- 03 Oct, 2018 21 commits
-
-
Venkata Sandeep Dhanalakota authored
Moving send completion code into rdmavt in order to have shared logic between qib and hfi1 drivers. Reviewed-by:
Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by:
Brian Welty <brian.welty@intel.com> Signed-off-by:
Venkata Sandeep Dhanalakota <venkata.s.dhanalakota@intel.com> Signed-off-by:
Harish Chegondi <harish.chegondi@intel.com> Signed-off-by:
Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by:
Jason Gunthorpe <jgg@mellanox.com>
-
Brian Welty authored
This patch moves hfi1_copy_sge() into rdmavt for sharing with qib. This patch also moves all the wss_*() functions into rdmavt as several wss_*() functions are called from hfi1_copy_sge() When SGE copy mode is adaptive, cacheless copy may be done in some cases for performance reasons. In those cases, X86 cacheless copy function is called since the drivers that use rdmavt and may set SGE copy mode to adaptive are X86 only. For this reason, this patch adds "depends on X86_64" to rdmavt/Kconfig. Reviewed-by:
Ashutosh Dixit <ashutosh.dixit@intel.com> Reviewed-by:
Michael J. Ruhl <michael.j.ruhl@intel.com> Reviewed-by:
-
Nathan Chancellor authored
Clang warns when one enumerated type is implicitly converted to another. drivers/infiniband/hw/mlx4/mad.c:1811:41: warning: implicit conversion from enumeration type 'enum mlx4_ib_qp_flags' to different enumeration type 'enum ib_qp_create_flags' [-Wenum-conversion] qp_init_attr.init_attr.create_flags = MLX4_IB_SRIOV_TUNNEL_QP; ~ ^~~~~~~~~~~~~~~~~~~~~~~ drivers/infiniband/hw/mlx4/mad.c:1819:41: warning: implicit conversion from enumeration type 'enum mlx4_ib_qp_flags' to different enumeration type 'enum ib_qp_create_flags' [-Wenum-conversion] qp_init_attr.init_attr.create_flags = MLX4_IB_SRIOV_SQP; ~ ^~~~~~~~~~~~~~~~~ The type mlx4_ib_qp_flags explicitly provides supplemental values to the type ib_qp_create_flags. Make that clear to Clang by changing the create_flags type to u32. Reported-by:Nick Desaulniers <ndesaulniers@google.com> Signed-off-by:
Nathan Chancellor <natechancellor@gmail.com> Signed-off-by:
-
Lijun Ou authored
The atomic operation not supported inline. Besides, the standard atomic operation only support a sge and the sge is placed in the wqe. Fix: 384f8818("RDMA/hns: Add atomic support") Signed-off-by:
Lijun Ou <oulijun@huawei.com> Signed-off-by:
-
Lijun Ou authored
In order to extend vlan device range, the design add two field of qp context for checking vlan packet in sender and in recevicer. Signed-off-by:
-
Lijun Ou authored
This patch adds local invalidate Memory Region (MR) support in the kernel space driver. Signed-off-by:
Yangyang Li <liyangyang20@huawei.com> Signed-off-by:
-
Lijun Ou authored
The hip08 hardware has two version. the version id are 0x20 and 0x21 according to the pci revision. It needs to adjust some fields for extending new features. The specific updates include: 1. Add some fields for supporting new features by enabling some reserved fields in 0x20 version. 2. remove some fields which the user is not visiable in order to support the extend features. 3. Init some fields with zero. These updates is compatible with 0x20 version. Signed-off-by:
-
Lijun Ou authored
According to hip08 limit, the buffer size of extend sge needs to be an integer wqe_sge_buf_page size. For example, the value of sge_shift field of qp context is greater or equal to eight when buffer page size is 4K size. The value of sge_shift field of qp context assigned by hr_qp->sge.sge_cnt. Signed-off-by:
-
Lijun Ou authored
According to the IB protocol definition, the driver needs to show the correct device information and the information will be queryed by device attribute. Signed-off-by:
-
Lijun Ou authored
In order to compatible with the third party RoCE device, The hardware modify the set method for the ecn field of ip header in new hip08 version. The high 6bit of tclass be assigned for dscp field of packet. Signed-off-by:
