On some Qualcomm Technologies Inc. SoCs like SC7180, there
exists a hardware errata where the APSS (Application Processor
SubSystem)/CPU watchdog counter is stopped when the trace unit
power up ETM register is set (TRCPDCR.PU = 1). Since the ETMs
share the same power domain as that of respective CPU cores,
they are powered on when the CPU core is powered on. So we can
skip powering up of trace unit after checking for this errata
via new property called "qcom,skip-power-up".
Signed-off-by: Tingwei Zhang <tingwei@codeaurora.org>
Co-developed-by: Sai Prakash Ranjan <saiprakash.ranjan@codeaurora.org>
Signed-off-by: Sai Prakash Ranjan <saiprakash.ranjan@codeaurora.org>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Link: https://lore.kernel.org/r/20200716175746.3338735-4-mathieu.poirier@linaro.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The current probe() function calls a pair of cpuhp_xxx API functions to
setup CPU hotplug handling. The hotplug lock is held for the duration of
the two calls and other CPU related code using cpus_read_lock() /
cpus_read_unlock() calls.
The problem is that on error states, goto: statements bypass the
cpus_read_unlock() call. This code has increased in complexity as the
driver has developed.
This patch introduces a pair of helper functions etm4_pm_setup_cpuslocked()
and etm4_pm_clear() which correct the issues above and group the PM code a
little better.
The two functions etm4_cpu_pm_register() and etm4_cpu_pm_unregister() are
dropped as these call cpu_pm_register_notifier() / ..unregister_notifier()
dependent on CONFIG_CPU_PM - but this define is used to nop these functions
out in the pm headers - so the wrapper functions are superfluous.
Fixes: f188b5e76a ("coresight: etm4x: Save/restore state across CPU low power states")
Fixes: e9f5d63f84 ("hwtracing/coresight-etm4x: Use cpuhp_setup_state_nocalls_cpuslocked()")
Fixes: 58eb457be0 ("hwtracing/coresight-etm4x: Convert to hotplug state machine")
Signed-off-by: Mike Leach <mike.leach@linaro.org>
Cc: stable <stable@vger.kernel.org>
Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Link: https://lore.kernel.org/r/20200701160852.2782823-3-mathieu.poirier@linaro.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Differing default states set on driver init / perf init and as a result
of a sysfs reset.
The ETMv4 can be programmed to trace the entire instruction address range
without the need to use address comparator filter resources.
(Described in the ETMv4.x technical reference manual)
sysfs reset was using this method, perf and default driver init were setup
with an address range comparator for the entire address range.
The perf / driver init has been altered to use the method without needing
any comparator address hardware.
Minor adjustment to the vinst_ctrl register initialisation to ensure
correct zero initialisation.
Signed-off-by: Mike Leach <mike.leach@linaro.org>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Link: https://lore.kernel.org/r/20200518180242.7916-15-mathieu.poirier@linaro.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Some of the newly added code in the etm4x driver is inside of an #ifdef,
and some other code is outside of it, leading to a harmless warning when
CONFIG_CPU_PM is disabled:
drivers/hwtracing/coresight/coresight-etm4x.c:68:13: error: 'etm4_os_lock' defined but not used [-Werror=unused-function]
static void etm4_os_lock(struct etmv4_drvdata *drvdata)
^~~~~~~~~~~~
To avoid the warning and simplify the the #ifdef checks, use
IS_ENABLED() instead, so the compiler can drop the unused functions
without complaining.
Fixes: f188b5e76a ("coresight: etm4x: Save/restore state across CPU low power states")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
[Fixed capital 'f' in title]
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Link: https://lore.kernel.org/r/20191213223107.1484-2-mathieu.poirier@linaro.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fixes the following issues when using the ETMv4 start-stop logic.
1) Setting a start or a stop address should not automatically set the
start-stop status to 'on'. The value set by the user in 'mode' must
be respected or start instances could be missed.
2) Missing API for controlling TRCVIPCSSCTLR - start stop control by
PE comparators.
3) Default ETM configuration sets a trace all range, and correctly sets
the start-stop status bit. This was not being correctly reflected in
the 'mode' parameter.
