From fe17a42e704a64477b15bb2cf8366fe3e5119aff Mon Sep 17 00:00:00 2001 From: Marc Zyngier Date: Fri, 12 May 2017 12:55:35 +0100 Subject: [PATCH 001/112] irqdomain: Let irq_domain_mapping display hierarchical domains Hierarchical domains seem to be hard to grasp, and a number of aspiring kernel hackers find them utterly discombobulating. In order to ease their pain, let's make them appear in /sys/kernel/debug/irq_domain_mapping, such as the following: 96 0x81808 MSI 0x (null) RADIX MSI 96+ 0x00063 GICv2m 0xffff8003ee116980 RADIX GICv2m 96+ 0x00063 GICv2 0xffff00000916bfd8 LINEAR GICv2 [output compressed to fit in a commit log] This shows that IRQ96 is implemented by a stack of three domains, the + sign indicating the stacking. Signed-off-by: Marc Zyngier Link: http://lkml.kernel.org/r/20170512115538.10767-2-marc.zyngier@arm.com Signed-off-by: Thomas Gleixner --- kernel/irq/irqdomain.c | 68 ++++++++++++++++++++++++++---------------- 1 file changed, 43 insertions(+), 25 deletions(-) diff --git a/kernel/irq/irqdomain.c b/kernel/irq/irqdomain.c index 31805f237396..1f6cd2cacf74 100644 --- a/kernel/irq/irqdomain.c +++ b/kernel/irq/irqdomain.c @@ -746,13 +746,54 @@ unsigned int irq_find_mapping(struct irq_domain *domain, EXPORT_SYMBOL_GPL(irq_find_mapping); #ifdef CONFIG_IRQ_DOMAIN_DEBUG +static void virq_debug_show_one(struct seq_file *m, struct irq_desc *desc) +{ + struct irq_domain *domain; + struct irq_data *data; + + domain = desc->irq_data.domain; + data = &desc->irq_data; + + while (domain) { + unsigned int irq = data->irq; + unsigned long hwirq = data->hwirq; + struct irq_chip *chip; + bool direct; + + if (data == &desc->irq_data) + seq_printf(m, "%5d ", irq); + else + seq_printf(m, "%5d+ ", irq); + seq_printf(m, "0x%05lx ", hwirq); + + chip = irq_data_get_irq_chip(data); + seq_printf(m, "%-15s ", (chip && chip->name) ? chip->name : "none"); + + seq_printf(m, data ? "0x%p " : " %p ", + irq_data_get_irq_chip_data(data)); + + seq_printf(m, " %c ", (desc->action && desc->action->handler) ? '*' : ' '); + direct = (irq == hwirq) && (irq < domain->revmap_direct_max_irq); + seq_printf(m, "%6s%-8s ", + (hwirq < domain->revmap_size) ? "LINEAR" : "RADIX", + direct ? "(DIRECT)" : ""); + seq_printf(m, "%s\n", domain->name); +#ifdef CONFIG_IRQ_DOMAIN_HIERARCHY + domain = domain->parent; + data = data->parent_data; +#else + domain = NULL; +#endif + } +} + static int virq_debug_show(struct seq_file *m, void *private) { unsigned long flags; struct irq_desc *desc; struct irq_domain *domain; struct radix_tree_iter iter; - void *data, **slot; + void **slot; int i; seq_printf(m, " %-16s %-6s %-10s %-10s %s\n", @@ -782,30 +823,7 @@ static int virq_debug_show(struct seq_file *m, void *private) continue; raw_spin_lock_irqsave(&desc->lock, flags); - domain = desc->irq_data.domain; - - if (domain) { - struct irq_chip *chip; - int hwirq = desc->irq_data.hwirq; - bool direct; - - seq_printf(m, "%5d ", i); - seq_printf(m, "0x%05x ", hwirq); - - chip = irq_desc_get_chip(desc); - seq_printf(m, "%-15s ", (chip && chip->name) ? chip->name : "none"); - - data = irq_desc_get_chip_data(desc); - seq_printf(m, data ? "0x%p " : " %p ", data); - - seq_printf(m, " %c ", (desc->action && desc->action->handler) ? '*' : ' '); - direct = (i == hwirq) && (i < domain->revmap_direct_max_irq); - seq_printf(m, "%6s%-8s ", - (hwirq < domain->revmap_size) ? "LINEAR" : "RADIX", - direct ? "(DIRECT)" : ""); - seq_printf(m, "%s\n", desc->irq_data.domain->name); - } - + virq_debug_show_one(m, desc); raw_spin_unlock_irqrestore(&desc->lock, flags); } From 2370c00dc7232d0c4af224e7730b4de031f3b1a0 Mon Sep 17 00:00:00 2001 From: Marc Zyngier Date: Fri, 12 May 2017 12:55:36 +0100 Subject: [PATCH 002/112] irqdomain: Let irq_domain_mapping display ACPI fwnode attributes If the system is using ACPI, there is no of_node to display. But ACPI can use a struct irqchip_fwid as a domain identifier, and it can be used to display the name contained in that structure. The output on such a system will look like this: pMSI 0 0 0 irqchip@00000000e1180000 MSI 37 0 0 irqchip@00000000e1180000 GICv2m 37 0 0 irqchip@00000000e1180000 GICv2 448 448 0 irqchip@ffff000008003000 Signed-off-by: Marc Zyngier Link: http://lkml.kernel.org/r/20170512115538.10767-3-marc.zyngier@arm.com Signed-off-by: Thomas Gleixner --- kernel/irq/irqdomain.c | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-) diff --git a/kernel/irq/irqdomain.c b/kernel/irq/irqdomain.c index 1f6cd2cacf74..70b9da72018b 100644 --- a/kernel/irq/irqdomain.c +++ b/kernel/irq/irqdomain.c @@ -801,15 +801,26 @@ static int virq_debug_show(struct seq_file *m, void *private) mutex_lock(&irq_domain_mutex); list_for_each_entry(domain, &irq_domain_list, link) { struct device_node *of_node; + const char *name; + int count = 0; + of_node = irq_domain_get_of_node(domain); + if (of_node) + name = of_node_full_name(of_node); + else if (is_fwnode_irqchip(domain->fwnode)) + name = container_of(domain->fwnode, struct irqchip_fwid, + fwnode)->name; + else + name = ""; + radix_tree_for_each_slot(slot, &domain->revmap_tree, &iter, 0) count++; seq_printf(m, "%c%-16s %6u %10u %10u %s\n", domain == irq_default_domain ? '*' : ' ', domain->name, domain->revmap_size + count, domain->revmap_size, domain->revmap_direct_max_irq, - of_node ? of_node_full_name(of_node) : ""); + name); } mutex_unlock(&irq_domain_mutex); From a97b852b4d4c2f8c50cab13c71566639f9a1a990 Mon Sep 17 00:00:00 2001 From: Marc Zyngier Date: Fri, 12 May 2017 12:55:37 +0100 Subject: [PATCH 003/112] genirq/msi: Populate the domain name if provided by the irqchip In order to ease debug, let's populate the domain name upfront, before any MSI gets requested. This allows the domain to appear in the irq_domain_mapping, and the user to easily find the expected data. Signed-off-by: Marc Zyngier Link: http://lkml.kernel.org/r/20170512115538.10767-4-marc.zyngier@arm.com Signed-off-by: Thomas Gleixner --- kernel/irq/msi.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/kernel/irq/msi.c b/kernel/irq/msi.c index ddc2f5427f75..fe4d48ec5bc4 100644 --- a/kernel/irq/msi.c +++ b/kernel/irq/msi.c @@ -265,13 +265,19 @@ struct irq_domain *msi_create_irq_domain(struct fwnode_handle *fwnode, struct msi_domain_info *info, struct irq_domain *parent) { + struct irq_domain *domain; + if (info->flags & MSI_FLAG_USE_DEF_DOM_OPS) msi_domain_update_dom_ops(info); if (info->flags & MSI_FLAG_USE_DEF_CHIP_OPS) msi_domain_update_chip_ops(info); - return irq_domain_create_hierarchy(parent, IRQ_DOMAIN_FLAG_MSI, 0, - fwnode, &msi_domain_ops, info); + domain = irq_domain_create_hierarchy(parent, IRQ_DOMAIN_FLAG_MSI, 0, + fwnode, &msi_domain_ops, info); + if (domain && info->chip && info->chip->name) + domain->name = info->chip->name; + + return domain; } int msi_domain_prepare_irqs(struct irq_domain *domain, struct device *dev, From 53286669392b2d888f4b78b0e7894e022e668d11 Mon Sep 17 00:00:00 2001 From: Marc Zyngier Date: Fri, 12 May 2017 12:55:38 +0100 Subject: [PATCH 004/112] Documentation: Update IRQ-domain.txt to document irq_domain_mapping irq_domain_mapping is a rather useful tool to understand how IRqs are mapped in irqdomains, and yet it is not documented anywhere. Let's address this. Signed-off-by: Marc Zyngier Link: http://lkml.kernel.org/r/20170512115538.10767-5-marc.zyngier@arm.com Signed-off-by: Thomas Gleixner --- Documentation/IRQ-domain.txt | 41 ++++++++++++++++++++++++++++++++++-- 1 file changed, 39 insertions(+), 2 deletions(-) diff --git a/Documentation/IRQ-domain.txt b/Documentation/IRQ-domain.txt index 82001a25a14b..1f246eb25ca5 100644 --- a/Documentation/IRQ-domain.txt +++ b/Documentation/IRQ-domain.txt @@ -231,5 +231,42 @@ needs to: 4) No need to implement irq_domain_ops.map and irq_domain_ops.unmap, they are unused with hierarchy irq_domain. -Hierarchy irq_domain may also be used to support other architectures, -such as ARM, ARM64 etc. +Hierarchy irq_domain is in no way x86 specific, and is heavily used to +support other architectures, such as ARM, ARM64 etc. + +=== Debugging === + +If you switch on CONFIG_IRQ_DOMAIN_DEBUG (which depends on +CONFIG_IRQ_DOMAIN and CONFIG_DEBUG_FS), you will find a new file in +your debugfs mount point, called irq_domain_mapping. This file +contains a live snapshot of all the IRQ domains in the system: + + name mapped linear-max direct-max devtree-node + pl061 8 8 0 /smb/gpio@e0080000 + pl061 8 8 0 /smb/gpio@e1050000 + pMSI 0 0 0 /interrupt-controller@e1101000/v2m@e0080000 + MSI 37 0 0 /interrupt-controller@e1101000/v2m@e0080000 + GICv2m 37 0 0 /interrupt-controller@e1101000/v2m@e0080000 + GICv2 448 448 0 /interrupt-controller@e1101000 + +it also iterates over the interrupts to display their mapping in the +domains, and makes the domain stacking visible: + + +irq hwirq chip name chip data active type domain + 1 0x00019 GICv2 0xffff00000916bfd8 * LINEAR GICv2 + 2 0x0001d GICv2 0xffff00000916bfd8 LINEAR GICv2 + 3 0x0001e GICv2 0xffff00000916bfd8 * LINEAR GICv2 + 4 0x0001b GICv2 0xffff00000916bfd8 * LINEAR GICv2 + 5 0x0001a GICv2 0xffff00000916bfd8 LINEAR GICv2 +[...] + 96 0x81808 MSI 0x (null) RADIX MSI + 96+ 0x00063 GICv2m 0xffff8003ee116980 RADIX GICv2m + 96+ 0x00063 GICv2 0xffff00000916bfd8 LINEAR GICv2 + 97 0x08800 MSI 0x (null) * RADIX MSI + 97+ 0x00064 GICv2m 0xffff8003ee116980 * RADIX GICv2m + 97+ 0x00064 GICv2 0xffff00000916bfd8 * LINEAR GICv2 + +Here, interrupts 1-5 are only using a single domain, while 96 and 97 +are build out of a stack of three domain, each level performing a +particular function. From 5a29ef22098874db79af7bf92a247a0f503bfa6e Mon Sep 17 00:00:00 2001 From: Vincent Legoll Date: Tue, 9 May 2017 10:34:09 +0200 Subject: [PATCH 005/112] genirq: Make early_irq_init() print out more informative The printk in early_irq_init() is cryptic and badly formatted: NR_IRQS:33024 nr_irqs:968 16 The last number is the number of preallocated interrupts, so add a prefix to it: NR_IRQS: 33024, nr_irqs: 968, preallocated irqs: 16 Cleanup the formatting for better readability as well. Signed-off-by: Vincent Legoll Signed-off-by: Thomas Gleixner Link: http://lkml.kernel.org/r/1494318849-6733-1-git-send-email-vincent.legoll@gmail.com --- kernel/irq/irqdesc.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 00bb0aeea1d0..09abce2ea8f0 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -480,7 +480,8 @@ int __init early_irq_init(void) /* Let arch update nr_irqs and return the nr of preallocated irqs */ initcnt = arch_probe_nr_irqs(); - printk(KERN_INFO "NR_IRQS:%d nr_irqs:%d %d\n", NR_IRQS, nr_irqs, initcnt); + printk(KERN_INFO "NR_IRQS: %d, nr_irqs: %d, preallocated irqs: %d\n", + NR_IRQS, nr_irqs, initcnt); if (WARN_ON(nr_irqs > IRQ_BITMAP_BITS)) nr_irqs = IRQ_BITMAP_BITS; @@ -516,7 +517,7 @@ int __init early_irq_init(void) init_irq_default_affinity(); - printk(KERN_INFO "NR_IRQS:%d\n", NR_IRQS); + printk(KERN_INFO "NR_IRQS: %d\n", NR_IRQS); desc = irq_desc; count = ARRAY_SIZE(irq_desc); From 201d7f47f34bd7cb19161d0426f13b141e381f30 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 31 May 2017 11:58:32 +0200 Subject: [PATCH 006/112] genirq: Handle NOAUTOEN interrupt setup proper If an interrupt is marked NOAUTOEN then request_irq() installs the action, but does not enable the interrupt via startup_irq(). The interrupt is enabled via enable_irq() later from the driver. enable_irq() calls irq_enable(). That means that for interrupts which have a irq_startup() callback this callback is never invoked. Neither is irq_domain_activate_irq() invoked for such interrupts. If an interrupt depends on irq_startup() or irq_domain_activate_irq() then the enable via irq_enable() is not enough. Add a status flag IRQD_IRQ_STARTED_UP and use this to select the proper mechanism in enable_irq(). Use the flag also to avoid pointless calls into the low level functions. Signed-off-by: Thomas Gleixner Acked-by: Marc Zyngier Cc: dianders@chromium.org Cc: jeffy Cc: Brian Norris Cc: tfiga@chromium.org Link: http://lkml.kernel.org/r/20170531100212.130986205@linutronix.de --- include/linux/irq.h | 6 ++++ kernel/irq/chip.c | 76 ++++++++++++++++++++++++++++++++------------- kernel/irq/manage.c | 12 +++++-- 3 files changed, 69 insertions(+), 25 deletions(-) diff --git a/include/linux/irq.h b/include/linux/irq.h index f887351aa80e..94d1ad6ffdd4 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -216,6 +216,7 @@ enum { IRQD_WAKEUP_ARMED = (1 << 19), IRQD_FORWARDED_TO_VCPU = (1 << 20), IRQD_AFFINITY_MANAGED = (1 << 21), + IRQD_IRQ_STARTED = (1 << 22), }; #define __irqd_to_state(d) ACCESS_PRIVATE((d)->common, state_use_accessors) @@ -329,6 +330,11 @@ static inline void irqd_clr_activated(struct irq_data *d) __irqd_to_state(d) &= ~IRQD_ACTIVATED; } +static inline bool irqd_is_started(struct irq_data *d) +{ + return __irqd_to_state(d) & IRQD_IRQ_STARTED; +} + #undef __irqd_to_state static inline irq_hw_number_t irqd_to_hwirq(struct irq_data *d) diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index c94da688ee9b..e0051d58c909 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -185,37 +185,64 @@ static void irq_state_set_masked(struct irq_desc *desc) irqd_set(&desc->irq_data, IRQD_IRQ_MASKED); } +static void irq_state_clr_started(struct irq_desc *desc) +{ + irqd_clear(&desc->irq_data, IRQD_IRQ_STARTED); +} + +static void irq_state_set_started(struct irq_desc *desc) +{ + irqd_set(&desc->irq_data, IRQD_IRQ_STARTED); +} + int irq_startup(struct irq_desc *desc, bool resend) { int ret = 0; - irq_state_clr_disabled(desc); desc->depth = 0; - irq_domain_activate_irq(&desc->irq_data); - if (desc->irq_data.chip->irq_startup) { - ret = desc->irq_data.chip->irq_startup(&desc->irq_data); - irq_state_clr_masked(desc); - } else { + if (irqd_is_started(&desc->irq_data)) { irq_enable(desc); + } else { + irq_domain_activate_irq(&desc->irq_data); + if (desc->irq_data.chip->irq_startup) { + ret = desc->irq_data.chip->irq_startup(&desc->irq_data); + irq_state_clr_disabled(desc); + irq_state_clr_masked(desc); + } else { + irq_enable(desc); + } + irq_state_set_started(desc); } + if (resend) check_irq_resend(desc); + return ret; } +static void __irq_disable(struct irq_desc *desc, bool mask); + void irq_shutdown(struct irq_desc *desc) { - irq_state_set_disabled(desc); - desc->depth = 1; - if (desc->irq_data.chip->irq_shutdown) - desc->irq_data.chip->irq_shutdown(&desc->irq_data); - else if (desc->irq_data.chip->irq_disable) - desc->irq_data.chip->irq_disable(&desc->irq_data); - else - desc->irq_data.chip->irq_mask(&desc->irq_data); + if (irqd_is_started(&desc->irq_data)) { + desc->depth = 1; + if (desc->irq_data.chip->irq_shutdown) { + desc->irq_data.chip->irq_shutdown(&desc->irq_data); + irq_state_set_disabled(desc); + irq_state_set_masked(desc); + } else { + __irq_disable(desc, true); + } + irq_state_clr_started(desc); + } + /* + * This must be called even if the interrupt was never started up, + * because the activation can happen before the interrupt is + * available for request/startup. It has it's own state tracking so + * it's safe to call it unconditionally. + */ irq_domain_deactivate_irq(&desc->irq_data); - irq_state_set_masked(desc); } void irq_enable(struct irq_desc *desc) @@ -228,6 +255,17 @@ void irq_enable(struct irq_desc *desc) irq_state_clr_masked(desc); } +static void __irq_disable(struct irq_desc *desc, bool mask) +{ + irq_state_set_disabled(desc); + if (desc->irq_data.chip->irq_disable) { + desc->irq_data.chip->irq_disable(&desc->irq_data); + irq_state_set_masked(desc); + } else if (mask) { + mask_irq(desc); + } +} + /** * irq_disable - Mark interrupt disabled * @desc: irq descriptor which should be disabled @@ -250,13 +288,7 @@ void irq_enable(struct irq_desc *desc) */ void irq_disable(struct irq_desc *desc) { - irq_state_set_disabled(desc); - if (desc->irq_data.chip->irq_disable) { - desc->irq_data.chip->irq_disable(&desc->irq_data); - irq_state_set_masked(desc); - } else if (irq_settings_disable_unlazy(desc)) { - mask_irq(desc); - } + __irq_disable(desc, irq_settings_disable_unlazy(desc)); } void irq_percpu_enable(struct irq_desc *desc, unsigned int cpu) diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 070be980c37a..57056109f176 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -533,9 +533,15 @@ void __enable_irq(struct irq_desc *desc) goto err_out; /* Prevent probing on this irq: */ irq_settings_set_noprobe(desc); - irq_enable(desc); - check_irq_resend(desc); - /* fall-through */ + /* + * Call irq_startup() not irq_enable() here because the + * interrupt might be marked NOAUTOEN. So irq_startup() + * needs to be invoked when it gets enabled the first + * time. If it was already started up, then irq_startup() + * will invoke irq_enable() under the hood. + */ + irq_startup(desc, true); + break; } default: desc->depth--; From 04c848d398797a626608ff48804d809ae6687163 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 31 May 2017 11:58:33 +0200 Subject: [PATCH 007/112] genirq: Warn when IRQ_NOAUTOEN is used with shared interrupts Shared interrupts do not go well with disabling auto enable: 1) The sharing interrupt might request it while it's still disabled and then wait for interrupts forever. 