From patchwork Tue Apr 12 13:32:30 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Alexandru Elisei X-Patchwork-Id: 12810843 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id D39D3C433F5 for ; Tue, 12 Apr 2022 13:36:16 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20210309; h=Sender: Content-Transfer-Encoding:Content-Type:List-Subscribe:List-Help:List-Post: List-Archive:List-Unsubscribe:List-Id:MIME-Version:References:In-Reply-To: Message-Id:Date:Subject:To:From:Reply-To:Cc:Content-ID:Content-Description: Resent-Date:Resent-From:Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID: List-Owner; bh=MtEYbF23lgW+hPcZuftI0km8YRiXrLkpVdBh2ZfxecA=; b=vHJccaYe3NVsVp 80L1+Yn1Qui3isglDBAYhoc7WFGttCuSaZS4JwSEqUVZDyJRdiksnqmlQdKwI/KXPzYLtnFb8zRLT Iy71+7M5qWoJBSIR8Jm63+7+Mc84uz3IEUAmRJ0YoF70RxHfY03TYBzNdB70qZhaGhJfcM4qITVW3 xZ6Z5OxQBUdriaKod/KUgWM37cT1GxgAq+8lcGCTVOumx3cBnt+IQpxKJgJ6blJrg5mnE1933h1Ss xWLxVCUrwBmnGhlpFea056B++2Lum0oZs+PTLs/7b4IVP1+7Y1WB7DnmxghXNJxGEzuc/gQTLYPdC 6q0am6TjSmdI/7GaJCZQ==; Received: from localhost ([::1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.94.2 #2 (Red Hat Linux)) id 1neGff-00Ed0z-5S; Tue, 12 Apr 2022 13:35:03 +0000 Received: from foss.arm.com ([217.140.110.172]) by bombadil.infradead.org with esmtp (Exim 4.94.2 #2 (Red Hat Linux)) id 1neGda-00Ebiw-8P for linux-arm-kernel@lists.infradead.org; Tue, 12 Apr 2022 13:32:56 +0000 Received: from usa-sjc-imap-foss1.foss.arm.com (unknown [10.121.207.14]) by usa-sjc-mx-foss1.foss.arm.com (Postfix) with ESMTP id DAE5D1516; Tue, 12 Apr 2022 06:32:53 -0700 (PDT) Received: from e121798.arm.com (unknown [10.57.11.98]) by usa-sjc-imap-foss1.foss.arm.com (Postfix) with ESMTPA id 747383F70D; Tue, 12 Apr 2022 06:32:52 -0700 (PDT) From: Alexandru Elisei To: will@kernel.org, julien.thierry.kdev@gmail.com, linux-arm-kernel@lists.infradead.org, kvmarm@lists.cs.columbia.edu, maz@kernel.org, james.morse@arm.com, suzuki.poulose@arm.com, mark.rutland@arm.com, andre.przywara@arm.com Subject: [PATCH v3 kvmtool 10/11] arm64: Add support for KVM_ARM_VCPU_PMU_V3_SET_PMU Date: Tue, 12 Apr 2022 14:32:30 +0100 Message-Id: <20220412133231.35355-11-alexandru.elisei@arm.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20220412133231.35355-1-alexandru.elisei@arm.com> References: <20220412133231.35355-1-alexandru.elisei@arm.com> MIME-Version: 1.0 X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20220412_063254_440225_9D034675 X-CRM114-Status: GOOD ( 25.10 ) X-BeenThere: linux-arm-kernel@lists.infradead.org X-Mailman-Version: 2.1.34 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: "linux-arm-kernel" Errors-To: linux-arm-kernel-bounces+linux-arm-kernel=archiver.kernel.org@lists.infradead.org The KVM_ARM_VCPU_PMU_V3_CTRL(KVM_ARM_VCPU_PMU_V3_SET_PMU) VCPU ioctl is used to assign a physical PMU to the events that KVM creates when emulating the PMU for that VCPU. This is useful on heterogeneous systems, when there is more than one hardware PMU present. All VCPUs