25 Commits

Author SHA1 Message Date
Juan Zhou
39730a19bc Two patchs are uploaded from rdma-core mainline
1.Remove unnecessary QP checks
2.Fix reference to uninitialized cq pointer

Signed-off-by: Juan Zhou <zhoujuan51@h-partners.com>
(cherry picked from commit 3ab0271a03e49392018be33a06b3078559250a1c)
2023-11-03 09:18:50 +08:00
Zhou Juan
ef2166f6b8 Support user to choose using UD sl or pktype to adapt MPI APP
According to Annex17_RoCEv2 (A17.4.5.2), for RoCEv2 UD, a CQE should
carry a flag that indicates if the received frame is an IPv4, IPv6 or
RoCE packet. But currently, the values of the flag corresponding to
these packet types haven't been defined yet in WC.

In UCX, 'sl' in ibv_wc for UD is used as the packet type flag, and the
packet type values have already been defined in the UCX patch of
ed28845b88

Therefore, to adapt UCX, add a create flag to hnsdv_create_qp() to allow
users to choose whether they use 'sl' in ibv_wc as service level or
packet type for UD. For the latter, obtain and translate the packet type
from CQE and fill it to 'sl' in ibv_wc.

Singed-off-by: Juan Zhou <zhoujuan51@h-partners.com>
(cherry picked from commit e102d4c9aa2992c125b26ad5cc237ae002bc6541)
2023-11-02 20:18:35 +08:00
Zhou Juan
dabf4d530e Backport bugfix for hns
1.Fix the owner bit error of sq in new io
2.Fix incorrect post-send with direct wqe of
3.Add a judgment to the congestion control algorithm

Singed-off-by: Juan Zhou <zhoujuan51@h-partners.com>
(cherry picked from commit 092143ba858a7aba0630fadd416faa2a4e7eaf06)
2023-10-31 19:20:53 +08:00
Zhou Juan
bade32c716 Fix the sge number related errors and remove local invalidate operation
1. The hns hardware logic requires wr->num_sge to be 1 when
performing atomic operations. The code does not judge this
condition, and the current patch adds this constraint.

2. In the sq inline scenario, when num_sge in post_send is not 1, sge
array appears in the for loop without rotation and directly copy
out of bounds.

3. Currently local invalidate operation don't work properly.
Disable it for the time being.
HIP08 and HIP09 hardware does not support this feature, so
delete the associated code.

Signed-off-by: Juan Zhou <zhoujuan51@h-partners.com>
(cherry picked from commit 43c14b73409cf6e63278d5ff68e2694e592e9015)
2023-10-31 16:20:15 +08:00
Zhou Juan
97581b9828 Add support for SVE Direct WQE
Some Kunpeng SoCs do not support the DWQE through NEON
instructions. In this case, the IO path works normally,
but the performance will deteriorate.

For these SoCs that do not support NEON DWQE, they support
DWQE through SVE instructions. This patch supports SVE DWQE
to guarantee the performance of these SoCs. In addition, in
this scenario, DWQE only supports acceleration through SVE's
ldr and str instructions. Other load and store instructions
also cause performance degradation.

Signed-off-by: Juan Zhou <zhoujuan51@h-partners.com>
(cherry picked from commit 268e25f9374021fc4c0d6dabd62e0f360193081f)
2023-07-28 14:50:22 +08:00
Zhou Juan
e73844b3ec Support congestion control algorithm configuration
Added the use of direct verbs to implement QP-level
user-configurable congestion control algorithms. Among them,
the user mode driver mainly provides interfaces for users to
choose, and the kernel mode driver is responsible for filling
the resources of different algorithms and providing the
supported algorithm types for user mode.

At the same time, provide a direct verbs interface for users to
query the type of congestion control algorithm.

Signed-off-by: Juan Zhou <zhoujuan51@h-partners.com>
(cherry picked from commit 48ccef6ba011468fe91056398aad863f998569d2)
2023-06-02 11:57:20 +08:00
Yixing Liu
8b56ab8b70 Support libhns stop sending db mechanism after reset
Add an interface to the user space, which is used to receive
the kernel reset state. After receiving the reset flag, the
user space stops sending db.