2) The interrupt might have been requested by the driver sharing the line before IRQ_NOAUTOEN has been set. So the driver which expects that disabled state after calling request_irq() will not get what it wants. Even worse, when it calls enable_irq() later, it will trigger the unbalanced enable_irq() warning. Reported-by: Brian Norris Signed-off-by: Thomas Gleixner Cc: dianders@chromium.org Cc: jeffy Cc: Marc Zyngier Cc: tfiga@chromium.org Link: http://lkml.kernel.org/r/20170531100212.210682135@linutronix.de --- kernel/irq/chip.c | 7 +++++++ kernel/irq/manage.c | 12 ++++++++++-- 2 files changed, 17 insertions(+), 2 deletions(-) diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index e0051d58c909..bc1331f84fb5 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -935,6 +935,13 @@ void irq_modify_status(unsigned int irq, unsigned long clr, unsigned long set) if (!desc) return; + + /* + * Warn when a driver sets the no autoenable flag on an already + * active interrupt. + */ + WARN_ON_ONCE(!desc->depth && (set & _IRQ_NOAUTOEN)); + irq_settings_clr_and_set(desc, clr, set); irqd_clear(&desc->irq_data, IRQD_NO_BALANCING | IRQD_PER_CPU | diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 57056109f176..49c37f1e71c0 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -1334,11 +1334,19 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) if (new->flags & IRQF_ONESHOT) desc->istate |= IRQS_ONESHOT; - if (irq_settings_can_autoenable(desc)) + if (irq_settings_can_autoenable(desc)) { irq_startup(desc, true); - else + } else { + /* + * Shared interrupts do not go well with disabling + * auto enable. The sharing interrupt might request + * it while it's still disabled and then wait for + * interrupts forever. + */ + WARN_ON_ONCE(new->flags & IRQF_SHARED); /* Undo nested disables: */ desc->depth = 1; + } /* Exclude IRQ from balancing if requested */ if (new->flags & IRQF_NOBALANCING) { From 707188f5f2421a304324e6ef3aaf4413cfab0f3d Mon Sep 17 00:00:00 2001 From: Bartosz Golaszewski Date: Wed, 31 May 2017 18:06:56 +0200 Subject: [PATCH 008/112] irq/generic-chip: Provide irq_free_generic_chip() Currently there's no way for users of irq_alloc_generic_chip() to free the allocated memory other than calling kfree() manually on the returned pointer. This may lead to errors if the internals of irq_alloc_generic_chip() ever change. Provide a routine to free the generic chip. Signed-off-by: Bartosz Golaszewski Signed-off-by: Thomas Gleixner Acked-by: Marc Zyngier Cc: linux-doc@vger.kernel.org Cc: Jonathan Corbet Link: http://lkml.kernel.org/r/1496246820-13250-2-git-send-email-brgl@bgdev.pl --- include/linux/irq.h | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/include/linux/irq.h b/include/linux/irq.h index 94d1ad6ffdd4..2c957fe5d9d7 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -22,6 +22,7 @@ #include #include #include +#include #include #include @@ -973,6 +974,11 @@ int __irq_alloc_domain_generic_chips(struct irq_domain *d, int irqs_per_chip, handler, clr, set, flags); \ }) +static inline void irq_free_generic_chip(struct irq_chip_generic *gc) +{ + kfree(gc); +} + static inline struct irq_chip_type *irq_data_get_chip_type(struct irq_data *d) { return container_of(d->chip, struct irq_chip_type, chip); From 32bb6cbb3b4ea5ca24e3fa13e11772c192616e04 Mon Sep 17 00:00:00 2001 From: Bartosz Golaszewski Date: Wed, 31 May 2017 18:06:57 +0200 Subject: [PATCH 009/112] irq/generic-chip: Provide irq_destroy_generic_chip() Most users of irq_alloc_generic_chip() call irq_setup_generic_chip() too. To simplify the cleanup provide a function that both removes a generic chip and frees its memory. Signed-off-by: Bartosz Golaszewski Signed-off-by: Thomas Gleixner Acked-by: Marc Zyngier Cc: linux-doc@vger.kernel.org Cc: Jonathan Corbet Link: http://lkml.kernel.org/r/1496246820-13250-3-git-send-email-brgl@bgdev.pl --- include/linux/irq.h | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/include/linux/irq.h b/include/linux/irq.h index 2c957fe5d9d7..dc63aa10ce70 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -979,6 +979,14 @@ static inline void irq_free_generic_chip(struct irq_chip_generic *gc) kfree(gc); } +static inline void irq_destroy_generic_chip(struct irq_chip_generic *gc, + u32 msk, unsigned int clr, + unsigned int set) +{ + irq_remove_generic_chip(gc, msk, clr, set); + irq_free_generic_chip(gc); +} + static inline struct irq_chip_type *irq_data_get_chip_type(struct irq_data *d) { return container_of(d->chip, struct irq_chip_type, chip); From f160203986a6ad23ab8077c4a25b260fe55d6e26 Mon Sep 17 00:00:00 2001 From: Bartosz Golaszewski Date: Wed, 31 May 2017 18:06:58 +0200 Subject: [PATCH 010/112] irq/generic-chip: Export irq_init_generic_chip() locally This function will be used in the devres variant of irq_alloc_generic_chip(). Signed-off-by: Bartosz Golaszewski Signed-off-by: Thomas Gleixner Acked-by: Marc Zyngier Cc: linux-doc@vger.kernel.org Cc: Jonathan Corbet Link: http://lkml.kernel.org/r/1496246820-13250-4-git-send-email-brgl@bgdev.pl --- kernel/irq/generic-chip.c | 7 +++---- kernel/irq/internals.h | 11 +++++++++++ 2 files changed, 14 insertions(+), 4 deletions(-) diff --git a/kernel/irq/generic-chip.c b/kernel/irq/generic-chip.c index ee32870079c9..f7086b78ad6e 100644 --- a/kernel/irq/generic-chip.c +++ b/kernel/irq/generic-chip.c @@ -201,10 +201,9 @@ static void irq_writel_be(u32 val, void __iomem *addr) iowrite32be(val, addr); } -static void -irq_init_generic_chip(struct irq_chip_generic *gc, const char *name, - int num_ct, unsigned int irq_base, - void __iomem *reg_base, irq_flow_handler_t handler) +void irq_init_generic_chip(struct irq_chip_generic *gc, const char *name, + int num_ct, unsigned int irq_base, + void __iomem *reg_base, irq_flow_handler_t handler) { raw_spin_lock_init(&gc->lock); gc->num_ct = num_ct; diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index bc226e783bd2..921a2419720c 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -226,3 +226,14 @@ irq_pm_install_action(struct irq_desc *desc, struct irqaction *action) { } static inline void irq_pm_remove_action(struct irq_desc *desc, struct irqaction *action) { } #endif + +#ifdef CONFIG_GENERIC_IRQ_CHIP +void irq_init_generic_chip(struct irq_chip_generic *gc, const char *name, + int num_ct, unsigned int irq_base, + void __iomem *reg_base, irq_flow_handler_t handler); +#else +static inline void +irq_init_generic_chip(struct irq_chip_generic *gc, const char *name, + int num_ct, unsigned int irq_base, + void __iomem *reg_base, irq_flow_handler_t handler) { } +#endif /* CONFIG_GENERIC_IRQ_CHIP */ From 1c3e36309fe2e94b8a889fa32cb5c871434f8ed6 Mon Sep 17 00:00:00 2001 From: Bartosz Golaszewski Date: Wed, 31 May 2017 18:06:59 +0200 Subject: [PATCH 011/112] irq/generic-chip: Provide devm_irq_alloc_generic_chip() Provide a resource managed variant of irq_alloc_generic_chip(). Signed-off-by: Bartosz Golaszewski Signed-off-by: Thomas Gleixner Acked-by: Marc Zyngier Cc: linux-doc@vger.kernel.org Cc: Jonathan Corbet Link: http://lkml.kernel.org/r/1496246820-13250-5-git-send-email-brgl@bgdev.pl --- Documentation/driver-model/devres.txt | 1 + include/linux/irq.h | 5 ++++ kernel/irq/devres.c | 34 +++++++++++++++++++++++++++ 3 files changed, 40 insertions(+) diff --git a/Documentation/driver-model/devres.txt b/Documentation/driver-model/devres.txt index e72587fe477d..d473be8c8781 100644 --- a/Documentation/driver-model/devres.txt +++ b/Documentation/driver-model/devres.txt @@ -311,6 +311,7 @@ IRQ devm_irq_alloc_desc_at() devm_irq_alloc_desc_from() devm_irq_alloc_descs_from() + devm_irq_alloc_generic_chip() LED devm_led_classdev_register() diff --git a/include/linux/irq.h b/include/linux/irq.h index dc63aa10ce70..64ae54673e08 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -958,6 +958,11 @@ int irq_setup_alt_chip(struct irq_data *d, unsigned int type); void irq_remove_generic_chip(struct irq_chip_generic *gc, u32 msk, unsigned int clr, unsigned int set); +struct irq_chip_generic * +devm_irq_alloc_generic_chip(struct device *dev, const char *name, int num_ct, + unsigned int irq_base, void __iomem *reg_base, + irq_flow_handler_t handler); + struct irq_chip_generic *irq_get_domain_generic_chip(struct irq_domain *d, unsigned int hw_irq); int __irq_alloc_domain_generic_chips(struct irq_domain *d, int irqs_per_chip, diff --git a/kernel/irq/devres.c b/kernel/irq/devres.c index 1613bfd48365..21ee0aebccfb 100644 --- a/kernel/irq/devres.c +++ b/kernel/irq/devres.c @@ -4,6 +4,8 @@ #include #include +#include "internals.h" + /* * Device resource management aware IRQ request/free implementation. */ @@ -198,3 +200,35 @@ int __devm_irq_alloc_descs(struct device *dev, int irq, unsigned int from, return base; } EXPORT_SYMBOL_GPL(__devm_irq_alloc_descs); + +#ifdef CONFIG_GENERIC_IRQ_CHIP +/** + * devm_irq_alloc_generic_chip - Allocate and initialize a generic chip + * for a managed device + * @dev: Device to allocate the generic chip for + * @name: Name of the irq chip + * @num_ct: Number of irq_chip_type instances associated with this + * @irq_base: Interrupt base nr for this chip + * @reg_base: Register base address (virtual) + * @handler: Default flow handler associated with this chip + * + * Returns an initialized irq_chip_generic structure. The chip defaults + * to the primary (index 0) irq_chip_type and @handler + */ +struct irq_chip_generic * +devm_irq_alloc_generic_chip(struct device *dev, const char *name, int num_ct, + unsigned int irq_base, void __iomem *reg_base, + irq_flow_handler_t handler) +{ + struct irq_chip_generic *gc; + unsigned long sz = sizeof(*gc) + num_ct * sizeof(struct irq_chip_type); + + gc = devm_kzalloc(dev, sz, GFP_KERNEL); + if (gc) + irq_init_generic_chip(gc, name, num_ct, + irq_base, reg_base, handler); + + return gc; +} +EXPORT_SYMBOL_GPL(devm_irq_alloc_generic_chip); +#endif /* CONFIG_GENERIC_IRQ_CHIP */ From 30fd8fc5c91973485705f83c7efe9588b8e6f371 Mon Sep 17 00:00:00 2001 From: Bartosz Golaszewski Date: Wed, 31 May 2017 18:07:00 +0200 Subject: [PATCH 012/112] irq/generic-chip: Provide devm_irq_setup_generic_chip() Provide a resource managed variant of irq_setup_generic_chip(). Signed-off-by: Bartosz Golaszewski Signed-off-by: Thomas Gleixner Acked-by: Marc Zyngier Cc: linux-doc@vger.kernel.org Cc: Jonathan Corbet Link: http://lkml.kernel.org/r/1496246820-13250-6-git-send-email-brgl@bgdev.pl --- Documentation/driver-model/devres.txt | 1 + include/linux/irq.h | 3 ++ kernel/irq/devres.c | 52 +++++++++++++++++++++++++++ 3 files changed, 56 insertions(+) diff --git a/Documentation/driver-model/devres.txt b/Documentation/driver-model/devres.txt index d473be8c8781..6a6618f34440 100644 --- a/Documentation/driver-model/devres.txt +++ b/Documentation/driver-model/devres.txt @@ -312,6 +312,7 @@ IRQ devm_irq_alloc_desc_from() devm_irq_alloc_descs_from() devm_irq_alloc_generic_chip() + devm_irq_setup_generic_chip() LED devm_led_classdev_register() diff --git a/include/linux/irq.h b/include/linux/irq.h index 64ae54673e08..d996314b6522 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -962,6 +962,9 @@ struct irq_chip_generic * devm_irq_alloc_generic_chip(struct device *dev, const char *name, int num_ct, unsigned int irq_base, void __iomem *reg_base, irq_flow_handler_t handler); +int devm_irq_setup_generic_chip(struct device *dev, struct irq_chip_generic *gc, + u32 msk, enum irq_gc_flags flags, + unsigned int clr, unsigned int set); struct irq_chip_generic *irq_get_domain_generic_chip(struct irq_domain *d, unsigned int hw_irq); diff --git a/kernel/irq/devres.c b/kernel/irq/devres.c index 21ee0aebccfb..194c506d9d20 100644 --- a/kernel/irq/devres.c +++ b/kernel/irq/devres.c @@ -231,4 +231,56 @@ devm_irq_alloc_generic_chip(struct device *dev, const char *name, int num_ct, return gc; } EXPORT_SYMBOL_GPL(devm_irq_alloc_generic_chip); + +struct irq_generic_chip_devres { + struct irq_chip_generic *gc; + u32 msk; + unsigned int clr; + unsigned int set; +}; + +static void devm_irq_remove_generic_chip(struct device *dev, void *res) +{ + struct irq_generic_chip_devres *this = res; + + irq_remove_generic_chip(this->gc, this->msk, this->clr, this->set); +} + +/** + * devm_irq_setup_generic_chip - Setup a range of interrupts with a generic + * chip for a managed device + * + * @dev: Device to setup the generic chip for + * @gc: Generic irq chip holding all data + * @msk: Bitmask holding the irqs to initialize relative to gc->irq_base + * @flags: Flags for initialization + * @clr: IRQ_* bits to clear + * @set: IRQ_* bits to set + * + * Set up max. 32 interrupts starting from gc->irq_base. Note, this + * initializes all interrupts to the primary irq_chip_type and its + * associated handler. + */ +int devm_irq_setup_generic_chip(struct device *dev, struct irq_chip_generic *gc, + u32 msk, enum irq_gc_flags flags, + unsigned int clr, unsigned int set) +{ + struct irq_generic_chip_devres *dr; + + dr = devres_alloc(devm_irq_remove_generic_chip, + sizeof(*dr), GFP_KERNEL); + if (!dr) + return -ENOMEM; + + irq_setup_generic_chip(gc, msk, flags, clr, set); + + dr->gc = gc; + dr->msk = msk; + dr->clr = clr; + dr->set = set; + devres_add(dev, dr); + + return 0; +} +EXPORT_SYMBOL_GPL(devm_irq_setup_generic_chip); #endif /* CONFIG_GENERIC_IRQ_CHIP */ From 9ce18f6f0b0c7463fcdce70c35db17a891adc90d Mon Sep 17 00:00:00 2001 From: Chen-Yu Tsai Date: Tue, 6 Jun 2017 13:59:23 +0800 Subject: [PATCH 013/112] irqchip/sunxi-nmi: Convert magic numbers to defines The sunxi-nmi driver has a bunch of raw register offsets and bit values. Convert them into define macros for better readability. Signed-off-by: Chen-Yu Tsai Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-sunxi-nmi.c | 34 +++++++++++++++++++++++---------- 1 file changed, 24 insertions(+), 10 deletions(-) diff --git a/drivers/irqchip/irq-sunxi-nmi.c b/drivers/irqchip/irq-sunxi-nmi.c index 668730c5cb66..177efb473c7d 100644 --- a/drivers/irqchip/irq-sunxi-nmi.c +++ b/drivers/irqchip/irq-sunxi-nmi.c @@ -25,6 +25,20 @@ #define SUNXI_NMI_SRC_TYPE_MASK 0x00000003 +#define SUNXI_NMI_IRQ_BIT BIT(0) + +#define SUN6I_NMI_CTRL 0x00 +#define SUN6I_NMI_PENDING 0x04 +#define SUN6I_NMI_ENABLE 0x34 + +#define SUN7I_NMI_CTRL 0x00 +#define SUN7I_NMI_PENDING 0x04 +#define SUN7I_NMI_ENABLE 0x08 + +#define SUN9I_NMI_CTRL 0x00 +#define SUN9I_NMI_ENABLE 0x04 +#define SUN9I_NMI_PENDING 0x08 + enum { SUNXI_SRC_TYPE_LEVEL_LOW = 0, SUNXI_SRC_TYPE_EDGE_FALLING, @@ -39,21 +53,21 @@ struct sunxi_sc_nmi_reg_offs { }; static struct sunxi_sc_nmi_reg_offs sun7i_reg_offs = { - .ctrl = 0x00, - .pend = 0x04, - .enable = 0x08, + .ctrl = SUN7I_NMI_CTRL, + .pend = SUN7I_NMI_PENDING, + .enable = SUN7I_NMI_ENABLE, }; static struct sunxi_sc_nmi_reg_offs sun6i_reg_offs = { - .ctrl = 0x00, - .pend = 0x04, - .enable = 0x34, + .ctrl = SUN6I_NMI_CTRL, + .pend = SUN6I_NMI_PENDING, + .enable = SUN6I_NMI_ENABLE, }; static struct sunxi_sc_nmi_reg_offs sun9i_reg_offs = { - .ctrl = 0x00, - .pend = 0x08, - .enable = 0x04, + .ctrl = SUN9I_NMI_CTRL, + .pend = SUN9I_NMI_PENDING, + .enable = SUN9I_NMI_ENABLE, }; static inline void sunxi_sc_nmi_write(struct irq_chip_generic *gc, u32 off, @@ -188,7 +202,7 @@ static int __init sunxi_sc_nmi_irq_init(struct device_node *node, gc->chip_types[1].handler = handle_edge_irq; sunxi_sc_nmi_write(gc, reg_offs->enable, 0); - sunxi_sc_nmi_write(gc, reg_offs->pend, 0x1); + sunxi_sc_nmi_write(gc, reg_offs->pend, SUNXI_NMI_IRQ_BIT); irq_set_chained_handler_and_data(irq, sunxi_sc_nmi_handle_irq, domain); From e3ece0d5a31a588ced938d09f30c870ed9b2a2a6 Mon Sep 17 00:00:00 2001 From: Chen-Yu Tsai Date: Tue, 6 Jun 2017 13:59:24 +0800 Subject: [PATCH 014/112] irqchip/sunxi-nmi: Document interrupt disabling and clearing at probe time The sunxi-nmi disables all its interrupts and clears any pending interrupts at probe time. Add comments documenting it, just to make it clear. Signed-off-by: Chen-Yu Tsai Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-sunxi-nmi.