- 07 Dec, 2018 11 commits
-
-
Mike Marciniszyn authored
Commit 4e045572 ("IB/hfi1: Add unique txwait_lock for txreq events") laid the ground work to support per resource waiting locking. This patch adds that with a lock unique to each sdma engine and pio sendcontext and makes necessary changes for verbs, PSM, and vnic to use the new locks. This is particularly beneficial for smaller messages that will exhaust resources at a faster rate. Fixes: 77241056 ("IB/hfi1: add driver files") Reviewed-by:
Gary Leshner <Gary.S.Leshner@intel.com> Signed-off-by:
Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by:
Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by:
Jason Gunthorpe <jgg@mellanox.com>
-
Mike Marciniszyn authored
The call to sdma_progress() is called outside the wait lock. In this case, there is a race condition where sdma_progress() can return false and the sdma_engine can idle. If that happens, there will be no more sdma interrupts to cause the wakeup and the vnic_sdma xmit will hang. Fix by moving the lock to enclose the sdma_progress() call. Also, delete the tx_retry. The need for this was removed by: commit bcad2913 ("IB/hfi1: Serve the most starved iowait entry first") Fixes: 64551ede ("IB/hfi1: VNIC SDMA support") Reviewed-by:
-
Mike Marciniszyn authored
This patch adds an interface to allow the driver to initialize the QP priv struct when the QP is created and after the qpn has been assigned. A field is added to the QP priv struct to reference the rcd and two new files are added to contain the function to initialize the rcd field so that more TID RDMA related code can be added here later. Signed-off-by:
Kaike Wan <kaike.wan@intel.com> Signed-off-by:
-
Kaike Wan authored
The OPFN and TID RDMA capability bits are added to allow users to control which feature is enabled and disabled. Reviewed-by:
Mitko Haralanov <mitko.haralanov@intel.com> Signed-off-by:
-
Kaike Wan authored
Currently, When a reserved operation is completed, its entry in the send queue will not be unreserved, which leads to the miscalculation of qp->s_avail and thus the triggering of a WARN_ON call trace. This patch fixes the problem by unreserving the reserved operation when it is completed. Fixes: 856cc4c2 ("IB/hfi1: Add the capability for reserved operations") Reviewed-by:
-
Ashutosh Dixit authored
Ingress packet check for 16B/bypass packets should consider the port LMC. Not doing this will result in packets sent to the LMC LIDs getting dropped. The check is implemented in HW for 9B packets. Reviewed-by:
Mike Ruhl <michael.j.ruhl@intel.com> Signed-off-by:
Ashutosh Dixit <ashutosh.dixit@intel.com> Signed-off-by:
-
Michael J. Ruhl authored
An incorrect sge sizing in the HFI PIO path will cause an OOPs similar to this: BUG: unable to handle kernel NULL pointer dereference at (null) IP: [] hfi1_verbs_send_pio+0x3d8/0x530 [hfi1] PGD 0 Oops: 0000 1 SMP Call Trace: ? hfi1_verbs_send_dma+0xad0/0xad0 [hfi1] hfi1_verbs_send+0xdf/0x250 [hfi1] ? make_rc_ack+0xa80/0xa80 [hfi1] hfi1_do_send+0x192/0x430 [hfi1] hfi1_do_send_from_rvt+0x10/0x20 [hfi1] rvt_post_send+0x369/0x820 [rdmavt] ib_uverbs_post_send+0x317/0x570 [ib_uverbs] ib_uverbs_write+0x26f/0x420 [ib_uverbs] ? security_file_permission+0x21/0xa0 vfs_write+0xbd/0x1e0 ? mntput+0x24/0x40 SyS_write+0x7f/0xe0 system_call_fastpath+0x16/0x1b Fix by adding the missing sizing check to correctly determine the sge length. Fixes: 77241056 ("IB/hfi1: add driver files") Reviewed-by:
-
Michael J. Ruhl authored
VNIC assumes that all SDMA engines have been configured for use. This is not necessarily true (i.e. if the count was constrained by the module parameter). Update VNICs usage to use the configured count, rather than the hardware count. Reviewed-by:
-
Mitko Haralanov authored
