is complementary and not mandatory.
>
> Back to eswitch analogy. All I hear from mlx is "at this stage
> it's not approriate ...". Well, I've been asking to do
> 'the right thing' for eswitch for months now and eswitch was
> in the driver for years. What are the chances that this fpga
> will ever be exposed? Once the code is merged there is no
> incentive for the vendor to do it differently.
>
"mlx5 eswitch == ethernet SRIOV" and today we already implement most
-if not all- of SRIOV VF ndos, and all are accessible from
"(ip link show) and (ip set vf xyz) " this is for the Legacy mode.
for the new switchdev mode, eswitch is managed by TC tool and
switchdev APIs only.
So we already have max kernel/user visibility in both modes.
can you clarify what bothers you and what is "the right thing"?
because i think we already agreed on the MLX5_ESWITCH/SRIOV kconfig
directive to eliminate eswitch for those who don't need it.
What exactly is missing ?
[1] http://www.mellanox.com/page/products_dyn?product_family=228&mtag=programmable_adapter_cards_
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* RE: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <20170601153704.GA1680-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
@ 2017-06-04 7:51 ` Ilan Tayari
[not found] ` <AM4PR0501MB19404B83A69B87AFB1326B45DBF50-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Ilan Tayari @ 2017-06-04 7:51 UTC (permalink / raw)
To: Jason Gunthorpe
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
Andy Shevchenko, linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
> -----Original Message-----
> From: Jason Gunthorpe [mailto:jgunthorpe@obsidianresearch.com]
> Subject: Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
>
> On Mon, May 29, 2017 at 04:09:06PM +0000, Ilan Tayari wrote:
>
> > > For IPSec, this is already in the kernel.
> > > See this patchset:
> > > http://www.mail-archive.com/netdev@vger.kernel.org/msg162876.html
> >
> > Sorry, I pointed at the RFC by mistake.
> >
> > This is the relevant pull request:
> > https://patchwork.ozlabs.org/patch/752707/
>
> This is connecting ipsec to a netdev, while Innova seems to be a
Jason,
"network connected ipsec accelerator configured using IP packets."
No. This is incorrect.
Where did you get that from?
It is an IPSec-capable NIC.
See here for details:
http://www.mellanox.com/page/products_dyn?product_family=249&mtag=programmable_network_adapters
The driver we are intending to submit for it, interfaces with the mentioned API in the pull-request (Netdev xdo callbacks)
So you configure it from userspace with regular IPSec 'ip xfrm state' commands or over netlink with your favorite IKE daemon.
>
> Those two things don't seem to be the same.
>
> Jason
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <AM4PR0501MB19404B83A69B87AFB1326B45DBF50-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
@ 2017-06-05 15:17 ` Jason Gunthorpe
[not found] ` <20170605151724.GA20182-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Jason Gunthorpe @ 2017-06-05 15:17 UTC (permalink / raw)
To: Ilan Tayari
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
Andy Shevchenko, linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
On Sun, Jun 04, 2017 at 07:51:24AM +0000, Ilan Tayari wrote:
> > From: Jason Gunthorpe [mailto:jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org]
> > Subject: Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
> >
> > On Mon, May 29, 2017 at 04:09:06PM +0000, Ilan Tayari wrote:
> >
> > > > For IPSec, this is already in the kernel.
> > > > See this patchset:
> > > > http://www.mail-archive.com/netdev-u79uwXL29TY76Z2rM5mHXA@public.gmane.org/msg162876.html
> > >
> > > Sorry, I pointed at the RFC by mistake.
> > >
> > > This is the relevant pull request:
> > > https://patchwork.ozlabs.org/patch/752707/
> >
> > This is connecting ipsec to a netdev, while Innova seems to be a
>
> Jason,
>
> "network connected ipsec accelerator configured using IP packets."
> No. This is incorrect.
> Where did you get that from?
That is what you described to me - you said the only way to configure
the FPGA was via IP packets in-band. It is not part of the NIC and the
NIC only loads the FPGA bitstream.
Maybe you should explain more how this works?
> So you configure it from userspace with regular IPSec 'ip xfrm
> state' commands or over netlink with your favorite IKE daemon.
But you don't give an ip xfrm configuration to the NIC when you submit
the work request? By your description it sounded liked the FPGA
pattern matches packets from the NIC side to apply the xfrm.
Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* RE: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <20170605151724.GA20182-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
@ 2017-06-06 6:52 ` Ilan Tayari
[not found] ` <AM4PR0501MB194008AAABEB6AAAA2ADFC82DBCB0-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
2017-06-06 16:17 ` Jason Gunthorpe
0 siblings, 2 replies; 72+ messages in thread
From: Ilan Tayari @ 2017-06-06 6:52 UTC (permalink / raw)
To: Jason Gunthorpe
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
Andy Shevchenko, linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="utf-8", Size: 3427 bytes --]
> -----Original Message-----
> From: Jason Gunthorpe [mailto:jgunthorpe@obsidianresearch.com]
> Subject: Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
>
> On Sun, Jun 04, 2017 at 07:51:24AM +0000, Ilan Tayari wrote:
> > > From: Jason Gunthorpe [mailto:jgunthorpe@obsidianresearch.com]
> > > Subject: Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for
> Innova
> > >
> > > On Mon, May 29, 2017 at 04:09:06PM +0000, Ilan Tayari wrote:
> > >
> > > > > For IPSec, this is already in the kernel.
> > > > > See this patchset:
> > > > > http://www.mail-archive.com/netdev@vger.kernel.org/msg162876.html
> > > >
> > > > Sorry, I pointed at the RFC by mistake.
> > > >
> > > > This is the relevant pull request:
> > > > https://patchwork.ozlabs.org/patch/752707/
> > >
> > > This is connecting ipsec to a netdev, while Innova seems to be a
> >
> > Jason,
> >
> > "network connected ipsec accelerator configured using IP packets."
> > No. This is incorrect.
> > Where did you get that from?
>
> That is what you described to me - you said the only way to configure
> the FPGA was via IP packets in-band. It is not part of the NIC and the
> NIC only loads the FPGA bitstream.
>
> Maybe you should explain more how this works?
I think this is just a misunderstanding.
The FPGA is part of this NIC, not something external.
The fact that we configure the FPGA using special inband packets isn't
changing anything. IMO, it might have been any other bus on the card,
standard or proprietary, and the arguments for how to design the driver
would stay the same.
Note that these packets are not generated as IP packets at the host.
They are RoCE packets. The command payload is generated by the driver,
then the QP and RoCE headers are generates by the ConnectX ASIC. The packet
is then consumed by the FPGA.
On the way back, the response is generated by the FPGA, and consumed by
the ConnectX ASIC, with the response payload delivered to the QP in the
driver.
So neither the host stack nor the network are aware of them.
They exist momentarily only on the internal traces on the board and not
anywhere else.
>
> > So you configure it from userspace with regular IPSec 'ip xfrm
> > state' commands or over netlink with your favorite IKE daemon.
>
> But you don't give an ip xfrm configuration to the NIC when you submit
> the work request? By your description it sounded liked the FPGA
> pattern matches packets from the NIC side to apply the xfrm.
>
> Jason
I didn't want to dig into this, because it's not submitted yet.
But here's an example configuration flow for creating an IPSec SA:
* User gives 'ip xfrm state add' command in shell, with 'offload' argument
* iproute2 sends XFRM_MSG_NEWSA netlink message with XFRMA_OFFLOAD_DEV attribute
* Xfrm stack generates a new xfrm_state object and calls the netdevice's xdo_dev_state_add callback
* mlx5e driver formats HW-specific ADD_SA command packet and sends it over RoCE QP to FPGA
* FPGA adds the SA to its offloaded SADB, and sends a response packet for 'success'
That's pretty much it.
