From patchwork Wed May 27 18:27:37 2015 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Duc Dang X-Patchwork-Id: 6492351 X-Patchwork-Delegate: bhelgaas@google.com Return-Path: X-Original-To: patchwork-linux-pci@patchwork.kernel.org Delivered-To: patchwork-parsemail@patchwork1.web.kernel.org Received: from mail.kernel.org (mail.kernel.org [198.145.29.136]) by patchwork1.web.kernel.org (Postfix) with ESMTP id 299889F399 for ; Wed, 27 May 2015 18:29:04 +0000 (UTC) Received: from mail.kernel.org (localhost [127.0.0.1]) by mail.kernel.org (Postfix) with ESMTP id 59B80206A9 for ; Wed, 27 May 2015 18:29:02 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id DBA2A20644 for ; Wed, 27 May 2015 18:28:59 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753229AbbE0S26 (ORCPT ); Wed, 27 May 2015 14:28:58 -0400 Received: from exprod5og122.obsmtp.com ([64.18.0.192]:56280 "EHLO mail-pa0-f50.google.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1752207AbbE0S2W (ORCPT ); Wed, 27 May 2015 14:28:22 -0400 Received: from mail-pa0-f50.google.com ([209.85.220.50]) (using TLSv1) by exprod5ob122.postini.com ([64.18.4.12]) with SMTP ID DSNKVWYMxkMeV++U7jlQgm9oqu09z+bq2ife@postini.com; Wed, 27 May 2015 11:28:22 PDT Received: by paza2 with SMTP id a2so3804627paz.3 for ; Wed, 27 May 2015 11:28:21 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:in-reply-to:references; bh=NfsFyscYXyANOm3zTy++KwvcKeMtiZSfaYk3D3f4rr8=; b=NfFH+onBu3ebOOywmLKIgm+lRNfOTemuB3AH3R0094IFfsub68WtGO3rJ9Ji4/ZaQh QrnaIO+rGm4LtTOnIImcyXU9zstQsG0tkP2muntI93WyjIXWsTZSJLL3NsLIpdIo9pXq ngkX+W0ROCHCDeBlTPfjyE+sara/Zr7ejrNrFvaT0wAKWf3UmkH3zM70ybuvZIVPUb1l DE9HeowOVVOjYcXWrKFJZ82wNdaylaKwDnFfrcjh9krszcJCXXuEyf2dXZ5OQEUKitHu 6HJpi+34pY59EDOXXMjB8Fwf7mtl17eAZU5nQoL9ROPX4RTwLVXYv/4EJTa5UazABqP0 tOAg== X-Gm-Message-State: ALoCoQmmzVjuOY4yj0tByljtHpes+fSp8dOUCz4KSCVzWpZCIdJQ25+Uk0JRr2vWHMFgsMpaI0HhCj/uOC02uBothi/sBUyCz7eGeqZoYdSuH/KGyYLADlP+lDdkyecGncHqPbvtuP59nskdnLMd3xtwMXSoG9yXVw== X-Received: by 10.68.135.233 with SMTP id pv9mr724114pbb.115.1432751301621; Wed, 27 May 2015 11:28:21 -0700 (PDT) X-Received: by 10.68.135.233 with SMTP id pv9mr724097pbb.115.1432751301492; Wed, 27 May 2015 11:28:21 -0700 (PDT) Received: from dhdang-Precision-WorkStation-T3400.amcc.com (67-207-112-226.static.wiline.com. [67.207.112.226]) by mx.google.com with ESMTPSA id hj11sm16887848pbd.33.2015.05.27.11.28.19 (version=TLSv1.2 cipher=ECDHE-RSA-AES128-SHA bits=128/128); Wed, 27 May 2015 11:28:20 -0700 (PDT) From: Duc Dang To: Bjorn Helgaas , Arnd Bergmann , Grant Likely , Liviu Dudau , Marc Zyngier , Thomas Gleixner , Jason Cooper , Mark Rutland Cc: linux-pci@vger.kernel.org, linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, Tanmay Inamdar , Loc Ho , Feng Kan , Duc Dang Subject: [PATCH v9 2/4] PCI: X-Gene: Add the APM X-Gene v1 PCIe MSI/MSIX termination driver Date: Wed, 27 May 2015 11:27:37 -0700 Message-Id: <8d5049d8c03f8f5883ea626596dafc7d3f1c99f1.1432747280.git.dhdang@apm.com> X-Mailer: git-send-email 1.9.1 In-Reply-To: References: In-Reply-To: References: <20150525125206.64d15861@arm.com> Sender: linux-pci-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-pci@vger.kernel.org X-Spam-Status: No, score=-6.9 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_HI, T_RP_MATCHES_RCVD, UNPARSEABLE_RELAY autolearn=unavailable version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on mail.