A CA is supposed to ignore FECN bits in multicast, ACK, and CNP packets. This patch corrects the behavior of the HFI1 driver in this regard by ignoring FECNs in those packet types. While fixing the above behavior, fix the extraction of the FECN and BECN bits from the packet headers for both 9B and 16B packets. Furthermore, this patch corrects the driver's response to a FECN in RDMA READ RESPONSE packets. Instead of sending an "empty" ACK, the driver now sends a CNP packet. While editing that code path, add the missing trace for CNP packets. Fixes: 88733e3b ("IB/hfi1: Add 16B UD support") Fixes: f59fb9e0 ("IB/hfi1: Fix handling of FECN marked multicast packet") Reviewed-by:
-
Kaike Wan authored
When it is requested to change its physical state back to Offline while in the process to go up, DC8051 will set the ERROR field in the DC8051_DBG_ERR_INFO_SET_BY_8051 register. This ERROR field will remain until the next time when DC8051 transitions from Offline to Polling. Subsequently, when the host requests DC8051 to change its physical state to Polling again, it may receive a DC8051 interrupt with the stale ERROR field still in DC8051_DBG_ERR_INFO_SET_BY_8051. If the host link state has been changed to Polling, this stale ERROR will force the host to transition to Offline state, resulting in a vicious cycle of Polling ->Offline->Polling->Offline. On the other hand, if the host link state is still Offline when the stale ERROR is received, the stale ERROR will be ignored, and the link will come up correctly. This patch implements the correct behavior by changing host link state to Polling only after DC8051 changes its physical state to Polling. Reviewed-by:
Krzysztof Goreczny <krzysztof.goreczny@intel.com> Signed-off-by:
-
Kaike Wan authored
This patch dumps the pio info for non-user send contexts to assist debugging in the field. Reviewed-by:
-
- 05 Dec, 2018 4 commits
-
-
Lijun Ou authored
This patch implements the process flow of SRQ asynchronous event. Signed-off-by:
Lijun Ou <oulijun@huawei.com> Signed-off-by:
-
Lijun Ou authored
This patch implements the SRQ(Share Receive Queue) verbs and update the poll cq verbs to deal with SRQ complentions. Signed-off-by:
-
Lijun Ou authored
This patch inits hem resource for SRQ table, includes SRQWQE and SRQWQE index resource. Signed-off-by:
-
Lijun Ou authored
This patch configures the flags for enabling the SRQ(Share Receive Queue) capacity as well as update the verb of querying device for setting srq specifications. Signed-off-by:
-
- 04 Dec, 2018 20 commits
-
-
Doug Ledford authored
From Yishai, ----------------------------------- This series enriches DEVX support in few aspects: it enables interoperability between DEVX and verbs and improves mechanism for controlling privileged DEVX commands. The first patch updates mlx5 ifc file. Next 3 patches enable modifying and querying verbs objects via the DEVX interface. To achieve that the core layer introduced the 'UVERBS_IDR_ANY_OBJECT' type to match any IDR object. Once it's used by some driver's method, the infrastructure skips checking for the IDR type and it becomes the driver handler responsibility. The DEVX methods of modify and query were changed to get any object type via the 'UVERBS_IDR_ANY_OBJECT' mechanism. The type checking is done per object as part of the driver code. The next 3 patches introduce more robust mechanism for controlling privileged DEVX commands. The responsibility to block/allow per command was moved to be done in the firmware based on the UID credentials that the driver reports upon user context creation. This enables more granularity per command based on the device security model and the user credentials. In addition, by introducing a valid range for 'general commands' we prevent the need to touch the driver's code any time that a new future command will be added. The last patch fixes the XRC verbs flow once a DEVX context is used. This is needed as XRCD is some shared kernel resource and as such a kernel UID (=0) should be used in its related resources. Thanks Yishai Hadas ----------------------------------- The top 6 patches are the mlx5-devx series, the remainder are from the mlx5-next tree as the mlx5-devx series depended on the mlx5-next mlx5_ifc file update. * mlx5-devx: (42 commits) IB/mlx5: Allow XRC usage via verbs in DEVX context IB/mlx5: Update the supported DEVX commands IB/mlx5: Enforce