Does this help to understand what goes on?
I don't mind explaining further, but I think you will just see it in the
patchset when we submit.
Ilan.
N‹§²æìr¸›yúèšØb²X¬¶Ç§vØ^–)Þº{.nÇ+‰·¥Š{±ÙšŠ{ayº\x1dʇڙë,j\a¢f£¢·hš‹»öì\x17/oSc¾™Ú³9˜uÀ¦æå‰È&jw¨®\x03(階ŠÝ¢j"ú\x1a¶^[m§ÿïêäz¹Þ–Šàþf£¢·hšˆ§~ˆmš
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <AM4PR0501MB194008AAABEB6AAAA2ADFC82DBCB0-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
@ 2017-06-06 15:50 ` David Miller
0 siblings, 0 replies; 72+ messages in thread
From: David Miller @ 2017-06-06 15:50 UTC (permalink / raw)
To: ilant-VPRAkNaXOzVWk0Htik3J/w
Cc: jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/,
alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w,
saeedm-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb,
dledford-H+wXaHxf7aLQT0dZR+AlfA, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
andy.shevchenko-Re5JQEeQqe8AvxtiuMwx3w,
linux-fpga-u79uwXL29TY76Z2rM5mHXA,
atull-yzvPICuk2ABMcg4IHK0kFoH6Mc4MB0Vx,
yi1.li-VuQAYsv1563Yd54FQh9/CA, borisp-VPRAkNaXOzVWk0Htik3J/w
From: Ilan Tayari <ilant-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Date: Tue, 6 Jun 2017 06:52:15 +0000
> The fact that we configure the FPGA using special inband packets isn't
> changing anything. IMO, it might have been any other bus on the card,
> standard or proprietary, and the arguments for how to design the driver
> would stay the same.
+1
> So neither the host stack nor the network are aware of them.
> They exist momentarily only on the internal traces on the board and not
> anywhere else.
+1
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-06-06 6:52 ` Ilan Tayari
[not found] ` <AM4PR0501MB194008AAABEB6AAAA2ADFC82DBCB0-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
@ 2017-06-06 16:17 ` Jason Gunthorpe
[not found] ` <20170606161709.GA8671-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
1 sibling, 1 reply; 72+ messages in thread
From: Jason Gunthorpe @ 2017-06-06 16:17 UTC (permalink / raw)
To: Ilan Tayari
Cc: Alexei Starovoitov, Saeed Mahameed, David S. Miller,
Doug Ledford, netdev, linux-rdma, jsorensen, Andy Shevchenko,
linux-fpga, Alan Tull, yi1.li, Boris Pismenny
On Tue, Jun 06, 2017 at 06:52:15AM +0000, Ilan Tayari wrote:
> So neither the host stack nor the network are aware of them.
> They exist momentarily only on the internal traces on the board and not
> anywhere else.
Is that really true? If you are creating rocee QPs' then the RDMA
stack sees this stuff and now we have buried a RDMA ULP inside an
ethernet driver which seems really wonky..
> I don't mind explaining further, but I think you will just see it in the
> patchset when we submit.
You described exactly what I thought.. I just disagree with you that
an ethernet connected and controlled IP accelerator is 'part of the
NIC', even if it happens to be colocated on the same circuit board.
Jason
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <20170606161709.GA8671-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
@ 2017-06-06 17:42 ` Alexei Starovoitov
[not found] ` <20170606174233.w377ctwtapzccsk7-+o4/htvd0TCa6kscz5V53/3mLCh9rsb+VpNB7YpNyf8@public.gmane.org>
2017-06-07 4:16 ` Saeed Mahameed
1 sibling, 1 reply; 72+ messages in thread
From: Alexei Starovoitov @ 2017-06-06 17:42 UTC (permalink / raw)
To: Jason Gunthorpe
Cc: Ilan Tayari, Saeed Mahameed, David S. Miller, Doug Ledford,
netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
jsorensen-b10kYP2dOMg, Andy Shevchenko,
linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
On Tue, Jun 06, 2017 at 10:17:09AM -0600, Jason Gunthorpe wrote:
> On Tue, Jun 06, 2017 at 06:52:15AM +0000, Ilan Tayari wrote:
>
> > So neither the host stack nor the network are aware of them.
> > They exist momentarily only on the internal traces on the board and not
> > anywhere else.
>
> Is that really true? If you are creating rocee QPs' then the RDMA
> stack sees this stuff and now we have buried a RDMA ULP inside an
> ethernet driver which seems really wonky..
>
> > I don't mind explaining further, but I think you will just see it in the
> > patchset when we submit.
>
> You described exactly what I thought.. I just disagree with you that
> an ethernet connected and controlled IP accelerator is 'part of the
> NIC', even if it happens to be colocated on the same circuit board.
+1
what Ilan described is a kernel bypass done by hw.
This is non starter in production. Same as eswitch this fpga is not
represented as a kernel object, there is no way to debug it.
NIC crafts roce packets back and forth?!
so it's like rdma, but without using kernel rdma stack?
When hw ipsec or tls will mysteriously drop or mangle the packets
how this can be debugged? Does fpga have attached ddr to
store/forward the packets? How memory issues will be reported?
No MCE errors ever? Buffer overflow? How many receive queues inside fpga?
How health check of fgpa itself will be done? Through roce packets?
I would buy the lack of kernel visibility if this fpga+nic combo
was a prototype, but it's being presented as a production device
with subsequent changes to core networking stack and that's where
I have a problem with its sw architecture.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <20170606174233.w377ctwtapzccsk7-+o4/htvd0TCa6kscz5V53/3mLCh9rsb+VpNB7YpNyf8@public.gmane.org>
@ 2017-06-06 17:47 ` David Miller
2017-06-06 18:34 ` Alexei Starovoitov
0 siblings, 1 reply; 72+ messages in thread
From: David Miller @ 2017-06-06 17:47 UTC (permalink / raw)
To: alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w
Cc: jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/,
ilant-VPRAkNaXOzVWk0Htik3J/w,
saeedm-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb,
dledford-H+wXaHxf7aLQT0dZR+AlfA, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
andy.shevchenko-Re5JQEeQqe8AvxtiuMwx3w,
linux-fpga-u79uwXL29TY76Z2rM5mHXA,
atull-yzvPICuk2ABMcg4IHK0kFoH6Mc4MB0Vx,
yi1.li-VuQAYsv1563Yd54FQh9/CA, borisp-VPRAkNaXOzVWk0Htik3J/w
From: Alexei Starovoitov <alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Date: Tue, 6 Jun 2017 10:42:35 -0700
> so it's like rdma, but without using kernel rdma stack?
No sockets here, just transformation rules. It's like offloading
a complex TC rule to hardware version of that transformation.
Yes, there is state, but I argue that it is no different than TC
offloading rules. What if TC had "hash" and "crypt" operations
and we attached them to appropriate u32 matches? You wouldn't
be able to tell the difference.
I think you are way over-obsessed with this FPGA offload thing,
quite frankly.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-06-06 17:47 ` David Miller
@ 2017-06-06 18:34 ` Alexei Starovoitov
2017-06-06 18:38 ` David Miller
0 siblings, 1 reply; 72+ messages in thread
From: Alexei Starovoitov @ 2017-06-06 18:34 UTC (permalink / raw)
To: David Miller
Cc: jgunthorpe, ilant, saeedm, dledford, netdev, linux-rdma,
jsorensen, andy.shevchenko, linux-fpga, atull, yi1.li, borisp
On Tue, Jun 06, 2017 at 01:47:26PM -0400, David Miller wrote:
> From: Alexei Starovoitov <alexei.starovoitov@gmail.com>
> Date: Tue, 6 Jun 2017 10:42:35 -0700
>
> > so it's like rdma, but without using kernel rdma stack?