-
Lijun Ou authored
The hip08 split two hardware version. The version id are 0x20 and 0x21 according to the PCI revison. The max size of extend sge of sq is limited to 2M for 0x20 version and 8M for 0x21 version. It may be exceeded to 2M according to the algorithm that compute the product of wqe count and extend sge number of every wqe. But the product always less than 8M. Signed-off-by:
-
Lijun Ou authored
It will print the warning when the MSB bit of SLID is not zero running cm_req_handler function that test CM. It needs to fixed zero when test RoCE device. Signed-off-by:
-
Lijun Ou authored
When user issues a RDMA read and enables sq inline, it needs to report a bad wr to user. Signed-off-by:
-
Lijun Ou authored
It needs to include two special qps for every port. The hip08 have four ports and the all reserved qp numbers are eight. Signed-off-by:
-
Leon Romanovsky authored
All users of rdma_nl_chk_listeners() are interested to get boolean answer if netlink socket has listeners, so update all places to boolean function. Signed-off-by:
Leon Romanovsky <leonro@mellanox.com> Reviewed-by:
-
Kamal Heib authored
The ll parameter is not used in ib_modify_qp_is_ok(), so remove it. Signed-off-by:
Kamal Heib <kamalheib1@gmail.com> Reviewed-by:
-
Kamal Heib authored
This function is not in use - delete it. Signed-off-by:
-
Zhu Yanjun authored
In rxe_queue_init, q and q->buf are allocated. In do_mmap_info, q->ip is allocated. When error occurs, rxe_srq_from_init and the later error handler do not free these allocated memories. This will make memory leak. Signed-off-by:
Zhu Yanjun <yanjun.zhu@oracle.com> Signed-off-by:
-
Wei Yongjun authored
Fix to return a negative error code from the mthca_cmd_init() error handling case instead of 0, as done elsewhere in this function. Fixes: 80fd8238 ("[PATCH] IB/mthca: Encapsulate command interface init") Signed-off-by:
Wei Yongjun <weiyongjun1@huawei.com> Signed-off-by:
-
Jason Gunthorpe authored
The uapi radix tree is a write-once data structure protected by kref. Once we get to the ioctl() fop it is not possible for anything else to be writing to it, so the access should use rcu_dereference_protected. Reported-by:
Matthew Wilcox <willy@infradead.org> Signed-off-by:
-
Jason Gunthorpe authored
The zap_vma_ptes() is declared but not defined on NOMMU kernels, causing a link error for the newly added uverbs code: drivers/infiniband/core/uverbs_main.o: In function `uverbs_user_mmap_disassociate': uverbs_main.c:(.text+0x114c): undefined reference to `zap_vma_ptes' drivers/infiniband/core/uverbs_main.o: In function `rdma_umap_open': uverbs_main.c:(.text+0x53c): undefined reference to `zap_vma_ptes' Since all user access for all of our drivers depend on remapping pages to user space disable USER_ACCESS when there is no mmu. Fixes: 5f9794dc ("RDMA/ucontext: Add a core API for mmaping driver IO memory") Reported-by:
Arnd Bergmann <arnd@arndb.de> Signed-off-by:
-
- 01 Oct, 2018 13 commits
-
-
Parav Pandit authored
When RDMA CM connect request arrives for IB transport, it already contains device, port, netdevice (optional). Instead of traversing all the cma devices, use the cma device already found by the cma_find_listener() for which a listener id is provided. iWarp devices doesn't need to derive RoCE GIDs, therefore drop RoCE specific checks from cma_acquire_dev() and rename it to cma_iw_acquire_dev(). Signed-off-by:
Parav Pandit <parav@mellanox.com> Reviewed-by:
Daniel Jurgens <danielj@mellanox.com> Signed-off-by:
-
Parav Pandit authored
Light weight version of cma_acquire_dev() just for binding with rdma device based on source IP(v4/v6) address. This simplifies cma_acquire_dev() to avoid listen_id specific checks and also for subsequent simplification for IB vs iWarp. Signed-off-by:
-
Parav Pandit authored
