diff options
| author | Linus Torvalds <torvalds@linux-foundation.org> | 2023-08-28 17:34:54 -0700 |
|---|---|---|
| committer | Linus Torvalds <torvalds@linux-foundation.org> | 2023-08-28 17:34:54 -0700 |
| commit | 542034175ca715d28522a59c1f88a58349f2765c (patch) | |
| tree | 9be4252d2769109fb0b49a653a6e754df963935b /drivers/perf/fsl_imx8_ddr_perf.c | |
| parent | bb2d9e5448ab9e86ecc79d8984dc6d3a8a9fba1e (diff) | |
| parent | e1df27213941725962f68c64349bbcc60ece0314 (diff) | |
| download | lwn-542034175ca715d28522a59c1f88a58349f2765c.tar.gz lwn-542034175ca715d28522a59c1f88a58349f2765c.zip | |
Merge tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Pull arm64 updates from Will Deacon:
"I think we have a bit less than usual on the architecture side, but
that's somewhat balanced out by a large crop of perf/PMU driver
updates and extensions to our selftests.
CPU features and system registers:
- Advertise hinted conditional branch support (FEAT_HBC) to userspace
- Avoid false positive "SANITY CHECK" warning when xCR registers
differ outside of the length field
Documentation:
- Fix macro name typo in SME documentation
Entry code:
- Unmask exceptions earlier on the system call entry path
Memory management:
- Don't bother clearing PTE_RDONLY for dirty ptes in pte_wrprotect()
and pte_modify()
Perf and PMU drivers:
- Initial support for Coresight TRBE devices on ACPI systems (the
coresight driver changes will come later)
- Fix hw_breakpoint single-stepping when called from bpf
- Fixes for DDR PMU on i.MX8MP SoC
- Add NUMA-awareness to Hisilicon PCIe PMU driver
- Fix locking dependency issue in Arm DMC620 PMU driver
- Workaround Hisilicon erratum 162001900 in the SMMUv3 PMU driver
- Add support for Arm CMN-700 r3 parts to the CMN PMU driver
- Add support for recent Arm Cortex CPU PMUs
- Update Hisilicon PMU maintainers
Selftests:
- Add a bunch of new features to the hwcap test (JSCVT, PMULL, AES,
SHA1, etc)
- Fix SSVE test to leave streaming-mode after grabbing the signal
context
- Add new test for SVE vector-length changes with SME enabled
Miscellaneous:
- Allow compiler to warn on suspicious looking system register
expressions
- Work around SDEI firmware bug by aborting any running handlers on a
kernel crash
- Fix some harmless warnings when building with W=1
- Remove some unused function declarations
- Other minor fixes and cleanup"
* tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux: (62 commits)
drivers/perf: hisi: Update HiSilicon PMU maintainers
arm_pmu: acpi: Add a representative platform device for TRBE
arm_pmu: acpi: Refactor arm_spe_acpi_register_device()
kselftest/arm64: Fix hwcaps selftest build
hw_breakpoint: fix single-stepping when using bpf_overflow_handler
arm64/sysreg: refactor deprecated strncpy
kselftest/arm64: add jscvt feature to hwcap test
kselftest/arm64: add pmull feature to hwcap test
kselftest/arm64: add AES feature check to hwcap test
kselftest/arm64: add SHA1 and related features to hwcap test
arm64: sysreg: Generate C compiler warnings on {read,write}_sysreg_s arguments
kselftest/arm64: build BTI tests in output directory
perf/imx_ddr: don't enable counter0 if none of 4 counters are used
perf/imx_ddr: speed up overflow frequency of cycle
drivers/perf: hisi: Schedule perf session according to locality
kselftest/arm64: fix a memleak in zt_regs_run()
perf/arm-dmc620: Fix dmc620_pmu_irqs_lock/cpu_hotplug_lock circular lock dependency
perf/smmuv3: Add MODULE_ALIAS for module auto loading
perf/smmuv3: Enable HiSilicon Erratum 162001900 quirk for HIP08/09
kselftest/arm64: Size sycall-abi buffers for the actual maximum VL
...
