- 05 Feb, 2019 26 commits
-
-
Kaike Wan authored
This patch adds a function to build TID RDMA WRITE DATA packet. Signed-off-by:
Mitko Haralanov <mitko.haralanov@intel.com> Signed-off-by:
Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by:
Ashutosh Dixit <ashutosh.dixit@intel.com> Signed-off-by:
Kaike Wan <kaike.wan@intel.com> Signed-off-by:
Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by:
Doug Ledford <dledford@redhat.com>
-
Kaike Wan authored
This patch adds a function to receive TID RDMA WRITE response. The TID entries will be stored for encoding TID RDMA WRITE DATA packet later. Signed-off-by:
-
Kaike Wan authored
This patch adds the TID resource timer, which is used by the responder to free any TID resources that are allocated for TID RDMA WRITE request and not returned by the requester after a reasonable time. Reviewed-by:
-
Kaike Wan authored
This patch adds the function to build TID RDMA WRITE response. The main role of the TID RDMA WRITE RESP packet is to send TID entries to the requester so that they can be used to encode TID RDMA WRITE DATA packet. Signed-off-by:
-
Kaike Wan authored
This patch adds the functions to receive TID RDMA WRITE request. The request will be stored in the QP's s_ack_queue. This patch also adds code to handle duplicate TID RDMA WRITE request and a function to allocate TID resources for data receiving on the responder side. Signed-off-by:
-
Kaike Wan authored
The s_ack_queue is managed by two pointers into the ring: r_head_ack_queue and s_tail_ack_queue. r_head_ack_queue is the index of where the next received request is going to be placed and s_tail_ack_queue is the entry of the request currently being processed. This works perfectly fine for normal Verbs as the requests are processed one at a time and the s_tail_ack_queue is not moved until the request that it points to is fully completed. In this fashion, s_tail_ack_queue constantly chases r_head_ack_queue and the two pointers can easily be used to determine "queue full" and "queue empty" conditions. The detection of these two conditions are imported in determining when an old entry can safely be overwritten with a new received request and the resources associated with the old request be safely released. When pipelined TID RDMA WRITE is introduced into this mix, things look very different. r_head_ack_queue is still the point at which a newly received request will be inserted, s_tail_ack_queue is still the currently processed request. However, with pipelined TID RDMA WRITE requests, s_tail_ack_queue moves to the next request once all TID RDMA WRITE responses for that request have been sent. The rest of the protocol for a particular request is managed by other pointers specific to TID RDMA - r_tid_tail and r_tid_ack - which point to the entries for which the next TID RDMA DATA packets are going to arrive and the request for which the next TID RDMA ACK packets are to be generated, respectively. What this means is that entries in the ring, which are "behind" s_tail_ack_queue (entries which s_tail_ack_queue has gone past) are no longer considered complete. This is where the problem is - a newly received request could potentially overwrite a still active TID RDMA WRITE request. The reason why the TID RDMA pointers trail s_tail_ack_queue is that the normal Verbs send engine uses s_tail_ack_queue as the pointer for the next response. Since TID RDMA WRITE responses are processed by the normal Verbs send engine, s_tail_ack_queue had to be moved to the next entry once all TID RDMA WRITE response packets were sent to get the desired pipelining between requests. Doing otherwise would mean that the normal Verbs send engine would not be able to send the TID RDMA WRITE responses for the next TID RDMA request until the current one is fully completed. This patch introduces the s_acked_ack_queue index to point to the next request to complete on the responder side. For requests other than TID RDMA WRITE, s_acked_ack_queue should always be kept in sync with s_tail_ack_queue. For TID RDMA WRITE request, it may fall behind s_tail_ack_queue. Reviewed-by:
-
Kaike Wan authored
The TID RDMA WRITE protocol differs from normal IB RDMA WRITE in that TID RDMA WRITE requests do require responses, not just ACKs. Therefore, TID RDMA WRITE requests need to be treated as RDMA READ requests from the point of view of the QPs' s_ack_queue. In other words, the QPs' need to allow for TID RDMA WRITE requests to be stored in their s_ack_queue. However, because the user does not know anything about the TID RDMA capability and/or protocols, these extra entries in the queue cannot be advertized to the user. Reviewed-by:
-
Kaike Wan authored
This patch adds the functions to build TID RDMA WRITE request. The work request opcode, packet opcode, and packet formats for TID RDMA WRITE protocol are also defined in this patch. Signed-off-by:
-
Doug Ledford authored