>
> No sockets here, just transformation rules. It's like offloading
> a complex TC rule to hardware version of that transformation.
>
> Yes, there is state, but I argue that it is no different than TC
> offloading rules. What if TC had "hash" and "crypt" operations
> and we attached them to appropriate u32 matches? You wouldn't
> be able to tell the difference.
there is huge difference in underlying hw.
fpga is a separate device with its own phy and mac layers, its
own queues, packet parsing and rdma logic.
Where as tc offload is happening within the same hw queues/memory/stats
management logic. My understanding that when I do 'ethtool -L' to
change number of queues or 'ethtool -G' it changes the memory layout
that tc offload is operating on as well.
When I do 'ethtool -S' it shows me the stats for the device
that tc offload rules are integral part of.
Whereas fpga is a different physical device with its own
buffers and such. We can add 'ethtool -G_fpga, -L_fpga', etc
but this type of discussion needs to happen _before_ the whole
thing is merged. It will never happen after the fact.
Just look at mlx responses, they still don't acknowledge the issue
and instead pushing for ipsec, tls (in other words: new features)
instead of addressing production issues that are obviously
not glamorous to work on and fix.
> I think you are way over-obsessed with this FPGA offload thing,
> quite frankly.
if we didn't have issues with eswitch that drops packets and
we don't even know how many, I wouldn't be complaining.
There is a discussion going on to add few counters for
eswitch visibility, but it's taking forever and it's not at
the point of exposing eswitch as a kernel object.
Why? because it's hard to refactor it now into something like devlink
or whatever new abstraction that would be needed.
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-06-06 18:34 ` Alexei Starovoitov
@ 2017-06-06 18:38 ` David Miller
[not found] ` <20170606.143824.717466091308335341.davem-fT/PcQaiUtIeIZ0/mPfg9Q@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: David Miller @ 2017-06-06 18:38 UTC (permalink / raw)
To: alexei.starovoitov
Cc: jgunthorpe, ilant, saeedm, dledford, netdev, linux-rdma,
jsorensen, andy.shevchenko, linux-fpga, atull, yi1.li, borisp
From: Alexei Starovoitov <alexei.starovoitov@gmail.com>
Date: Tue, 6 Jun 2017 11:34:59 -0700
> fpga is a separate device with its own phy and mac layers, its
> own queues, packet parsing and rdma logic.
Because that's how they bolted it onto the ASIC in current
implementation, it might not always be that way and be fully
integrated in the future.
And I stress the word "implementation" as in "implementation detail"
the visible behavior is going to be the same, the difference is how
the thing is hooked up and maybe how you program it.
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <20170606.143824.717466091308335341.davem-fT/PcQaiUtIeIZ0/mPfg9Q@public.gmane.org>
@ 2017-06-06 18:55 ` Alexei Starovoitov
[not found] ` <20170606185532.2byjdonwsyan2asl-+o4/htvd0TCa6kscz5V53/3mLCh9rsb+VpNB7YpNyf8@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Alexei Starovoitov @ 2017-06-06 18:55 UTC (permalink / raw)
To: David Miller
Cc: jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/,
ilant-VPRAkNaXOzVWk0Htik3J/w,
saeedm-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb,
dledford-H+wXaHxf7aLQT0dZR+AlfA, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
andy.shevchenko-Re5JQEeQqe8AvxtiuMwx3w,
linux-fpga-u79uwXL29TY76Z2rM5mHXA, yi1.li-VuQAYsv1563Yd54FQh9/CA,
borisp-VPRAkNaXOzVWk0Htik3J/w
On Tue, Jun 06, 2017 at 02:38:24PM -0400, David Miller wrote:
> From: Alexei Starovoitov <alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> Date: Tue, 6 Jun 2017 11:34:59 -0700
>
> > fpga is a separate device with its own phy and mac layers, its
> > own queues, packet parsing and rdma logic.
>
> Because that's how they bolted it onto the ASIC in current
> implementation, it might not always be that way and be fully
> integrated in the future.
>
> And I stress the word "implementation" as in "implementation detail"
> the visible behavior is going to be the same, the difference is how
> the thing is hooked up and maybe how you program it.
whether fpga is a separate chip or part of the same asic makes no difference.
They are still different devices from sw point of view.
If in the future mlx will make it into the nic in a way that encryption shares
all memory management logic and there is no fpga at all then it indeed will
be similar to tc offload. Right now it's not and needs different sw architecture.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <20170606185532.2byjdonwsyan2asl-+o4/htvd0TCa6kscz5V53/3mLCh9rsb+VpNB7YpNyf8@public.gmane.org>
@ 2017-06-06 19:01 ` David Miller
[not found] ` <20170606.150151.1650636686526694540.davem-fT/PcQaiUtIeIZ0/mPfg9Q@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: David Miller @ 2017-06-06 19:01 UTC (permalink / raw)
To: alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w
Cc: jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/,
ilant-VPRAkNaXOzVWk0Htik3J/w,
saeedm-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb,
dledford-H+wXaHxf7aLQT0dZR+AlfA, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
andy.shevchenko-Re5JQEeQqe8AvxtiuMwx3w,
linux-fpga-u79uwXL29TY76Z2rM5mHXA, yi1.li-VuQAYsv1563Yd54FQh9/CA,
borisp-VPRAkNaXOzVWk0Htik3J/w
From: Alexei Starovoitov <alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Date: Tue, 6 Jun 2017 11:55:33 -0700
> If in the future mlx will make it into the nic in a way that
> encryption shares all memory management logic and there is no fpga
> at all then it indeed will be similar to tc offload. Right now it's
> not and needs different sw architecture.
If the visible effect is identical, I fundamentally disagree with you.
I don't care if there is a frog sitting on the PHY that transforms
the packets, it's all the same if the visible behavior is identical.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <20170606.150151.1650636686526694540.davem-fT/PcQaiUtIeIZ0/mPfg9Q@public.gmane.org>
@ 2017-06-06 22:44 ` Alexei Starovoitov
2017-06-07 0:48 ` Andrew Lunn
2017-06-07 3:47 ` Saeed Mahameed
0 siblings, 2 replies; 72+ messages in thread
From: Alexei Starovoitov @ 2017-06-06 22:44 UTC (permalink / raw)
To: David Miller
Cc: jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/,
ilant-VPRAkNaXOzVWk0Htik3J/w,
saeedm-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb,
dledford-H+wXaHxf7aLQT0dZR+AlfA, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
andy.shevchenko-Re5JQEeQqe8AvxtiuMwx3w,
linux-fpga-u79uwXL29TY76Z2rM5mHXA, yi1.li-VuQAYsv1563Yd54FQh9/CA,
borisp-VPRAkNaXOzVWk0Htik3J/w
On Tue, Jun 06, 2017 at 03:01:51PM -0400, David Miller wrote:
> From: Alexei Starovoitov <alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> Date: Tue, 6 Jun 2017 11:55:33 -0700
>
> > If in the future mlx will make it into the nic in a way that
> > encryption shares all memory management logic and there is no fpga
> > at all then it indeed will be similar to tc offload. Right now it's
> > not and needs different sw architecture.
>
> If the visible effect is identical, I fundamentally disagree with you.
>
> I don't care if there is a frog sitting on the PHY that transforms
> the packets, it's all the same if the visible behavior is identical.
that frog is a good example why we disagree.
I need to check the pulse of that frog and last time it ate.
In production I cannot have magical creatures do stuff for me.
I need to monitor all components, debug and mitigate the issues.