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/irqchip/irq-sunxi-nmi.c b/drivers/irqchip/irq-sunxi-nmi.c index 177efb473c7d..9a7f6971cc3a 100644 --- a/drivers/irqchip/irq-sunxi-nmi.c +++ b/drivers/irqchip/irq-sunxi-nmi.c @@ -201,7 +201,10 @@ static int __init sunxi_sc_nmi_irq_init(struct device_node *node, gc->chip_types[1].regs.type = reg_offs->ctrl; gc->chip_types[1].handler = handle_edge_irq; + /* Disable any active interrupts */ sunxi_sc_nmi_write(gc, reg_offs->enable, 0); + + /* Clear any pending NMI interrupts */ sunxi_sc_nmi_write(gc, reg_offs->pend, SUNXI_NMI_IRQ_BIT); irq_set_chained_handler_and_data(irq, sunxi_sc_nmi_handle_irq, domain); From c81a248069cd2f48f6326c35e2cdbd181cbc403c Mon Sep 17 00:00:00 2001 From: Chen-Yu Tsai Date: Tue, 6 Jun 2017 13:59:25 +0800 Subject: [PATCH 015/112] irqchip/sunxi-nmi: Reorder sunxi_sc_nmi_reg_offs' in ascending order This is a pure code move to reorder the various sunxi_sc_nmi_reg_offs' by family and alphabetical order. No functionality changes. Signed-off-by: Chen-Yu Tsai Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-sunxi-nmi.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/drivers/irqchip/irq-sunxi-nmi.c b/drivers/irqchip/irq-sunxi-nmi.c index 9a7f6971cc3a..76f2431b6b7b 100644 --- a/drivers/irqchip/irq-sunxi-nmi.c +++ b/drivers/irqchip/irq-sunxi-nmi.c @@ -52,18 +52,18 @@ struct sunxi_sc_nmi_reg_offs { u32 enable; }; -static struct sunxi_sc_nmi_reg_offs sun7i_reg_offs = { - .ctrl = SUN7I_NMI_CTRL, - .pend = SUN7I_NMI_PENDING, - .enable = SUN7I_NMI_ENABLE, -}; - static struct sunxi_sc_nmi_reg_offs sun6i_reg_offs = { .ctrl = SUN6I_NMI_CTRL, .pend = SUN6I_NMI_PENDING, .enable = SUN6I_NMI_ENABLE, }; +static struct sunxi_sc_nmi_reg_offs sun7i_reg_offs = { + .ctrl = SUN7I_NMI_CTRL, + .pend = SUN7I_NMI_PENDING, + .enable = SUN7I_NMI_ENABLE, +}; + static struct sunxi_sc_nmi_reg_offs sun9i_reg_offs = { .ctrl = SUN9I_NMI_CTRL, .pend = SUN9I_NMI_PENDING, From 11b345ab79a1986fd9da365f646621d9ced9dcfb Mon Sep 17 00:00:00 2001 From: Chen-Yu Tsai Date: Tue, 6 Jun 2017 13:59:26 +0800 Subject: [PATCH 016/112] irqchip/sunxi-nmi: Const-ify sunxi_sc_nmi_reg_offs structures The sunxi_sc_nmi_reg_offs, which hold the register offsets for the various variants, is never modified, and only used at init time within the init functions referenced by IRQCHIP_DECLARE, which themselves are tagged __init. Const-ify the sunxi_sc_nmi_reg_offs structures, and tag them as __initconst. Signed-off-by: Chen-Yu Tsai Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-sunxi-nmi.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/drivers/irqchip/irq-sunxi-nmi.c b/drivers/irqchip/irq-sunxi-nmi.c index 76f2431b6b7b..e713ec66322b 100644 --- a/drivers/irqchip/irq-sunxi-nmi.c +++ b/drivers/irqchip/irq-sunxi-nmi.c @@ -52,19 +52,19 @@ struct sunxi_sc_nmi_reg_offs { u32 enable; }; -static struct sunxi_sc_nmi_reg_offs sun6i_reg_offs = { +static const struct sunxi_sc_nmi_reg_offs sun6i_reg_offs __initconst = { .ctrl = SUN6I_NMI_CTRL, .pend = SUN6I_NMI_PENDING, .enable = SUN6I_NMI_ENABLE, }; -static struct sunxi_sc_nmi_reg_offs sun7i_reg_offs = { +static const struct sunxi_sc_nmi_reg_offs sun7i_reg_offs __initconst = { .ctrl = SUN7I_NMI_CTRL, .pend = SUN7I_NMI_PENDING, .enable = SUN7I_NMI_ENABLE, }; -static struct sunxi_sc_nmi_reg_offs sun9i_reg_offs = { +static const struct sunxi_sc_nmi_reg_offs sun9i_reg_offs __initconst = { .ctrl = SUN9I_NMI_CTRL, .pend = SUN9I_NMI_PENDING, .enable = SUN9I_NMI_ENABLE, @@ -142,7 +142,7 @@ static int sunxi_sc_nmi_set_type(struct irq_data *data, unsigned int flow_type) } static int __init sunxi_sc_nmi_irq_init(struct device_node *node, - struct sunxi_sc_nmi_reg_offs *reg_offs) + const struct sunxi_sc_nmi_reg_offs *reg_offs) { struct irq_domain *domain; struct irq_chip_generic *gc; From cd647575ac2f0c72dfded033f1d8c36c4cf13aa6 Mon Sep 17 00:00:00 2001 From: Chen-Yu Tsai Date: Tue, 6 Jun 2017 13:59:27 +0800 Subject: [PATCH 017/112] dt-bindings/interrupt-controller: sunxi-nmi: Add compatible for A31 R_INTC The A31 and later have an R_INTC block which handles the NMI interrupt pin on the SoC. This interrupt pin is used by the external PMIC to signal interrupts to the SoC. While this hardware block is undocumented, the interrupt offsets combined with the register regions for the existing "sun6i-a31-sc-nmi" compatible line up with the old interrupt controller found on the A10. Experiments show that only the first 32 interrupt lines can be enabled, and only the first (NMI) interrupt is actually connected. This patch adds a new, properly named compatible for the A31 R_INTC block, which requires the register region to be properly aligned to the block boundary. For comparison, the old "sun6i-a31-sc-nmi" compatible had its register region aligned with the first used register. This didn't match up with the memory map in the SoC's datasheet/user manual. Since the new compatible supercedes the old one, deprecate the old one. Signed-off-by: Chen-Yu Tsai Signed-off-by: Marc Zyngier --- .../bindings/interrupt-controller/allwinner,sunxi-nmi.txt | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/Documentation/devicetree/bindings/interrupt-controller/allwinner,sunxi-nmi.txt b/Documentation/devicetree/bindings/interrupt-controller/allwinner,sunxi-nmi.txt index 81cd3692405e..4ae553eb333d 100644 --- a/Documentation/devicetree/bindings/interrupt-controller/allwinner,sunxi-nmi.txt +++ b/Documentation/devicetree/bindings/interrupt-controller/allwinner,sunxi-nmi.txt @@ -3,8 +3,11 @@ Allwinner Sunxi NMI Controller Required properties: -- compatible : should be "allwinner,sun7i-a20-sc-nmi" or - "allwinner,sun6i-a31-sc-nmi" or "allwinner,sun9i-a80-nmi" +- compatible : should be one of the following: + - "allwinner,sun7i-a20-sc-nmi" + - "allwinner,sun6i-a31-sc-nmi" (deprecated) + - "allwinner,sun6i-a31-r-intc" + - "allwinner,sun9i-a80-nmi" - reg : Specifies base physical address and size of the registers. - interrupt-controller : Identifies the node as an interrupt controller - #interrupt-cells : Specifies the number of cells needed to encode an From 173bda53b3403d7278d74c4dabedf3d4325a2f89 Mon Sep 17 00:00:00 2001 From: Chen-Yu Tsai Date: Tue, 6 Jun 2017 13:59:28 +0800 Subject: [PATCH 018/112] irqchip/sunxi-nmi: Support sun6i-a31-r-intc compatible The R_INTC on the A31 is undocumented. It was previously supported by the sun6i-a31-sc-nmi compatible. This compatible however required the register region to start at the first used register, rather than the boundaries laid out in the SoC's memory map. The new compatible fixes the alignment, while also naming it properly. Since the only difference between the old and new compatibles are a fixed offset for the registers, and since the old one is deprecated, this patch adds a set of register defines for the new compatible, while modifying the old set to reference the new set minus a fixed offset. Signed-off-by: Chen-Yu Tsai Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-sunxi-nmi.c | 29 ++++++++++++++++++++++++++--- 1 file changed, 26 insertions(+), 3 deletions(-) diff --git a/drivers/irqchip/irq-sunxi-nmi.c b/drivers/irqchip/irq-sunxi-nmi.c index e713ec66322b..a412b5d5d0fa 100644 --- a/drivers/irqchip/irq-sunxi-nmi.c +++ b/drivers/irqchip/irq-sunxi-nmi.c @@ -27,9 +27,18 @@ #define SUNXI_NMI_IRQ_BIT BIT(0) -#define SUN6I_NMI_CTRL 0x00 -#define SUN6I_NMI_PENDING 0x04 -#define SUN6I_NMI_ENABLE 0x34 +#define SUN6I_R_INTC_CTRL 0x0c +#define SUN6I_R_INTC_PENDING 0x10 +#define SUN6I_R_INTC_ENABLE 0x40 + +/* + * For deprecated sun6i-a31-sc-nmi compatible. + * Registers are offset by 0x0c. + */ +#define SUN6I_R_INTC_NMI_OFFSET 0x0c +#define SUN6I_NMI_CTRL (SUN6I_R_INTC_CTRL - SUN6I_R_INTC_NMI_OFFSET) +#define SUN6I_NMI_PENDING (SUN6I_R_INTC_PENDING - SUN6I_R_INTC_NMI_OFFSET) +#define SUN6I_NMI_ENABLE (SUN6I_R_INTC_ENABLE - SUN6I_R_INTC_NMI_OFFSET) #define SUN7I_NMI_CTRL 0x00 #define SUN7I_NMI_PENDING 0x04 @@ -52,6 +61,12 @@ struct sunxi_sc_nmi_reg_offs { u32 enable; }; +static const struct sunxi_sc_nmi_reg_offs sun6i_r_intc_reg_offs __initconst = { + .ctrl = SUN6I_R_INTC_CTRL, + .pend = SUN6I_R_INTC_PENDING, + .enable = SUN6I_R_INTC_ENABLE, +}; + static const struct sunxi_sc_nmi_reg_offs sun6i_reg_offs __initconst = { .ctrl = SUN6I_NMI_CTRL, .pend = SUN6I_NMI_PENDING, @@ -217,6 +232,14 @@ fail_irqd_remove: return ret; } +static int __init sun6i_r_intc_irq_init(struct device_node *node, + struct device_node *parent) +{ + return sunxi_sc_nmi_irq_init(node, &sun6i_r_intc_reg_offs); +} +IRQCHIP_DECLARE(sun6i_r_intc, "allwinner,sun6i-a31-r-intc", + sun6i_r_intc_irq_init); + static int __init sun6i_sc_nmi_irq_init(struct device_node *node, struct device_node *parent) { From 53ccf3310d271facca17a2766bd4020811ada6a4 Mon Sep 17 00:00:00 2001 From: Tobias Klauser Date: Fri, 2 Jun 2017 10:20:53 +0200 Subject: [PATCH 019/112] irqchip/i8259: Constify irq_domain_ops struct irq_domain_ops is not modified, so it can be made const. Signed-off-by: Tobias Klauser Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-i8259.