On Thu, Dec 24, 2015 at 10:25:08AM +0200, Or Gerlitz wrote:
> On 12/24/2015 9:31 AM, Faisal Latif wrote:
> >On Wed, Dec 23, 2015 at 08:42:01AM -0800, Or Gerlitz wrote:
> >>On 12/22/2015 1:13 AM, Faisal Latif wrote:
> >>>+
> >>>+enum i40iw_memreg_type {
> >>>+ IW_MEMREG_TYPE_MEM = 0x,
> >>>+
On 12/24/2015 11:22 AM, eran ben elisha wrote:
> On Mon, Dec 21, 2015 at 4:20 PM, Christoph Lameter wrote:
>> V2->V3: Add check for NOIETF mode and create special table
>> for that case.
>>
>> Check if the extended counters are available and if so
>> create the proper extended
On Tue, Dec 22, 2015 at 06:27:57PM -0800, gre...@linuxfoundation.org wrote:
> On Tue, Dec 22, 2015 at 02:15:08PM -0500, ira.weiny wrote:
> > On Mon, Dec 21, 2015 at 05:01:48PM -0800, gre...@linuxfoundation.org wrote:
[snip]
> > >
> > > No, git is good :)
> > >
> > > > How do we handle changes
Dispatch only port event to IB stack when port state changes.
Don't explicitly modify qps to error. Let application listen to
port events on async event queue or let QP fail with retry-exceeded
completion error.
Signed-off-by: Padmanabh Ratnakar
Signed-off-by:
Recently Dough Ledford reported a deadlock happening
between ocrdma-load sequence and NetworkManager service
issuing "open" on be2net interface.
The deadlock happens when any be2net hook (e.g. open/close) is called
in parallel to insmod ocrdma.ko.
A. be2net is sending administrative open/close
This patch series fixes for critical bugs introduced due to recent
changes in linux Infiniband stack.
Patch 0001 fixes a vlan-id assignment issues in presence of PFC.
This bug was introduced recently got introduced and it fixes
dbf727de7440 ('IB/core: Use GID table in AH creation and dmac
Recently Dough Ledford reported a deadlock happening
between ocrdma-load sequence and NetworkManager service
issueing "open" on be2net interface.
The deadlock happens when any be2net hook (e.g. open/close) is called
in parallel to insmod ocrdma.ko.
A. be2net is sending administrative open/close
vlan-id is wrongly getting as 0 when PFC is enabled.
Set vlan-id configured by user in QP parameters.
In case vlan interface is not used, flash a warning to
user to configure vlan and assign vlan-id as 0 in qp params.
Fixes: dbf727de7440 ('IB/core: Use GID table in AH creation and dmac
On 12/24/2015 05:41 AM, Or Gerlitz wrote:
> On 12/24/2015 12:00 PM, Christoph Hellwig wrote:
>> On Thu, Dec 24, 2015 at 10:02:29AM +0200, Or Gerlitz wrote:
>>> We had consensus among the reviewers that the 1st patch ("IB/core: Align
>>> coding style of ib_device_cap_flags structure") is wrong
On Thu, Dec 24, 2015 at 06:22:14PM +0200, eran ben elisha wrote:
> On Mon, Dec 21, 2015 at 4:20 PM, Christoph Lameter wrote:
[snip]
> >
> > +/*
> > + * Figure out which counter table to use depending on
> > + * the device capabilities.
> > + */
> > +static struct attribute_group
On 12/24/2015 9:05 AM, Faisal Latif wrote:
>Why the iwarp port mapper implementationhas to be repeated in each
>driver? can you join your code in a common place and avoid the duplication?
>
>root@r-dcs58 hw]# git grep RDMA_NL_ nes
>nes/nes.c: [RDMA_NL_IWPM_REG_PID] = {.dump =
On 12/24/2015 9:31 AM, Faisal Latif wrote:
On Wed, Dec 23, 2015 at 08:42:01AM -0800, Or Gerlitz wrote:
On 12/22/2015 1:13 AM, Faisal Latif wrote:
+
+enum i40iw_memreg_type {
+ IW_MEMREG_TYPE_MEM = 0x,
+ IW_MEMREG_TYPE_QP = 0x0001,
+ IW_MEMREG_TYPE_CQ = 0x0002,
+
On 12/24/2015 9:57 AM, Matan Barak wrote:
I totally agree that it's better to use the local IP address and not
just get a random device by using 127.0.0.1. You could get a specific
device by binding it, but then - use its local IP instead of
127.0.0.1.