If encryption is done by the nic, I get all the monitoring and
debugging as part of the standard tools. When it's a frog
hidden by the nic, I cannot do much when the fire erupts,
hence frog and production environment don't mix.
To move things forward...
how about marking the whole thing CONFIG_EXPERIMENTAL instead of revert?
Right now it's effectively non-production==experimental code and
I want to make it clear.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-06-06 22:44 ` Alexei Starovoitov
@ 2017-06-07 0:48 ` Andrew Lunn
2017-06-07 3:47 ` Saeed Mahameed
1 sibling, 0 replies; 72+ messages in thread
From: Andrew Lunn @ 2017-06-07 0:48 UTC (permalink / raw)
To: Alexei Starovoitov
Cc: David Miller, jgunthorpe, ilant, saeedm, dledford, netdev,
linux-rdma, jsorensen, andy.shevchenko, linux-fpga, yi1.li,
borisp
On Tue, Jun 06, 2017 at 03:44:53PM -0700, Alexei Starovoitov wrote:
> On Tue, Jun 06, 2017 at 03:01:51PM -0400, David Miller wrote:
> > From: Alexei Starovoitov <alexei.starovoitov@gmail.com>
> > Date: Tue, 6 Jun 2017 11:55:33 -0700
> >
> > > If in the future mlx will make it into the nic in a way that
> > > encryption shares all memory management logic and there is no fpga
> > > at all then it indeed will be similar to tc offload. Right now it's
> > > not and needs different sw architecture.
> >
> > If the visible effect is identical, I fundamentally disagree with you.
> >
> > I don't care if there is a frog sitting on the PHY that transforms
> > the packets, it's all the same if the visible behavior is identical.
>
> that frog is a good example why we disagree.
> I need to check the pulse of that frog and last time it ate.
It is probably over-engineered for a single frog, but maybe you could
use a modified RFC 2795?
Andrew
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-06-06 22:44 ` Alexei Starovoitov
2017-06-07 0:48 ` Andrew Lunn
@ 2017-06-07 3:47 ` Saeed Mahameed
1 sibling, 0 replies; 72+ messages in thread
From: Saeed Mahameed @ 2017-06-07 3:47 UTC (permalink / raw)
To: Alexei Starovoitov
Cc: David Miller, Jason Gunthorpe, Ilan Tayari, Doug Ledford,
Linux Netdev List, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
Jes Sorensen, andy.shevchenko-Re5JQEeQqe8AvxtiuMwx3w,
linux-fpga-u79uwXL29TY76Z2rM5mHXA, yi1.li-VuQAYsv1563Yd54FQh9/CA,
borisp-VPRAkNaXOzVWk0Htik3J/w
On Wed, Jun 7, 2017 at 1:44 AM, Alexei Starovoitov
<alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
> On Tue, Jun 06, 2017 at 03:01:51PM -0400, David Miller wrote:
>> From: Alexei Starovoitov <alexei.starovoitov-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
>> Date: Tue, 6 Jun 2017 11:55:33 -0700
>>
>> > If in the future mlx will make it into the nic in a way that
>> > encryption shares all memory management logic and there is no fpga
>> > at all then it indeed will be similar to tc offload. Right now it's
>> > not and needs different sw architecture.
>>
>> If the visible effect is identical, I fundamentally disagree with you.
>>
>> I don't care if there is a frog sitting on the PHY that transforms
>> the packets, it's all the same if the visible behavior is identical.
>
> that frog is a good example why we disagree.
> I need to check the pulse of that frog and last time it ate.
> In production I cannot have magical creatures do stuff for me.
> I need to monitor all components, debug and mitigate the issues.
Every HW vendor has his own magical creatures. you don't want just to
have a kernel object representative for every HW unit!
that's just not scalable.
> If encryption is done by the nic, I get all the monitoring and
> debugging as part of the standard tools. When it's a frog
> hidden by the nic, I cannot do much when the fire erupts,
> hence frog and production environment don't mix.
> To move things forward...
Let's assume there was no FPGA but the ASIC provided the encryption
feature, and still a fire erupts,
what would you have done differently ? Only the vendor will know how
to debug regardless of what creature went nuts!
Frog or a Cat the kernel shouldn't care much "it is all implementation
details (HW implementation)", we should use the entry point to this
device (PCI device/netdevice)
as an abstraction point, and standardizing visibility/debuggability
should be per 'kernel<->stack<->driver<->HW'
features/flows/transactions.
Having a full driver<->HW visibility/standardization eliminates the
need of vendor specific drivers and each vendor should implement only
standard HWs that can work with generic drivers.
Re debuggability in ConnectX architecture, there is a well defined
health reporting and monitoring mechanism between FW and driver.
and it is up to the FW to check the pulse of every frog it is
babysitting and report back to the driver.
believe me you don't want the driver to know about every single wire
in the chip.
> how about marking the whole thing CONFIG_EXPERIMENTAL instead of revert?
> Right now it's effectively non-production==experimental code and
> I want to make it clear.
>
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <20170606161709.GA8671-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2017-06-06 17:42 ` Alexei Starovoitov
@ 2017-06-07 4:16 ` Saeed Mahameed
2017-06-07 15:48 ` Jason Gunthorpe
1 sibling, 1 reply; 72+ messages in thread
From: Saeed Mahameed @ 2017-06-07 4:16 UTC (permalink / raw)
To: Jason Gunthorpe
Cc: Ilan Tayari, Alexei Starovoitov, David S. Miller, Doug Ledford,
netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
jsorensen-b10kYP2dOMg, Andy Shevchenko,
linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
On Tue, Jun 6, 2017 at 7:17 PM, Jason Gunthorpe
<jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org> wrote:
> On Tue, Jun 06, 2017 at 06:52:15AM +0000, Ilan Tayari wrote:
>
>> So neither the host stack nor the network are aware of them.
>> They exist momentarily only on the internal traces on the board and not
>> anywhere else.
>
> Is that really true? If you are creating rocee QPs' then the RDMA
> stack sees this stuff and now we have buried a RDMA ULP inside an
> ethernet driver which seems really wonky..
It is not an ethernet driver, mlx5_core provides both RDMA and
ethernet interfaces to both mlx5_ib and the mlx5e netdevice.
so it is perfectly capable of creating QPs on its own, after all it is
the one creating QPs for the RDMA stack :).
rdma_create_qp->mlx5_ib_create_qp->mlx5_core_create_qp.
>
>> I don't mind explaining further, but I think you will just see it in the
>> patchset when we submit.
>
> You described exactly what I thought.. I just disagree with you that
> an ethernet connected and controlled IP accelerator is 'part of the
> NIC', even if it happens to be colocated on the same circuit board.
>
> Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-06-07 4:16 ` Saeed Mahameed
@ 2017-06-07 15:48 ` Jason Gunthorpe
2017-06-07 19:13 ` Saeed Mahameed
0 siblings, 1 reply; 72+ messages in thread
From: Jason Gunthorpe @ 2017-06-07 15:48 UTC (permalink / raw)
To: Saeed Mahameed
Cc: Ilan Tayari, Alexei Starovoitov, David S. Miller, Doug Ledford,
netdev, linux-rdma, jsorensen, Andy Shevchenko, linux-fpga,
Alan Tull, yi1.li, Boris Pismenny
On Wed, Jun 07, 2017 at 07:16:42AM +0300, Saeed Mahameed wrote:
> On Tue, Jun 6, 2017 at 7:17 PM, Jason Gunthorpe
> <jgunthorpe@obsidianresearch.com> wrote:
> > On Tue, Jun 06, 2017 at 06:52:15AM +0000, Ilan Tayari wrote:
> >
> >> So neither the host stack nor the network are aware of them.
