- 25 Oct, 2017 20 commits
-
-
Somnath Kotur authored
When host is shutting down, it invokes the shutdown hook of the L2 driver where it would attempt to free the MSI-X vectors, but would fail because some vectors are held by the RoCE driver. Implement the new hook in the L2 -> RoCE interface which will be invoked so that the RoCE driver can unregister the device and free up the MSI-X vectors it had claimed so that L2 can proceed with it's shutdown without failure. Signed-off-by:
Somnath Kotur <somnath.kotur@broadcom.com> Signed-off-by:
Doug Ledford <dledford@redhat.com>
-
Leon Romanovsky authored
The scqe.stag is actually __b32, fix it. drivers/infiniband/hw/cxgb4/cq.c:754:52: warning: cast to restricted __be32 Cc: Steve Wise <swise@opengridcomputing.com> Signed-off-by:
Leon Romanovsky <leon@kernel.org> Reviewed-by:
Steve Wise <swise@opengridcomputing.com> Signed-off-by:
-
Kees Cook authored
In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() to pass the timer pointer explicitly. Cc: Moni Shoua <monis@mellanox.com> Cc: Doug Ledford <dledford@redhat.com> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: linux-rdma@vger.kernel.org Signed-off-by:
Kees Cook <keescook@chromium.org> Signed-off-by:
-
Parav Pandit authored
In recent code, two path record entries are alwasy cleared while allocated could be either one or two path record entries. This leads to zero out of unallocated memory. This fix initializes alternative path record only when alternative path is set. While we are at it, path record allocation doesn't check for OPA alternative path, but rest of the code checks for OPA alternative path. Path record allocation code doesn't check for OPA alternative LID. This can further lead to memory corruption when only one path record is allocated, but there is actually alternative OPA path record present in CM request. Cc: <stable@vger.kernel.org> # v4.12+ Fixes: 9fdca4da ("IB/SA: Split struct sa_path_rec based on IB and ROCE specific fields") Signed-off-by:
Parav Pandit <parav@mellanox.com> Reviewed-by:
Moni Shoua <monis@mellanox.com> Signed-off-by:
-
Maor Gottlieb authored
Some user space application would like to do RSS on the inner packet fields instead on the outer. When MLX5_RX_HASH_INNER is set with one or more of the other hash fields, then the RSS will be done using the inner packet. Signed-off-by:
Maor Gottlieb <maorg@mellanox.com> Reviewed-by:
Mark Bloch <markb@mellanox.com> Signed-off-by:
-
Maor Gottlieb authored
The device can support receive Stateless Offloads for the inner packet's fields only when the packet is processed by TIR which is enabled to support tunneling. Otherwise, the device treats the packet as an ordinary non-tunneling packet and receive offloads can be done only for the outer packet's field. In order to enable receive Stateless Offloading support for incoming tunneling traffic the TIR should be created with tunneled_offload_en. Tunneling offloads is supported only be raw ethernet QP. This patch includes: * New QP creation flag for tunneling offloads. * Reports device capabilities. Signed-off-by:
-
Maor Gottlieb authored
This patch updates the mlx5_ifc with the following: - Fix tunnel_stateless_gre typo. - max_geneve_opt_len - Maximum geneve options length. - tunnel_stateless_geneve_rx - If set, receive Stateless Offloads for Geneve tunneled (inner) packets are supported. Signed-off-by:
-
Guy Levi authored
In some benchmarks and some CPU architectures, writing the CQE on a full cache line size improves performance by saving memory access operations (read-modify-write) relative to partial cache line change. This patch lets the user to configure the device to pad the CQE up to 128B in case its content is less than 128B. Currently the driver supports only padding for a CQE size of 128B. Signed-off-by:
Guy Levi <guyle@mellanox.com> Reviewed-by:
-
Guy Levi authored
