This patch adds the interrupt handler function, the NAPI poll
function, and its associated helper functions for receiving
accelerated ipoib packets. While we are here, fix the formats
of two error printouts.
Link: https://lore.kernel.org/r/20200511160637.173205.64890.stgit@awfm-01.aw.intel.com
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Sadanand Warrier <sadanand.warrier@intel.com>
Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Currently the driver doesn't support completion vectors. These
are used to indicate which sets of CQs should be grouped together
into the same vector. A vector is a CQ processing thread that
runs on a specific CPU.
If an application has several CQs bound to different completion
vectors, and each completion vector runs on different CPUs, then
the completion queue workload is balanced. This helps scale as more
nodes are used.
Implement CQ completion vector support using a global workqueue
where a CQ entry is queued to the CPU corresponding to the CQ's
completion vector. Since the workqueue is global, it's guaranteed
to always be there when queueing CQ entries; Therefore, the RCU
locking for cq->rdi->worker in the hot path is superfluous.
Each completion vector is assigned to a different CPU. The number of
completion vectors available is computed by taking the number of
online, physical CPUs from the local NUMA node and subtracting the
CPUs used for kernel receive queues and the general interrupt.
Special use cases:
* If there are no CPUs left for completion vectors, the same CPU
for the general interrupt is used; Therefore, there would only
be one completion vector available.
* For multi-HFI systems, the number of completion vectors available
for each device is the total number of completion vectors in
the local NUMA node divided by the number of devices in the same
NUMA node. If there's a division remainder, the first device to
get initialized gets an extra completion vector.
Upon a CQ creation, an invalid completion vector could be specified.
Handle it as follows:
* If the completion vector is less than 0, set it to 0.
* Set the completion vector to the result of the passed completion
vector moded with the number of device completion vectors
available.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Sebastian Sanchez <sebastian.sanchez@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
To improve the readability of function prototypes, give the parameters
names in the affinity module.
Reviewed-by: Sebastian Sanchez <sebastian.sanchez@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
- Shared mlx5 updates with net stack (will drop out on merge if Dave's
tree has already been merged)
- Driver updates: cxgb4, hfi1, hns-roce, i40iw, mlx4, mlx5, qedr, rxe
- Debug cleanups
- New connection rejection helpers
- SRP updates
- Various misc fixes
- New paravirt driver from vmware
-----BEGIN PGP SIGNATURE-----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=QeGy
-----END PGP SIGNATURE-----
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma
Pull rdma updates from Doug Ledford:
"This is the complete update for the rdma stack for this release cycle.
Most of it is typical driver and core updates, but there is the
entirely new VMWare pvrdma driver. You may have noticed that there
were changes in DaveM's pull request to the bnxt Ethernet driver to
support a RoCE RDMA driver. The bnxt_re driver was tentatively set to
be pulled in this release cycle, but it simply wasn't ready in time
and was dropped (a few review comments still to address, and some
multi-arch build issues like prefetch() not working across all
arches).
Summary:
- shared mlx5 updates with net stack (will drop out on merge if
Dave's tree has already been merged)
- driver updates: cxgb4, hfi1, hns-roce, i40iw, mlx4, mlx5, qedr, rxe
- debug cleanups
- new connection rejection helpers
- SRP updates
- various misc fixes
- new paravirt driver from vmware"
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma: (210 commits)
IB: Add vmw_pvrdma driver
IB/mlx4: fix improper return value
IB/ocrdma: fix bad initialization
infiniband: nes: return value of skb_linearize should be handled
MAINTAINERS: Update Intel RDMA RNIC driver maintainers
MAINTAINERS: Remove Mitesh Ahuja from emulex maintainers
IB/core: fix unmap_sg argument
qede: fix general protection fault may occur on probe
IB/mthca: Replace pci_pool_alloc by pci_pool_zalloc
mlx5, calc_sq_size(): Make a debug message more informative
mlx5: Remove a set-but-not-used variable
mlx5: Use { } instead of { 0 } to init struct
IB/srp: Make writing the add_target sysfs attr interruptible
IB/srp: Make mapping failures easier to debug
IB/srp: Make login failures easier to debug
IB/srp: Introduce a local variable in srp_add_one()
IB/srp: Fix CONFIG_DYNAMIC_DEBUG=n build
IB/multicast: Check ib_find_pkey() return value
IPoIB: Avoid reading an uninitialized member variable
IB/mad: Fix an array index check
...
Do not allocate credit return base and DMA memory for
NUMA nodes without CPUs.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Harish Chegondi <harish.chegondi@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The struct hfi1_affinity is not used anymore.
We use the struct hfi1_affinity_node and hfi1_affinity_node_list
instead.