> >> They exist momentarily only on the internal traces on the board and not
> >> anywhere else.
> >
> > Is that really true? If you are creating rocee QPs' then the RDMA
> > stack sees this stuff and now we have buried a RDMA ULP inside an
> > ethernet driver which seems really wonky..
>
> It is not an ethernet driver, mlx5_core provides both RDMA and
> ethernet interfaces to both mlx5_ib and the mlx5e netdevice.
>
> so it is perfectly capable of creating QPs on its own, after all it is
> the one creating QPs for the RDMA stack :).
>
> rdma_create_qp->mlx5_ib_create_qp->mlx5_core_create_qp.
Wait, so you built a RDMA ULP inside your driver without using the
RDMA API?
This keep getting more ugly :(
What about security? What if user space sends some raw packets to the
FPGA - can it reprogram the ISPEC settings or worse?
Jason
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-06-07 15:48 ` Jason Gunthorpe
@ 2017-06-07 19:13 ` Saeed Mahameed
2017-06-07 19:21 ` Jason Gunthorpe
0 siblings, 1 reply; 72+ messages in thread
From: Saeed Mahameed @ 2017-06-07 19:13 UTC (permalink / raw)
To: Jason Gunthorpe
Cc: Ilan Tayari, Alexei Starovoitov, David S. Miller, Doug Ledford,
netdev, linux-rdma, jsorensen, Andy Shevchenko, linux-fpga,
Alan Tull, yi1.li, Boris Pismenny
On Wed, Jun 7, 2017 at 6:48 PM, Jason Gunthorpe
<jgunthorpe@obsidianresearch.com> wrote:
> On Wed, Jun 07, 2017 at 07:16:42AM +0300, Saeed Mahameed wrote:
>> On Tue, Jun 6, 2017 at 7:17 PM, Jason Gunthorpe
>> <jgunthorpe@obsidianresearch.com> wrote:
>> > On Tue, Jun 06, 2017 at 06:52:15AM +0000, Ilan Tayari wrote:
>> >
>> >> So neither the host stack nor the network are aware of them.
>> >> They exist momentarily only on the internal traces on the board and not
>> >> anywhere else.
>> >
>> > Is that really true? If you are creating rocee QPs' then the RDMA
>> > stack sees this stuff and now we have buried a RDMA ULP inside an
>> > ethernet driver which seems really wonky..
>>
>> It is not an ethernet driver, mlx5_core provides both RDMA and
>> ethernet interfaces to both mlx5_ib and the mlx5e netdevice.
>>
>> so it is perfectly capable of creating QPs on its own, after all it is
>> the one creating QPs for the RDMA stack :).
>>
>> rdma_create_qp->mlx5_ib_create_qp->mlx5_core_create_qp.
>
> Wait, so you built a RDMA ULP inside your driver without using the
> RDMA API?
>
No !!
I am just showing you that the ib_core eventually will end up calling
mlx5_core to create a QP.
so mlx5_core can create the QP it self since it is the one eventually
creating QPs.
we just call mlx5_core_create_qp directly.
> This keep getting more ugly :(
>
> What about security? What if user space sends some raw packets to the
> FPGA - can it reprogram the ISPEC settings or worse?
>
No such thing. This QP is only for internal driver/HW communications,
as it is faster from the existing command interface.
it is not meant to be exposed for any raw user space usages at all,
without proper standard API adapter of course.
> Jason
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-06-07 19:13 ` Saeed Mahameed
@ 2017-06-07 19:21 ` Jason Gunthorpe
[not found] ` <20170607192132.GA10929-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
0 siblings, 1 reply; 72+ messages in thread
From: Jason Gunthorpe @ 2017-06-07 19:21 UTC (permalink / raw)
To: Saeed Mahameed
Cc: Ilan Tayari, Alexei Starovoitov, David S. Miller, Doug Ledford,
netdev, linux-rdma, jsorensen, Andy Shevchenko, linux-fpga,
Alan Tull, yi1.li, Boris Pismenny
On Wed, Jun 07, 2017 at 10:13:43PM +0300, Saeed Mahameed wrote:
> On Wed, Jun 7, 2017 at 6:48 PM, Jason Gunthorpe
> <jgunthorpe@obsidianresearch.com> wrote:
> > On Wed, Jun 07, 2017 at 07:16:42AM +0300, Saeed Mahameed wrote:
> >> On Tue, Jun 6, 2017 at 7:17 PM, Jason Gunthorpe
> >> <jgunthorpe@obsidianresearch.com> wrote:
> >> > On Tue, Jun 06, 2017 at 06:52:15AM +0000, Ilan Tayari wrote:
> >> >
> >> >> So neither the host stack nor the network are aware of them.
> >> >> They exist momentarily only on the internal traces on the board and not
> >> >> anywhere else.
> >> >
> >> > Is that really true? If you are creating rocee QPs' then the RDMA
> >> > stack sees this stuff and now we have buried a RDMA ULP inside an
> >> > ethernet driver which seems really wonky..
> >>
> >> It is not an ethernet driver, mlx5_core provides both RDMA and
> >> ethernet interfaces to both mlx5_ib and the mlx5e netdevice.
> >>
> >> so it is perfectly capable of creating QPs on its own, after all it is
> >> the one creating QPs for the RDMA stack :).
> >>
> >> rdma_create_qp->mlx5_ib_create_qp->mlx5_core_create_qp.
> >
> > Wait, so you built a RDMA ULP inside your driver without using the
> > RDMA API?
> >
>
> No !!
> I am just showing you that the ib_core eventually will end up calling
> mlx5_core to create a QP.
> so mlx5_core can create the QP it self since it is the one eventually
> creating QPs.
> we just call mlx5_core_create_qp directly.
Which is building a RDMA ULP inside a driver without using the core
code :(
> > This keep getting more ugly :(
> >
> > What about security? What if user space sends some raw packets to the
> > FPGA - can it reprogram the ISPEC settings or worse?
> >
>
> No such thing. This QP is only for internal driver/HW communications,
> as it is faster from the existing command interface.
> it is not meant to be exposed for any raw user space usages at all,
> without proper standard API adapter of course.
I'm not asking about the QP, I'm asking what happens after the NIC
part. You use ROCE packets to control the FPGA. What prevents
userspace from forcibly constructing roce packets and sending them to
the FPGA. How does the FPGA know for certain the packet came from the
kernel QP and not someplace else.
This is especially true for mlx nics as there are many raw packet
bypass mechanisms available to userspace.
Jason
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <20170607192132.GA10929-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
@ 2017-06-09 22:24 ` Doug Ledford
[not found] ` <1497047041.7171.234.camel-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2017-06-11 5:59 ` Ilan Tayari
1 sibling, 1 reply; 72+ messages in thread
From: Doug Ledford @ 2017-06-09 22:24 UTC (permalink / raw)
To: Jason Gunthorpe, Saeed Mahameed
Cc: Ilan Tayari, Alexei Starovoitov, David S. Miller,
netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
jsorensen-b10kYP2dOMg, Andy Shevchenko,
linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
On Wed, 2017-06-07 at 13:21 -0600, Jason Gunthorpe wrote:
> On Wed, Jun 07, 2017 at 10:13:43PM +0300, Saeed Mahameed wrote:
> >
> > No !!
> > I am just showing you that the ib_core eventually will end up
> > calling
> > mlx5_core to create a QP.
> > so mlx5_core can create the QP it self since it is the one
> > eventually
> > creating QPs.
> > we just call mlx5_core_create_qp directly.
>
> Which is building a RDMA ULP inside a driver without using the core
> code :(
Aren't the transmit/receive queues of the Ethernet netdevice on
mlx4/mlx5 hardware QPs too? Those bypass the RDMA subsystem entirely.