DEVX privilege by firmware IB/mlx5: Enable modify and query verbs objects via DEVX IB/core: Enable getting an object type from a given uobject IB/core: Introduce UVERBS_IDR_ANY_OBJECT net/mlx5: Update mlx5_ifc with DEVX UCTX capabilities bits RDMA/mlx5: Unfold modify RMP function RDMA/mlx5: Unfold create RMP function RDMA/mlx5: Initialize SRQ tables on mlx5_ib RDMA/mlx5: Update SRQ functions signatures to mlx5_ib format RDMA/mlx5: Use stages for callback to setup and release DEVX RDMA/mlx5: Remove SRQ signature global flag net/mlx5: Move SRQ functions to RDMA part net/mlx5: Remove references to local mlx5_core functions net/mlx5: Remove not-used lib/eq.h header file net/mlx5: Remove dead transobj code net/mlx5: Align SRQ licenses and copyright information net/mlx5: Debug print for forwarded async events net/mlx5: Forward SRQ resource events ... Signed-off-by:Doug Ledford <dledford@redhat.com>
-
Yishai Hadas authored
Allows XRC usage from the verbs flow in a DEVX context. As XRCD is some shared kernel resource between processes it should be created with UID=0 to point on that. As a result once XRC QP/SRQ are created they must be used as well with UID=0 so that firmware will allow the XRCD usage. Signed-off-by:
Yishai Hadas <yishaih@mellanox.com> Reviewed-by:
Artemy Kovalyov <artemyko@mellanox.com> Signed-off-by:
Leon Romanovsky <leonro@mellanox.com> Signed-off-by:
-
Yishai Hadas authored
Update the supported DEVX commands, it includes adding to the query/modify command's list and to the encoding handling. In addition, a valid range for general commands was added to be used for future commands. Signed-off-by:
-
Yishai Hadas authored
Enforce DEVX privilege by firmware, this enables future device functionality without the need to make driver changes unless a new privilege type will be introduced. Signed-off-by:
-
Yishai Hadas authored
Enables modify and query verbs objects via the DEVX interface. To support this the above DEVX handlers were changed to get any object type via the UVERBS_IDR_ANY_OBJECT mechanism. The type checking and handling is done per object as part of the driver code. Signed-off-by:
-
Yishai Hadas authored
Enable getting an object type from a given uobject, the type is saved upon tree merging and is returned as part of some helper function. Signed-off-by:
-
Yishai Hadas authored
Introduce the UVERBS_IDR_ANY_OBJECT type to match any IDR object. Once used, the infrastructure skips checking for the IDR type, it becomes the driver handler responsibility. This enables drivers to get in a given method an object from various of types. Signed-off-by:
-
Doug Ledford authored
The enhanced devx support series needs commit: 9d43faac ("net/mlx5: Update mlx5_ifc with DEVX UCTX capabilities bits") Signed-off-by:
-
Yishai Hadas authored
Expose device capabilities for DEVX user context, it includes which caps the device is supported and a matching bit to set as part of user context creation. Signed-off-by:
-
Leon Romanovsky authored
There is no need to perform modify_rmp in two separate function, while one of them uses stack as a placeholder for data while other allocates it dynamically. Combine those two functions to one call instead of two. Reviewed-by:
Mark Bloch <markb@mellanox.com> Signed-off-by:
-
Leon Romanovsky authored
There is no need to perform create_rmp in two separate function, while one of them uses stack as a placeholder for data while other allocates it dynamically. Combine those two functions to one instead of two. Reviewed-by:
-
Leon Romanovsky authored
Transfer initialization and cleanup from mlx5_priv struct of mlx5_core_dev to be part of mlx5_ib_dev. This completes removal of SRQ from mlx5_core. Reviewed-by:
-
Leon Romanovsky authored
Reflect the change of moving SRQ code from mlx5_core to mlx5_ib by updating function signatures do not require mlx5_core_dev as an input, because all operations in mlx5_ib are supposed to use mlx5_ib_dev. Reviewed-by:
-
Leon Romanovsky authored
