>
>
> I assume by user-space driver we are talking about libverbs? We have
> separate libraries for ipath/qib and hfi. We should probably coalesce these
> into a single library but that is a separate issue. PSM is also unrelated to
> the work here since PSM is not verbs.
>
Not libibverbs but lib (e
I was writing my own objection to this but Jason beat me too it. More comments
below.
>
> On Wed, Oct 14, 2015 at 11:29:42AM +0300, Haggai Eran wrote:
> > respect the pkey_index in ib_send_wr.ud for GSI packets. Apparently
> > having the pkey_index in a work request isn't required by the IBA
>
Hi Linus,
We have four batched up patches for the current rc kernel.
Two of them are small fixes that are obvious.
One of them is larger than I would like for a late stage rc pull, but we
found an issue in the namespace lookup code related to RoCE and this
works around the issue for now (we allo
On Tue, Oct 13, 2015 at 08:32:25AM +0300, Moni Shoua wrote:
We initially thought to implement a shared library that contains the
transport logic.
However, it seems that a SW Verbs transport driver would allow better
code sharing.
In fact, the VT driver would need only a single user-space driver
On Wed, Oct 14, 2015 at 11:29:42AM +0300, Haggai Eran wrote:
> respect the pkey_index in ib_send_wr.ud for GSI packets. Apparently having
> the pkey_index in a work request isn't required by the IBA
> specifications,
I disagree. The spec is very clear, 13.5.3.2.1 perscribes how GMP
replies must be
On Mon, Oct 12, 2015 at 11:41:45AM -0700, Christoph Hellwig wrote:
>
> Hi Ted,
>
> while we have a high level agreement there's still some discussion
> needed. I can prepare a few slides for 10 minute discussion and then
> take it to the hallways with the interested people.
Thanks,
The Tech se
> On Oct 14, 2015, at 9:32 AM, Devesh Sharma
> wrote:
>
> Hi Chuck,
>
> With the server crash fix in place, ocrdma is passing iozone on this series.
>
> Series Tested-By: Devesh Sharma
Thanks to you and Sagi. A v3 is coming soon, which should be more
or less ready to merge.
>
> On Tue, O
On Wed, 26 Aug 2015, eran ben elisha wrote:
> > Do you have this in a git tree somewhere for testing?
>
> Yes,
> please pull from https://github.com/eranbenelisha/linux branch
> rebased-for-4.3/lb_prev
>
> If you with to get the user space as well for testing, please use:
> https://github.com/eran
On Wed, Oct 14, 2015 at 5:17 PM, Marciniszyn, Mike
wrote:
>> > Subject: [PATCH] fix return value error
>> >
>>
>> I checked returns in configfs (-ENOMEM), proc (-ENOENT), proc-sys (-
>> ENOMEM), ramfs (-ENOSPC), vfs (-ENOMEM).
>>
>> Not entirely consistent but this matches the majority.
>>
>> I ag
> > Subject: [PATCH] fix return value error
> >
>
> I checked returns in configfs (-ENOMEM), proc (-ENOENT), proc-sys (-
> ENOMEM), ramfs (-ENOSPC), vfs (-ENOMEM).
>
> Not entirely consistent but this matches the majority.
>
> I agree -EPERM is pretty misleading.
>
> Acked-by: Mike Marciniszyn
> Subject: [PATCH] fix return value error
>
I checked returns in configfs (-ENOMEM), proc (-ENOENT), proc-sys (-ENOMEM),
ramfs (-ENOSPC), vfs (-ENOMEM).
Not entirely consistent but this matches the majority.
I agree -EPERM is pretty misleading.
Acked-by: Mike Marciniszyn
--
To unsubscribe fr
On 10/14/2015 3:56 PM, Doug Ledford wrote:
Yes Or, I'm working on it. A new 4.3-rc pull request should come out
today and an update on the for next area probably as well.
Excellent. Re the next area, as was communicated to you multiple times,
we really need the next bits to be in
your kernel.