Just because something uses a QP on hardware that does *everything*
via QPs doesn't necessarily mean it must go through the RDMA subsystem.
Now, the fact that the content of the packets is basically a RoCE
packet does make things a bit fuzzier, but if their packets are
specially crafted RoCE packets that aren't really intended to be fully
RoCE spec compliant (maybe they don't support all the options as normal
RoCE QPs), then I can see hiding them from the larger RoCE portion of
the RDMA stack.
> >
> > >
> > > This keep getting more ugly :(
> > >
> > > What about security? What if user space sends some raw packets to
> > > the
> > > FPGA - can it reprogram the ISPEC settings or worse?
> > >
> >
> > No such thing. This QP is only for internal driver/HW
> > communications,
> > as it is faster from the existing command interface.
> > it is not meant to be exposed for any raw user space usages at all,
> > without proper standard API adapter of course.
>
> I'm not asking about the QP, I'm asking what happens after the NIC
> part. You use ROCE packets to control the FPGA. What prevents
> userspace from forcibly constructing roce packets and sending them to
> the FPGA. How does the FPGA know for certain the packet came from the
> kernel QP and not someplace else.
This is a valid concern.
> This is especially true for mlx nics as there are many raw packet
> bypass mechanisms available to userspace.
Right. The question becomes: Does the firmware filter outgoing raw ETH
QPs such that a nefarious user could not send a crafted RoCE packet
that the bump on the wire would intercept and accept?
--
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
GPG KeyID: B826A3330E572FDD
Key fingerprint = AE6B 1BDA 122B 23B4 265B 1274 B826 A333 0E57 2FDD
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <1497047041.7171.234.camel-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
@ 2017-06-10 14:11 ` Majd Dibbiny
2017-06-12 16:17 ` Jason Gunthorpe
0 siblings, 1 reply; 72+ messages in thread
From: Majd Dibbiny @ 2017-06-10 14:11 UTC (permalink / raw)
To: Doug Ledford
Cc: Jason Gunthorpe, Saeed Mahameed, Ilan Tayari, Alexei Starovoitov,
David S. Miller, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
Andy Shevchenko, linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
> On Jun 10, 2017, at 1:24 AM, Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
>
>> On Wed, 2017-06-07 at 13:21 -0600, Jason Gunthorpe wrote:
>>> On Wed, Jun 07, 2017 at 10:13:43PM +0300, Saeed Mahameed wrote:
>>>
>>> No !!
>>> I am just showing you that the ib_core eventually will end up
>>> calling
>>> mlx5_core to create a QP.
>>> so mlx5_core can create the QP it self since it is the one
>>> eventually
>>> creating QPs.
>>> we just call mlx5_core_create_qp directly.
>>
>> Which is building a RDMA ULP inside a driver without using the core
>> code :(
>
> Aren't the transmit/receive queues of the Ethernet netdevice on
> mlx4/mlx5 hardware QPs too? Those bypass the RDMA subsystem entirely.
> Just because something uses a QP on hardware that does *everything*
> via QPs doesn't necessarily mean it must go through the RDMA subsystem.
>
> Now, the fact that the content of the packets is basically a RoCE
> packet does make things a bit fuzzier, but if their packets are
> specially crafted RoCE packets that aren't really intended to be fully
> RoCE spec compliant (maybe they don't support all the options as normal
> RoCE QPs), then I can see hiding them from the larger RoCE portion of
> the RDMA stack.
>
>>>
>>>>
>>>> This keep getting more ugly :(
>>>>
>>>> What about security? What if user space sends some raw packets to
>>>> the
>>>> FPGA - can it reprogram the ISPEC settings or worse?
>>>>
>>>
>>> No such thing. This QP is only for internal driver/HW
>>> communications,
>>> as it is faster from the existing command interface.
>>> it is not meant to be exposed for any raw user space usages at all,
>>> without proper standard API adapter of course.
>>
>> I'm not asking about the QP, I'm asking what happens after the NIC
>> part. You use ROCE packets to control the FPGA. What prevents
>> userspace from forcibly constructing roce packets and sending them to
>> the FPGA. How does the FPGA know for certain the packet came from the
>> kernel QP and not someplace else.
>
> This is a valid concern.
>
>> This is especially true for mlx nics as there are many raw packet
>> bypass mechanisms available to userspace.
>
All of the Raw packet bypass mechanisms are restricted to CAP_NET_RAW, and thus malicious users can't simply open a RAW Packet QP and send it to the FPGA..
> Right. The question becomes: Does the firmware filter outgoing raw ETH
> QPs such that a nefarious user could not send a crafted RoCE packet
> that the bump on the wire would intercept and accept?
>
> --
> Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
> GPG KeyID: B826A3330E572FDD
>
> Key fingerprint = AE6B 1BDA 122B 23B4 265B 1274 B826 A333 0E57 2FDD
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* RE: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <20170607192132.GA10929-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2017-06-09 22:24 ` Doug Ledford
@ 2017-06-11 5:59 ` Ilan Tayari
[not found] ` <AM4PR0501MB19401208254971445E61367EDBCC0-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
1 sibling, 1 reply; 72+ messages in thread
From: Ilan Tayari @ 2017-06-11 5:59 UTC (permalink / raw)
To: Jason Gunthorpe, Saeed Mahameed
Cc: Alexei Starovoitov, David S. Miller, Doug Ledford,
netdev-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA,
jsorensen-b10kYP2dOMg, Andy Shevchenko,
linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
> -----Original Message-----
> From: Jason Gunthorpe [mailto:jgunthorpe@obsidianresearch.com]
> Subject: Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
>
> > > This keep getting more ugly :(
> > >
> > > What about security? What if user space sends some raw packets to the
> > > FPGA - can it reprogram the ISPEC settings or worse?
> > >
> >
> > No such thing. This QP is only for internal driver/HW communications,
> > as it is faster from the existing command interface.
> > it is not meant to be exposed for any raw user space usages at all,
> > without proper standard API adapter of course.
>
> I'm not asking about the QP, I'm asking what happens after the NIC
> part. You use ROCE packets to control the FPGA. What prevents
> userspace from forcibly constructing roce packets and sending them to
> the FPGA. How does the FPGA know for certain the packet came from the
> kernel QP and not someplace else.
>
> This is especially true for mlx nics as there are many raw packet
> bypass mechanisms available to userspace.
Hi Jason,
The device uses internal signaling that ensures that no entity other than the mlx5 driver can talk over the FPGA channel.
This is also the reason why this is not a "ULP in a driver", but rather an internal bus that happens to use some of our existing HW features.
As explained earlier, this "bus" is an internal device implementation issue, and has nothing to do with the network or RDMA stack.
Ilan.
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
[not found] ` <AM4PR0501MB19401208254971445E61367EDBCC0-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
@ 2017-06-12 16:14 ` Jason Gunthorpe
0 siblings, 0 replies; 72+ messages in thread
From: Jason Gunthorpe @ 2017-06-12 16:14 UTC (permalink / raw)
To: Ilan Tayari
Cc: Saeed Mahameed, Alexei Starovoitov, David S. Miller,
Doug Ledford, netdev-u79uwXL29TY76Z2rM5mHXA,
linux-rdma-u79uwXL29TY76Z2rM5mHXA, jsorensen-b10kYP2dOMg,
Andy Shevchenko, linux-fpga-u79uwXL29TY76Z2rM5mHXA, Alan Tull,
yi1.li-VuQAYsv1563Yd54FQh9/CA, Boris Pismenny
On Sun, Jun 11, 2017 at 05:59:04AM +0000, Ilan Tayari wrote:
> > This is especially true for mlx nics as there are many raw packet
> > bypass mechanisms available to userspace.
