On Thu, May 22, 2014 at 03:55:17PM -0400, bfields wrote:
> On Mon, May 19, 2014 at 01:40:22PM -0400, Chuck Lever wrote:
> > An NFS/RDMA client's source port is meaningless for RDMA transports.
> > The transport layer typically sets the source port value on the
> > connection to a random ephemeral p
On Mon, May 19, 2014 at 01:40:22PM -0400, Chuck Lever wrote:
> An NFS/RDMA client's source port is meaningless for RDMA transports.
> The transport layer typically sets the source port value on the
> connection to a random ephemeral port.
>
> Currently, NFS server administrators must specify the "
For user QPs, the creation process does not currently initialize the fields:
* qp->rq.offset
* qp->sq.offset
* qp->sq.wqe_shift
Signed-off-by: Haggai Eran
---
drivers/infiniband/hw/mlx5/qp.c | 4
1 file changed, 4 insertions(+)
diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infinib
For memory regions that are allocated using reg_umr, the suffix of
mlx5_core_create_mkey isn't being called. Instead the creation is completed in
a callback function (reg_mr_callback). This means that these MRs aren't being
added to the MR radix tree. The patch add them in the callback.
Signed-off
The patch stores iova, pd and size during mr creation and after UMRs that
modify them. It removes the unused access flags field.
Signed-off-by: Haggai Eran
---
drivers/infiniband/hw/mlx5/mr.c | 4
drivers/net/ethernet/mellanox/mlx5/core/mr.c | 4
include/linux/mlx5/driver.
From: Shachar Raindel
Instead of having the UMR context part of each memory region,
allocate a struct on the stack. This allows queuing multiple UMRs that access
the same memory region.
Signed-off-by: Shachar Raindel
Signed-off-by: Haggai Eran
---
drivers/infiniband/hw/mlx5/mlx5_ib.h | 13 +++
Hi,
This patchset contains fixes and minor improvements to the mlx5 driver.
The first patch fixes a memory leak in an error flow in the reg_umr
function. The second adds newly created memory regions to the MR radix
tree when they are created through reg_umr. Patches 3-4 fill some
fields that wer
If ib_post_send fails when posting the UMR work request in reg_umr, the code
doesn't release the temporary pas buffer allocated, and doesn't dma_unmap it.
Signed-off-by: Haggai Eran
---
drivers/infiniband/hw/mlx5/mr.c | 31 ---
1 file changed, 16 insertions(+), 15 del
On 21/5/2014 11:00 PM, Jason Gunthorpe wrote:
On Sun, May 18, 2014 at 12:39:11PM +0300, Or Gerlitz wrote:
From: Matan Barak
In order to implement IP based addressing for UD QPs, we need a way to
resolve the addresses internally.
The L2 params are passed to the provider driver using an extensio
On 21/5/2014 11:10 PM, Jason Gunthorpe wrote:
On Sun, May 18, 2014 at 12:38:48PM +0300, Or Gerlitz wrote:
+enum ibv_query_port_ex_attr_mask {
+ IBV_QUERY_PORT_EX_STATE = 1 << 0,
+ IBV_QUERY_PORT_EX_MAX_MTU = 1 << 1,
+ IBV_QUERY_PORT_EX_ACTIVE_MTU
On 21/5/2014 10:55 PM, Jason Gunthorpe wrote:
On Sun, May 18, 2014 at 12:38:46PM +0300, Or Gerlitz wrote:
From: Matan Barak
In order to support IP based addressing, one needs to pass the L2
parameters to the provider drive. This is done through a new extendable
On 22/05/2014 11:00, Or Gerlitz wrote:
sorry for the spam, I forgot to add Mike Christie, the iscsi
maintainer, so here you are CC-ed Mike,
I preferred doing it with a single reply vs. a whole new post, Mike if
you need the actual patch
for the sake of review/looking it's here
http://marc.inf
From: Ariel Nahum
iSER relies on refcounting to manage iser connections
establishment and teardown.
Following commit 39ff05d "IB/iser: Enhance disconnection logic
for multi-pathing", iser connection maintain 3 references:
- iscsi_endpoint (at creation stage)
- cma_id (at connection request stag
From: Roi Dayan
Logging messages need terminating newlines to avoid
possible message interleaving. Add them.
Signed-off-by: Roi Dayan
Signed-off-by: Joe Perches
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/iser/iscsi_iser.c |8
1 files changed, 4 insertions(+), 4 deleti
From: Or Gerlitz
This patchset address several resiliency issues in the area of
connection establishment and teardown (especially in error flows).
Patch #1 introduces a re-design of iSER connection management done
by Ariel and Sagi. It comes to solve few error-flow hangs/oops's
we've seen in re
From: Or Gerlitz
Signed-off-by: Sagi Grimberg
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/iser/iscsi_iser.h |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.h
b/drivers/infiniband/ulp/iser/iscsi_iser.h
index d309620..97c
From: Ariel Nahum
In some circumstances (multiple targets), RDMA_CM ESTABLISHED event and
ep_disconnect may race. In this case, the iser connection state may
transition to UP (after ep_disconnect transitioned it to TERMINATING),
while the connection is being teared down.
Upon RDMA_CM event ESTAB
17 matches
Mail list logo
