Implement the IDXD performance monitor capability (named 'perfmon' in
the DSA (Data Streaming Accelerator) spec [1]), which supports the
collection of information about key events occurring during DSA and
IAX (Intel Analytics Accelerator) device execution, to assist in
performance tuning and debugging.
The idxd perfmon support is implemented as part of the IDXD driver and
interfaces with the Linux perf framework. It has several features in
common with the existing uncore pmu support:
- it does not support sampling
- does not support per-thread counting
However it also has some unique features not present in the core and
uncore support:
- all general-purpose counters are identical, thus no event constraints
- operation is always system-wide
While the core perf subsystem assumes that all counters are by default
per-cpu, the uncore pmus are socket-scoped and use a cpu mask to
restrict counting to one cpu from each socket. IDXD counters use a
similar strategy but expand the scope even further; since IDXD
counters are system-wide and can be read from any cpu, the IDXD perf
driver picks a single cpu to do the work (with cpu hotplug notifiers
to choose a different cpu if the chosen one is taken off-line).
More specifically, the perf userspace tool by default opens a counter
for each cpu for an event. However, if it finds a cpumask file
associated with the pmu under sysfs, as is the case with the uncore
pmus, it will open counters only on the cpus specified by the cpumask.
Since perfmon only needs to open a single counter per event for a
given IDXD device, the perfmon driver will create a sysfs cpumask file
for the device and insert the first cpu of the system into it. When a
user uses perf to open an event, perf will open a single counter on
the cpu specified by the cpu mask. This amounts to the default
system-wide rather than per-cpu counting mentioned previously for
perfmon pmu events. In order to keep the cpu mask up-to-date, the
driver implements cpu hotplug support for multiple devices, as IDXD
usually enumerates and registers more than one idxd device.
The perfmon driver implements basic perfmon hardware capability
discovery and configuration, and is initialized by the IDXD driver's
probe function. During initialization, the driver retrieves the total
number of supported performance counters, the pmu ID, and the device
type from idxd device, and registers itself under the Linux perf
framework.
The perf userspace tool can be used to monitor single or multiple
events depending on the given configuration, as well as event groups,
which are also supported by the perfmon driver. The user configures
events using the perf tool command-line interface by specifying the
event and corresponding event category, along with an optional set of
filters that can be used to restrict counting to specific work queues,
traffic classes, page and transfer sizes, and engines (See [1] for
specifics).
With the configuration specified by the user, the perf tool issues a
system call passing that information to the kernel, which uses it to
initialize the specified event(s). The event(s) are opened and
started, and following termination of the perf command, they're
stopped. At that point, the perfmon driver will read the latest count
for the event(s), calculate the difference between the latest counter
values and previously tracked counter values, and display the final
incremental count as the event count for the cycle. An overflow
handler registered on the IDXD irq path is used to account for counter
overflows, which are signaled by an overflow interrupt.
Below are a couple of examples of perf usage for monitoring DSA events.
The following monitors all events in the 'engine' category. Becuuse
no filters are specified, this captures all engine events for the
workload, which in this case is 19 iterations of the work generated by
the kernel dmatest module.
Details describing the events can be found in Appendix D of [1],
Performance Monitoring Events, but briefly they are:
event 0x1: total input data processed, in 32-byte units
event 0x2: total data written, in 32-byte units
event 0x4: number of work descriptors that read the source
event 0x8: number of work descriptors that write the destination
event 0x10: number of work descriptors dispatched from batch descriptors
event 0x20: number of work descriptors dispatched from work queues
# perf stat -e dsa0/event=0x1,event_category=0x1/,
dsa0/event=0x2,event_category=0x1/,
dsa0/event=0x4,event_category=0x1/,
dsa0/event=0x8,event_category=0x1/,
dsa0/event=0x10,event_category=0x1/,
dsa0/event=0x20,event_category=0x1/
modprobe dmatest channel=dma0chan0 timeout=2000
iterations=19 run=1 wait=1
Performance counter stats for 'system wide':
5,332 dsa0/event=0x1,event_category=0x1/
5,327 dsa0/event=0x2,event_category=0x1/
19 dsa0/event=0x4,event_category=0x1/
19 dsa0/event=0x8,event_category=0x1/
0 dsa0/event=0x10,event_category=0x1/
19 dsa0/event=0x20,event_category=0x1/
21.977436186 seconds time elapsed
The command below illustrates filter usage with a simple example. It
specifies that MEM_MOVE operations should be counted for the DSA
device dsa0 (event 0x8 corresponds to the EV_MEM_MOVE event - Number
of Memory Move Descriptors, which is part of event category 0x3 -
Operations. The detailed category and event IDs are available in
Appendix D, Performance Monitoring Events, of [1]). In addition to
the event and event category, a number of filters are also specified
(the detailed filter values are available in Chapter 6.4 (Filter
Support) of [1]), which will restrict counting to only those events
that meet all of the filter criteria. In this case, the filters
specify that only MEM_MOVE operations that are serviced by work queue
wq0 and specifically engine number engine0 and traffic class tc0
having sizes between 0 and 4k and page size of between 0 and 1G result
in a counter hit; anything else will be filtered out and not appear in
the final count. Note that filters are optional - any filter not
specified is assumed to be all ones and will pass anything.