Signed-off-by: Mike Leach <mike.leach@linaro.org>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Link: https://lore.kernel.org/r/20191104181251.26732-8-mathieu.poirier@linaro.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Some hardware will ignore bit TRCPDCR.PU which is used to signal
to hardware that power should not be removed from the trace unit.
Let's mitigate against this by conditionally saving and restoring
the trace unit state when the CPU enters low power states.
This patchset introduces a firmware property named
'arm,coresight-loses-context-with-cpu' - when this is present the
hardware state will be conditionally saved and restored.
A module parameter 'pm_save_enable' is also introduced which can
be configured to override the firmware property. This can be set
to never allow save/restore or to conditionally allow it (only for
self-hosted). The default value is determined by firmware.
We avoid saving the hardware state when self-hosted coresight isn't
in use to reduce PM latency - we can't determine this by reading the
claim tags (TRCCLAIMCLR) as these are 'trace' registers which need
power and clocking, something we can't easily provide in the PM
context. Therefore we rely on the existing drvdata->mode internal
state that is set when self-hosted coresight is used (and powered).
Signed-off-by: Andrew Murray <andrew.murray@arm.com>
Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Link: https://lore.kernel.org/r/20191104181251.26732-2-mathieu.poirier@linaro.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Synchronization is recommended before disabling the trace registers
to prevent any start or stop points being speculative at the point
of disabling the unit (section 7.3.77 of ARM IHI 0064D).
Synchronization is also recommended after programming the trace
registers to ensure all updates are committed prior to normal code
resuming (section 4.3.7 of ARM IHI 0064D).
Let's ensure these syncronization points are present in the code
and clearly commented.
Note that we could rely on the barriers in CS_LOCK and
coresight_disclaim_device_unlocked or the context switch to user
space - however coresight may be of use in the kernel.
On armv8 the mb macro is defined as dsb(sy) - Given that the etm4x is
only used on armv8 let's directly use dsb(sy) instead of mb(). This
removes some ambiguity and makes it easier to correlate the code with
the TRM.
Signed-off-by: Andrew Murray <andrew.murray@arm.com>
Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
[Fixed capital letter for "use" in title]
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Link: https://lore.kernel.org/r/20190829202842.580-11-mathieu.poirier@linaro.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
So far we have reused the name of the "platform" device for
the CoreSight device. But this is not very intuitive when
we move to ACPI. Also, the ACPI device names have ":" in them
(e.g, ARMHC97C:01), which the perf tool doesn't like very much.
This patch introduces a generic naming scheme, givin more intuitive
names for the devices that appear on the CoreSight bus.
The names follow the pattern "prefix" followed by "index" (e.g, etm5).
We maintain a list of allocated devices per "prefix" to make sure
we don't allocate a new name when it is reprobed (e.g, due to
unsatisifed device dependencies). So, we maintain the list
of "fwnodes" of the parent devices to allocate a consistent name.
All devices except the ETMs get an index allocated in the order
of probing. ETMs get an index based on the CPU they are attached to.
TMC devices are named using "tmc_etf", "tmc_etb", and "tmc_etr"
prefixes depending on the configuration of the device.
The replicators and funnels are not classified as dynamic/static
anymore. One could easily figure that out by checking the presence
of "mgmt" registers under sysfs.
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We are about to introduce methods to clean up the platform data
as we switch to tracking the device reference from "name" to "fwnode
handles" for device connections. This requires us to drop the fwnode
handle references when the data is no longer required - i.e, when
the device probe fails or the device gets unregistered.
In order to consolidate the invocation of the cleanup, we delay the
platform probing to the very last minute, possibly before invoking
the coresight_register. Then, we leave the coresight core code to
do the clean up. i.e, if the coresight_register fails, it takes
care of freeing the data. Otherwise, coresight_unregister will
do the necessary operations.
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We are about to use a name independent of the parent AMBA device
name. As such, there is no need to have it in the platform description.
Let us move this to coresight description instead.
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
CPU field is only used by ETMs and there is a separate API
for fetching the same. So, let us use that instead of using
the common platform probing helper. Also, remove it from the
platform_data.
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
So far we have hard coded the DT platform parsing code in
every driver. Introduce generic helper to parse the information
provided by the firmware in a platform agnostic manner, in preparation
for the ACPI support.