This is the series for adding TID RDMA read. Kaike put in a lot of effort into making this more consumable for review so special thanks to him. Allocating resources and tracing are separated out followed by patches which build up the read request. Then we have the patches to receive incoming TID RDMA read requests and handle integration with the RC protocol. See the cover letter of the original posting for more of a detailed overview of TID. https://www.spinics.net/lists/linux-rdma/msg66611.html * tid-read: IB/hfi1: Add static trace for TID RDMA READ protocol IB/hfi1: Enable TID RDMA READ protocol IB/hfi1: Add interlock between a TID RDMA request and other requests IB/hfi1: Integrate TID RDMA READ protocol into RC protocol IB/hfi1: Increment the retry timeout value for TID RDMA READ request IB/hfi1: Add functions for restarting TID RDMA READ request IB/hfi1: Add TID RDMA handlers IB/hfi1: Add functions to receive TID RDMA READ response IB/hfi1: Add a function to build TID RDMA READ response IB/hfi1: Add functions to receive TID RDMA READ request IB/hfi1: Set PbcInsertHcrc for TID RDMA packets IB/hfi1: Add functions to build TID RDMA READ request IB/hfi1: Add static trace for flow and TID management functions IB/hfi1: Add the counter n_tidwait IB/hfi1: TID RDMA RcvArray programming and TID allocation IB/hfi1: TID RDMA flow allocation IB/hfi: Move RC functions into a header file Signed-off-by:
-
Kaike Wan authored
This patch makes the following changes to the static trace: 1. Adds the decoding of TID RDMA READ packets in IB header trace; 2. Tracks qpriv->s_flags and iow_flags in qpsleepwakeup trace; 3. Adds a new event to track RC ACK receiving; 4. Adds trace events for various stages of the TID RDMA READ protocol. These events provide a fine-grained control for monitoring and debugging the hfi1 driver in the filed. Reviewed-by:
-
Kaike Wan authored
This patch enables TID RDMA READ protocol by converting a qualified RDMA READ request into a TID RDMA READ request internally: (1) The TID RDMA capability must be enabled; (2) The request must start on a 4K page boundary and all receiving buffers must start on 4K page boundaries; (3) The request length must be a multiple of 4K and must be larger or equal to 256K. Each receiving buffer length must be a multiple of 4K. Signed-off-by:
-
Kaike Wan authored
This locking mechanism is designed to provent vavious memory corruption scenarios from occurring when requests are pipelined, especially when RDMA READ/WRITE requests are interleaved with TID RDMA READ/WRITE requests: 1. READ-AFTER-READ; 2. READ-AFTER-WRITE; 3. WRITE-AFTER-READ; When memory corruption is likely, a request will be held back until previous requests have been completed. Reviewed-by:
-
Kaike Wan authored
This patch integrates the TID RDMA READ protocol into the IB RC protocol. This protocol is an end-to-end protocol between the hfi1 drivers on two OPA nodes that converts a qualified RDMA READ request into a TID RDMA READ request to avoid data copying on the requester side. The following codes are added in this patch: - Send the TID RDMA READ request; - Complete the TID RDMA READ send request; - Send the TID RDMA READ response; - Complete the TID RDMA READ request; Reviewed-by:
-
Kaike Wan authored
The RC retry timeout value is based on the estimated time for the response packet to come back. However, for TID RDMA READ request, due to the use of header suppression, the driver is normally not notified for each incoming response packet until the last TID RDMA READ response packet. Consequently, the retry timeout value should be extended to cover the transaction time for the entire length of a segment (default 256K) instead of that for a single packet. This patch addresses the issue by introducing new retry timer functions to account for multiple packets and wrapper functions for backward compatibility. Reviewed-by:
-
Kaike Wan authored
This patch adds functions to retry TID RDMA READ request. Since TID RDMA READ request could be retried from any segment boundary, it requires a number of tracking fields in various structures and those fields should be reset properly. The qp->s_num_rd_atomic field is reset before retry and therefore should be incremented for each new or retried RDMA READ or atomic request. Reviewed-by:
-
Kaike Wan authored
This commit adds the TID RDMA READ pointers to the receiving opcode handlers. It also adds TID RDMA READ header sizes to header size table. A function to print the RHF EFLAGS errors is created so that it can be shared by both IB and TID RDMA receiving functions. Reviewed-by:
-
Kaike Wan authored
