From f146c7bc85a51f95c05e0f8173d484eb301ece78 Mon Sep 17 00:00:00 2001 From: Danilo Krummrich Date: Thu, 13 Aug 2026 18:52:04 +0200 Subject: PCI: Add pci_irq_type() to query the allocated interrupt type Add a helper that returns PCI_IRQ_MSIX, PCI_IRQ_MSI, or PCI_IRQ_INTX based on the interrupt type the PCI core selected after pci_alloc_irq_vectors(). Several drivers already open-code this check against pdev->msix_enabled and pdev->msi_enabled, or even open code this helper [1]. A common helper avoids the duplication and keeps drivers from accessing the bitfield directly (see also [2]). Acked-by: Bjorn Helgaas Tested-by: John Hubbard Link: https://elixir.bootlin.com/linux/v7.1/source/drivers/net/ethernet/aquantia/atlantic/aq_pci_func.c#L196 [1] Inspired-by: John Hubbard Link: https://lore.kernel.org/all/DKKG2QM3YJYB.Z2H2B2UXJ75N@kernel.org/ [2] Reviewed-by: Gary Guo Link: https://patch.msgid.link/20260813165234.620555-5-dakr@kernel.org [ Add missing pci_irq_type() stub for CONFIG_PCI=n. ] Signed-off-by: Danilo Krummrich --- tools/perf/util/affinity.c | 103 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 103 insertions(+) create mode 100644 tools/perf/util/affinity.c (limited to 'tools/perf/util/affinity.c') diff --git a/tools/perf/util/affinity.c b/tools/perf/util/affinity.c new file mode 100644 index 000000000..6c64b5f69 --- /dev/null +++ b/tools/perf/util/affinity.c @@ -0,0 +1,103 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Manage affinity to optimize IPIs inside the kernel perf API. */ +#define _GNU_SOURCE 1 +#include +#include +#include +#include +#include +#include "perf.h" +#include "cpumap.h" +#include "affinity.h" + +static int get_cpu_set_size(void) +{ + int sz = cpu__max_cpu().cpu + 8 - 1; + /* + * sched_getaffinity doesn't like masks smaller than the kernel. + * Hopefully that's big enough. + */ + if (sz < 4096) + sz = 4096; + return sz / 8; +} + +int affinity__setup(struct affinity *a) +{ + int cpu_set_size = get_cpu_set_size(); + + a->orig_cpus = bitmap_zalloc(cpu_set_size * 8); + if (!a->orig_cpus) + return -1; + sched_getaffinity(0, cpu_set_size, (cpu_set_t *)a->orig_cpus); + a->sched_cpus = bitmap_zalloc(cpu_set_size * 8); + if (!a->sched_cpus) { + zfree(&a->orig_cpus); + return -1; + } + bitmap_zero((unsigned long *)a->sched_cpus, cpu_set_size); + a->changed = false; + return 0; +} + +/* + * perf_event_open does an IPI internally to the target CPU. + * It is more efficient to change perf's affinity to the target + * CPU and then set up all events on that CPU, so we amortize + * CPU communication. + */ +void affinity__set(struct affinity *a, int cpu) +{ + int cpu_set_size = get_cpu_set_size(); + + /* + * Return: + * - if cpu is -1 + * - restrict out of bound access to sched_cpus + */ + if (cpu == -1 || ((cpu >= (cpu_set_size * 8)))) + return; + + a->changed = true; + __set_bit(cpu, a->sched_cpus); + /* + * We ignore errors because affinity is just an optimization. + * This could happen for example with isolated CPUs or cpusets. + * In this case the IPIs inside the kernel's perf API still work. + */ + sched_setaffinity(0, cpu_set_size, (cpu_set_t *)a->sched_cpus); + __clear_bit(cpu, a->sched_cpus); +} + +static void __affinity__cleanup(struct affinity *a) +{ + int cpu_set_size = get_cpu_set_size(); + + if (a->changed) + sched_setaffinity(0, cpu_set_size, (cpu_set_t *)a->orig_cpus); + zfree(&a->sched_cpus); + zfree(&a->orig_cpus); +} + +void affinity__cleanup(struct affinity *a) +{ + if (a != NULL) + __affinity__cleanup(a); +} + +void cpu_map__set_affinity(const struct perf_cpu_map *cpumap) +{ + int cpu_set_size = get_cpu_set_size(); + unsigned long *cpuset = bitmap_zalloc(cpu_set_size * 8); + struct perf_cpu cpu; + unsigned int idx; + + if (!cpuset) + return; + + perf_cpu_map__for_each_cpu_skip_any(cpu, idx, cpumap) + __set_bit(cpu.cpu, cpuset); + + sched_setaffinity(0, cpu_set_size, (cpu_set_t *)cpuset); + zfree(&cpuset); +} -- cgit v1.3.1