>
> The device uses internal signaling that ensures that no entity other
> than the mlx5 driver can talk over the FPGA channel. This is also
> the reason why this is not a "ULP in a driver", but rather an
> internal bus that happens to use some of our existing HW features.
You are taking both positions at once - that this is *just* a 'bump on
the wire' and everything is done via ethernet packets, and also that
there is a special internal bus that nothing needs to know about.
It is hard to be both ways, if it is ethernet packets there there are
ways to create those packets outside the driver's control and security
is a big question.
Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-06-10 14:11 ` Majd Dibbiny
@ 2017-06-12 16:17 ` Jason Gunthorpe
2017-06-13 16:05 ` Saeed Mahameed
0 siblings, 1 reply; 72+ messages in thread
From: Jason Gunthorpe @ 2017-06-12 16:17 UTC (permalink / raw)
To: Majd Dibbiny
Cc: Doug Ledford, Saeed Mahameed, Ilan Tayari, Alexei Starovoitov,
David S. Miller, netdev, linux-rdma, jsorensen, Andy Shevchenko,
linux-fpga, Alan Tull, yi1.li, Boris Pismenny
On Sat, Jun 10, 2017 at 02:11:13PM +0000, Majd Dibbiny wrote:
> >> This is especially true for mlx nics as there are many raw packet
> >> bypass mechanisms available to userspace.
> All of the Raw packet bypass mechanisms are restricted to
> CAP_NET_RAW, and thus malicious users can't simply open a RAW Packet
> QP and send it to the FPGA..
It is big expansion of CAP_NET_RAW to also basically also include
reconfiguring ipsec xfrm.
Plus, if someone configures ethernet bridging (eg in a VM situation)
then could a hacked VM reconfigure this FPGA?
Jason
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova
2017-06-12 16:17 ` Jason Gunthorpe
@ 2017-06-13 16:05 ` Saeed Mahameed
0 siblings, 0 replies; 72+ messages in thread
From: Saeed Mahameed @ 2017-06-13 16:05 UTC (permalink / raw)
To: Jason Gunthorpe
Cc: Majd Dibbiny, Doug Ledford, Ilan Tayari, Alexei Starovoitov,
David S. Miller, netdev, linux-rdma, jsorensen, Andy Shevchenko,
linux-fpga, Alan Tull, yi1.li, Boris Pismenny
On Mon, Jun 12, 2017 at 7:17 PM, Jason Gunthorpe
<jgunthorpe@obsidianresearch.com> wrote:
> On Sat, Jun 10, 2017 at 02:11:13PM +0000, Majd Dibbiny wrote:
>
>> >> This is especially true for mlx nics as there are many raw packet
>> >> bypass mechanisms available to userspace.
>
>> All of the Raw packet bypass mechanisms are restricted to
>> CAP_NET_RAW, and thus malicious users can't simply open a RAW Packet
>> QP and send it to the FPGA..
>
> It is big expansion of CAP_NET_RAW to also basically also include
> reconfiguring ipsec xfrm.
>
> Plus, if someone configures ethernet bridging (eg in a VM situation)
> then could a hacked VM reconfigure this FPGA?
>
> Jason
Hi Jason
As we mentioned earlier, the device ensures that only the FPGA
component in the driver can configure the FPGA regardless of any type
of standard traffic.
Thanks,
Saeed.
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23
2017-05-25 16:02 ` [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23 David Miller
2017-06-01 22:57 ` Doug Ledford
@ 2017-06-14 19:30 ` Doug Ledford
2017-06-14 19:44 ` David Miller
1 sibling, 1 reply; 72+ messages in thread
From: Doug Ledford @ 2017-06-14 19:30 UTC (permalink / raw)
To: David Miller, saeedm; +Cc: netdev, linux-rdma, ilant
On Thu, 2017-05-25 at 12:02 -0400, David Miller wrote:
> From: Saeed Mahameed <saeedm@mellanox.com>
> Date: Tue, 23 May 2017 14:43:58 +0300
>
> > Hi Dave and Doug,
> >
> > This series introduces some small updates and FPGA support to the
> mlx5
> > core/ethernet and IB drivers.
> >
> > For more details please see below.
> >
> > Please pull and let me know if there's any problem.
>
> Ok, I've pulled this into net-next.
>
> Doug let me know if there are any merge hassles we need to coordinate
> on.
Turns out that you had pulled this prior to your net-next tree making
it up to v4.12-rc3, while I had my -rc branch based on v4.12-rc3, so
when I pulled just up to your merge commit for this pull request, it
meant a later merge of my -rc branch would pull in a bunch of stuff
between here and -rc3. So I ended up pulling your latest net-next as
of today as the easiest way to resolve that issue. Now I can merge my
-rc branch in and it won't cause any extra noise in the merge. It
seems to be a fairly regular pattern that I'm going to have to hold my
pull request to Linus until after your tree is pulled, so I might just
start planning on that from now on ;-)
--
Doug Ledford <dledford@redhat.com>
GPG KeyID: B826A3330E572FDD
Key fingerprint = AE6B 1BDA 122B 23B4 265B 1274 B826 A333 0E57 2FDD
^ permalink raw reply [flat|nested] 72+ messages in thread
* Re: [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23
2017-06-14 19:30 ` Doug Ledford
@ 2017-06-14 19:44 ` David Miller
0 siblings, 0 replies; 72+ messages in thread
From: David Miller @ 2017-06-14 19:44 UTC (permalink / raw)
To: dledford; +Cc: saeedm, netdev, linux-rdma, ilant
From: Doug Ledford <dledford@redhat.com>
Date: Wed, 14 Jun 2017 15:30:13 -0400
> On Thu, 2017-05-25 at 12:02 -0400, David Miller wrote:
>> From: Saeed Mahameed <saeedm@mellanox.com>
>> Date: Tue, 23 May 2017 14:43:58 +0300
>>
>> > Hi Dave and Doug,
>> >
>> > This series introduces some small updates and FPGA support to the
>> mlx5
>> > core/ethernet and IB drivers.
>> >
>> > For more details please see below.
>> >
>> > Please pull and let me know if there's any problem.
>>
>> Ok, I've pulled this into net-next.
>>
>> Doug let me know if there are any merge hassles we need to coordinate
>> on.