must have the same PMU assigned. The assumption that is made in the implementation is that the user will pin the kvmtool process on a set of CPUs that share the same PMU. This allows kvmtool to set the same PMU for all VCPUs from the main thread, instead of in the individual VCPU threads. If a VCPU thread migrates to a CPU which has a different a PMU than the CPU on which the main thread was executing when the PMU was set, the KVM_RUN ioctl will fail with kvm_run.exit_reason set to KVM_EXIT_FAIL_ENTRY, and kvm_run.fail_entry will be populated with the physical CPU ID on which the VCPU tried to execute. Signed-off-by: Alexandru Elisei --- arm/aarch64/include/kvm/kvm-config-arch.h | 4 +- arm/aarch64/pmu.c | 147 +++++++++++++++++++++- 2 files changed, 148 insertions(+), 3 deletions(-) diff --git a/arm/aarch64/include/kvm/kvm-config-arch.h b/arm/aarch64/include/kvm/kvm-config-arch.h index 3d39cd6d1f4e..b08ee60ee1c4 100644 --- a/arm/aarch64/include/kvm/kvm-config-arch.h +++ b/arm/aarch64/include/kvm/kvm-config-arch.h @@ -5,7 +5,9 @@ OPT_BOOLEAN('\0', "aarch32", &(cfg)->aarch32_guest, \ "Run AArch32 guest"), \ OPT_BOOLEAN('\0', "pmu", &(cfg)->has_pmuv3, \ - "Create PMUv3 device"), \ + "Create PMUv3 device. The emulated PMU will be" \ + " set to the PMU associated with the" \ + " main thread"), \ OPT_BOOLEAN('\0', "disable-mte", &(cfg)->mte_disabled, \ "Disable Memory Tagging Extension"), \ OPT_U64('\0', "kaslr-seed", &(cfg)->kaslr_seed, \ diff --git a/arm/aarch64/pmu.c b/arm/aarch64/pmu.c index ac5b7bcd6ca9..869a150cd6d3 100644 --- a/arm/aarch64/pmu.c +++ b/arm/aarch64/pmu.c @@ -1,3 +1,9 @@ +#include +#include + +#include "linux/cpumask.h" +#include "linux/err.h" + #include "kvm/fdt.h" #include "kvm/kvm.h" #include "kvm/kvm-cpu.h" @@ -7,6 +13,17 @@ #include "asm/pmu.h" +static bool pmu_has_attr(struct kvm_cpu *vcpu, u64 attr) +{ + struct kvm_device_attr pmu_attr = { + .group = KVM_ARM_VCPU_PMU_V3_CTRL, + .attr = attr, + }; + int ret = ioctl(vcpu->vcpu_fd, KVM_HAS_DEVICE_ATTR, &pmu_attr); + + return ret == 0; +} + static void set_pmu_attr(struct kvm_cpu *vcpu, void *addr, u64 attr) { struct kvm_device_attr pmu_attr = { @@ -16,8 +33,7 @@ static void set_pmu_attr(struct kvm_cpu *vcpu, void *addr, u64 attr) }; int ret; - ret = ioctl(vcpu->vcpu_fd, KVM_HAS_DEVICE_ATTR, &pmu_attr); - if (!ret) { + if (pmu_has_attr(vcpu, attr)) { ret = ioctl(vcpu->vcpu_fd, KVM_SET_DEVICE_ATTR, &pmu_attr); if (ret) die_perror("PMU KVM_SET_DEVICE_ATTR"); @@ -26,11 +42,124 @@ static void set_pmu_attr(struct kvm_cpu *vcpu, void *addr, u64 attr) } } +#define SYS_EVENT_SOURCE "/sys/bus/event_source/devices/" +/* + * int is 32 bits and INT_MAX translates in decimal to 2 * 10^9. + * Make room for newline and NUL. + */ +#define PMU_ID_MAXLEN 12 + +/* + * In the case of homogeneous systems, there only one hardware PMU, and all + * VCPUs will use the same PMU, regardless of where the attribute gets set. + * + * For heterogeneous systems, the assumption is that the user has