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/irqchip/irq-i8259.c b/drivers/irqchip/irq-i8259.c index 1aec12c6d9ac..7aafbb091b67 100644 --- a/drivers/irqchip/irq-i8259.c +++ b/drivers/irqchip/irq-i8259.c @@ -307,7 +307,7 @@ static int i8259A_irq_domain_map(struct irq_domain *d, unsigned int virq, return 0; } -static struct irq_domain_ops i8259A_ops = { +static const struct irq_domain_ops i8259A_ops = { .map = i8259A_irq_domain_map, .xlate = irq_domain_xlate_onecell, }; From dcbbefceb27d53379be96979480a7e7d5de59ab8 Mon Sep 17 00:00:00 2001 From: Tobias Klauser Date: Fri, 2 Jun 2017 10:20:54 +0200 Subject: [PATCH 020/112] irqchip/irq-imx-gpcv2: Constify irq_domain_ops struct irq_domain_ops is not modified, so it can be made const. Signed-off-by: Tobias Klauser Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-imx-gpcv2.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/irqchip/irq-imx-gpcv2.c b/drivers/irqchip/irq-imx-gpcv2.c index 9463f3557e82..bb36f572e322 100644 --- a/drivers/irqchip/irq-imx-gpcv2.c +++ b/drivers/irqchip/irq-imx-gpcv2.c @@ -200,7 +200,7 @@ static int imx_gpcv2_domain_alloc(struct irq_domain *domain, &parent_fwspec); } -static struct irq_domain_ops gpcv2_irqchip_data_domain_ops = { +static const struct irq_domain_ops gpcv2_irqchip_data_domain_ops = { .translate = imx_gpcv2_domain_translate, .alloc = imx_gpcv2_domain_alloc, .free = irq_domain_free_irqs_common, From e183c2a3ea1996932fef7d5867df387793b1941f Mon Sep 17 00:00:00 2001 From: Tobias Klauser Date: Fri, 2 Jun 2017 10:20:55 +0200 Subject: [PATCH 021/112] irqchip/irq-mbigen: Constify irq_domain_ops struct irq_domain_ops is not modified, so it can be made const. Signed-off-by: Tobias Klauser Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-mbigen.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/irqchip/irq-mbigen.c b/drivers/irqchip/irq-mbigen.c index 31d6b5a582d2..567b29c47608 100644 --- a/drivers/irqchip/irq-mbigen.c +++ b/drivers/irqchip/irq-mbigen.c @@ -228,7 +228,7 @@ static int mbigen_irq_domain_alloc(struct irq_domain *domain, return 0; } -static struct irq_domain_ops mbigen_domain_ops = { +static const struct irq_domain_ops mbigen_domain_ops = { .translate = mbigen_domain_translate, .alloc = mbigen_irq_domain_alloc, .free = irq_domain_free_irqs_common, From 0b7e815aaa3bb7c89b05a4f092835a7391c0724c Mon Sep 17 00:00:00 2001 From: Tobias Klauser Date: Fri, 2 Jun 2017 10:20:56 +0200 Subject: [PATCH 022/112] irqchip/irq-mips-gic: Constify irq_domain_ops struct irq_domain_ops is not modified, so it can be made const. Signed-off-by: Tobias Klauser Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-mips-gic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/irqchip/irq-mips-gic.c b/drivers/irqchip/irq-mips-gic.c index eb7fbe159963..65642a78b8a3 100644 --- a/drivers/irqchip/irq-mips-gic.c +++ b/drivers/irqchip/irq-mips-gic.c @@ -874,7 +874,7 @@ int gic_ipi_domain_match(struct irq_domain *d, struct device_node *node, } } -static struct irq_domain_ops gic_ipi_domain_ops = { +static const struct irq_domain_ops gic_ipi_domain_ops = { .xlate = gic_ipi_domain_xlate, .alloc = gic_ipi_domain_alloc, .free = gic_ipi_domain_free, From 580393cc10f9a05a4f3b2e09fb066456f97d5b8d Mon Sep 17 00:00:00 2001 From: Tobias Klauser Date: Fri, 2 Jun 2017 10:20:57 +0200 Subject: [PATCH 023/112] irqchip/irq-renesas-h8300h: Constify irq_domain_ops struct irq_domain_ops is not modified, so it can be made const. Cc: Yoshinori Sato Cc: uclinux-h8-devel@lists.sourceforge.jp Signed-off-by: Tobias Klauser Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-renesas-h8300h.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/irqchip/irq-renesas-h8300h.c b/drivers/irqchip/irq-renesas-h8300h.c index c378768d75b3..b8327590ae52 100644 --- a/drivers/irqchip/irq-renesas-h8300h.c +++ b/drivers/irqchip/irq-renesas-h8300h.c @@ -67,7 +67,7 @@ static int irq_map(struct irq_domain *h, unsigned int virq, return 0; } -static struct irq_domain_ops irq_ops = { +static const struct irq_domain_ops irq_ops = { .map = irq_map, .xlate = irq_domain_xlate_onecell, }; From c9262475fa592c1f7777edbc2689589a5ce7266e Mon Sep 17 00:00:00 2001 From: Tobias Klauser Date: Fri, 2 Jun 2017 10:20:58 +0200 Subject: [PATCH 024/112] irqchip/irq-renesas-h8s: Constify irq_domain_ops struct irq_domain_ops is not modified, so it can be made const. Cc: Yoshinori Sato Cc: uclinux-h8-devel@lists.sourceforge.jp Signed-off-by: Tobias Klauser Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-renesas-h8s.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/irqchip/irq-renesas-h8s.c b/drivers/irqchip/irq-renesas-h8s.c index af8c6c61c824..71d8139be26c 100644 --- a/drivers/irqchip/irq-renesas-h8s.c +++ b/drivers/irqchip/irq-renesas-h8s.c @@ -73,7 +73,7 @@ static __init int irq_map(struct irq_domain *h, unsigned int virq, return 0; } -static struct irq_domain_ops irq_ops = { +static const struct irq_domain_ops irq_ops = { .map = irq_map, .xlate = irq_domain_xlate_onecell, }; From 3403b0259d152c3ca67b32bb710f826bc17c0d16 Mon Sep 17 00:00:00 2001 From: Robin Murphy Date: Wed, 31 May 2017 18:52:27 +0100 Subject: [PATCH 025/112] irqchip/gic-v3-its: Fix MSI alias accounting The call to pci_for_each_dma_alias() in the ITS PCI code has aroused suspicion in the past, and upon closer inspection does turn out to be completely backwards. Rather than iterating through each RID alias of the given device, what we actually want to be doing here is iterating through all the *other* devices which may also alias the same RID, in order to size the table for the worst case. Do the right thing by ignoring the initial DMA aliases themselves and just using that walk to detect an aliasing bridge, then walking back down the bus topology as necessary to actually count everything else. Our alias handling still isn't perfect, since we don't account for the cases of certain bridges only taking ownership of transactions under particular circumstances, but without completely reworking the ITS code to cope with the notion of multiple DevIDs per device, it'll have to do. Signed-off-by: Robin Murphy Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-gic-v3-its-pci-msi.c | 35 ++++++++++++------------ 1 file changed, 17 insertions(+), 18 deletions(-) diff --git a/drivers/irqchip/irq-gic-v3-its-pci-msi.c b/drivers/irqchip/irq-gic-v3-its-pci-msi.c index aee1c60d7ab5..77931214d954 100644 --- a/drivers/irqchip/irq-gic-v3-its-pci-msi.c +++ b/drivers/irqchip/irq-gic-v3-its-pci-msi.c @@ -41,27 +41,22 @@ static struct irq_chip its_msi_irq_chip = { .irq_write_msi_msg = pci_msi_domain_write_msg, }; -struct its_pci_alias { - struct pci_dev *pdev; - u32 count; -}; - -static int its_pci_msi_vec_count(struct pci_dev *pdev) +static int its_pci_msi_vec_count(struct pci_dev *pdev, void *data) { - int msi, msix; + int msi, msix, *count = data; msi = max(pci_msi_vec_count(pdev), 0); msix = max(pci_msix_vec_count(pdev), 0); + *count += max(msi, msix); - return max(msi, msix); + return 0; } static int its_get_pci_alias(struct pci_dev *pdev, u16 alias, void *data) { - struct its_pci_alias *dev_alias = data; + struct pci_dev **alias_dev = data; - if (pdev != dev_alias->pdev) - dev_alias->count += its_pci_msi_vec_count(pdev); + *alias_dev = pdev; return 0; } @@ -69,9 +64,9 @@ static int its_get_pci_alias(struct pci_dev *pdev, u16 alias, void *data) static int its_pci_msi_prepare(struct irq_domain *domain, struct device *dev, int nvec, msi_alloc_info_t *info) { - struct pci_dev *pdev; - struct its_pci_alias dev_alias; + struct pci_dev *pdev, *alias_dev; struct msi_domain_info *msi_info; + int alias_count = 0; if (!dev_is_pci(dev)) return -EINVAL; @@ -79,16 +74,20 @@ static int its_pci_msi_prepare(struct irq_domain *domain, struct device *dev, msi_info = msi_get_domain_info(domain->parent); pdev = to_pci_dev(dev); - dev_alias.pdev = pdev; - dev_alias.count = nvec; - - pci_for_each_dma_alias(pdev, its_get_pci_alias, &dev_alias); + /* + * If pdev is downstream of any aliasing bridges, take an upper + * bound of how many other vectors could map to the same DevID. + */ + pci_for_each_dma_alias(pdev, its_get_pci_alias, &alias_dev); + if (alias_dev != pdev && alias_dev->subordinate) + pci_walk_bus(alias_dev->subordinate, its_pci_msi_vec_count, + &alias_count); /* ITS specific DeviceID, as the core ITS ignores dev. */ info->scratchpad[0].ul = pci_msi_domain_get_msi_rid(domain, pdev); return msi_info->ops->msi_prepare(domain->parent, - dev, dev_alias.count, info); + dev, max(nvec, alias_count), info); } static struct msi_domain_ops its_pci_msi_ops = { From 8b8d94a72fd4e773320173efbc745ed8fe677ac5 Mon Sep 17 00:00:00 2001 From: MaJun Date: Thu, 18 May 2017 16:19:13 +0800 Subject: [PATCH 026/112] irqchip/gicv3-its: Skip irq affinity setting when target cpu is the same as current setting Just skip the irq affinity setting when the target cpu is the same as current setting. This is a small optimization for irq affinity setting logic. Signed-off-by: MaJun Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-gic-v3-its.