When IP failover is used between multiple ports of a given rdma device, allow accepting CM requests from either of the ports. This is applicable for IPv4 and IPv6 non link local addressing scheme. IPv6 link local addresses are bound. IP failover requests for listen cm_ids bound to specific netdev interfaces cannot be supported. (Similar to traditional sockets). Signed-off-by:
-
Kaike Wan authored
Subnet Management Packets (SMP) should exclusively use VL15 and their SL is ignored (IBTA v1.3, Section 3.5.8.2). Therefore, when an SMP is posted, the SL in the address handle can be set to 0 by a user application. Consequently, when an address handle is created by the IB core, some fields in struct rvt_ah may not be set correctly by using the SL2SC and SC2VL tables at the time. Subsequently, when the request is post sent, the incoming swqe may fail the validation check, resulting in the rejection of the send request. This patch fixes the problem by using VL15 for any validation, ignoring the SL in the address handle. Reviewed-by:
Kaike Wan <kaike.wan@intel.com> Signed-off-by:
-
Alex Estrin authored
Since Virtual Lanes BCT credits and MTU are set through separate MADs, we have to ensure both are valid, and data VLs are ready for transmission before we allow port transition to Armed state. Fixes: 5e2d6764 ("IB/hfi1: Verify port data VLs credits on transition to Armed") Reviewed-by:
Alex Estrin <alex.estrin@intel.com> Signed-off-by:
-
Dennis Dalessandro authored
The dlid assignment made by looking into the u_ucast_dlid array does not do an explicit check for the size of the array. The code path to arrive at def_port, the index value is long and complicated so its best to just have an explicit check here. Signed-off-by:
-
Kaike Wan authored
This patch adds the static trace for resource wait. Reviewed-by:
Mitko Haralanov <mitko.haralanov@intel.com> Signed-off-by:
-
Dennis Dalessandro authored
Current implementation allows each qp to have only one send engine. As such, each qp has only one list to queue prebuilt packets when send engine resources are not available. To improve performance, it is desired to support multiple send engines for each qp. This patch creates the framework to support two send engines (two legs) for each qp for the TID RDMA protocol, which can be easily extended to support more send engines. It achieves the goal by creating a leg specific struct, iowait_work in the iowait struct, to hold the work_struct and the tx_list as well as a pointer to the parent iowait struct. The hfi1_pkt_state now has an additional field to record the current legs work structure and that is now passed to all egress waiters to determine the leg that needs to wait via a new iowait helper. The APIs are adjusted to use the new leg specific struct as required. Many new and modified helpers are added to support this change. Reviewed-by:
-
Kaike Wan authored
The driver-provided function check_send_wqe allows the hardware driver to check and set up the incoming send wqe before it is inserted into the swqe ring. This patch will rename it as setup_wqe to better reflect its usage. In addition, this function is only called when all setup is complete in rdmavt. Reviewed-by:
-
YueHaibing authored
Fixes gcc '-Wunused-but-set-variable' warning: drivers/infiniband/hw/hns/hns_roce_hw_v2.c: In function 'hns_roce_v2_post_send': drivers/infiniband/hw/hns/hns_roce_hw_v2.c:194:35: warning: variable 'dseg' set but not used [-Wunused-but-set-variable] Signed-off-by:
YueHaibing <yuehaibing@huawei.com> Signed-off-by:
-
Nathan Chancellor authored
Clang warns when one enumerated type is explicitly converted to another. drivers/infiniband/hw/qedr/qedr_roce_cm.c:198:28: warning: implicit conversion from enumeration type 'enum qed_roce_ll2_tx_dest' to different enumeration type 'enum qed_ll2_tx_dest' [-Wenum-conversion] ll2_tx_pkt.tx_dest = pkt->tx_dest; ~ ~~~~~^~~~~~~ 1 warning generated. Turns out that QED_ROCE_LL2_TX_DEST_NW and QED_ROCE_LL2_TX_DEST_LB are only used once in the whole tree and QED_ROCE_LL2_TX_DEST_MAX is used nowhere. Remove them and use the equivalent values from qed_ll2_tx_dest in their place. Reported-by:Michal Kalderon <michal.kalderon@cavium.com> Signed-off-by:
-
Michael J. Ruhl authored
If a MAD packet has incorrect header information, the logic uses the reply path to report the error. The reply path expects *resp_len to be set prior to return. Unfortunately, *resp_len is set to 0 for this path. This causes an incorrect response packet. Fix by ensuring that the *resp_len is defaulted to the incoming packet size (wc->bytes_len - sizeof(GRH)). Reviewed-by:
-
Zhu Yanjun authored
The buf is allocated by vmalloc_user in the function rxe_queue_init. So it is better to free it by vfree. Fixes: 8700e3e7 ("Soft RoCE driver") Reviewed-by:
-
- 28 Sep, 2018 1 commit
-
-
Israel Rukshin authored
In case target remote invalidates bogus rkey and signature is not used, pi_ctx is NULL deref. The commit also fails the connection on bogus remote invalidation. Fixes: 59caaed7 ("IB/iser: Support the remote invalidation exception") Signed-off-by:
Israel Rukshin <israelr@mellanox.com> Reviewed-by:
Max Gurtovoy <maxg@mellanox.com> Reviewed-by:
Sagi Grimberg <sagi@grimberg.me> Signed-off-by:
-
- 27 Sep, 2018 5 commits
-
-
Yishai Hadas authored
IB has additional protections with SELinux that cannot be extended to the DEVX domain. SELinux can restrict access to pkeys. The first version of DEVX blocked IB entirely until this could be understood. Since DEVX requires CAP_NET_RAW, it supersedes the SELinux restriction and allows userspace to form arbitrary packets with arbitrary pkeys. Thus we enable IB for DEVX when CAP_NET_RAW is given. Signed-off-by:
Yishai Hadas <yishaih@mellanox.com> Signed-off-by:
-
Yishai Hadas authored
Enable DEVX white list commands without the need for CAP_NET_RAW. DEVX uid must exist from the ucontext or the device so that the firmware will mask unprivileged capabilities. Signed-off-by:
-
Yishai Hadas authored
Manage device uid for DEVX white list commands. The created device uid will be used on white list commands if the user didn't supply its own uid. This will enable the firmware to filter out non privileged functionality as of the recognition of the uid. Signed-off-by:
-
Yishai Hadas authored
Expose RAW QP device handles to user space by extending the UHW part of mlx5_ib_create_qp_resp. This data is returned only when DEVX context is used where it may be applicable. Signed-off-by:
-
Parav Pandit authored
Currently mmap_sem is read locked while pinning the memory. In a multi-threaded application of a process, holding mmap_sem lock creates contention with other threads who might be either registering memory, creating QPs or simply doing mmap() as such operations also require to hold the mmap_sem write lock. All such operation cannot make forward progress until one memory pin operation is completed. It becomes more worse if the memory is unpinned and/or memory registration is large (in GB range). Therefore, instead of holding mmap_sem for too long (for whole region pinning), acquire and release the lock for every few pages. For example on x86 with 4K page size, acquire and release mmap_sem for every 2Mbytes memory chunk. This allows other competing threads to make progress who might wish to hold mmap_sem for shorter duration. When memory registration latency is measured using [1] for memory sizes ranging from 4K to 48GB, <= 1% or 0.5% degradation is noticed. In many runs no difference is seen other than run-to-run variance. In other targeted tests of users with large memory, desired improvements are seen due to reduced contention of mmap_sem. [1] https://github.com/paravmellanox/rtool $ rdma_resource_lat -c 1 -s 48G -a -u L -i 500 -A It registers pinned memory from 4K to 48GB size with 500 iterations for each memory size. $ rdma_resource_lat -c 1 -s 12G -a -u L -i 500 -t 4 4 competing threads pin memory, each of 12GB size with 500 iterations. Signed-off-by:
-