>
> Turns out that you had pulled this prior to your net-next tree making
> it up to v4.12-rc3, while I had my -rc branch based on v4.12-rc3, so
> when I pulled just up to your merge commit for this pull request, it
> meant a later merge of my -rc branch would pull in a bunch of stuff
> between here and -rc3. So I ended up pulling your latest net-next as
> of today as the easiest way to resolve that issue. Now I can merge my
> -rc branch in and it won't cause any extra noise in the merge. It
> seems to be a fairly regular pattern that I'm going to have to hold my
> pull request to Linus until after your tree is pulled, so I might just
> start planning on that from now on ;-)
Hehe, ok :)
^ permalink raw reply [flat|nested] 72+ messages in thread
end of thread, other threads:[~2017-06-14 19:45 UTC | newest]
Thread overview: 72+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-05-23 11:43 [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23 Saeed Mahameed
2017-05-23 11:44 ` [for-next 2/6] net/mlx5: Update the list of the PCI supported devices Saeed Mahameed
2017-05-23 11:44 ` [for-next 3/6] net/mlx5: Introduce trigger_health_work function Saeed Mahameed
2017-05-23 11:44 ` [for-next 4/6] net/mlx5: FPGA, Add basic support for Innova Saeed Mahameed
[not found] ` <20170523114404.20387-5-saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
2017-05-24 16:51 ` Alexei Starovoitov
2017-05-25 5:20 ` Ilan Tayari
[not found] ` <AM4PR0501MB1940885C2F1CEF4DDE4EA8D1DBFF0-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
2017-05-25 10:40 ` Saeed Mahameed
[not found] ` <CALzJLG-B_tAmASn_SMmPNiucq-tTpywHniRTkb4N32oGF6Y3Ng-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-25 20:48 ` Jes Sorensen
2017-05-26 8:29 ` Saeed Mahameed
[not found] ` <CALzJLG9YNpagdJAcrh6O0jJhZWtsck6KigRtVxyjkArTm=82ew-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-26 18:31 ` Jes Sorensen
2017-05-28 7:24 ` Ilan Tayari
2017-06-02 20:31 ` Jes Sorensen
[not found] ` <4c164e09-0103-7daf-e9f8-9260223ada08-b10kYP2dOMg@public.gmane.org>
2017-06-02 20:33 ` Doug Ledford
2017-05-26 3:07 ` Alexei Starovoitov
2017-05-26 8:59 ` Saeed Mahameed
[not found] ` <CALzJLG98D=3yMJV_q4sjVNG41AERFRU+6rwqQJsxnRuVeDTPdA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-26 17:56 ` Alexei Starovoitov
2017-05-26 18:15 ` Jason Gunthorpe
[not found] ` <20170526181517.GA3860-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2017-05-28 7:22 ` Ilan Tayari
[not found] ` <AM4PR0501MB1940330F0EBAA819C87C5278DBF20-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
2017-05-29 15:31 ` Jason Gunthorpe
[not found] ` <20170529153131.GB7924-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2017-05-29 15:58 ` Ilan Tayari
[not found] ` <AM4PR0501MB1940D05A19F098286B99EAD0DBF30-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
2017-05-29 16:02 ` Jason Gunthorpe
2017-05-29 16:05 ` Ilan Tayari
[not found] ` <AM4PR0501MB194037FF8F17466BC9ECC73DDBF30-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
2017-05-29 16:09 ` Ilan Tayari
[not found] ` <AM4PR0501MB19409139227E11A4A7F82F0FDBF30-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
2017-06-01 15:37 ` Jason Gunthorpe
[not found] ` <20170601153704.GA1680-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2017-06-04 7:51 ` Ilan Tayari
[not found] ` <AM4PR0501MB19404B83A69B87AFB1326B45DBF50-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
2017-06-05 15:17 ` Jason Gunthorpe
[not found] ` <20170605151724.GA20182-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2017-06-06 6:52 ` Ilan Tayari
[not found] ` <AM4PR0501MB194008AAABEB6AAAA2ADFC82DBCB0-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
2017-06-06 15:50 ` David Miller
2017-06-06 16:17 ` Jason Gunthorpe
[not found] ` <20170606161709.GA8671-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2017-06-06 17:42 ` Alexei Starovoitov
[not found] ` <20170606174233.w377ctwtapzccsk7-+o4/htvd0TCa6kscz5V53/3mLCh9rsb+VpNB7YpNyf8@public.gmane.org>
2017-06-06 17:47 ` David Miller
2017-06-06 18:34 ` Alexei Starovoitov
2017-06-06 18:38 ` David Miller
[not found] ` <20170606.143824.717466091308335341.davem-fT/PcQaiUtIeIZ0/mPfg9Q@public.gmane.org>
2017-06-06 18:55 ` Alexei Starovoitov
[not found] ` <20170606185532.2byjdonwsyan2asl-+o4/htvd0TCa6kscz5V53/3mLCh9rsb+VpNB7YpNyf8@public.gmane.org>
2017-06-06 19:01 ` David Miller
[not found] ` <20170606.150151.1650636686526694540.davem-fT/PcQaiUtIeIZ0/mPfg9Q@public.gmane.org>
2017-06-06 22:44 ` Alexei Starovoitov
2017-06-07 0:48 ` Andrew Lunn
2017-06-07 3:47 ` Saeed Mahameed
2017-06-07 4:16 ` Saeed Mahameed
2017-06-07 15:48 ` Jason Gunthorpe
2017-06-07 19:13 ` Saeed Mahameed
2017-06-07 19:21 ` Jason Gunthorpe
[not found] ` <20170607192132.GA10929-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2017-06-09 22:24 ` Doug Ledford
[not found] ` <1497047041.7171.234.camel-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2017-06-10 14:11 ` Majd Dibbiny
2017-06-12 16:17 ` Jason Gunthorpe
2017-06-13 16:05 ` Saeed Mahameed
2017-06-11 5:59 ` Ilan Tayari
[not found] ` <AM4PR0501MB19401208254971445E61367EDBCC0-dp/nxUn679gfNUYDR5dMTsDSnupUy6xnnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org>
2017-06-12 16:14 ` Jason Gunthorpe
2017-05-28 12:33 ` Or Gerlitz
2017-05-26 3:58 ` please revert. Was: " Alexei Starovoitov
2017-05-26 4:13 ` David Miller
2017-05-26 4:40 ` Alexei Starovoitov
2017-05-26 14:51 ` David Miller
2017-05-23 11:44 ` [for-next 5/6] net/mlx5: Bump driver version Saeed Mahameed
[not found] ` <20170523114404.20387-6-saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
2017-05-26 12:56 ` Dennis Dalessandro
2017-05-26 16:35 ` Saeed Mahameed
[not found] ` <CALzJLG_ha-XiPAMnoKrUgm_EwPx2yH0T2y4EBRfrWNYSZi1cTg-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-26 16:55 ` Dennis Dalessandro
[not found] ` <ee23ad82-4a2e-8546-d41b-11f979b127bb-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
2017-05-26 18:10 ` Leon Romanovsky
2017-05-26 21:53 ` Jakub Kicinski
[not found] ` <20170526145318.7fd8c8e2-68UzVGuGftmUSpRRplVxJ1aTQe2KTcn/@public.gmane.org>
2017-05-29 5:47 ` Leon Romanovsky
[not found] ` <20170523114404.20387-1-saeedm-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
2017-05-23 11:43 ` [for-next 1/6] {net, IB}/mlx5: Replace mlx5_vzalloc with kvzalloc Saeed Mahameed
2017-05-23 11:44 ` [for-next 6/6] IB/mlx5: Bump driver version Saeed Mahameed
2017-05-25 16:02 ` [pull request][for-next 0/6] Mellanox mlx5 updates 2017-05-23 David Miller
2017-06-01 22:57 ` Doug Ledford
[not found] ` <1496357879.7171.76.camel-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2017-06-02 15:39 ` Leon Romanovsky
[not found] ` <20170602153940.GX5406-U/DQcQFIOTAAJjI8aNfphQ@public.gmane.org>
2017-06-02 16:06 ` Alexei Starovoitov
[not found] ` <20170602160641.ylowbobe5v72ui7g-+o4/htvd0TCa6kscz5V53/3mLCh9rsb+VpNB7YpNyf8@public.gmane.org>
2017-06-02 16:08 ` David Miller
[not found] ` <20170602.120839.1394660754953676217.davem-fT/PcQaiUtIeIZ0/mPfg9Q@public.gmane.org>
2017-06-02 16:57 ` Alexei Starovoitov
[not found] ` <20170602165736.nwunidodmu6xsmuv-+o4/htvd0TCa6kscz5V53/3mLCh9rsb+VpNB7YpNyf8@public.gmane.org>
2017-06-03 19:46 ` Or Gerlitz
2017-06-03 22:45 ` Saeed Mahameed
2017-06-14 19:30 ` Doug Ledford
2017-06-14 19:44 ` David Miller
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).