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c index 45ea193325d2..b33528026c5c 100644 --- a/drivers/irqchip/irq-gic-v3-its.c +++ b/drivers/irqchip/irq-gic-v3-its.c @@ -644,9 +644,12 @@ static int its_set_affinity(struct irq_data *d, const struct cpumask *mask_val, if (cpu >= nr_cpu_ids) return -EINVAL; - target_col = &its_dev->its->collections[cpu]; - its_send_movi(its_dev, target_col, id); - its_dev->event_map.col_map[id] = cpu; + /* don't set the affinity when the target cpu is same as current one */ + if (cpu != its_dev->event_map.col_map[id]) { + target_col = &its_dev->its->collections[cpu]; + its_send_movi(its_dev, target_col, id); + its_dev->event_map.col_map[id] = cpu; + } return IRQ_SET_MASK_OK_DONE; } From 9a234c9c0a8de4245d5db788bcbbbaa35dcbe0a6 Mon Sep 17 00:00:00 2001 From: Thomas Petazzoni Date: Thu, 18 May 2017 10:07:37 +0200 Subject: [PATCH 027/112] irqchip/armada-370-xp: Re-order register definitions In order to clarify to which register base the various register definitions apply, this commit re-orders them, and adds a comment that clearly indicate which registers are relative to "main_int_base" and which registers are relative to "per_cpu_int_base". Acked-by: Gregory CLEMENT Signed-off-by: Thomas Petazzoni Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-armada-370-xp.c | 21 ++++++++++----------- 1 file changed, 10 insertions(+), 11 deletions(-) diff --git a/drivers/irqchip/irq-armada-370-xp.c b/drivers/irqchip/irq-armada-370-xp.c index eb0d4d41b156..76147dfdfccd 100644 --- a/drivers/irqchip/irq-armada-370-xp.c +++ b/drivers/irqchip/irq-armada-370-xp.c @@ -34,25 +34,24 @@ #include #include -/* Interrupt Controller Registers Map */ -#define ARMADA_370_XP_INT_SET_MASK_OFFS (0x48) -#define ARMADA_370_XP_INT_CLEAR_MASK_OFFS (0x4C) -#define ARMADA_370_XP_INT_FABRIC_MASK_OFFS (0x54) -#define ARMADA_370_XP_INT_CAUSE_PERF(cpu) (1 << cpu) - +/* Registers relative to main_int_base */ #define ARMADA_370_XP_INT_CONTROL (0x00) +#define ARMADA_370_XP_SW_TRIG_INT_OFFS (0x04) #define ARMADA_370_XP_INT_SET_ENABLE_OFFS (0x30) #define ARMADA_370_XP_INT_CLEAR_ENABLE_OFFS (0x34) #define ARMADA_370_XP_INT_SOURCE_CTL(irq) (0x100 + irq*4) #define ARMADA_370_XP_INT_SOURCE_CPU_MASK 0xF #define ARMADA_370_XP_INT_IRQ_FIQ_MASK(cpuid) ((BIT(0) | BIT(8)) << cpuid) -#define ARMADA_370_XP_CPU_INTACK_OFFS (0x44) +/* Registers relative to per_cpu_int_base */ +#define ARMADA_370_XP_IN_DRBEL_CAUSE_OFFS (0x08) +#define ARMADA_370_XP_IN_DRBEL_MSK_OFFS (0x0c) #define ARMADA_375_PPI_CAUSE (0x10) - -#define ARMADA_370_XP_SW_TRIG_INT_OFFS (0x4) -#define ARMADA_370_XP_IN_DRBEL_MSK_OFFS (0xc) -#define ARMADA_370_XP_IN_DRBEL_CAUSE_OFFS (0x8) +#define ARMADA_370_XP_CPU_INTACK_OFFS (0x44) +#define ARMADA_370_XP_INT_SET_MASK_OFFS (0x48) +#define ARMADA_370_XP_INT_CLEAR_MASK_OFFS (0x4C) +#define ARMADA_370_XP_INT_FABRIC_MASK_OFFS (0x54) +#define ARMADA_370_XP_INT_CAUSE_PERF(cpu) (1 << cpu) #define ARMADA_370_XP_MAX_PER_CPU_IRQS (28) From 054ea4ce0016bcb560a25e6e94de95538954a4a8 Mon Sep 17 00:00:00 2001 From: Thomas Petazzoni Date: Thu, 18 May 2017 10:07:38 +0200 Subject: [PATCH 028/112] irqchip/armada-370-xp: Document the overall driver logic Since the overall logic of the driver to handle the global and per-CPU masking of the interrupts is far from trivial, this commit adds a long comment detailing how the hardware operates and what strategy the driver implements on top of that. Acked-by: Gregory CLEMENT Signed-off-by: Thomas Petazzoni Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-armada-370-xp.c | 80 +++++++++++++++++++++++++++++ 1 file changed, 80 insertions(+) diff --git a/drivers/irqchip/irq-armada-370-xp.c b/drivers/irqchip/irq-armada-370-xp.c index 76147dfdfccd..1f7dea613d15 100644 --- a/drivers/irqchip/irq-armada-370-xp.c +++ b/drivers/irqchip/irq-armada-370-xp.c @@ -34,6 +34,86 @@ #include #include +/* + * Overall diagram of the Armada XP interrupt controller: + * + * To CPU 0 To CPU 1 + * + * /\ /\ + * || || + * +---------------+ +---------------+ + * | | | | + * | per-CPU | | per-CPU | + * | mask/unmask | | mask/unmask | + * | CPU0 | | CPU1 | + * | | | | + * +---------------+ +---------------+ + * /\ /\ + * || || + * \\_______________________// + * || + * +-------------------+ + * | | + * | Global interrupt | + * | mask/unmask | + * | | + * +-------------------+ + * /\ + * || + * interrupt from + * device + * + * The "global interrupt mask/unmask" is modified using the + * ARMADA_370_XP_INT_SET_ENABLE_OFFS and + * ARMADA_370_XP_INT_CLEAR_ENABLE_OFFS registers, which are relative + * to "main_int_base". + * + * The "per-CPU mask/unmask" is modified using the + * ARMADA_370_XP_INT_SET_MASK_OFFS and + * ARMADA_370_XP_INT_CLEAR_MASK_OFFS registers, which are relative to + * "per_cpu_int_base". This base address points to a special address, + * which automatically accesses the registers of the current CPU. + * + * The per-CPU mask/unmask can also be adjusted using the global + * per-interrupt ARMADA_370_XP_INT_SOURCE_CTL register, which we use + * to configure interrupt affinity. + * + * Due to this model, all interrupts need to be mask/unmasked at two + * different levels: at the global level and at the per-CPU level. + * + * This driver takes the following approach to deal with this: + * + * - For global interrupts: + * + * At ->map() time, a global interrupt is unmasked at the per-CPU + * mask/unmask level. It is therefore unmasked at this level for + * the current CPU, running the ->map() code. This allows to have + * the interrupt unmasked at this level in non-SMP + * configurations. In SMP configurations, the ->set_affinity() + * callback is called, which using the + * ARMADA_370_XP_INT_SOURCE_CTL() readjusts the per-CPU mask/unmask + * for the interrupt. + * + * The ->mask() and ->unmask() operations only mask/unmask the + * interrupt at the "global" level. + * + * So, a global interrupt is enabled at the per-CPU level as soon + * as it is mapped. At run time, the masking/unmasking takes place + * at the global level. + * + * - For per-CPU interrupts + * + * At ->map() time, a per-CPU interrupt is unmasked at the global + * mask/unmask level. + * + * The ->mask() and ->unmask() operations mask/unmask the interrupt + * at the per-CPU level. + * + * So, a per-CPU interrupt is enabled at the global level as soon + * as it is mapped. At run time, the masking/unmasking takes place + * at the per-CPU level. + */ + /* Registers relative to main_int_base */ #define ARMADA_370_XP_INT_CONTROL (0x00) #define ARMADA_370_XP_SW_TRIG_INT_OFFS (0x04) From 0fa4ce746d1d8c8aa3d09fbc675497fa4c4a5475 Mon Sep 17 00:00:00 2001 From: Thomas Petazzoni Date: Thu, 18 May 2017 10:07:39 +0200 Subject: [PATCH 029/112] irqchip/armada-370-xp: Re-enable per-CPU interrupts at resume time Commit d17cab4451df1 ("irqchip: Kill off set_irq_flags usage") changed the code of armada_370_xp_mpic_irq_map() from using set_irq_flags() to irq_set_probe(). While the commit log seems to imply that there are no functional changes, there are indeed functional changes introduced by this commit: the IRQ_NOAUTOEN flag is no longer cleared. This functional change caused a regression on Armada XP, which no longer works properly after suspend/resume because per-CPU interrupts remain disabled. This regression was temporarly worked around in commit 353d6d6c82e5d ("irqchip/armada-370-xp: Fix regression by clearing IRQ_NOAUTOEN"), but it is not the most satisfying solution. This commit implements the solution that