For some HCAs, ib_modify_qp() is an expensive operation running
virtualized.
For both the active and passive side, the QP returned by the CM has the
state set to RTS, so no need for this excess RTS -> RTS transition. With
IB Core's ability to set the RNR Retry timer, we use this interface to
shave off another ib_modify_qp().
Fixes: ec16227e14 ("RDS/IB: Infiniband transport")
Link: https://lore.kernel.org/r/1617216194-12890-3-git-send-email-haakon.bugge@oracle.com
Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Introduce the ability for kernel ULPs to adjust the minimum RNR Retry
timer. The INIT -> RTR transition executed by RDMA CM will be used for
this adjustment. This avoids an additional ib_modify_qp() call.
rdma_set_min_rnr_timer() must be called before the call to rdma_connect()
on the active side and before the call to rdma_accept() on the passive
side.
The default value of RNR Retry timer is zero, which translates to 655
ms. When the receiver is not ready to accept a send messages, it encodes
the RNR Retry timer value in the NAK. The requestor will then wait at
least the specified time value before retrying the send.
The 5-bit value to be supplied to the rdma_set_min_rnr_timer() is
documented in IBTA Table 45: "Encoding for RNR NAK Timer Field".
Link: https://lore.kernel.org/r/1617216194-12890-2-git-send-email-haakon.bugge@oracle.com
Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com>
Acked-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Saeed Mahameed says:
====================
This pr contains changes from mlx5-next branch,
already reviewed on netdev and rdma mailing lists, links below.
1) From Leon, Dynamically assign MSI-X vectors count
Already Acked by Bjorn Helgaas.
https://patchwork.kernel.org/project/netdevbpf/cover/20210314124256.70253-1-leon@kernel.org/
2) Cleanup series:
https://patchwork.kernel.org/project/netdevbpf/cover/20210311070915.321814-1-saeed@kernel.org/
From Mark, E-Switch cleanups and refactoring, and the addition
of single FDB mode needed HW bits.
From Mikhael, Remove unused struct field
From Saeed, Cleanup W=1 prototype warning
From Zheng, Esw related cleanup
From Tariq, User order-0 page allocation for EQs
====================
* mlx5-next:
net/mlx5: Implement sriov_get_vf_total_msix/count() callbacks
net/mlx5: Dynamically assign MSI-X vectors count
net/mlx5: Add dynamic MSI-X capabilities bits
PCI/IOV: Add sysfs MSI-X vector assignment interface
net/mlx5: Use order-0 allocations for EQs
net/mlx5: Add IFC bits needed for single FDB mode
net/mlx5: E-Switch, Refactor send to vport to be more generic
RDMA/mlx5: Use representor E-Switch when getting netdev and metadata
net/mlx5: E-Switch, Add eswitch pointer to each representor
net/mlx5: E-Switch, Add match on vhca id to default send rules
net/mlx5: Remove unused mlx5_core_health member recover_work
net/mlx5: simplify the return expression of mlx5_esw_offloads_pair()
net/mlx5: Cleanup prototype warning
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
To update xlt (during mlx5_ib_reg_user_mr()), the driver can request up to
1 MB (order-8) memory, depending on the size of the MR. This costly
allocation can sometimes take very long to return (a few seconds). This
causes the calling application to hang for a long time, especially when
the system is fragmented. To avoid these long latency spikes, the calls
the higher order allocations need to fail faster in case they are not
available.
In order to acheive this we need __GFP_NORETRY flag in the gfp_mask before
during fetching the free pages. Allow the algorithm to automatically fall
back to smaller page sizes.
Link: https://lore.kernel.org/r/1617425635-35631-1-git-send-email-praveen.kannoju@oracle.com
Signed-off-by: Praveen Kumar Kannoju <praveen.kannoju@oracle.com>
Acked-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
The number of MSI-X vectors is a PCI property visible through lspci. The
field is read-only and configured by the device. The mlx5 devices work in
a static or dynamic assignment mode.
Static assignment means that all newly created VFs have a preset number of
MSI-X vectors determined by device configuration parameters. This can
result in some VFs having too many or too few MSI-X vectors. Till now this
has been the only means of fine-tuning the MSI-X vector count and it was
acceptable for small numbers of VFs.
With dynamic assignment the inefficiency of having a fixed number of MSI-X
vectors can be avoided with each VF having exactly the required
vectors. Userspace will provide this information while provisioning the VF
for use, based on the intended use. For instance if being used with a VM,
the MSI-X vector count might be matched to the CPU count of the VM.
For compatibility mlx5 continues to start up with MSI-X vector assignment,
but the kernel can now access a larger dynamic vector pool and assign more
vectors to created VFs.
Link: https://lore.kernel.org/linux-pci/20210314124256.70253-4-leon@kernel.org
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
These new fields declare the number of MSI-X vectors that is possible to
allocate on the VF through PF configuration.
Value must be in range defined by min_dynamic_vf_msix_table_size and
max_dynamic_vf_msix_table_size.
The driver should continue to query its MSI-X table through PCI
configuration header.
Link: https://lore.kernel.org/linux-pci/20210314124256.70253-3-leon@kernel.org
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
A typical cloud provider SR-IOV use case is to create many VFs for use by
guest VMs. The VFs may not be assigned to a VM until a customer requests a
VM of a certain size, e.g., number of CPUs. A VF may need MSI-X vectors
proportional to the number of CPUs in the VM, but there is no standard way
to change the number of MSI-X vectors supported by a VF.
Some Mellanox ConnectX devices support dynamic assignment of MSI-X vectors
to SR-IOV VFs. This can be done by the PF driver after VFs are enabled,
and it can be done without affecting VFs that are already in use. The
hardware supports a limited pool of MSI-X vectors that can be assigned to
the PF or to individual VFs. This is device-specific behavior that
requires support in the PF driver.
Add a read-only "sriov_vf_total_msix" sysfs file for the PF and a writable
"sriov_vf_msix_count" file for each VF. Management software may use these
to learn how many MSI-X vectors are available and to dynamically assign
them to VFs before the VFs are passed through to a VM.
If the PF driver implements the ->sriov_get_vf_total_msix() callback,
"sriov_vf_total_msix" contains the total number of MSI-X vectors available
for distribution among VFs.
If no driver is bound to the VF, writing "N" to "sriov_vf_msix_count" uses
the PF driver ->sriov_set_msix_vec_count() callback to assign "N" MSI-X
vectors to the VF. When a VF driver subsequently reads the MSI-X Message
Control register, it will see the new Table Size "N".
Link: https://lore.kernel.org/linux-pci/20210314124256.70253-2-leon@kernel.org
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
HIP08 supports both normal and record doorbell mode for RQ and CQ, SQ
record doorbell for userspace is also supported by the software for
flushing CQE process. As now the capability of HIP08 are exposed to the
user and are configurable, the support of normal doorbell should be added
back.
Note that, if switching to normal doorbell, the kernel will report "flush
cqe is unsupported" if modify qp to error status as the flush is based on
record doorbell.
Link: https://lore.kernel.org/r/1616840738-7866-2-git-send-email-liweihang@huawei.com
Signed-off-by: Yixian Liu <liuyixian@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