pinned the VM + * (via taskset or similar) to a set of CPUs that share the same hardware PMU. + * This simplifies things for kvmtool, as correctness is not affected by setting + * the PMU for each VCPU from the main thread, instead of setting it from each + * individual VCPU thread. + */ +static int find_pmu(void) +{ + char buf[PMU_ID_MAXLEN]; + struct dirent *dirent; + char *cpulist, *path; + int pmu_id = -ENXIO; + unsigned long val; + cpumask_t cpumask; + ssize_t fd_sz; + int this_cpu; + int fd, ret; + DIR *dir; + + memset(buf, 0, PMU_ID_MAXLEN); + + this_cpu = sched_getcpu(); + if (this_cpu < 0) + return -errno; + + cpulist = calloc(1, PAGE_SIZE); + if (!cpulist) + die_perror("calloc"); + + path = calloc(1, PAGE_SIZE); + if (!path) + die_perror("calloc"); + + dir = opendir(SYS_EVENT_SOURCE); + if (!dir) { + pmu_id = -errno; + goto out_free; + } + + /* Make the compiler happy by copying the NUL terminating byte. */ + strncpy(path, SYS_EVENT_SOURCE, strlen(SYS_EVENT_SOURCE) + 1); + + while ((dirent = readdir(dir))) { + if (dirent->d_type != DT_LNK) + continue; + + strcat(path, dirent->d_name); + strcat(path, "/cpus"); + fd = open(path, O_RDONLY); + if (fd < 0) + goto next_dir; + + fd_sz = read_file(fd, cpulist, PAGE_SIZE); + if (fd_sz < 0) { + pmu_id = -errno; + goto out_free; + } + close(fd); + + ret = cpulist_parse(cpulist, &cpumask); + if (ret) { + pmu_id = ret; + goto out_free; + } + + if (!cpumask_test_cpu(this_cpu, &cpumask)) + goto next_dir; + + strcpy(&path[strlen(path) - 4], "type"); + fd = open(path, O_RDONLY); + if (fd < 0) + goto next_dir; + + fd_sz = read_file(fd, buf, PMU_ID_MAXLEN - 1); + if (fd_sz < 0) { + pmu_id = -errno; + goto out_free; + } + close(fd); + + val = strtoul(buf, NULL, 10); + if (val > INT_MAX) { + pmu_id = -EOVERFLOW; + goto out_free; + } + pmu_id = (int)val; + pr_debug("Using PMU %s (id %d)", dirent->d_name, pmu_id); + break; + +next_dir: + /* Reset path. */ + memset(&path[strlen(SYS_EVENT_SOURCE)], '\0', + strlen(path) - strlen(SYS_EVENT_SOURCE)); + } + +out_free: + free(path); + free(cpulist); + return pmu_id; +} + void pmu__generate_fdt_nodes(void *fdt, struct kvm *kvm) { const char compatible[] = "arm,armv8-pmuv3"; int irq = KVM_ARM_PMUv3_PPI; struct kvm_cpu *vcpu; + int pmu_id = -ENXIO; int i; u32 cpu_mask = (((1 << kvm->nrcpus) - 1) << GIC_FDT_IRQ_PPI_CPU_SHIFT) \ @@ -44,9 +173,23 @@ void pmu__generate_fdt_nodes(void *fdt, struct kvm *kvm) if (!kvm->cfg.arch.has_pmuv3) return; + if (pmu_has_attr(kvm->cpus[0], KVM_ARM_VCPU_PMU_V3_SET_PMU)) { + pmu_id = find_pmu(); + if (pmu_id < 0) { + pr_debug("Failed to find a PMU (errno: %d), " + "PMU events might not work", -pmu_id); + } + } + for (i = 0; i < kvm->nrcpus; i++) { vcpu = kvm->cpus[i]; set_pmu_attr(vcpu, &irq, KVM_ARM_VCPU_PMU_V3_IRQ); + /* + * PMU IDs 0-5 are reserved; a positive value means a PMU was + * found. + */ + if (pmu_id > 0) + set_pmu_attr(vcpu, &pmu_id, KVM_ARM_VCPU_PMU_V3_SET_PMU); set_pmu_attr(vcpu, NULL, KVM_ARM_VCPU_PMU_V3_INIT); }