This patch adds the functions to receive TID RDMA READ response. The TID resource information in the KDETH packet header will direct the hardware to deliver the packet payload to the user buffer automatically and the software will handle the packet header for the last packet of a segment as all other packet headers are suppressed by default. The TID entries will be freed when all packets for a segment have been received. This patch also adds the functions to handle KDETH eflag errors, including flow sequence and generation errors, when a TID RDMA READ response packet is received . The flow sequence error can be recovered by software checking of the flow sequence and will disappear when the hardware flow is programmed with a new generation number. Reviewed-by:
-
Kaike Wan authored
This patch adds the function to build TID RDMA READ response packet. The previously received TID resource information will be used to build the KDETH packet, which will direct the delivery of packet payload by hardware. Reviewed-by:
-
Kaike Wan authored
This patch adds the functions to receive TID RDMA READ request. The TID resource information will be stored and tracked. Duplicate request will also be handled properly. Reviewed-by:
-
Kaike Wan authored
All TID RDMA packets are in KDETH packet format and therefore the PbcInsertHcrc must be set properly before sending the packet to hardware. Otherwise, the packets will be dropped by the receiver. By default, HCRC is not inserted for 9B packets without KDETH, and this patch adds that back for TID RDMA packets. Reviewed-by:
-
Kaike Wan authored
This patch adds the helper functions to build the TID RDMA READ request on the requester side. The key is to allocate TID resources (TID flow and TID entries) and send the resource information to the responder side along with the read request. Since the TID resources are limited, each TID RDMA READ request has to be split into segments with a default segment size of 256K. A software flow is allocated to track the data transaction for each segment. The work request opcode, packet opcode, and packet formats for TID RDMA READ protocol are also defined in this patch. Reviewed-by:
-
Kaike Wan authored
This patch adds the static trace for the flow and TID management functions to help debugging in the filed. Reviewed-by:
-
Kaike Wan authored
This patch adds the counter n_tidwait to count the number of times the TID resource allocator has to wait for TID resources. Reviewed-by:
-
Kaike Wan authored
TID entries are used by hfi1 hardware to receive data payload from incoming packets directly into a user buffer and thus avoid data copying by software. This patch implements the functions for TID allocation, freeing, and programming TID RcvArray entries in hardware for kernel clients. TID entries are managed via lists of TID groups similar to PSM. Furthermore, to track TID resource allocation for each request, software flows are also allocated and freed as needed. Since software flows consume large amount of memory for tracking TID allocation and freeing, it is generally desirable to allocate them dynamically in the send queue and only for TID RDMA requests, but pre-allocate them for receive queue because the send queue could have thousands of entries while the receive queue has only a limited number of entries. Signed-off-by:
-
Kaike Wan authored
The hfi1 hardware flow is a hardware flow-control mechanism for a KDETH data packet that is received on a hfi1 port. It validates the packet by checking both the generation and sequence. Each QP that uses the TID RDMA mechanism will allocate a hardware flow from its receiving context for any incoming KDETH data packets. This patch implements: (1) a function to allocate hardware flow (2) a function to free hardware flow (3) a function to initialize hardware flow generation for a receiving context (4) a wait mechanism if the hardware flow is not available (4) a function to remove the qp from the wait queue for hardware flow when the qp is reset or destroyed. Signed-off-by: -
Kaike Wan authored
This patch moves some RC helper functions into a header file so that they can be called from both RC and TID RDMA functions. In addition, a common function for rewinding a request is created in rdmavt so that it can be shared between qib and hfi1 driver. Reviewed-by:
-
- 31 Jan, 2019 7 commits
-
-
Doug Ledford authored
This series adds the OPFN feature, which is used as the negotiation protocol by TID RDMA. This adds a totally hidden, in-band negotiation transfer that happens on the consumer's queue pair but without the consumer's knowledge. For that reason, things like completions for OPFN transfers must be filtered out of the completion queue and not sent to the consumer. This feature does not impact any consumer APIs, but does impact the driver/driver wire API. At a high level OPFN enables exchanging parameters between two hosts using IB compare and swap requests to a special virtual address. The request uses a reserved IB work request opcode (see patch 3). * opfn: IB/hfi1: Add static trace for OPFN IB/hfi1: Integrate OPFN into RC transactions IB/hfi1, IB/rdmavt: Allow for extending of QP's s_ack_queue IB/hfi1: OPFN interface IB/hfi1: Add OPFN helper functions for TID RDMA feature IB/hfi1: OPFN support discovery Signed-off-by: -