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP APM X-Gene v1 SoC supports its own implementation of MSI, which is not compliant to GIC V2M specification for MSI Termination. There is single MSI block in X-Gene v1 SOC which serves all 5 PCIe ports. This MSI block supports 2048 MSI termination ports coalesced into 16 physical HW IRQ lines and shared across all 5 PCIe ports. As there are only 16 HW IRQs to serve 2048 MSI vectors, to support set_affinity correctly for each MSI vectors, the 16 HW IRQs are statically allocated to 8 X-Gene v1 cores (2 HW IRQs for each cores). To steer MSI interrupt to target CPU, MSI vector is moved around these HW IRQs lines. With this approach, the total MSI vectors this driver supports is reduced to 256. Signed-off-by: Duc Dang Signed-off-by: Tanmay Inamdar Reviewed-by: Marc Zyngier Signed-off-by: Duc Dang Signed-off-by: Tanmay Inamdar Signed-off-by: Bjorn Helgaas Reviewed-by: Marc Zyngier --- drivers/pci/host/Kconfig | 10 + drivers/pci/host/Makefile | 1 + drivers/pci/host/pci-xgene-msi.c | 595 +++++++++++++++++++++++++++++++++++++++ drivers/pci/host/pci-xgene.c | 21 ++ 4 files changed, 627 insertions(+) create mode 100644 drivers/pci/host/pci-xgene-msi.c diff --git a/drivers/pci/host/Kconfig b/drivers/pci/host/Kconfig index 1dfb567..a93fda8 100644 --- a/drivers/pci/host/Kconfig +++ b/drivers/pci/host/Kconfig @@ -89,11 +89,21 @@ config PCI_XGENE depends on ARCH_XGENE depends on OF select PCIEPORTBUS + select PCI_MSI_IRQ_DOMAIN if PCI_MSI help Say Y here if you want internal PCI support on APM X-Gene SoC. There are 5 internal PCIe ports available. Each port is GEN3 capable and have varied lanes from x1 to x8. +config PCI_XGENE_MSI + bool "X-Gene v1 PCIe MSI feature" + depends on PCI_XGENE && PCI_MSI + default y + help + Say Y here if you want PCIE MSI support for APM X-Gene v1 SoC. + This MSI driver will provide MSI support for 5 PCIe ports of + APM X-Gene v1 SoC + config PCI_LAYERSCAPE bool "Freescale Layerscape PCIe controller" depends on OF && ARM diff --git a/drivers/pci/host/Makefile b/drivers/pci/host/Makefile index f733b4e..1957431 100644 --- a/drivers/pci/host/Makefile +++ b/drivers/pci/host/Makefile @@ -11,6 +11,7 @@ obj-$(CONFIG_PCIE_SPEAR13XX) += pcie-spear13xx.o obj-$(CONFIG_PCI_KEYSTONE) += pci-keystone-dw.o pci-keystone.o obj-$(CONFIG_PCIE_XILINX) += pcie-xilinx.o obj-$(CONFIG_PCI_XGENE) += pci-xgene.o +obj-$(CONFIG_PCI_XGENE_MSI) += pci-xgene-msi.o obj-$(CONFIG_PCI_LAYERSCAPE) += pci-layerscape.o obj-$(CONFIG_PCI_VERSATILE) += pci-versatile.o obj-$(CONFIG_PCIE_IPROC) += pcie-iproc.o diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c new file mode 100644 index 0000000..461cb11 --- /dev/null +++ b/drivers/pci/host/pci-xgene-msi.c @@ -0,0 +1,595 @@ +/* + * APM X-Gene MSI Driver + * + * Copyright (c) 2014, Applied Micro Circuits Corporation + * Author: Tanmay Inamdar + * Duc Dang + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2 of the License, or (at your + * option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + */ +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#define MSI_IR0 0x000000 +#define MSI_INT0 0x800000 +#define IDX_PER_GROUP 8 +#define IRQS_PER_IDX 16 +#define NR_HW_IRQS 16 +#define NR_MSI_VEC (IDX_PER_GROUP * IRQS_PER_IDX * NR_HW_IRQS) + +struct xgene_msi_group { + struct xgene_msi *msi; + int gic_irq; + u32 msi_grp; +}; + +struct xgene_msi { + struct device_node *node; + struct msi_controller mchip; + struct irq_domain *domain; + u64 msi_addr; + void __iomem *msi_regs; + unsigned long *bitmap; + struct mutex bitmap_lock; + struct xgene_msi_group *msi_groups; + int num_cpus; +}; + +/* Global data */ +static struct xgene_msi xgene_msi_ctrl; + +static struct irq_chip xgene_msi_top_irq_chip = { + .name = "X-Gene1 MSI", + .irq_enable = pci_msi_unmask_irq, + .irq_disable = pci_msi_mask_irq, + .irq_mask = pci_msi_mask_irq, + .irq_unmask = pci_msi_unmask_irq, +}; + +static struct msi_domain_info xgene_msi_domain_info = { + .flags = (MSI_FLAG_USE_DEF_DOM_OPS | MSI_FLAG_USE_DEF_CHIP_OPS | + MSI_FLAG_PCI_MSIX), + .chip = &xgene_msi_top_irq_chip, +}; + +/* + * X-Gene v1 has 16 groups of MSI termination registers MSInIRx, where + * n is group number (0..F), x is index of registers in each group (0..7) + * The registers layout is like following: + * MSI0IR0 base_addr + * MSI0IR1 base_addr + 0x10000 + * ... ... + * MSI0IR6 base_addr + 0x60000 + * MSI0IR7 base_addr + 0x70000 + * MSI1IR0 base_addr + 0x80000 + * MSI1IR1 base_addr + 0x90000 + * ... ... + * MSI1IR7 base_addr + 0xF0000 + * MSI2IR0 base_addr + 0x100000 + * ... ... + * MSIFIR0 base_addr + 0x780000 + * MSIFIR1 base_addr + 0x790000 + * ... ... + * MSIFIR7 base_addr + 0x7F0000 + * MSIINT0 base_addr + 0x800000 + * MSIINT1 base_addr + 0x810000 + * ... ... + * MSIINTF base_addr + 0x8F0000 + * + * Each index register support 16 MSI vectors (0..15) to generate interrupt. + * There are total 16 GIC IRQs assigned for these 16 groups of MSI termination + * registers. + * + * Each MSI termination group has 1 MSIINTn register (n is 0..15) to indicate + * the MSI pending status caused by 1 of its 8 index registers. + */ + +/* MSInIRx read helper */ +static inline u32 xgene_msi_ir_read(struct xgene_msi *msi, + u32 msi_grp, u32 msir_idx) +{ + return readl_relaxed(msi->msi_regs + MSI_IR0 + + (msi_grp << 19) + (msir_idx << 16)); +} + +/* MSIINTn read helper */ +static inline u32 xgene_msi_int_read(struct xgene_msi *msi, u32 msi_grp) +{ + return readl_relaxed(msi->msi_regs + MSI_INT0 + (msi_grp << 16)); +} + +/* With 2048 MSI vectors supported, the MSI message can be construct using + * following scheme: + * - Divide into 8 256-vector groups + * Group 0: 0-255 + * Group 1: 256-511 + * Group 2: 512-767 + * ... + * Group 7: 1792-2047 + * - Each 256-vector group is divided into 16 16-vector groups + * As an example: 16 16-vector groups for 256-vector group 0-255 is + * Group 0: 0-15 + * Group 1: 16-32 + * ... + * Group 15: 240-255 + * - The termination address of MSI vector in 256-vector group n and 16-vector + * group x is the address of MSIxIRn + * - The data for MSI vector in 16-vector group x is x + */ +static inline u32 hwirq_to_reg_set(unsigned long hwirq) +{ + return (hwirq / (NR_HW_IRQS * IRQS_PER_IDX)); +} + +static inline u32 hwirq_to_group(unsigned long hwirq) +{ + return (hwirq % NR_HW_IRQS); +} + +static inline u32 hwirq_to_msi_data(unsigned long hwirq) +{ + return ((hwirq / NR_HW_IRQS) % IRQS_PER_IDX); +} + +static void xgene_compose_msi_msg(struct irq_data *data, struct msi_msg *msg) +{ + struct xgene_msi *msi = irq_data_get_irq_chip_data(data); + u32 reg_set = hwirq_to_reg_set(data->hwirq); + u32 group = hwirq_to_group(data->hwirq); + u64 target_addr = msi->msi_addr + (((8 * group) + reg_set) << 16); + + msg->address_hi = upper_32_bits(target_addr); + msg->address_lo = lower_32_bits(target_addr); + msg->data = hwirq_to_msi_data(data->hwirq); +} + +/* + * X-Gene v1 only has 16 MSI GIC IRQs for 2048 MSI vectors. + * To maintain the expected behaviour of .set_affinity for each MSI + * interrupt, the 16 MSI GIC IRQs are statically allocated to 8 X-Gene + * v1 cores (2 GIC IRQs for each core). The MSI vector is moved fom 1 + * MSI GIC IRQ to another MSI GIC IRQ to steer its MSI interrupt to + * correct X-Gene v1 core. As a consequence, the total MSI vectors that + * X-Gene v1 supports will be reduced to 256 (2048/8) vectors. + */ +static inline int hwirq_to_cpu(unsigned long hwirq) +{ + return ((hwirq % NR_HW_IRQS) % xgene_msi_ctrl.num_cpus); +} + +static inline unsigned long hwirq_to_canonical_hwirq(unsigned long hwirq) +{ + return (hwirq - (hwirq % xgene_msi_ctrl.num_cpus)); +} + +static int xgene_msi_set_affinity(struct irq_data *irq_data, + const struct cpumask *mask, bool force) +{ + int target_cpu = cpumask_first(mask); + int curr_cpu; + + curr_cpu = hwirq_to_cpu(irq_data->hwirq); + if (curr_cpu == target_cpu) + return IRQ_SET_MASK_OK_DONE; + + /* Update MSI number to target the new CPU */ + irq_data->hwirq = irq_data->hwirq + (target_cpu - curr_cpu); + + return IRQ_SET_MASK_OK; +} + +static struct irq_chip xgene_msi_bottom_irq_chip = { + .name = "MSI", + .irq_set_affinity = xgene_msi_set_affinity, + .irq_compose_msi_msg = xgene_compose_msi_msg, +}; + +static int xgene_irq_domain_alloc(struct irq_domain *domain, unsigned int virq, + unsigned int nr_irqs, void *args) +{ + struct xgene_msi *msi = domain->host_data; + int msi_irq; + + mutex_lock(&msi->bitmap_lock); + + msi_irq = bitmap_find_next_zero_area(msi->bitmap, NR_MSI_VEC, 0, + msi->num_cpus, 0); + if (msi_irq < NR_MSI_VEC) + bitmap_set(msi->bitmap, msi_irq, msi->num_cpus); + else + msi_irq = -ENOSPC; + + mutex_unlock(&msi->bitmap_lock); + + if (msi_irq < 0) + return msi_irq; + + irq_domain_set_info(domain, virq, msi_irq, + &xgene_msi_bottom_irq_chip, domain->host_data, + handle_simple_irq, NULL, NULL); + set_irq_flags(virq, IRQF_VALID); + + return 0; +} + +static void xgene_irq_domain_free(struct irq_domain *domain, + unsigned int virq, unsigned int nr_irqs) +{ + struct irq_data *d = irq_domain_get_irq_data(domain, virq); + struct xgene_msi *msi = irq_data_get_irq_chip_data(d); + u32 hwirq; + + mutex_lock(&msi->bitmap_lock); + + hwirq = hwirq_to_canonical_hwirq(d->hwirq); + bitmap_clear(msi->bitmap, hwirq, msi->num_cpus); + + mutex_unlock(&msi->bitmap_lock); + + irq_domain_free_irqs_parent(domain, virq, nr_irqs); +} + +static const struct irq_domain_ops msi_domain_ops = { + .alloc = xgene_irq_domain_alloc, + .free = xgene_irq_domain_free, +}; + +static int xgene_allocate_domains(struct xgene_msi *msi) +{ + msi->domain = irq_domain_add_linear(NULL, NR_MSI_VEC, + &msi_domain_ops, msi); + if (!msi->domain) + return -ENOMEM; + + msi->mchip.domain = pci_msi_create_irq_domain(msi->mchip.of_node, + &xgene_msi_domain_info, + msi->domain); + + if (!msi->mchip.domain) { + irq_domain_remove(msi->domain); + return -ENOMEM; + } + + return 0; +} + +static void xgene_free_domains(struct xgene_msi *msi) +{ + if (msi->mchip.domain) + irq_domain_remove(msi->mchip.domain); + if (msi->domain) + irq_domain_remove(msi->domain); +} + +static int xgene_msi_init_allocator(struct xgene_msi *xgene_msi) +{ + int size = BITS_TO_LONGS(NR_MSI_VEC) * sizeof(long); + + xgene_msi->bitmap = kzalloc(size, GFP_KERNEL); + if (!xgene_msi->bitmap) + return -ENOMEM; + + mutex_init(&xgene_msi->bitmap_lock); + + xgene_msi->msi_groups = kcalloc(NR_HW_IRQS, + sizeof(struct xgene_msi_group), + GFP_KERNEL); + if (!xgene_msi->msi_groups) + return -ENOMEM; + + return 0; +} + +static void xgene_msi_isr(unsigned int irq, struct irq_desc *desc) +{ + struct irq_chip *chip = irq_desc_get_chip(desc); + struct xgene_msi_group *msi_groups; + struct xgene_msi *xgene_msi; + unsigned int virq; + int msir_index, msir_val, hw_irq; + u32 intr_index, grp_select, msi_grp; + + chained_irq_enter(chip, desc); + + msi_groups = irq_desc_get_handler_data(desc); + xgene_msi = msi_groups->msi; + msi_grp = msi_groups->msi_grp; + + /* + * MSIINTn (n is 0..F) indicates if there is a pending MSI interrupt + * If bit x of this register is set (x is 0..7), one or more interupts + * corresponding to MSInIRx is set. + */ + grp_select = xgene_msi_int_read(xgene_msi, msi_grp); + while (grp_select) { + msir_index = ffs(grp_select) - 1; + /* + * Calculate MSInIRx address to read to check for interrupts + * (refer to termination address and data assignment + * described in xgene_compose_msi_msg function) + */ + msir_val = xgene_msi_ir_read(xgene_msi, msi_grp, msir_index); + while (msir_val) { + intr_index = ffs(msir_val) - 1; + /* + * Calculate MSI vector number (refer to the termination + * address and data assignment described in + * xgene_compose_msi_msg function) + */ + hw_irq = (((msir_index * IRQS_PER_IDX) + intr_index) * + NR_HW_IRQS) + msi_grp; + /* + * As we have multiple hw_irq that maps to single MSI, + * always look up the virq using the hw_irq as seen from + * CPU0 + */ + hw_irq = hwirq_to_canonical_hwirq(hw_irq); + virq = irq_find_mapping(xgene_msi->domain, hw_irq); + WARN_ON(!virq); + if (virq != 0) + generic_handle_irq(virq); + msir_val &= ~(1 << intr_index); + } + grp_select &= ~(1 << msir_index); + + if (!grp_select) { + /* + * We handled all interrupts happened in this group, + * resample this group MSI_INTx register in case + * something else has been made pending in the meantime + */ + grp_select = xgene_msi_int_read(xgene_msi, msi_grp); + } + } + + chained_irq_exit(chip, desc); +} + +static int xgene_msi_remove(struct platform_device *pdev) +{ + int virq, i; + struct xgene_msi *msi = platform_get_drvdata(pdev); + + for (i = 0; i < NR_HW_IRQS; i++) { + virq = msi->msi_groups[i].gic_irq; + if (virq != 0) { + irq_set_chained_handler(virq, NULL); + irq_set_handler_data(virq, NULL); + } + } + kfree(msi->msi_groups); + + kfree(msi->bitmap); + msi->bitmap = NULL; + + xgene_free_domains(msi); + + return 0; +} + +static int xgene_msi_hwirq_alloc(unsigned int cpu) +{ + struct xgene_msi *msi = &xgene_msi_ctrl; + struct xgene_msi_group *msi_group; + cpumask_var_t mask; + int i; + int err; + + for (i = cpu; i < NR_HW_IRQS; i += msi->num_cpus) { + msi_group = &msi->msi_groups[i]; + if (!msi_group->gic_irq) + continue; + + irq_set_chained_handler(msi_group->gic_irq, + xgene_msi_isr); + err = irq_set_handler_data(msi_group->gic_irq, msi_group); + if (err) { + pr_err("failed to register GIC IRQ handler\n"); + return -EINVAL; + } + /* + * Statically allocate MSI GIC IRQs to each CPU core + * With 8-core X-Gene v1, 2 MSI GIC IRQs are allocated + * to each core. + */ + if (alloc_cpumask_var(&mask, GFP_KERNEL)) { + cpumask_clear(mask); + cpumask_set_cpu(cpu, mask); + err = irq_set_affinity(msi_group->gic_irq, mask); + if (err) + pr_err("failed to set affinity for GIC IRQ"); + free_cpumask_var(mask); + } else { + pr_err("failed to alloc CPU mask for affinity\n"); + err = -EINVAL; + } + + if (err) { + irq_set_chained_handler(msi_group->gic_irq, NULL); + irq_set_handler_data(msi_group->gic_irq, NULL); + return err; + } + } + + return 0; +} + +static void xgene_msi_hwirq_free(unsigned int cpu) +{ + struct xgene_msi *msi = &xgene_msi_ctrl; + struct xgene_msi_group *msi_group; + int i; + + for (i = cpu; i < NR_HW_IRQS; i += msi->num_cpus) { + msi_group = &msi->msi_groups[i]; + if (!msi_group->gic_irq) + continue; + + irq_set_chained_handler(msi_group->gic_irq, NULL); + irq_set_handler_data(msi_group->gic_irq, NULL); + } +} + +static int xgene_msi_cpu_callback(struct notifier_block *nfb, + unsigned long action, void *hcpu) +{ + unsigned cpu = (unsigned long)hcpu; + + switch (action) { + case CPU_ONLINE: + case CPU_ONLINE_FROZEN: + xgene_msi_hwirq_alloc(cpu); + break; + case CPU_DEAD: + case CPU_DEAD_FROZEN: + xgene_msi_hwirq_free(cpu); + break; + default: + break; + } + + return NOTIFY_OK; +} + +static struct notifier_block xgene_msi_cpu_notifier = { + .notifier_call = xgene_msi_cpu_callback, +}; + +static const struct of_device_id xgene_msi_match_table[] = { + {.compatible = "apm,xgene1-msi"}, + {}, +}; + +static int xgene_msi_probe(struct platform_device *pdev) +{ + struct resource *res; + int rc, irq_index; + struct xgene_msi *xgene_msi; + unsigned int cpu; + int virt_msir; + u32 msi_val, msi_idx; + + xgene_msi = &xgene_msi_ctrl; + + platform_set_drvdata(pdev, xgene_msi); + + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + xgene_msi->msi_regs = devm_ioremap_resource(&pdev->dev, res); + if (IS_ERR(xgene_msi->msi_regs)) { + dev_err(&pdev->dev, "no reg space\n"); + rc = -EINVAL; + goto error; + } + xgene_msi->msi_addr = res->start; + + xgene_msi->num_cpus = num_possible_cpus(); + + rc = xgene_msi_init_allocator(xgene_msi); + if (rc) { + dev_err(&pdev->dev, "Error allocating MSI bitmap\n"); + goto error; + } + + rc = xgene_allocate_domains(xgene_msi); + if (rc) { + dev_err(&pdev->dev, "Failed to allocate MSI domain\n"); + goto error; + } + + for (irq_index = 0; irq_index < NR_HW_IRQS; irq_index++) { + virt_msir = platform_get_irq(pdev, irq_index); + if (virt_msir < 0) { + dev_err(&pdev->dev, "Cannot translate IRQ index %d\n", + irq_index); + rc = -EINVAL; + goto error; + } + xgene_msi->msi_groups[irq_index].gic_irq = virt_msir; + xgene_msi->msi_groups[irq_index].msi_grp = irq_index; + xgene_msi->msi_groups[irq_index].msi = xgene_msi; + } + + /* + * MSInIRx registers are read-to-clear, before registering interrupt + * handlers, reading all of them to clear all spurious interrupts + * that may occur before the driver is probed. + */ + for (irq_index = 0; irq_index < NR_HW_IRQS; irq_index++) { + for (msi_idx = 0; msi_idx < IDX_PER_GROUP; msi_idx++) + msi_val = xgene_msi_ir_read(xgene_msi, irq_index, + msi_idx); + /* Read MSIINTn to confirm */ + msi_val = xgene_msi_int_read(xgene_msi, irq_index); + if (msi_val) { + dev_err(&pdev->dev, "Failed to clear spurious IRQ\n"); + rc = -EINVAL; + goto error; + } + } + + cpu_notifier_register_begin(); + + for_each_online_cpu(cpu) + if (xgene_msi_hwirq_alloc(cpu)) { + dev_err(&pdev->dev, "failed to regiser MSI handlers\n"); + cpu_notifier_register_done(); + goto error; + } + + rc = __register_hotcpu_notifier(&xgene_msi_cpu_notifier); + if (rc) { + dev_err(&pdev->dev, "failed to add CPU MSI notifier\n"); + cpu_notifier_register_done(); + goto error; + } + + cpu_notifier_register_done(); + + xgene_msi->mchip.of_node = pdev->dev.of_node; + rc = of_pci_msi_chip_add(&xgene_msi->mchip); + if (rc) { + dev_err(&pdev->dev, "failed to add MSI controller chip\n"); + goto error_notifier; + } + + dev_info(&pdev->dev, "APM X-Gene PCIe MSI driver loaded\n"); + + return 0; + +error_notifier: + unregister_hotcpu_notifier(&xgene_msi_cpu_notifier); +error: + xgene_msi_remove(pdev); + return rc; +} + +static struct platform_driver xgene_msi_driver = { + .driver = { + .name = "xgene-msi", + .owner = THIS_MODULE, + .of_match_table = xgene_msi_match_table, + }, + .probe = xgene_msi_probe, + .remove = xgene_msi_remove, +}; + +static int __init xgene_pcie_msi_init(void) +{ + return platform_driver_register(&xgene_msi_driver); +} +subsys_initcall(xgene_pcie_msi_init); diff --git a/drivers/pci/host/pci-xgene.c b/drivers/pci/host/pci-xgene.c index ee082c0..a9b4f16 100644 --- a/drivers/pci/host/pci-xgene.c +++ b/drivers/pci/host/pci-xgene.c @@ -468,6 +468,23 @@ static int xgene_pcie_setup(struct xgene_pcie_port *port, return 0; } +static int xgene_pcie_msi_enable(struct pci_bus *bus) +{ + struct device_node *msi_node; + + msi_node = of_parse_phandle(bus->dev.of_node, + "msi-parent", 0); + if (!msi_node) + return -ENODEV; + + bus->msi = of_pci_find_msi_chip_by_node(msi_node); + if (bus->msi) + bus->msi->dev = &bus->dev; + else + return -ENODEV; + return 0; +} + static int xgene_pcie_probe_bridge(struct platform_device *pdev) { struct device_node *dn = pdev->dev.of_node; @@ -504,6 +521,10 @@ static int xgene_pcie_probe_bridge(struct platform_device *pdev) if (!bus) return -ENOMEM; + if (IS_ENABLED(CONFIG_PCI_MSI)) + if (xgene_pcie_msi_enable(bus)) + dev_info(port->dev, "failed to enable MSI\n"); + pci_scan_child_bus(bus); pci_assign_unassigned_bus_resources(bus); pci_bus_add_devices(bus);