Yes guys, it might be better but the user
On 12/24/2015 5:31 AM, Doug Ledford wrote:
On 12/20/2015 12:16 PM, Leon Romanovsky wrote:
Leon Romanovsky (3):
IB/core: Align coding style of ib_device_cap_flags structure
IB/core: Add cross-channel support
IB/mlx5: Add driver cross-channel support
From: Leon Romanovsky
The following set of patches implements cross-channel (CC) support
in the RDMA core and in the mlx5 infiniband driver.
The cross-channel feature allows to execute WQEs that involve
cross-channel synchronization of IO operations’ on different QPs.
From: Leon Romanovsky
Add support of cross-channel functionality to mlx5
driver. This includes ability to ignore overrun for CQ
which intended for cross-channel, export device capability and
configure the QP to be sync master/slave queues.
The cross-channel enabled QP
From: Leon Romanovsky
The cross-channel feature allows to execute WQEs that involve
synchronization of IO operations’ on different QPs.
This capability enables to program complex flows with a single
function call, hereby significantly reducing overhead associated
with IO
mmap routines
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe_mmap.c | 173
1 file changed, 173 insertions(+)
create mode 100644
Default/initial rxe device parameter settings.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe_param.h | 177 +++
1 file changed, 177
Add declarations for data structures used to hold per opcode
and per work request opcode tables.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe_hdr.h | 950
From: Amir Vadai
Things todo in order to get out of staging subtree.
Signed-off-by: Amir Vadai
Signed-off-by: Kamal Heib
---
drivers/staging/rdma/rxe/TODO | 18 ++
1 file changed, 18 insertions(+)
create mode
Functions to manipulate QP objects.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
---
drivers/staging/rdma/rxe/rxe_qp.c | 835 ++
1 file changed, 835 insertions(+)
create mode 100644
From: Amir Vadai
Adding a macro for RoCEv2 UDP destination port.
Signed-off-by: Amir Vadai
Signed-off-by: Kamal Heib
---
include/rdma/ib_verbs.h | 2 ++
1 file changed, 2 insertions(+)
diff --git a/include/rdma/ib_verbs.h
Manage and allocate pool of objects with given limit on number of
elements. Gets parameters from rxe_type_info. Pool elements are
allocated out of a slab cache. Objects that are using this facility
are: PD, QP, SRQ, CQ, MR, FMR, MW, etc.
Signed-off-by: Kamal Heib
QP request logic.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
---
drivers/staging/rdma/rxe/rxe_req.c | 679 +
1 file changed, 679 insertions(+)
create mode 100644 drivers/staging/rdma/rxe/rxe_req.c
diff
Handling of Work Completions.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
---
drivers/staging/rdma/rxe/rxe_comp.c | 728
1 file changed, 728 insertions(+)
create mode 100644
Functions to manipulate Address Vector.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe_av.c | 87 +++
1 file changed, 87
Useful information about work request opcodes and pkt opcodes in table
form.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe_opcode.c | 961
Doug and list Hi,
This patchset introduces Soft RoCE driver.
Some background on the driver: The original Soft-RoCE driver was implemented by
Bob Pearson from SFW. Bob started the submission process [1], but his work was
abandoned after v2.
Mellanox decided to pick it up and continue the
header bit mask definitions and header lengths declaration of the
rxe_opcode_info struct and rxe_wr_opcode_info struct.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
Intorduce Add SEND_LAST_INV and SEND_ONLY_INV opcodes in ib_pack.h to be
used by RXE for RC.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
include/rdma/ib_pack.h | 4
1 file changed, 4
Compute ICRC for UDP/IP/BTH headers
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
---
drivers/staging/rdma/rxe/rxe_icrc.c | 96 +
1 file changed, 96 insertions(+)
create mode 100644
Functions to be called by the networking layer.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe.h | 70 ++
1 file changed, 70
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe_loc.h | 291 +
1 file changed, 291 insertions(+)
create mode 100644
A 'task' is a short function that returns 0 as long as it needs to be
called again. rxe tasks are based on the kernel's tasklet infrastructure.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
Kconfig and Makefiles for RXE driver
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/Kconfig | 2 ++
drivers/staging/rdma/Makefile | 1 +
QP response logic.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
---
drivers/staging/rdma/rxe/rxe_resp.c | 1368 +++
1 file changed, 1368 insertions(+)
create mode 100644 drivers/staging/rdma/rxe/rxe_resp.c
diff
Dummy DMA processing for RXE device.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe_dma.c | 166 +
1 file changed, 166