# perf stat -e dsa0/filter_wq=0x1,filter_tc=0x1,filter_sz=0x7,
filter_eng=0x1,event=0x8,event_category=0x3/
modprobe dmatest channel=dma0chan0 timeout=2000
iterations=19 run=1 wait=1
Performance counter stats for 'system wide':
19 dsa0/filter_wq=0x1,filter_tc=0x1,filter_sz=0x7,
filter_eng=0x1,event=0x8,event_category=0x3/
21.865914091 seconds time elapsed
The output above reflects that the unspecified workload resulted in
the counting of 19 MEM_MOVE operation events that met the filter
criteria.
[1]: https://software.intel.com/content/www/us/en/develop/download/intel-data-streaming-accelerator-preliminary-architecture-specification.html
[ Based on work originally by Jing Lin. ]
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Link: https://lore.kernel.org/r/0c5080a7d541904c4ad42b848c76a1ce056ddac7.1619276133.git.zanussi@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>
DSA spec states that when Request Interrupt Handle and Release Interrupt
Handle command bits are set in the CMDCAP register, these device commands
must be supported by the driver.
The interrupt handle is programmed in a descriptor. When Request Interrupt
Handle is not supported, the interrupt handle is the index of the desired
entry in the MSI-X table. When the command is supported, driver must use
the command to obtain a handle to be programmed in the submitted
descriptor.
A requested handle may be revoked. After the handle is revoked, any use of
the handle will result in Invalid Interrupt Handle error.
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/161894439422.3202472.17579543737810265471.stgit@djiang5-desk3.ch.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The read-only configuration mode is defined by the DSA spec as a mode of
the device WQ configuration. When GENCAP register bit 31 is set to 0,
the device is in RO mode and group configuration and some fields of the
workqueue configuration registers are read-only and reflect the fixed
configuration of the device. Add support for RO mode. The driver will
load the values from the registers directly setup all the internally
cached data structures based on the device configuration.
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/161894438847.3202472.6317563824045432727.stgit@djiang5-desk3.ch.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Current submission path has no way to restrict the submitter from
stop submiting on shutdown path or wq disable path. This provides a way to
quiesce the submission path.
Modeling after 'struct reqeust_queue' usage of percpu_ref. One of the
abilities of per_cpu reference counting is the ability to stop new
references from being taken while awaiting outstanding references to be
dropped. On wq shutdown, we want to block any new submissions to the kernel
workqueue and quiesce before disabling. The percpu_ref allows us to block
any new submissions and wait for any current submission calls to finish
submitting to the workqueue.
A percpu_ref is embedded in each idxd_wq context to allow control for
individual wq. The wq->wq_active counter is elevated before calling
movdir64b() or enqcmds() to submit a descriptor to the wq and dropped once
the submission call completes. The function is gated by
percpu_ref_tryget_live(). On shutdown with percpu_ref_kill() called, any
new submission would be blocked from acquiring a ref and failed. Once all
references are dropped for the wq, shutdown can continue.
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/161894438293.3202472.14894701611500822232.stgit@djiang5-desk3.ch.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
There is no need to have an additional bus for the IAX device. The removal
of IAX will change user ABI as /sys/bus/iax will no longer exist.
The iax device will be moved to the dsa bus. The device id for dsa and
iax will now be combined rather than unique for each device type in order
to accommodate the iax devices. This is in preparation for fixing the
sub-driver code for idxd. There's no hardware deployment for Sapphire
Rapids platform yet, which means that users have no reason to have
developed scripts against this ABI. There is some exposure to
released versions of accel-config, but those are being fixed up and
an accel-config upgrade is reasonable to get IAX support. As far as
accel-config is concerned IAX support starts when these devices appear
under /sys/bus/dsa, and old accel-config just assumes that an empty /
missing /sys/bus/iax just means a lack of platform support.
Fixes: f25b463883 ("dmaengine: idxd: add IAX configuration support in the IDXD driver")
Suggested-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/161852988298.2203940.4529909758034944428.stgit@djiang5-desk3.ch.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The char device setup and cleanup has device lifetime issues regarding when
parts are initialized and cleaned up. The initialization of struct device is
done incorrectly. device_initialize() needs to be called on the 'struct
device' and then additional changes can be added. The ->release() function
needs to be setup via device_type before dev_set_name() to allow proper
cleanup. The change re-parents the cdev under the wq->conf_dev to get
natural reference inheritance. No known dependency on the old device path exists.
Reported-by: Jason Gunthorpe <jgg@nvidia.com>
Fixes: 42d279f913 ("dmaengine: idxd: add char driver to expose submission portal to userland")
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Link: https://lore.kernel.org/r/161852987721.2203940.1478218825576630810.stgit@djiang5-desk3.ch.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The devm managed lifetime is incompatible with 'struct device' objects that
resides in idxd context. This is one of the series that clean up the idxd
driver 'struct device' lifetime. Fix idxd->conf_dev 'struct device'
lifetime. Address issues flagged by CONFIG_DEBUG_KOBJECT_RELEASE.