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Set the proper bit in the configuration register when contextID tracing
has been requested by user space. That way PE_CONTEXT elements are
generated by the tracers when a process is installed on a CPU.
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Tested-by: Leo Yan <leo.yan@linaro.org>
Tested-by: Robert Walker <robert.walker@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Pull ARM updates from Russell King:
- An improvement from Ard Biesheuvel, who noted that the identity map
setup was taking a long time due to flush_cache_louis().
- Update a comment about dma_ops from Wolfram Sang.
- Remove use of "-p" with ld, where this flag has been a no-op since
2004.
- Remove the printing of the virtual memory layout, which is no longer
useful since we hide pointers.
- Correct SCU help text.
- Remove legacy TWD registration method.
- Add pgprot_device() implementation for mapping PCI sysfs resource
files.
- Initialise PFN limits earlier for kmemleak.
- Fix argument count to match macro definition (affects clang builds)
- Use unified assembler language almost everywhere for clang, and other
clang improvements (from Stefan Agner, Nathan Chancellor).
- Support security extension for noMMU and other noMMU cleanups (from
Vladimir Murzin).
- Remove unnecessary SMP bringup code (which was incorrectly copy'n'
pasted from the ARM platform implementations) and remove it from the
arch code to discourge further copys of it appearing.
- Add Cortex A9 erratum preventing kexec working on some SoCs.
- AMBA bus identification updates from Mike Leach.
- More use of raw spinlocks to avoid -RT kernel issues (from Yang Shi
and Sebastian Andrzej Siewior).
- MCPM hyp/svc mode mismatch fixes from Marek Szyprowski.
* tag 'for-linus' of git://git.armlinux.org.uk/~rmk/linux-arm: (32 commits)
ARM: 8849/1: NOMMU: Fix encodings for PMSAv8's PRBAR4/PRLAR4
ARM: 8848/1: virt: Align GIC version check with arm64 counterpart
ARM: 8847/1: pm: fix HYP/SVC mode mismatch when MCPM is used
ARM: 8845/1: use unified assembler in c files
ARM: 8844/1: use unified assembler in assembly files
ARM: 8843/1: use unified assembler in headers
ARM: 8841/1: use unified assembler in macros
ARM: 8840/1: use a raw_spinlock_t in unwind
ARM: 8839/1: kprobe: make patch_lock a raw_spinlock_t
ARM: 8837/1: coresight: etmv4: Update ID register table to add UCI support
ARM: 8836/1: drivers: amba: Update component matching to use the CoreSight UCI values.
ARM: 8838/1: drivers: amba: Updates to component identification for driver matching.
ARM: 8833/1: Ensure that NEON code always compiles with Clang
ARM: avoid Cortex-A9 livelock on tight dmb loops
ARM: smp: remove arch-provided "pen_release"
ARM: actions: remove boot_lock and pen_release
ARM: oxnas: remove CPU hotplug implementation
ARM: qcom: remove unnecessary boot_lock
ARM: 8832/1: NOMMU: Limit visibility for CONFIG_FLASH_{MEM_BASE,SIZE}
ARM: 8831/1: NOMMU: pmsa-v8: remove unneeded semicolon
...
SDM845 has ETMv4.2 and can use the existing etm4x driver.
But the current etm driver checks only for ETMv4.0 and
errors out for other etm4x versions. This patch adds this
missing support to enable SoC's with ETMv4x to use same
driver by checking only the ETM architecture major version
number.