On 10/14/2015 2:59 PM, Leon Romanovsky wrote:
On Wed, Oct 14, 2015 at 11:17 AM, Heloise NH wrote:
>Signed-off-by: Heloise NH
The patch is a correct one, however can you update the subject and
description to be more informative?
Please add that new_inode() function can fail for allocation only.
Hi Chuck,
With the server crash fix in place, ocrdma is passing iozone on this series.
Series Tested-By: Devesh Sharma
On Tue, Oct 6, 2015 at 8:28 PM, Chuck Lever wrote:
> Introduce client-side support for bi-directional RPC/RDMA.
> Bi-directional RPC/RDMA is a pre-requisite for NFSv4.1 on RDM
On 10/14/2015 12:51 AM, Or Gerlitz wrote:
> On Sun, Oct 11, 2015 at 11:58 PM, Or Gerlitz wrote:
>> Hi Doug,
>> As we're soon on 4.3-rc5, could you please update how things look for 4.4?
>>
>> The mutlicast loopback prevention patches from Eran were intially
>> posted ~two months ago and went throu
On Wed, Oct 14, 2015 at 11:17 AM, Heloise NH wrote:
> Signed-off-by: Heloise NH
The patch is a correct one, however can you update the subject and
description to be more informative?
Please add that new_inode() function can fail for allocation only.
> ---
> drivers/infiniband/hw/ipath/ipath_fs.
On devices that do not support sending different P_Keys on a single GSI QP,
create a different UD QP for each non-zero P_Key in the P_Key table. When
transmitting a MAD with a specified P_Key, send it using the right QP.
Receive packets from all the QPs through the common SRQ.
Signed-off-by: Hagga
ib_mad_agent currently exposes an ib_qp and an ib_device unnecessarily.
Replace these fields with a single ib_pd, and use helper functions to get
the device and pd instead of accessing the fields directly.
Signed-off-by: Haggai Eran
---
drivers/infiniband/core/agent.c | 4 +-
drivers/in
Add a device event handler to capture P_Key table change events. For
devices that don't support setting the P_Key index per work request, update
the per-P_Key QP table in the MAD layer, creating QPs as
needed.
The code currently doesn't destroy created QPs when their pkeys are
cleared. This can be
As a preparation for supporting multiple transmission QPs for each GSI QP,
add a SRQ that will be used for all the receive buffers of these QPs.
Signed-off-by: Haggai Eran
---
drivers/infiniband/core/mad.c | 58 ++
drivers/infiniband/core/mad_priv.h | 1
The IB specs do not require per work-request control over the P_Key used in
GMPs. Most drivers have implemented this feature, but some drivers such as
ipath and mlx5 do not support it.
Add a capability bit, and turn it on in drivers that lack support for
per-WR P_Key.
Cc: Mike Marciniszyn
Cc: El
Hi,
Patch 4c21b5bcef73 ("IB/cma: Add net_dev and private data checks to RDMA
CM") started checking the BTH P_Key for RDMA CM requests, and this
uncovered some issues with P_Key handling. In mlx5 the pkey_index in ib_wc
wasn't set correctly [1]. In addition, we found out that mlx5 doesn't
respect t
In preparation for having an array of QPs in each ib_mad_qp_info, add the
qp_type and qp_num parameters to the ib_mad_qp_info struct so that clients
won't need to access the QPs themselves for this information.
Signed-off-by: Haggai Eran
---
drivers/infiniband/core/mad.c | 16 ++
Signed-off-by: Heloise NH
---
drivers/infiniband/hw/ipath/ipath_fs.c | 2 +-
drivers/infiniband/hw/qib/qib_fs.c | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/infiniband/hw/ipath/ipath_fs.c
b/drivers/infiniband/hw/ipath/ipath_fs.c
index 25422a3..da753bc 100644
On Tue, Oct 13, 2015 at 7:18 PM, Hefty, Sean wrote:
>> On Mon, Oct 12, 2015 at 7:42 PM, Hefty, Sean wrote:
>> >> When IP based addressing was introduced, ib_create_ah_from_wc was
>> >> changed in order to support a suitable AH. Since this AH should
>> >> now contains the DMAC (which isn't a simpl
25 matches
Mail list logo