Add release functions in order to free the allocated memory at the
appropriate time.
Reported-by: Jason Gunthorpe <jgg@nvidia.com>
Fixes: bfe1d56091 ("dmaengine: idxd: Init and probe for Intel data accelerators")
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/161852985319.2203940.4650791514462735368.stgit@djiang5-desk3.ch.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The devm managed lifetime is incompatible with 'struct device' objects that
resides in idxd context. This is one of the series that clean up the idxd
driver 'struct device' lifetime. Remove pcim_* management of the PCI device
and the ioremap of MMIO BAR and replace with unmanaged versions. This is
for consistency of removing all the pcim/devm based calls.
Reported-by: Jason Gunthorpe <jgg@nvidia.com>
Fixes: bfe1d56091 ("dmaengine: idxd: Init and probe for Intel data accelerators")
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Link: https://lore.kernel.org/r/161852984150.2203940.8043988289748519056.stgit@djiang5-desk3.ch.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The devm managed lifetime is incompatible with 'struct device' objects that
resides in idxd context. This is one of the series that clean up the idxd
driver 'struct device' lifetime. Remove embedding of dma_device and dma_chan
in idxd since it's not the only interface that idxd will use. The freeing of
the dma_device will be managed by the ->release() function.
Reported-by: Jason Gunthorpe <jgg@nvidia.com>
Fixes: bfe1d56091 ("dmaengine: idxd: Init and probe for Intel data accelerators")
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Link: https://lore.kernel.org/r/161852983001.2203940.14817017492384561719.stgit@djiang5-desk3.ch.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
There are calls to idxd_cmd_exec that pass a null status pointer however
a recent commit has added an assignment to *status that can end up
with a null pointer dereference. The function expects a null status
pointer sometimes as there is a later assignment to *status where
status is first null checked. Fix the issue by null checking status
before making the assignment.
Addresses-Coverity: ("Explicit null dereferenced")
Fixes: 89e3becd8f ("dmaengine: idxd: check device state before issue command")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/20210415110654.1941580-1-colin.king@canonical.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Nikhil reported the misc interrupt handler can sometimes miss handling
the command interrupt when an error interrupt happens near the same time.
Have the irq handling thread continue to process the misc interrupts until
all interrupts are processed. This is a low usage interrupt and is not
expected to handle high volume traffic. Therefore there is no concern of
this thread running for a long time.
Fixes: 0d5c10b4c8 ("dmaengine: idxd: add work queue drain support")
Reported-by: Nikhil Rao <nikhil.rao@intel.com>
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/161074755329.2183844.13295528344116907983.stgit@djiang5-desk3.ch.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Pull dmaengine updates from Vinod Koul:
"The last dmaengine updates for this year :)
This contains couple of new drivers, new device support and updates to
bunch of drivers.
New drivers/devices:
- Qualcomm ADM driver
- Qualcomm GPI driver
- Allwinner A100 DMA support
- Microchip Sama7g5 support
- Mediatek MT8516 apdma
Updates:
- more updates to idxd driver and support for IAX config
- runtime PM support for dw driver
- TI drivers"
* tag 'dmaengine-5.11-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine: (75 commits)
soc: ti: k3-ringacc: Use correct error casting in k3_ringacc_dmarings_init
dmaengine: ti: k3-udma-glue: Add support for K3 PKTDMA
dmaengine: ti: k3-udma: Initial support for K3 PKTDMA
dmaengine: ti: k3-udma: Add support for BCDMA channel TPL handling
dmaengine: ti: k3-udma: Initial support for K3 BCDMA
soc: ti: k3-ringacc: add AM64 DMA rings support.
dmaengine: ti: Add support for k3 event routers
dmaengine: ti: k3-psil: Add initial map for AM64
dmaengine: ti: k3-psil: Extend psil_endpoint_config for K3 PKTDMA
dt-bindings: dma: ti: Add document for K3 PKTDMA
dt-bindings: dma: ti: Add document for K3 BCDMA
dmaengine: dmatest: Use dmaengine_get_dma_device
dmaengine: doc: client: Update for dmaengine_get_dma_device() usage
dmaengine: Add support for per channel coherency handling
dmaengine: of-dma: Add support for optional router configuration callback
dmaengine: ti: k3-udma-glue: Configure the dma_dev for rings
dmaengine: ti: k3-udma-glue: Get the ringacc from udma_dev
dmaengine: ti: k3-udma-glue: Add function to get device pointer for DMA API
dmaengine: ti: k3-udma: Add support for second resource range from sysfw
dmaengine: ti: k3-udma: Wait for peer teardown completion if supported
...
Add support to allow configuration of Intel Analytics Accelerator (IAX) in
addition to the Intel Data Streaming Accelerator (DSA). The IAX hardware
has the same configuration interface as DSA. The main difference
is the type of operations it performs. We can support the DSA and
IAX devices on the same driver with some tweaks.
IAX has a 64B completion record that needs to be 64B aligned, as opposed to
a 32B completion record that is 32B aligned for DSA. IAX also does not
support token management.
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/160564555488.1834439.4261958859935360473.stgit@djiang5-desk3.ch.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