Functions to manipulate CQ.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe_cq.c | 165 ++
1 file changed, 165 insertions(+)
create
sysfs interface for ib_rxe
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe_sysfs.c | 168 +++
1 file changed, 168 insertions(+)
create
Linux netdev related code
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
---
drivers/staging/rdma/rxe/rxe_net.c | 729 +
drivers/staging/rdma/rxe/rxe_net.h | 78
2 files changed, 807 insertions(+)
create
From: Amir Vadai
Objects used by the userspace to post work requests.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
include/uapi/rdma/Kbuild | 1 +
Multicast groups handling.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe_mcast.c | 190 +++
1 file changed, 190 insertions(+)
create
Add maintainer for rxe driver
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
MAINTAINERS | 9 +
1 file changed, 9 insertions(+)
diff --git a/MAINTAINERS b/MAINTAINERS
index
Handles receiving new packets which are sent to either request or
response processing.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe_recv.c | 371
A simple circular buffer that can optionally be shared between user
space and the kernel and can be resized.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe_queue.c |
Functions to manipulate SRQ.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe_srq.c | 195 +
1 file changed, 195 insertions(+)
Module main for ib_rxe
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
---
drivers/staging/rdma/rxe/rxe.c | 436 +
drivers/staging/rdma/rxe/rxe.h | 2 +
2
rxe interface to rdma/core
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
Reviewed-by: Haggai Eran
Signed-off-by: Kamal Heib
---
drivers/staging/rdma/rxe/rxe_verbs.c | 1423
MR objects handling.
Signed-off-by: Kamal Heib
Signed-off-by: Amir Vadai
---
drivers/staging/rdma/rxe/rxe_mr.c | 764 ++
1 file changed, 764 insertions(+)
create mode 100644 drivers/staging/rdma/rxe/rxe_mr.c
diff
>
>
> There were discussions, and Mellanox even contributed code to the effort.
> See Kamal's patches in the patch set I provided.
>
As far as I see it discussions were shallow and never produced an
agreement. Kamal's patches should not be considered as as such.
>>
> I'm not sure I understand what you mean. Denny has posted several high level
> emails similar to this one and has asked for public feedback. We have been
> addressing all the feedback as it has come in and we continue to work toward
> this common layer.
>
> To that end we have been asking for
On Thu, Dec 24, 2015 at 4:07 PM, Matan Barak wrote:
> On Thu, Dec 24, 2015 at 2:38 PM, Or Gerlitz wrote:
>> On 12/24/2015 12:42 PM, Sagi Grimberg wrote:
>>>
>>>
> This patch seems to generate a list corruption [1] when I test
> with
On Thu, Dec 24, 2015 at 05:43:11PM +0200, Moni Shoua wrote:
There were discussions, and Mellanox even contributed code to the effort.
See Kamal's patches in the patch set I provided.
As far as I see it discussions were shallow and never produced an
agreement. Kamal's patches should not be
Based on profiling, UD performance drops in case of processes
in a single client due to excess context switches when
the progress workqueue is scheduled.
This is solved by modifying the heuristic to select the
direct progress instead of the scheduling progress via
the workqueue when UD-like
On Thu, Dec 24, 2015 at 10:02:29AM +0200, Or Gerlitz wrote:
> We had consensus among the reviewers that the 1st patch ("IB/core: Align
> coding style of ib_device_cap_flags structure") is wrong cleanup which
> basically is (1) unneeded (2) creates more damage (git blame and such,
> non-applicable
> Applied to target-pending/for-next as v4.5-rc1 material, along with
> Reviewed-by tags from HCH.
So this is both in your and Dougs now it seems. Given the non-trivial
merge with the other RDMA updates I'd suggest to drop it from the
target tree as Doug already sorted out the merge.
--
To
Applied to target-pending/for-next as v4.5-rc1 material, along with
Reviewed-by tags from HCH.
So this is both in your and Dougs now it seems. Given the non-trivial
merge with the other RDMA updates I'd suggest to drop it from the
target tree as Doug already sorted out the merge.
Yea, this
On 12/24/2015 12:12 PM, Sagi Grimberg wrote:
This patch seems to generate a list corruption [1] when I test
with Doug's for-4.5 tree. Eran, care to take a look at this?
This patch is part from a series that was introduced in 4.3-rc1 [1], did
4.4-rc5/6 worked for you before you uploaded there
I’m taking the responsibility from Mellanox side on libmlx5, the formal
tree to be used: git://openfabrics.org/~yishaih/libmlx5.git
Thanks Eli for creating and maintaining it from day one.
Yishai
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a
This patch seems to generate a list corruption [1] when I test
with Doug's for-4.5 tree.