In commit 1cbe6fc8 ("IB/mlx5: Add support for CQE compressing") the concept of CQE compression was introduced and added a support for 64B CQE size. This change update the code to support 128B CQE size as well. Signed-off-by:
-
Guy Levi authored
Adding new bits in mlx5_ifc_cmd_hca_cap to get the hardware capabilities for: - compression_128: Support 128B CQE compression - cqe_128_always: Support 128B CQE padding Signed-off-by:
-
Noa Osherovich authored
Allow creation of a multi-packet receive queue. In order to create a multi-packet RQ, the following fields in the mlx5_ib_rwq should be set: - log_num_strides: Log of number of strides per WQE - single_stride_log_num_of_bytes: Log of a single stride size - two_byte_shift_en: When enabled, hardware pads 2 bytes of zeros before writing the message to memory (e.g. for the IP alignment). Signed-off-by:
Noa Osherovich <noaos@mellanox.com> Reviewed-by:
Majd Dibbiny <majd@mellanox.com> Signed-off-by:
-
Noa Osherovich authored
This patch reports the device's striding RQ capabilities to the user-space: - min/max_single_stride_log_num_of_bytes: Log of min/max number of bytes in a single stride. - min/max_single_wqe_log_num_of_strides: Log of min/max number of strides in a single WQE. - supported_qpts: A bit mask to know which QP types support multi- packet RQ, for now only Raw Packet QPs. Signed-off-by:
-
oulijun authored
It is needed to call modify cq API for modifying cq context fields for controlling event generation moderations. This patch mainly adds it. Signed-off-by:
Lijun Ou <oulijun@huawei.com> Signed-off-by:
-
Wei Hu(Xavier) authored
This patch updates the PD specification to 16M for hip08. And it updates the numbers of mtt and cqe segments for the buddy. As the CQE supports hop num 1 addressing, the CQE specification is 64k. This patch updates to set the CQE specification to 64k. Signed-off-by:
Shaobo Xu <xushaobo2@huawei.com> Signed-off-by:
Wei Hu (Xavier) <xavier.huwei@huawei.com> Signed-off-by:
-
Wei Hu(Xavier) authored
As the increase of the IRRL specification in hip08, the IRRL table chunk size needs to be updated. This patch updates the IRRL table chunk size to 256k for hip08. Signed-off-by:
-
Wei Hu(Xavier) authored
This patch updates to support WQE, CQE and PBL page size configurable feature, which includes base address page size and buffer page size. Signed-off-by:
-
Erez Shitrit authored
NAPI budget is 64 packets, while maximum polling size for the send CQ is 16. Let's bring them in sync, so the NAPI budget will be reused completely. Cc: Kamal Heib <kamalh@mellanox.com> Signed-off-by:
Erez Shitrit <erezsh@mellanox.com> Reviewed-by:
Alex Vesker <valex@mellanox.com> Signed-off-by:
Leon Romanovsky <leonro@mellanox.com> Signed-off-by:
-
Erez Shitrit authored
Instead of explicit call to poll_cq of the tx ring, use the NAPI mechanism to handle the completions of each packet that has been sent to the HW. The next major changes were taken: * The driver init completion function in the creation of the send CQ, that function triggers the napi scheduling. * The driver uses CQ for RX for both modes UD and CM, and CQ for TX for CM and UD. Cc: Kamal Heib <kamalh@mellanox.com> Signed-off-by:
-
Erez Shitrit authored
The first step toward using NAPI in the UD/TX flow is to separate between two flows, the NAPI and the xmit, meaning no use of shared variables between both flows. This patch takes out the tx_outstanding variable that was used in both flows and instead the driver uses the 2 cyclic ring variables: tx_head and tx_tail, tx_head used in the xmit flow and tx_tail in the NAPI flow. Cc: Kamal Heib <kamalh@mellanox.com> Signed-off-by:
-
Leon Romanovsky authored
RDMA subsystem has one active maintainer who can apply patches - Doug Ledford, but the RDMA entries in MAINTAINER file are not stating it. The following patch removes Sean's and Hal's emails from the maintainers list. It will allow clean get_maintaner.pl output which is needed for people outside of our community and various semi-automatic tools. Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal@mellanox.com> Signed-off-by:
Sean Hefty <sean.hefty@intel.com> Signed-off-by:
-
- 18 Oct, 2017 20 commits
-
-
Bob Sharp authored
Control QP (CQP) command backlog list is initialized at device initialization time. It is not reinitialized in the reset flow. Move the initialization to CQP creation time so the list can be initialized correctly for reset as well. Fixes: 86dbcd0f ("i40iw: add file to handle cqp calls") Signed-off-by:
Bob Sharp <Robert.O.Sharp@intel.com> Signed-off-by:
Shiraz Saleem <shiraz.saleem@intel.com> Signed-off-by:
-
Ivan Barrera authored
If User-space Direct Access (UDA) QP creation fails, the QP entry is not removed from QoS list. Fix this by removing QP from QoS list if create QP fails. Fixes: 0fc2dc58 ("i40iw: Add Quality of Service support") Signed-off-by:
Ivan Barrera <ivan.d.barrera@intel.com> Signed-off-by:
-
Christopher Bednarz authored
Clear Control Queue Pair (CQP) Head/Tail on CQP initialization as during an adapter reset, these values are not reinitialized. Tail is cleared by writing 0 to CQP's tail register. Head is cleared by writing 0 to CQP's doorbell register. Fixes: 86dbcd0f ("i40iw: add file to handle cqp calls") Signed-off-by:
Christopher Bednarz <christopher.n.bednarz@intel.com> Signed-off-by:
-
Shiraz Saleem authored
Queue depth calculations use a mix of work requests and actual number of bytes. Consolidate all calculations using minimum WQE size to avoid confusion. Signed-off-by:
-
Shiraz Saleem authored
On a netdev MTU change event, the iWARP Exception Queue (IEQ) buffers may not be sized properly to handle the new MTU. Reinitialize the IEQ with new MTU size on MTU change event. Also, add define for the max ethernet frame size field in IEQ QP context instead of the snd_mss define which is for iWARP QPs' MSS field. Signed-off-by:
-
Shiraz Saleem authored
Completion Event Queues are created and destroyed on a per device basis as opposed to per User-space Direct Access resource. Move ceq_valid to the correct place in i40iw_sc_dev from i40iw_puda_rsrc. Signed-off-by:
-
Shiraz Saleem authored
The IPv6 header size is not subtracted from MTU when MSS is set for QPs. Save MTU opposed to MSS in the vsi struct during initialization and calculate the MSS based on IPv4 vs IPv6 connection. Fixes: f27b4746 ("i40iw: add connection management code") Signed-off-by:
-
Mustafa Ismail authored
Some structures for post SQ operation are not used. Remove the following: i40iw_post_send_w_inv i40iw_post_inline_send_w_inv send_w_sol send_w_inv send_w_sol_inv inline_send_w_sol inline_send_w_inv inline_send_w_sol_inv Signed-off-by:
Mustafa Ismail <mustafa.ismail@intel.com> Signed-off-by:
-
Mustafa Ismail authored
Consolidate exception_lan_queue under VSI structure where it belongs. Remove it from device and QP structures. Signed-off-by:
-
Mustafa Ismail authored
The field static_rsrc in i40iw_create_qp_info is unused and not needed. Remove it. Signed-off-by:
-
Mustafa Ismail authored
The AE source field in Asynchronous Event Queue Entry (AEQE) is not set by the hardware for some AEs, but the code assumes it does. This results in incorrect processing of some AEs. Fix this by setting ae_src to I40IW_AE_SOURCE_RSVD for those AEs. Fixes: 86dbcd0f ("i40iw: add file to handle cqp calls") Signed-off-by:
-
Mustafa Ismail authored
Remove unimplemented Asynchronous Events (AE) and correct names. Signed-off-by:
-
Doug Ledford authored
Conflicts: drivers/infiniband/hw/cxgb4/cm.c drivers/infiniband/hw/qib/qib_driver.c drivers/infiniband/hw/qib/qib_mad.c There were minor fixups needed in these files. Just minor context diffs due to patches from independent sources touching the same basic area. Signed-off-by: -
Doug Ledford authored
The early for-next branch was based on v4.14-rc2, while the shared pull request I got from Mellanox used a v4.14-rc4 base. I'm making the branch that was the shared Mellanox pull request the new for-next branch and merging the early for-next branch into it. Signed-off-by: -
