On Tue, Jan 2, 2018 at 9:29 AM, Or Gerlitz wrote:
> On Sun, Dec 31, 2017 at 2:28 PM, Or Gerlitz wrote:
>> On Sun, Dec 31, 2017 at 1:16 PM, Denis Drozdov wrote:
>>> IP link was broken due to the changes in IPoIB for the rdma_netdev
>>> support after commit cd565b4b51e5
>>> ("IB/IPoIB: Support acc
On Tue, Apr 25, 2017 at 2:14 PM, Or Gerlitz wrote:
> On Tue, Apr 25, 2017 at 2:11 PM, Erez Shitrit
> wrote:
>> On Tue, Apr 25, 2017 at 1:32 PM, Or Gerlitz wrote:
>>> On Tue, Apr 25, 2017 at 12:55 PM, Honggang LI wrote:
>>>> From: Honggang Li
>>
On Tue, Apr 25, 2017 at 1:32 PM, Or Gerlitz wrote:
> On Tue, Apr 25, 2017 at 12:55 PM, Honggang LI wrote:
>> From: Honggang Li
>>
>> Minimal hard_header_len set by bond_compute_features is ETH_HLEN, which
>> is smaller than IPOIB_HARD_LEN. ipoib_hard_header should check the
>> size of headroom t
On Mon, Mar 13, 2017 at 10:01 PM, Jason Gunthorpe
wrote:
> On Mon, Mar 13, 2017 at 08:31:15PM +0200, Erez Shitrit wrote:
>
>> diff --git a/include/rdma/ib_ipoib_accel_ops.h
>> b/include/rdma/ib_ipoib_accel_ops.h
>> new file mode 100644
>> index ..148a
On Tue, Mar 14, 2017 at 6:10 PM, Jason Gunthorpe
wrote:
> On Tue, Mar 14, 2017 at 04:53:24PM +0200, Erez Shitrit wrote:
>
>> > Why isn't this stuff in open/close?
>>
>> According to ipoib control flows, there is a different between
>> open/close to init/
On Mon, Mar 13, 2017 at 10:27 PM, Jason Gunthorpe
wrote:
> On Mon, Mar 13, 2017 at 08:31:36PM +0200, Erez Shitrit wrote:
>
>> +struct net_device *mlx5_alloc_rdma_netdev(struct ib_device *hca,
>> + u8 port_num,
>> +
On Tue, Mar 14, 2017 at 8:35 AM, Vishwanathapura, Niranjana
wrote:
> On Mon, Mar 13, 2017 at 08:31:16PM +0200, Erez Shitrit wrote:
>>
>> +static struct net_device *ipoib_create_netdev_default(struct ib_device
>> *hca,
>> +
On Tue, Mar 14, 2017 at 9:01 AM, Vishwanathapura, Niranjana
wrote:
> On Mon, Mar 13, 2017 at 02:01:36PM -0600, Jason Gunthorpe wrote:
>>>
>>> + /* multicast */
>>> + int (*attach_mcast)(struct net_device *dev, struct ib_device
>>> *hca,
>>> + union ib_gid *gid
Now IB device is able to call open or close for its net device.
TBD:
There is one change that is waiting for the new channels api, till then
used an "if" in the code.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 1 +
drivers/net/ethernet/mel
mlx5i_attach that creates the resources of IB network device.
mlx5i_detach cleans resources for IB device.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 122 +++---
1 file changed, 87 insertions(+), 35 deletions(-)
diff --git a/drivers/net
1. Change the mlx5e_detach_netdev api.
2. Let that function to be called after the rtnl_lock is already held,
like done in IB link.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 18
Let the IB link to call it directly with the relevant priv.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 4 +--
.../net/ethernet/mellanox/mlx5/core/en_ethtool.c | 24 -
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 31
New function to handle RSS table for IB link type.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en.h| 2 ++
drivers/net/ethernet/mellanox/mlx5/core/en_fs.c | 41 +
2 files changed, 43 insertions(+)
diff --git a/drivers/net/ethernet
Use priv object instead of netdev object, will give tha ability to use
it for IB link.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 8
drivers/net/ethernet/mellanox/mlx5/core/en_rep.c | 4
Enable IB link to use the same code, by default the underlay_qp is zero
for ETH link.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 2 ++
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 1 +
2 files changed, 3 insertions(+)
diff --git a/drivers/net
In order to have the ability to support IB link with the same base code.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 6 ++
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 7 +++
drivers/net/ethernet/mellanox/mlx5/core/en_rx.c | 10
Get the relevant capabilities if supports ipoib_enhanced_offloads and
init the flow steering table accordingly.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/fs_core.c | 10 +-
drivers/net/ethernet/mellanox/mlx5/core/fw.c | 3 ++-
2 files changed, 7
Now we can use it for non-ethernet type packets (like IB).