Diffstat (limited to 'drivers/perf/fsl_imx8_ddr_perf.c')
| -rw-r--r-- | drivers/perf/fsl_imx8_ddr_perf.c | 48 |
1 files changed, 31 insertions, 17 deletions
diff --git a/drivers/perf/fsl_imx8_ddr_perf.c b/drivers/perf/fsl_imx8_ddr_perf.c index 5222ba1e79d0..92611c98120f 100644 --- a/drivers/perf/fsl_imx8_ddr_perf.c +++ b/drivers/perf/fsl_imx8_ddr_perf.c @@ -10,10 +10,9 @@ #include <linux/io.h> #include <linux/module.h> #include <linux/of.h> -#include <linux/of_address.h> -#include <linux/of_device.h> #include <linux/of_irq.h> #include <linux/perf_event.h> +#include <linux/platform_device.h> #include <linux/slab.h> #define COUNTER_CNTL 0x0 @@ -28,6 +27,8 @@ #define CNTL_CLEAR_MASK 0xFFFFFFFD #define CNTL_OVER_MASK 0xFFFFFFFE +#define CNTL_CP_SHIFT 16 +#define CNTL_CP_MASK (0xFF << CNTL_CP_SHIFT) #define CNTL_CSV_SHIFT 24 #define CNTL_CSV_MASK (0xFFU << CNTL_CSV_SHIFT) @@ -35,6 +36,8 @@ #define EVENT_CYCLES_COUNTER 0 #define NUM_COUNTERS 4 +/* For removing bias if cycle counter CNTL.CP is set to 0xf0 */ +#define CYCLES_COUNTER_MASK 0x0FFFFFFF #define AXI_MASKING_REVERT 0xffff0000 /* AXI_MASKING(MSB 16bits) + AXI_ID(LSB 16bits) */ #define to_ddr_pmu(p) container_of(p, struct ddr_pmu, pmu) @@ -101,6 +104,7 @@ struct ddr_pmu { const struct fsl_ddr_devtype_data *devtype_data; int irq; int id; + int active_counter; }; static ssize_t ddr_perf_identifier_show(struct device *dev, @@ -427,6 +431,17 @@ static void ddr_perf_counter_enable(struct ddr_pmu *pmu, int config, writel(0, pmu->base + reg); val = CNTL_EN | CNTL_CLEAR; val |= FIELD_PREP(CNTL_CSV_MASK, config); + + /* + * On i.MX8MP we need to bias the cycle counter to overflow more often. + * We do this by initializing bits [23:16] of the counter value via the + * COUNTER_CTRL Counter Parameter (CP) field. + */ + if (pmu->devtype_data->quirks & DDR_CAP_AXI_ID_FILTER_ENHANCED) { + if (counter == EVENT_CYCLES_COUNTER) + val |= FIELD_PREP(CNTL_CP_MASK, 0xf0); + } + writel(val, pmu->base + reg); } else { /* Disable counter */ @@ -466,6 +481,12 @@ static void ddr_perf_event_update(struct perf_event *event) int ret; new_raw_count = ddr_perf_read_counter(pmu, counter); + /* Remove the bias applied in ddr_perf_counter_enable(). */ + if (pmu->devtype_data->quirks & DDR_CAP_AXI_ID_FILTER_ENHANCED) { + if (counter == EVENT_CYCLES_COUNTER) + new_raw_count &= CYCLES_COUNTER_MASK; + } + local64_add(new_raw_count, &event->count); /* @@ -495,6 +516,10 @@ static void ddr_perf_event_start(struct perf_event *event, int flags) ddr_perf_counter_enable(pmu, event->attr.config, counter, true); + if (!pmu->active_counter++) + ddr_perf_counter_enable(pmu, EVENT_CYCLES_ID, + EVENT_CYCLES_COUNTER, true); + hwc->state = 0; } @@ -548,6 +573,10 @@ static void ddr_perf_event_stop(struct perf_event *event, int flags) ddr_perf_counter_enable(pmu, event->attr.config, counter, false); ddr_perf_event_update(event); + if (!--pmu->active_counter) + ddr_perf_counter_enable(pmu, EVENT_CYCLES_ID, + EVENT_CYCLES_COUNTER, false); + hwc->state |= PERF_HES_STOPPED; } @@ -565,25 +594,10 @@ static void ddr_perf_event_del(struct perf_event *event, int flags) static void ddr_perf_pmu_enable(struct pmu *pmu) { - struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu); - - /* enable cycle counter if cycle is not active event list */ - if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL) - ddr_perf_counter_enable(ddr_pmu, - EVENT_CYCLES_ID, - EVENT_CYCLES_COUNTER, - true); } static void ddr_perf_pmu_disable(struct pmu *pmu) { - struct ddr_pmu *ddr_pmu = to_ddr_pmu(pmu); - - if (ddr_pmu->events[EVENT_CYCLES_COUNTER] == NULL) - ddr_perf_counter_enable(ddr_pmu, - EVENT_CYCLES_ID, - EVENT_CYCLES_COUNTER, - false); } static int ddr_perf_init(struct ddr_pmu *pmu, void __iomem *base, |