Kaike Wan authored
This patch adds the static trace to the OPFN code and moves tid related static trace code into a new header file. Reviewed-by:
-
Kaike Wan authored
OPFN parameter negotiation allows a pair of connected RC QPs to exchange a set of parameters in succession. This negotiation does not commence till the first ULP request. Because OPFN operations are operations private to the driver, they do not generate user completions or put the QP into error when they run out of retries. This patch integrates the OPFN protocol into the transactions of an RC QP. Reviewed-by:
-
Kaike Wan authored
The OPFN protocol uses the COMPARE_SWAP request to exchange data between the requester and the responder and therefore needs to be stored in the QP's s_ack_queue when the request is received on the responder side. However, because the user does not know anything about the OPFN protocol, this extra entry in the queue cannot be advertised to the user. This patch adds an extra entry in a QP's s_ack_queue. Reviewed-by:
-
Kaike Wan authored
OPFN allows a pair of connected RC QPs to exchange a set of parameters in succession. The parameter exchange itself is done using the IB compare and swap request with a special virtual address. The request is triggered using a reserved IB work request opcode. This patch implements the OPFN interface to initialize, start, process, and reset the OPFN request. Reviewed-by:
-
Kaike Wan authored
This patch adds the OPFN helper functions to initialize, encode, decode, and reset OPFN parameters for the TID RDMA feature. Reviewed-by:
-
Mitko Haralanov authored
OPFN (Omni Path Feature Negotiation) support discovery allows a RC QP to announce that it supports OPFN and also discover if OPFN is supported by the peer QP. OPFN parameter negotiation is skipped unless OPFN support is first discovered. OPFN support is announced by claiming what was the reserved bit in dword 1 of OmniPath modified base transport header in requests and responses. Reviewed-by:
-
- 30 Jan, 2019 4 commits
-
-
Michael J. Ruhl authored
Several locations for manipulating sges use an open coded sequence that is covered by helper functions. Use the appropriate helper functions. Signed-off-by:
Michael J. Ruhl <michael.j.ruhl@intel.com> Signed-off-by:
-
Michael J. Ruhl authored
Sge sizing is done in several places using an open coded method. This can cause maintenance issues. The open coded method is encapsulated in a helper routine. The helper was introduced with commit: 1198fcea ("IB/hfi1, rdmavt: Move SGE state helper routines into rdmavt") Update all call sites that have the open coded path with the helper routine. Reviewed-by:
-
Kamal Heib authored
The function ipoib_intercept_dev_id_attr() is only used in ipoib_main.c Fixes: f6350da4 ("IB/ipoib: Log sysfs 'dev_id' accesses from userspace") Signed-off-by:
Kamal Heib <kamalheib1@gmail.com> Signed-off-by:
Jason Gunthorpe <jgg@mellanox.com>
-
Adit Ranadive authored
Update the driver to use the new device capability to report 64-bit UAR PFNs. Reviewed-by:
Jorgen Hansen <jhansen@vmware.com> Signed-off-by:
Adit Ranadive <aditr@vmware.com> Reviewed-by:
Vishnu Dasa <vdasa@vmware.com> Signed-off-by:
-
- 29 Jan, 2019 3 commits
-
-
Jason Gunthorpe authored
Yishai Hadas says: Enable DEVX asynchronous query commands This series enables querying a DEVX object in an asynchronous mode. The userspace application won't block when calling the firmware and it will be able to get the response back once that it will be ready. To enable the above functionality: - DEVX asynchronous command completion FD object was introduced. - The applicable file operations were implemented to enable using it by the user application. - Query asynchronous method was added to the DEVX object, it will call the firmware asynchronously and manages the response on the given input FD. - Hot unplug support was added for the FD to work properly upon unbind/disassociate. - mlx5 core fence for asynchronous commands was implemented and used to prevent racing upon unbind/disassociate. This branch is based on mlx5-next & v5.0-rc2 due to dependencies, from git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux * branch 'devx-async': IB/mlx5: Implement DEVX hot unplug for async command FD IB/mlx5: Implement the file ops of DEVX async command FD IB/mlx5: Introduce async DEVX obj query API IB/mlx5: Introduce MLX5_IB_OBJECT_DEVX_ASYNC_CMD_FD Signed-off-by:
-
Yishai Hadas authored
Implement DEVX hot unplug for the async command FD. This is done by managing a list of the inflight commands and wait until all launched work is completed as part of devx_hot_unplug_async_cmd_event_file. Signed-off-by:
Yishai Hadas <yishaih@mellanox.com> Signed-off-by:
Leon Romanovsky <leonro@mellanox.com> Signed-off-by:
-
Yishai Hadas authored
Implement the file ops of the DEVX async command FD, this enables using the FD for reading the events and manage other options on the FD. Signed-off-by:
-