Eran, care to take a look at this?
[1]:
mlx4_core: Mellanox ConnectX core driver v2.2-1 (Feb, 2014)
mlx4_core: Initializing :04:00.0
mlx4_core :04:00.0: PCIe link speed is 8.0GT/s, device
My functional test suite includes Cthon, iozone, dbench, fio,
multi-threaded builds of git and the Linux kernel, and xfstests.
This patch series passes with NFSv3, NFSv4.0, and now NFSv4.1.
--
Chuck Lever
> On Dec 23, 2015, at 21:00, J. Bruce Fields wrote:
>
>> On Wed,
On Thu, Dec 24, 2015 at 11:17:46AM +0200, Kamal Heib wrote:
> We've located the driver in the staging subtree. This follows a requirement
> to implement an IB transport library - Soft RoCE is in the same boat like the
> hfi1
> driver. We need to define and implement a lib to prevent those code
>
Doug,
I'm also can't load mlx5 drivers in your tree [1] but
I don't know where it's from, it can come from pretty much everything...
Now I'm left with no useable HW to test with :(
[1]:
mlx5_core :06:00.0: firmware version: 12.14.74
mlx5_core :06:00.1: firmware version: 12.14.74
On 12/24/2015 12:00 PM, Christoph Hellwig wrote:
On Thu, Dec 24, 2015 at 10:02:29AM +0200, Or Gerlitz wrote:
We had consensus among the reviewers that the 1st patch ("IB/core: Align
coding style of ib_device_cap_flags structure") is wrong cleanup which
basically is (1) unneeded (2) creates more
This patch seems to generate a list corruption [1] when I test
with Doug's for-4.5 tree. Eran, care to take a look at this?
This patch is part from a series that was introduced in 4.3-rc1 [1],
Then something else broke it. Can people check their patches on doug's
tree? At the moment it's
On Thu, Dec 24, 2015 at 02:58:10PM +0200, Or Gerlitz wrote:
> On Thu, Dec 24, 2015 at 12:02 PM, Christoph Hellwig
> wrote:
> > On Thu, Dec 24, 2015 at 11:17:46AM +0200, Kamal Heib wrote:
> >> We've located the driver in the staging subtree. This follows a requirement
> >> to
On 12/24/2015 12:42 PM, Sagi Grimberg wrote:
This patch seems to generate a list corruption [1] when I test
with Doug's for-4.5 tree. Eran, care to take a look at this?
This patch is part from a series that was introduced in 4.3-rc1 [1],
Then something else broke it. Can people check their
On Thu, Dec 24, 2015 at 12:02 PM, Christoph Hellwig wrote:
> On Thu, Dec 24, 2015 at 11:17:46AM +0200, Kamal Heib wrote:
>> We've located the driver in the staging subtree. This follows a requirement
>> to implement an IB transport library - Soft RoCE is in the same boat like
On Thu, Dec 24, 2015 at 2:38 PM, Or Gerlitz wrote:
> On 12/24/2015 12:42 PM, Sagi Grimberg wrote:
>>
>>
This patch seems to generate a list corruption [1] when I test
with Doug's for-4.5 tree. Eran, care to take a look at this?
>>>
>>>
>>> This patch is part from
Expose the device capability flags which can be queried through uverbs in
the uapi headers.
Signed-off-by: Christoph Hellwig
---
include/rdma/ib_verbs.h | 94 +++-
include/uapi/rdma/ib_verbs.h | 66 +++
2
This exposes the WC opcodes supported by uverbs as part of the uapi
headers. It follows the same scheme as the WR opcodes.
Signed-off-by: Christoph Hellwig
---
include/rdma/ib_verbs.h | 29 +
include/uapi/rdma/ib_verbs.h | 16
2
This exposes the WC flags supported by uverbs as part of the uapi
headers. It follows the same scheme as the WR opcodes.
Signed-off-by: Christoph Hellwig
---
include/rdma/ib_verbs.h | 14 +++---
include/uapi/rdma/ib_verbs.h | 10 ++
2 files changed, 17
This exposes the send WR flags supported by uverbs as part of the uapi
headers. It follows the same scheme as the WR opcodes.
Signed-off-by: Christoph Hellwig
---
drivers/infiniband/hw/mlx5/mlx5_ib.h | 6 +++---
include/rdma/ib_verbs.h | 14 ++
Currently very little of the uverbs user interface is actually exposed in
uapi headers, and it's a constant struggle to figure out what's kernel
internal and what is actually exposed in public. This series starts
sorting this out by creating the infrastructure for a uapi header shared
between
75 matches
Mail list logo