After changing the order of the skb processing the function now can
check the skb->protocol field to check if it is ETH_P_IP/ETH_P_IPV6
packet.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
Signed-off-by: Erez Shitrit
---
drivers/infiniband/hw/mlx5/Makefile | 2 +-
drivers/infiniband/hw/mlx5/main.c | 10 +
drivers/infiniband/hw/mlx5/mlx5_ipoib_ops.c | 289
3 files changed, 300 insertions(+), 1 deletion(-)
create mode 100644
The function will be used in IB link in order to send packets.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en_tx.c | 55 +
include/linux/mlx5/driver.h | 5 ++-
2 files changed, 59 insertions(+), 1 deletion(-)
diff --git
New function that parse and build the skb for IPoIB traffic.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 8
drivers/net/ethernet/mellanox/mlx5/core/en_rx.c | 49 +++
2 files changed, 57 insertions(+)
diff --git a/drivers/net
IB link needs the the underlay_qp to support flow-steering, so change
the API of the flow-steering creation for supporting both types in the
same set of functions.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c | 12 +++-
drivers/net/ethernet/mellanox/mlx5
prepare of IB link type packets sending.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en.h| 4 +
drivers/net/ethernet/mellanox/mlx5/core/en_tx.c | 237 ++--
2 files changed, 141 insertions(+), 100 deletions(-)
diff --git a/drivers/net
mlx5_wqe_eth_pad will be used for TX flow for IB link type.
enhanced-ipoib-qp-mode for QP creation.
Signed-off-by: Erez Shitrit
---
drivers/infiniband/hw/mlx5/qp.c | 4
include/linux/mlx5/qp.h | 8
2 files changed, 8 insertions(+), 4 deletions(-)
diff --git a/drivers
Implement required interface that will able the IB link to be run on top
of the ETH data structures.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 98 +++
include/linux/mlx5/driver.h | 12 +++
2 files changed, 110
the driver implementation as it
was without HW vendor support.
TODO: We added remote qkey to ipoib_send in order to match send op
signature.
In accel mode this param will be used but in regular mode this param is
redundant. Need to think about better solution.
Signed-off-by: Erez Shitrit
Signed-off-by: Erez Shitrit
---
drivers/infiniband/ulp/ipoib/ipoib.h| 2 +-
drivers/infiniband/ulp/ipoib/ipoib_ib.c | 8
2 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/drivers/infiniband/ulp/ipoib/ipoib.h
b/drivers/infiniband/ulp/ipoib/ipoib.h
index b3900b253ad5
-off-by: Erez Shitrit
---
include/rdma/ib_ipoib_accel_ops.h | 59 +++
include/rdma/ib_verbs.h | 36
2 files changed, 95 insertions(+)
create mode 100644 include/rdma/ib_ipoib_accel_ops.h
diff --git a/include/rdma
cleanup flows not to be specific for ETH link.
Signed-off-by: Erez Shitrit
---
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 120 ++
1 file changed, 80 insertions(+), 40 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
b/drivers/net/ethernet
resources.
Signed-off-by: Erez Shitrit
---
drivers/infiniband/ulp/ipoib/ipoib.h | 5 +-
drivers/infiniband/ulp/ipoib/ipoib_ib.c| 39 +++---
drivers/infiniband/ulp/ipoib/ipoib_main.c | 116 -
drivers/infiniband/ulp/ipoib/ipoib_verbs.c | 40 +-
4
Add to the TIS and flow_table objects, and to hca capability table.
1. New capability bit: ipoib_enhanced_offloads, indicates new ability for UD
QP to do RSS.
2. In order to support SET_ROOT cmd, to connect between QP to FS table.
Signed-off-by: Erez Shitrit
---
include/linux/mlx5/mlx5_ifc.h
Add check for bit IB_QP_CREATE_NETIF_QP while creating QP.
Signed-off-by: Erez Shitrit
---
drivers/infiniband/hw/mlx5/qp.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c
index dc0ea63900c1..41e14d57fec9 100644
--- a/drivers
he
use of exported function from the mlx5_core layer to the ib layer.
Changes fron v0:
---
1. Use the vnic/hfi API as a base for the new design/impl.
2. Change the low level driver to support the new struct.
Erez Shitrit (25):
IB/ipoib: Separate control and data related initializa
ipoib_stop.
Signed-off-by: Erez Shitrit
---
drivers/infiniband/ulp/ipoib/ipoib.h | 2 +-
drivers/infiniband/ulp/ipoib/ipoib_ib.c | 228 +-
drivers/infiniband/ulp/ipoib/ipoib_main.c | 2 +-
3 files changed, 129 insertions(+), 103 deletions(-)
diff --git a/drivers
make sure to set the parent pointer prior to
invoking register_netdevice(), this allows the new ndo to be called
by the stack immediately after the child device is registered.
Fixes: 5aa7add8f14b ('infiniband/ipoib: implement ndo_get_iflink')
Reported-by: Honggang Li
Signed-off-by: Erez
make sure to set the parent pointer prior to
invoking register_netdevice(), this allows the new ndo to be called
by the stack immediately after the child device is registered.
Fixes: 5aa7add8f14b ('infiniband/ipoib: implement ndo_get_iflink')
Reported-by: Honggang Li
Signed-off-by: Erez
On Wed, Apr 15, 2015 at 7:06 PM, Jason Gunthorpe
wrote:
> On Wed, Apr 15, 2015 at 09:17:14AM +0300, Erez Shitrit wrote:
>> >>+ /* parent interface */
>> >>+ if (!test_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags))
>> >>+ ret
37 matches
Mail list logo