Jérémy Lefaure authored
Using the ARRAY_SIZE macro improves the readability of the code. Found with Coccinelle with the following semantic patch: @r depends on (org || report)@ type T; T[] E; position p; @@ ( (sizeof(E)@p /sizeof(*E)) | (sizeof(E)@p /sizeof(E[...])) | (sizeof(E)@p /sizeof(T)) ) Signed-off-by:
Jérémy Lefaure <jeremy.lefaure@lse.epita.fr> Signed-off-by:
-
Parav Pandit authored
The original code only took into consideration the largest header possible after the IB_BTH_BYTES. This was incorrect, as the largest possible header size is the largest possible combination of headers we might run into. The new code accounts for all possible headers in the largest possible combination and subtracts that from the MTU to make sure that all packets will fit on the wire. Link: https://www.spinics.net/lists/linux-rdma/msg54558.html Fixes: 3c86aa70 ("RDMA/cm: Add RDMA CM support for IBoE devices") Signed-off-by:
Daniel Jurgens <danielj@mellanox.com> Reported-by:
Roland Dreier <roland@purestorage.com> Signed-off-by:
-
Parav Pandit authored
The IB/core provides address resolution service and invokes callback handler when address resolve request completes of requester in worker thread context. Such caller might allocate or free memory in callback handler depending on the completion status to make further progress or to terminate a connection. Most ULPs resolve route which involves allocating route entry and path record elements in callback event handler. It has been noticed that WQ_MEM_RECLAIM flag should not be used for workers that tend to allocate memory in this [1] thread discussion. In order to mitigate this situation, WQ_MEM_RECLAIM flag was dropped for other such WQs in this [2] patch. Similar problem might arise with address resolution path, though its not yet noticed. The ib_addr workqueue is not memory reclaim path due to its nature of invoking callback that might allocate memory or don't free any memory under memory pressure. [1] https://www.spinics.net/lists/linux-rdma/msg53239.html [2] https://www.spinics.net/lists/linux-rdma/msg53416.html Fixes: f5481626 ("IB/addr: Remove deprecated create_singlethread_workqueue") Fixes: 5fff41e1 ("IB/core: Fix race condition in resolving IP to MAC") Signed-off-by:
-
Parav Pandit authored
This patch fixes the case where 'lifespan' entry of the hw_counters is not writable. Currently write callback is not exposed for for the hw_counters sysfs operation. Due to this, modifying lifespan value results into permission denied error in below example. echo 10 > /sys/class/infiniband/mlx5_0/ports/1/hw_counters/lifespan -bash: /sys/class/infiniband/mlx5_0/ports/1/hw_counters/lifespan: Permission denied This patch adds the hook to modify any attribute which implements store() operation. Fixes: b40f4757 ("IB/core: Make device counter infrastructure dynamic") Signed-off-by:
-
Parav Pandit authored
Since IB/core resolves the destination mac address for user and kernel consumers, avoid resolving in multiple provider drivers. Only ib_core resolves DMAC now, therefore resolve_eth_dmac is removed as exported symbol. Signed-off-by:
-
Parav Pandit authored
Introduce rdma_create_user_ah API which allows passing udata to provider driver and additionally which resolves DMAC for RoCE. ib_resolve_eth_dmac() resolves destination mac address for unicast, multicast, link local ipv4 mapped ipv6 and ipv6 destination gid entry. This allows all RoCE provider drivers to avoid duplicating such code. Such change brings consistency where IB core always resolves dmac and pass it to RoCE provider drivers for user and kernel consumers, with this ah_attr->roce.dmac is always an input field for provider drivers. This uniformity avoids exporting ib_resolve_eth_dmac symbol to providers or other modules. Therefore its removed as exported symbol at later in the patch series. Now uverbs and umad both makes use of rdma_create_user_ah API which fixes the issue where umad has invalid DMAC for address. Signed-off-by:
-