Signed-off-by: Yixing Liu <liuyixing1@huawei.com>
2022-12-14 20:23:39 +08:00
Chengchang Tang
b88a370b79 Support hns roce DCA
DCA(Dynamic context attachment) support many RC QPs to share the WQE
buffer in a memory pool, this help reducing the memory consumption
when there are many QPs are inactive.

Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
2022-11-30 20:34:42 +08:00
Yixing Liu
1c38175fa1 Add supoort libhns td unlock
This patch add libhns td unlock function.

Signed-off-by: Yixing Liu <liuyixing1@huawei.com>
2022-11-29 14:53:10 +08:00
Guofeng Yue
648d17f1ef Support hns RoH mode
These patches support running the roce function in hns roh mode

Signed-off-by: Guofeng Yue <yueguofeng@hisilicon.com>
2022-11-29 14:25:05 +08:00
Chengchang Tang
6f27f67e51 Backport patches from 41.1
Backport patches from rdma-core 41.1.

And bugfix patches reported by #I5Q3S5 has also been included.

Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
2022-11-06 23:17:09 +08:00
Chengchang Tang
cc1006c917 Fix missing patch list in spec
Patches for hns DSCP was forgotten to add to be added to the spec. And
some mistake in DSCP patch is fixed.

Fixes: 68f61fd0a1a8 ("Add support for hns DSCP")
Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
2022-11-01 20:59:41 +08:00
Chengchang Tang
484504d625 Add support for hns DSCP
Support DSCP for hns RoCE.

Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
2022-10-31 14:28:29 +08:00
Luoyouming
0759ff83eb Bugfix for sge num and support inline feature
Fix sge num bug, add compatibility for rq inline, support cqe inline

Signed-off-by: Luoyouming <luoyouming@huawei.com>
2022-10-31 14:28:19 +08:00
Chengchang Tang
060786cc7e Update to 41.0
Update rdma-core version from 35.1 to 41.0.

Version 41.0 is the latest version in community until
2022/7/27. It includes some new bugfixes and new features,
we choose this version to facilitate future development.

The patches added to this repo has already included in the
new version, so remove them.

Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
2022-07-27 18:26:26 +08:00
zhengfeng luo
1725c90650 Backport bugfix for hns SRQ and SGE from rdma-core V36.
Bugfix for hns SRQ and SGE. And also added all related cleanups and
refactorings.

Signed-off-by: zhengfeng luo <luozhengfeng@h-partners.com>
Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
2022-07-14 09:13:04 +08:00
peijiankang
38d2b22692 modify changlog error 2022-06-08 15:53:56 +08:00
Chengchang Tang
4238b9d0b7 Support hns DWQE
Add support for hns DWQE. All its dependent patches are also added.

Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
(cherry picked from commit 964c2273a107ddba9cbeea1425b3b9ab73ffe2ca)
2022-01-13 15:24:47 +08:00
eaglegai
7421d0165a update to 35.1 2021-12-09 19:24:40 +08:00
seuzw
ea6d72236d use -e para to make transferred meaning take effect 2021-09-26 17:05:44 +08:00
zhongxuan
bc23749c6c increase maximum number of cpus
Signed-off-by: zhongxuan <zhongxuan2@huawei.com>
2021-09-24 11:02:11 +08:00
Yangyang Li
e696cb029a Update to 35.0
update rdma-core version from 28.1 to 35.0, version 35.0 is
the latest version in community until 2021/7/5. It includes
some bugfixes and new features, we choose this version to
facilitate future development.

Signed-off-by: Yangyang Li <liyangyang20@huawei.com>
2021-07-05 11:14:43 +08:00
cuibb1
19b916d795 update to 28.1 2020-07-27 18:25:57 +08:00
zhuchunyi
2dc66b67f4 update code 2019-11-06 19:51:11 +08:00
overweight
bf1981a0b4 Package init 2019-09-30 11:16:12 -04:00