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The IRQ affinity entry is not needed after the irq notifier patch has been
added to the hfi1 driver.
The irq affinity settings for SDMA engine should be set using the standard
/proc/irq/<N>/ interface.
Reviewed-by: Jianxin Xiong <jianxin.xiong@intel.com>
Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Existing locking scheme in affinity.c file using the
&node_affinity.lock spinlock is not very elegant.
We acquire the lock to get hfi1_affinity_node entry,
unlock, and then use the entry without the lock held.
With more functions being added, which access and
modify the entries, this can lead to race conditions.
This patch makes this locking scheme more consistent.
It changes the spinlock to mutex. Since all the code
is executed in a user process context there is no need
for a spinlock. This also allows to keep the lock
not only while we look up for the node affinity entry,
but over the whole section where the entry is being used.
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sebastian Sanchez <sebastian.sanchez@intel.com>
Reviewed-by: Jianxin Xiong <jianxin.xiong@intel.com>
Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Add sysfs entry to allow user to override affinity for SDMA
engine interrupts.
Reviewed-by: Dean Luick <dean.luick@intel.com>
Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
When performing process affinity recommendations for MPI ranks, the current
algorithm doesn't take into account multiple HFI units. Also, real
cores and HT cores are not distinguished from one another. Therefore,
all HT cores are recommended to be assigned first within the local NUMA
node before recommending the assignments of cores in other NUMA nodes.
It's ideal to assign all real cores across all NUMA nodes first, then all
HT 1 cores, then all HT 2 cores, and so on to balance CPU workload. CPU
cores in other NUMA nodes could be running interrupt handlers, and this is
not taken into account.
To balance the CPU workload for user processes, the following
recommendation algorithm is used:
For each user process that is opening a context on HFI Y:
a) If all cores are assigned to user processes, start assignments all
over from the first core
b) Assign real cores first, then HT cores (First set of HT cores on
all physical cores, then second set of HT cores, and, so on) in the
following order:
1. Same NUMA node as HFI Y and not running an IRQ handler
2. Same NUMA node as HFI Y and running an IRQ handler
3. Different NUMA node to HFI Y and not running an IRQ handler
4. Different NUMA node to HFI Y and running an IRQ handler
c) Mark core as assigned in the global affinity structure. As user
processes are done, remove core assignments from global affinity
structure.
This implementation allows an arbitrary number of HT cores and provides
support for multiple HFIs.
This is being included in the kernel rather than user space due to the
fact that user space has no way of knowing the CPU recommendations for
contexts running as part of other jobs.
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Mitko Haralanov <mitko.haralanov@intel.com>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Sebastian Sanchez <sebastian.sanchez@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Kernel receive queues oversubscribe CPU cores on multi-HFI systems.
To prevent this, the kernel receive queues are separated onto
different cores, and the SDMA engine interrupts are constrained to
a lesser number of cores.
hfi1s_on_numa_node*krcvqs is the number of CPU cores that are
reserved for kernel receive queues for all HFIs. Each HFI initializes
its kernel receive queues to one of the reserved CPU cores. If there
ends up being 0 CPU cores leftover for SDMA engines, use the same
CPU cores as receive contexts.
In addition, general and control contexts are assigned to their own
CPU core, however, both types of contexts tend to have low traffic.
To save CPU cores, collapse general and control contexts to one CPU
core for all HFI units. This change prevents SDMA engine interrupts
from wrapping around general contexts.
Reviewed-by: Dean Luick <dean.luick@intel.com>
Signed-off-by: Sebastian Sanchez <sebastian.sanchez@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
When HFI units get initialized, they each use their own mask copy for
affinity assignments. On a multi-HFI system, affinity assignments
overbook CPU cores as each HFI doesn't have knowledge of affinity
assignments for other HFI units. Therefore, some CPU cores are never
used for interrupt handlers in systems with high number of CPU cores
per NUMA node.
For multi-HFI systems, SDMA engine interrupt assignments start all over
from the first CPU in the local NUMA node after the first HFI
initialization. This change allows assignments to continue where the
last HFI unit left off.
Add global structure for affinity assignments for multiple HFIs to share
affinity mask.
Reviewed-by: Jianxin Xiong <jianxin.xiong@intel.com>
Reviewed-by: Jubin John <jubin.john@intel.com>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Sebastian Sanchez <sebastian.sanchez@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The TODO list for the hfi1 driver was completed during 4.6. In addition
other objections raised (which are far beyond what was in the TODO list)
have been addressed as well. It is now time to remove the driver from
staging and into the drivers/infiniband sub-tree.
Reviewed-by: Jubin John <jubin.john@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Grzegorz Andrejczuk