This at least kicks the user mode applications that are watching
for device events.
Signed-off-by: Steve Wise
---
drivers/infiniband/hw/cxgb4/device.c | 10 +-
1 files changed, 9 insertions(+), 1 deletions(-)
diff --git a/drivers/infiniband/hw/cxgb4/device.c
b/drivers/infiniband/hw/
This avoids the CIDX_INC overflow issue with T4A2 when running
kernel RDMA applications.
Signed-off-by: Steve Wise
---
drivers/infiniband/hw/cxgb4/t4.h |8 +++-
1 files changed, 7 insertions(+), 1 deletions(-)
diff --git a/drivers/infiniband/hw/cxgb4/t4.h b/drivers/infiniband/hw/cxgb4/
Unloading iw_cxgb4 can crash due to the unload code trying to use
db_drop_task, which is uninitialized. So remove this dead code.
Signed-off-by: Steve Wise
---
drivers/infiniband/hw/cxgb4/device.c |1 -
drivers/infiniband/hw/cxgb4/iw_cxgb4.h |1 -
2 files changed, 0 insertions(+), 2
Set the ULP mode for initial RDMA connection setup to
the proper DDP mode. This avoids wasting some HW
resources while in streaming mode.
Signed-off-by: Steve Wise
---
drivers/infiniband/hw/cxgb4/cm.c |2 ++
drivers/net/cxgb4/t4_msg.h |1 +
2 files changed, 3 insertions(+), 0 del
Signed-off-by: Steve Wise
---
drivers/infiniband/hw/cxgb4/qp.c |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/infiniband/hw/cxgb4/qp.c b/drivers/infiniband/hw/cxgb4/qp.c
index 4f0be25..70a5a3c 100644
--- a/drivers/infiniband/hw/cxgb4/qp.c
+++ b/drivers/infini
Set the default to on.
Signed-off-by: Steve Wise
---
drivers/infiniband/hw/cxgb4/cm.c |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/infiniband/hw/cxgb4/cm.c b/drivers/infiniband/hw/cxgb4/cm.c
index 8b00e6c..65d3fe6 100644
--- a/drivers/infiniband/hw/cxgb4/c
- Show whether the SQ is in onchip memory or not.
- Dump both SQ and RQ QIDs.
Signed-off-by: Steve Wise
---
drivers/infiniband/hw/cxgb4/device.c | 13 +
1 files changed, 9 insertions(+), 4 deletions(-)
diff --git a/drivers/infiniband/hw/cxgb4/device.c
b/drivers/infiniband/hw/cxg
Hey Roland,
Here are some bug fixes and enhancements for the next merge.
---
Steve Wise (7):
RDMA/cxgb4: Dispatch FATAL event on EEH errors.
RDMA/cxgb4: Do CIDX_INC updates every 1/16 cq depth cqe reaps.
RDMA/cxgb4: remove db_drop_task.
RDMA/cxgb4: Use ULP_MODE_TCPDDP.
On Fri, Mar 11, 2011 at 09:55:00PM +, Alan Cook wrote:
> Roland Dreier writes:
> >
> > How are you setting up the connection?
> >
> > My understanding is that the RDMA CM (ie librdmacm) does not handle UC,
> > so if you are using that, it shouldn't work. Sean could confirm this.
> >
> > -
Jason Gunthorpe writes:
>
> That is quite likely, actually. You are using the RDMA CM and I have
> never heard of anyone using UC that way.
>
> Raw verbs UC definitely works.
>
> Jason
I will look into switching to using raw verbs. Thanks everyone for the
assistance.
--
To unsubscribe from t
Roland Dreier writes:
>
> How are you setting up the connection?
>
> My understanding is that the RDMA CM (ie librdmacm) does not handle UC,
> so if you are using that, it shouldn't work. Sean could confirm this.
>
> - R.
I am using RDMA CM. I have not seen any documentation that states it
On Fri, Mar 11, 2011 at 02:52:24PM -0500, Alan Cook wrote:
> I am using the vanilla 2.6.27.57 kernel. If the problem is a kernel
> bug, I would expect the kernel to not support UC at all, given the
> simplicity of the application.
That is quite likely, actually. You are using the RDMA CM and I hav
On Fri, Mar 11, 2011 at 10:29 AM, Alan Cook
wrote:
> I currently have a working RDMA InfiniBand based client-server application
> using
> Reliable Connection (RC) service. I would like to change the application to
> use
> Unreliable Connection (UC) service. I am not performing RDMA reads, so I
Jason Gunthorpe writes:
> I think what you are hitting is some code is calling ibv_modify_qp
> with an invalid qp_attr_mask for one of the state transitions. UC and
> RC have different requirements. Assuming you are not calling
> ibv_modify_qp in your app it is probably a kernel bug?
>
No calls
On Fri, Mar 11, 2011 at 06:29:19PM +, Alan Cook wrote:
> I currently have a working RDMA InfiniBand based client-server application
> using
> Reliable Connection (RC) service. I would like to change the application to
> use
> Unreliable Connection (UC) service. I am not performing RDMA read
> > I think the spec is pretty sound on this point. The SA is expected to
> > return SA queries for SMP records that match what is present in the
> > fabric if the requestor did the SMP query itself. The constraints IBA
> > places on a NodeInfo SMP query are such that localPortNum and
> > portGUID
I currently have a working RDMA InfiniBand based client-server application using
Reliable Connection (RC) service. I would like to change the application to use
Unreliable Connection (UC) service. I am not performing RDMA reads, so I should
be able to use UC.
I was under the impression that the
Fellows, I'd like to understand better the consequences of using or
not ipoib on an infiniband cluster used for running mpi jobs. Openmpi
can use the openib BTL without any ip configuration at the ib0 device.
Mys infiniband cluster has this configuration at the moment. Is such
configuration recomme
On 00:47 Thu 02 Sep , Hal Rosenstock wrote:
>
> Most of what the osmtest SA client is doing is not RMPP.
> For example, any SA GetTable request is not an RMPP operation
> but is requesting an RMPP response.
>
> SA AttributeOffset only needs to be set in an RMPP DATA packet
> so this is not ne
On 10:11 Wed 09 Mar , Jason Gunthorpe wrote:
> On Wed, Mar 09, 2011 at 05:47:44PM +0200, Alex Netes wrote:
> > Hi Jason,
> >
> > On 17:29 Tue 01 Mar , Jason Gunthorpe wrote:
> > > This value must match the portGUID, so it needs to vary on a per port
> > > basis like portGUID and not simply
20 matches
Mail list logo