Without this change, we get below error during etm probe:
/ # dmesg | grep etm
[ 6.660093] coresight-etm4x: probe of 7040000.etm failed with error -22
[ 6.666902] coresight-etm4x: probe of 7140000.etm failed with error -22
[ 6.673708] coresight-etm4x: probe of 7240000.etm failed with error -22
[ 6.680511] coresight-etm4x: probe of 7340000.etm failed with error -22
[ 6.687313] coresight-etm4x: probe of 7440000.etm failed with error -22
[ 6.694113] coresight-etm4x: probe of 7540000.etm failed with error -22
[ 6.700914] coresight-etm4x: probe of 7640000.etm failed with error -22
[ 6.707717] coresight-etm4x: probe of 7740000.etm failed with error -22
With this change, etm probe is successful:
/ # dmesg | grep etm
[ 6.659198] coresight-etm4x 7040000.etm: CPU0: ETM v4.2 initialized
[ 6.665848] coresight-etm4x 7140000.etm: CPU1: ETM v4.2 initialized
[ 6.672493] coresight-etm4x 7240000.etm: CPU2: ETM v4.2 initialized
[ 6.679129] coresight-etm4x 7340000.etm: CPU3: ETM v4.2 initialized
[ 6.685770] coresight-etm4x 7440000.etm: CPU4: ETM v4.2 initialized
[ 6.692403] coresight-etm4x 7540000.etm: CPU5: ETM v4.2 initialized
[ 6.699024] coresight-etm4x 7640000.etm: CPU6: ETM v4.2 initialized
[ 6.705646] coresight-etm4x 7740000.etm: CPU7: ETM v4.2 initialized
Signed-off-by: Sai Prakash Ranjan <saiprakash.ranjan@codeaurora.org>
Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add support for handling errors in enabling the component.
The ETM is enabled via cross call to owner CPU. Make
necessary changes to report the error back from the cross
call.
Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
For non-VHE systems host kernel runs at EL1 and jumps to EL2 whenever
hypervisor code should be executed. In this case ETM4x driver must
restrict configuration to EL1 when it setups kernel tracing.
However, there is no separate hypervisor privilege level when VHE
is enabled, the host kernel runs at EL2.
This patch fixes configuration of TRCACATRn register for VHE systems
so that ETM_EXLEVEL_NS_HYP bit is used instead of ETM_EXLEVEL_NS_OS
to on/off kernel tracing. At the same time, it moves common code
to new helper.
Signed-off-by: Tomasz Nowicki <tnowicki@caviumnetworks.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Convert component enable/disable messages from dev_info to dev_dbg.
When used with perf, the components in the paths are enabled/disabled
during each schedule of the run, which can flood the dmesg with these
messages. Moreover, they are only useful for debug purposes. So,
convert such messages to dev_dbg() which can be turned on as
needed.
Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
amba_id are not supposed to change at runtime. All functions
working with const amba_id. So mark the non-const structs as const.
Signed-off-by: Arvind Yadav <arvind.yadav.cs@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Adds handling to program the return stack option into ETMv4 hardware if
specified in the perf command line.
If option is not supported by the hardware then it will be ignored.
This allows capture to move between core/ETM combinations that have the
hardware support to those that do not.
Signed-off-by: Mike Leach <mike.leach@linaro.org>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When using the ETM4x tracers from the perf interface two trace options are
available: cycle accurate and timestamp.
Enabling the timestamp feature is done by setting TRCCONFIGR.TS (bit 11).
The position of the timestamp bit in that register coincidentally happens
to be the same as what was chosen to enable timestamping from the 'mode'
sysFS entry. The code does the right thing but the semantic is wrong.
This patch sets TRCCONFIGR.TS explicitly, as it is done from the sysFS
interface. That way timestamps are set the same way from both perf and
sysFS and there is no misunderstanding as to what is intended.
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Using perf record 'cyclacc' option in cs_etm event was not setting up cycle
accurate trace correctly.
Corrects bit set in TRCCONFIGR to enable cycle accurate trace.
Programs TRCCCCTLR with a valid threshold value as required by ETMv4 spec.
Signed-off-by: Mike Leach <mike.leach@linaro.org>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When the state names got added a script was used to add the extra argument
to the calls. The script basically converted the state constant to a
string, but the cleanup to convert these strings into meaningful ones did
not happen.
Replace all the useless strings with 'subsys/xxx/yyy:state' strings which
are used in all the other places already.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Siewior <bigeasy@linutronix.de>
Link: http://lkml.kernel.org/r/20161221192112.085444152@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
With this patch we add start/stop filtering as specified on
the perf cmd line. When the IP matches the start address
trace generation gets triggered. The stop condition is
achieved when the IP matches the stop address.
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch adds the capability to specify address ranges from
the perf cmd line using the --filter option. If the IP
falls within the range(s) program flow traces are generated.
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The include/exclude function of a tracer is applicable to address
range and start/stop filters. To avoid duplication and reuse code
moving the include/exclude configuration to a function of its own.
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
