From patchwork Wed Jun 17 09:37:25 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Stefan Hajnoczi X-Patchwork-Id: 11609499 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 418DD90 for ; Wed, 17 Jun 2020 09:37:52 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id 23FEC208B3 for ; Wed, 17 Jun 2020 09:37:52 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (1024-bit key) header.d=redhat.com header.i=@redhat.com header.b="SuR/U9qP" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1725964AbgFQJhv (ORCPT ); Wed, 17 Jun 2020 05:37:51 -0400 Received: from us-smtp-delivery-1.mimecast.com ([207.211.31.120]:30781 "EHLO us-smtp-1.mimecast.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1726798AbgFQJhu (ORCPT ); Wed, 17 Jun 2020 05:37:50 -0400 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1592386668; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=7MAdMlImgumCZT3LfGm29F+L+iTUaSnuaS5P4ZPaAz4=; b=SuR/U9qPNrwF5QSDM5jfQe+nDE7W+QCFlpvAsqR1FJHdia/AhR/hx/KWWNlyhODi/AUwOs /PSUwd6xeQEMKgNN5Dovm+rEbNpNosy5jxOVriby5wg5VK7yAQH6I5ZFGQqtMFzxb/sgix Fbkjh8GJqtP1gRnse/frykS2v04pJe8= Received: from mimecast-mx01.redhat.com (mimecast-mx01.redhat.com [209.132.183.4]) (Using TLS) by relay.mimecast.com with ESMTP id us-mta-481-oikxqi3UO3CV5AoofoRoFg-1; Wed, 17 Jun 2020 05:37:45 -0400 X-MC-Unique: oikxqi3UO3CV5AoofoRoFg-1 Received: from smtp.corp.redhat.com (int-mx06.intmail.prod.int.phx2.redhat.com [10.5.11.16]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mimecast-mx01.redhat.com (Postfix) with ESMTPS id E3D6480B71D; Wed, 17 Jun 2020 09:37:43 +0000 (UTC) Received: from localhost (ovpn-114-151.ams2.redhat.com [10.36.114.151]) by smtp.corp.redhat.com (Postfix) with ESMTP id 7BA5D5C1D2; Wed, 17 Jun 2020 09:37:43 +0000 (UTC) From: Stefan Hajnoczi To: linux-kernel@vger.kernel.org Cc: Marcelo Tosatti , linux-pci@vger.kernel.org, Thomas Gleixner , Bjorn Helgaas , "Michael S. Tsirkin" , Stefan Hajnoczi Subject: [RFC 2/2] genirq/matrix: take NUMA into account for managed IRQs Date: Wed, 17 Jun 2020 10:37:25 +0100 Message-Id: <20200617093725.1725569-3-stefanha@redhat.com> In-Reply-To: <20200617093725.1725569-1-stefanha@redhat.com> References: <20200617093725.1725569-1-stefanha@redhat.com> MIME-Version: 1.0 X-Scanned-By: MIMEDefang 2.79 on 10.5.11.16 Sender: linux-pci-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-pci@vger.kernel.org Select CPUs from the IRQ's NUMA node in preference over other CPUs. This ensures that managed IRQs are assigned to the same NUMA node as the device. Signed-off-by: Stefan Hajnoczi --- include/linux/irq.h | 2 +- arch/x86/kernel/apic/vector.c | 3 ++- kernel/irq/matrix.c | 16 ++++++++++++---- 3 files changed, 15 insertions(+), 6 deletions(-) diff --git a/include/linux/irq.h b/include/linux/irq.h index 8d5bc2c237d7..bdc3faa3c280 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -1202,7 +1202,7 @@ void irq_matrix_assign_system(struct irq_matrix *m, unsigned int bit, bool repla int irq_matrix_reserve_managed(struct irq_matrix *m, const struct cpumask *msk); void irq_matrix_remove_managed(struct irq_matrix *m, const struct cpumask *msk); int irq_matrix_alloc_managed(struct irq_matrix *m, const struct cpumask *msk, - unsigned int *mapped_cpu); + int node, unsigned int *mapped_cpu); void irq_matrix_reserve(struct irq_matrix *m); void irq_matrix_remove_reserved(struct irq_matrix *m); int irq_matrix_alloc(struct irq_matrix *m, const struct cpumask *msk, diff --git a/arch/x86/kernel/apic/vector.c b/arch/x86/kernel/apic/vector.c index 67768e54438b..8eb10b0d981d 100644 --- a/arch/x86/kernel/apic/vector.c +++ b/arch/x86/kernel/apic/vector.c @@ -309,6 +309,7 @@ assign_managed_vector(struct irq_data *irqd, const struct cpumask *dest) { const struct cpumask *affmsk = irq_data_get_affinity_mask(irqd); struct apic_chip_data *apicd = apic_chip_data(irqd); + int node = irq_data_get_node(irqd); int vector, cpu; cpumask_and(vector_searchmask, dest, affmsk); @@ -317,7 +318,7 @@ assign_managed_vector(struct irq_data *irqd, const struct cpumask *dest) if (apicd->vector && cpumask_test_cpu(apicd->cpu, vector_searchmask)) return 0; vector = irq_matrix_alloc_managed(vector_matrix, vector_searchmask, - &cpu); + node, &cpu); trace_vector_alloc_managed(irqd->irq, vector, vector); if (vector < 0) return vector; diff --git a/kernel/irq/matrix.c b/kernel/irq/matrix.c index 30cc217b8631..ee35b6172b64 100644 --- a/kernel/irq/matrix.c +++ b/kernel/irq/matrix.c @@ -148,7 +148,8 @@ static unsigned int matrix_find_best_cpu(struct irq_matrix *m, /* Find the best CPU which has the lowest number of managed IRQs allocated */ static unsigned int matrix_find_best_cpu_managed(struct irq_matrix *m, - const struct cpumask *msk) + const struct cpumask *msk, + int node) { unsigned int cpu, best_cpu, allocated = UINT_MAX; struct cpumap *cm; @@ -156,6 +157,9 @@ static unsigned int matrix_find_best_cpu_managed(struct irq_matrix *m, best_cpu = UINT_MAX; for_each_cpu(cpu, msk) { + if (node != NUMA_NO_NODE && cpu_to_node(cpu) != node) + continue; + cm = per_cpu_ptr(m->maps, cpu); if (!cm->online || cm->managed_allocated > allocated) @@ -280,10 +284,12 @@ void irq_matrix_remove_managed(struct irq_matrix *m, const struct cpumask *msk) /** * irq_matrix_alloc_managed - Allocate a managed interrupt in a CPU map * @m: Matrix pointer - * @cpu: On which CPU the interrupt should be allocated + * @mask: The mask of CPUs on which the interrupt can be allocated + * @node: The preferred NUMA node + * @mapped_cpu: The resulting CPU on which the interrupt should be allocated */ int irq_matrix_alloc_managed(struct irq_matrix *m, const struct cpumask *msk, - unsigned int *mapped_cpu) + int node, unsigned int *mapped_cpu) { unsigned int bit, cpu, end = m->alloc_end; struct cpumap *cm; @@ -291,7 +297,9 @@ int irq_matrix_alloc_managed(struct irq_matrix *m, const struct cpumask *msk, if (cpumask_empty(msk)) return -EINVAL; - cpu = matrix_find_best_cpu_managed(m, msk); + cpu = matrix_find_best_cpu_managed(m, msk, node); + if (cpu == UINT_MAX) + cpu = matrix_find_best_cpu_managed(m, msk, NUMA_NO_NODE); if (cpu == UINT_MAX) return -ENOSPC;