Reuse existing infrastructure to initialize and release DEVX uid. The DevX interface is intended for user space access, so it is supposed to be initialized before ib_register_device(). Also it isn't supported in switchdev mode and don't need to initialize it in that mode. Fixes: 76dc5a84 ("IB/mlx5: Manage device uid for DEVX white list commands") Reviewed-by:
-
Leon Romanovsky authored
SRQ signature is not supported, hence no need for special static global variable to announce it. Reviewed-by:
-
Leon Romanovsky authored
There is no need to keep SRQ which is RDMA object in mlx5_core. In this patch, we partially move the execution code, while next patches will move table initialization/release logic too. Reviewed-by:
-
Leon Romanovsky authored
As a preparation to move SRQ functionality to RDMA, drop all references to mlx5_core logic and make SRQ be dependent on shared code only. Most of the time, we are interested to know if events are working/not working and it is possible with previous commit ("net/mlx5: Debug print for forwarded async events"). Reviewed-by: -
Leon Romanovsky authored
lib/eq.h is needed for EQ manipulation which are not performed in SRQ. Reviewed-by:
-
Leon Romanovsky authored
Delete functions which are not called and not needed. Reviewed-by:
-
Leon Romanovsky authored
Ensure that both RDMA and netdev parts of SRQ implementation has same copyright and license information annotated by SPDX tags. Reviewed-by:
-
- 03 Dec, 2018 5 commits
-
-
Leon Romanovsky authored
[leonro@server ~]$ rdma res show 1: mlx5_0: pd 3 cq 5 qp 4 cm_id 0 mr 0 ctx 0 Signed-off-by:
-
Leon Romanovsky authored
Add restrack annotations to track allocations of ucontexts. Signed-off-by:
-
Leon Romanovsky authored
Add ability to track allocated ib_ucontext, which are limited resource and worth to be visible by users. Signed-off-by:
-
Doug Ledford authored
Make all of the write() handlers use a consistent flow From Jason, This series unifies all the write handlers to use a flow that is very similar to the ioctl handler flow, including having the same basic assumptions about extensible buffer handling and the same handler function call signature. Along the way this consolidates all the copy_to/from_user into a small set of safe buffer accessor functions tailored to the usage here. These accessors use the new dispatcher-controlled calling convention for ucore data, and support a placement of the response that does not rely on the cmd.response value. Overall this brings in in strong bounds checking to all the write() handlers and consistent enforcement of the zero-fill/zero-check methodology for buffer extension. The end result is a significant complexity reduction for all of the handlers and creates a high degree of uniformity between the write, write_ex, and ioctl handlers and dispatch flow. Thanks Jason Gunthorpe (12): RDMA/uverbs: Remove out_len checks that are now done by the core RDMA/uverbs: Use uverbs_attr_bundle to pass ucore for write/write_ex RDMA/uverbs: Get rid of the 'callback' scheme in the compat path RDMA/uverbs: Use uverbs_response() for remaining response copying RDMA/uverbs: Use uverbs_request() for request copying RDMA/uverbs: Use uverbs_request() and core for write_ex handlers RDMA/uverbs: Fill in the response for IB_USER_VERBS_EX_CMD_MODIFY_QP RDMA/uverbs: Simplify ib_uverbs_ex_query_device RDMA/uverbs: Add a simple iterator interface for reading the command RDMA/uverbs: Use the iterator for ib_uverbs_unmarshall_recv() RDMA/uverbs: Do not check the input length on create_cq/qp paths RDMA/uverbs: Use only attrs for the write() handler signature drivers/infiniband/core/rdma_core.h | 5 +- drivers/infiniband/core/uverbs_cmd.c | 1165 ++++++++++--------------- drivers/infiniband/core/uverbs_main.c | 23 +- drivers/infiniband/core/uverbs_uapi.c | 23 +- include/rdma/uverbs_ioctl.h | 9 +- 5 files changed, 479 insertions(+), 746 deletions(-) Signed-off-by: -
Jason Gunthorpe authored
All of the old arguments can be derived from the uverbs_attr_bundle structure, so get rid of the redundant arguments. Most of the prior work has been removing users of the arguments to allow this to be a simple patch. Signed-off-by:
-