was initially discussed with Thomas Gleixner. Due to how the hardware registers work, the irq-armada-370-xp cannot simply save/restore a bunch of registers at suspend/resume to make sure that the interrupts remain in the same state after resuming. Therefore, it relies on the kernel to say whether the interrupt is disabled or not, using the irqd_irq_disabled() function. This was all working fine while the IRQ_NOAUTOEN flag was cleared. With the change introduced by Rob Herring in d17cab4451df1, the IRQ_NOAUTOEN flag is now set for all interrupts. irqd_irq_disabled() returns false for per-CPU interrupts, and therefore our per-CPU interrupts are no longer re-enabled after resume. This commit fixes that by using irqd_irq_disabled() only for global interrupts, and using the newly introduced irq_percpu_is_enabled() for per-CPU interrupts. Also, it fixes a related problems that per-CPU interrupts were only re-enabled on the boot CPU and not other CPUs. Until now this wasn't a problem since on this platform, only the local timers are using per-CPU interrupts and the local timers of secondary CPUs are turned off/on during CPU hotplug before suspend, after after resume. However, since Linux 4.4, we are also be using per-CPU interrupts for the network controller, so we need to properly restore the per-CPU interrupts on secondary CPUs as well. Signed-off-by: Thomas Petazzoni Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-armada-370-xp.c | 46 +++++++++++++++++++++++++---- 1 file changed, 40 insertions(+), 6 deletions(-) diff --git a/drivers/irqchip/irq-armada-370-xp.c b/drivers/irqchip/irq-armada-370-xp.c index 1f7dea613d15..1d4de6e73d5d 100644 --- a/drivers/irqchip/irq-armada-370-xp.c +++ b/drivers/irqchip/irq-armada-370-xp.c @@ -360,7 +360,6 @@ static int armada_370_xp_mpic_irq_map(struct irq_domain *h, irq_set_percpu_devid(virq); irq_set_chip_and_handler(virq, &armada_370_xp_irq_chip, handle_percpu_devid_irq); - } else { irq_set_chip_and_handler(virq, &armada_370_xp_irq_chip, handle_level_irq); @@ -424,16 +423,40 @@ static void armada_mpic_send_doorbell(const struct cpumask *mask, ARMADA_370_XP_SW_TRIG_INT_OFFS); } +static void armada_xp_mpic_reenable_percpu(void) +{ + unsigned int irq; + + /* Re-enable per-CPU interrupts that were enabled before suspend */ + for (irq = 0; irq < ARMADA_370_XP_MAX_PER_CPU_IRQS; irq++) { + struct irq_data *data; + int virq; + + virq = irq_linear_revmap(armada_370_xp_mpic_domain, irq); + if (virq == 0) + continue; + + data = irq_get_irq_data(virq); + + if (!irq_percpu_is_enabled(virq)) + continue; + + armada_370_xp_irq_unmask(data); + } +} + static int armada_xp_mpic_starting_cpu(unsigned int cpu) { armada_xp_mpic_perf_init(); armada_xp_mpic_smp_cpu_init(); + armada_xp_mpic_reenable_percpu(); return 0; } static int mpic_cascaded_starting_cpu(unsigned int cpu) { armada_xp_mpic_perf_init(); + armada_xp_mpic_reenable_percpu(); enable_percpu_irq(parent_irq, IRQ_TYPE_NONE); return 0; } @@ -581,16 +604,27 @@ static void armada_370_xp_mpic_resume(void) if (virq == 0) continue; - if (!is_percpu_irq(irq)) + data = irq_get_irq_data(virq); + + if (!is_percpu_irq(irq)) { + /* Non per-CPU interrupts */ writel(irq, per_cpu_int_base + ARMADA_370_XP_INT_CLEAR_MASK_OFFS); - else + if (!irqd_irq_disabled(data)) + armada_370_xp_irq_unmask(data); + } else { + /* Per-CPU interrupts */ writel(irq, main_int_base + ARMADA_370_XP_INT_SET_ENABLE_OFFS); - data = irq_get_irq_data(virq); - if (!irqd_irq_disabled(data)) - armada_370_xp_irq_unmask(data); + /* + * Re-enable on the current CPU, + * armada_xp_mpic_reenable_percpu() will take + * care of secondary CPUs when they come up. + */ + if (irq_percpu_is_enabled(virq)) + armada_370_xp_irq_unmask(data); + } } /* Reconfigure doorbells for IPIs and MSIs */ From 1957b7bfa65893ee282e9901b2132d87c2720e07 Mon Sep 17 00:00:00 2001 From: Thomas Petazzoni Date: Thu, 18 May 2017 10:07:40 +0200 Subject: [PATCH 030/112] Revert "irqchip/armada-370-xp: Fix regression by clearing IRQ_NOAUTOEN" This reverts commit 353d6d6c82e5d2533ba22e7f9fb081582bf50dc2, which is no longer needed, now that the irq-armada-370-xp driver properly re-enables per-CPU interrupt on both the boot CPU and secondary CPUs after resume. Signed-off-by: Thomas Petazzoni Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-armada-370-xp.c | 1 - 1 file changed, 1 deletion(-) diff --git a/drivers/irqchip/irq-armada-370-xp.c b/drivers/irqchip/irq-armada-370-xp.c index 1d4de6e73d5d..577b42831de8 100644 --- a/drivers/irqchip/irq-armada-370-xp.c +++ b/drivers/irqchip/irq-armada-370-xp.c @@ -365,7 +365,6 @@ static int armada_370_xp_mpic_irq_map(struct irq_domain *h, handle_level_irq); } irq_set_probe(virq); - irq_clear_status_flags(virq, IRQ_NOAUTOEN); return 0; } From 13dcf12bf877b9ed8845c7a68390702e368ec622 Mon Sep 17 00:00:00 2001 From: Tobias Klauser Date: Fri, 2 Jun 2017 10:20:52 +0200 Subject: [PATCH 031/112] irqchip/aspeed-vic: Constify irq_domain_ops struct irq_domain_ops is not modified, so it can be made const. Cc: Joel Stanley Signed-off-by: Tobias Klauser Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-aspeed-vic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/irqchip/irq-aspeed-vic.c b/drivers/irqchip/irq-aspeed-vic.c index d24451d5bf8a..88e139e4c1f9 100644 --- a/drivers/irqchip/irq-aspeed-vic.c +++ b/drivers/irqchip/irq-aspeed-vic.c @@ -186,7 +186,7 @@ static int avic_map(struct irq_domain *d, unsigned int irq, return 0; } -static struct irq_domain_ops avic_dom_ops = { +static const struct irq_domain_ops avic_dom_ops = { .map = avic_map, .xlate = irq_domain_xlate_onetwocell, }; From fa776a3f2b1b62970179bcefc55fae85e5700746 Mon Sep 17 00:00:00 2001 From: Andrew Jeffery Date: Tue, 16 May 2017 15:57:47 +0800 Subject: [PATCH 032/112] irqchip/aspeed-vic: Add AST2500 compatible string In addition to introducing the new compatible string the bindings description is reworked to be more generic. Acked-by: Rob Herring Signed-off-by: Andrew Jeffery Signed-off-by: Marc Zyngier --- .../bindings/interrupt-controller/aspeed,ast2400-vic.txt | 9 +++++---- drivers/irqchip/irq-aspeed-vic.c | 3 ++- 2 files changed, 7 insertions(+), 5 deletions(-) diff --git a/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt b/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt index 6c6e85324b9d..e3fea0758d25 100644 --- a/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt +++ b/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-vic.txt @@ -1,12 +1,13 @@ Aspeed Vectored Interrupt Controller -These bindings are for the Aspeed AST2400 interrupt controller register layout. -The SoC has an legacy register layout, but this driver does not support that -mode of operation. +These bindings are for the Aspeed interrupt controller. The AST2400 and +AST2500 SoC families include a legacy register layout before a re-designed +layout, but the bindings do not prescribe the use of one or the other. Required properties: -- compatible : should be "aspeed,ast2400-vic". +- compatible : "aspeed,ast2400-vic" + "aspeed,ast2500-vic" - interrupt-controller : Identifies the node as an interrupt controller - #interrupt-cells : Specifies the number of cells needed to encode an diff --git a/drivers/irqchip/irq-aspeed-vic.c b/drivers/irqchip/irq-aspeed-vic.c index 88e139e4c1f9..03ba477ea0d0 100644 --- a/drivers/irqchip/irq-aspeed-vic.c +++ b/drivers/irqchip/irq-aspeed-vic.c @@ -227,4 +227,5 @@ static int __init avic_of_init(struct device_node *node, return 0; } -IRQCHIP_DECLARE(aspeed_new_vic, "aspeed,ast2400-vic", avic_of_init); +IRQCHIP_DECLARE(ast2400_vic, "aspeed,ast2400-vic", avic_of_init); +IRQCHIP_DECLARE(ast2500_vic, "aspeed,ast2500-vic", avic_of_init); From 0a56f9eebe6320980b68c60c852436cbf2a14b61 Mon Sep 17 00:00:00 2001 From: Brendan Higgins Date: Fri, 2 Jun 2017 18:29:48 -0700 Subject: [PATCH 033/112] irqchip/aspeed-i2c-ic: Add binding docs for Aspeed I2C Interrupt Controller Added device tree binding documentation for Aspeed I2C Interrupt Controller. Signed-off-by: Brendan Higgins Acked-by: Rob Herring Signed-off-by: Marc Zyngier --- .../aspeed,ast2400-i2c-ic.txt | 25 +++++++++++++++++++ 1 file changed, 25 insertions(+) create mode 100644 Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-i2c-ic.txt diff --git a/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-i2c-ic.txt b/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-i2c-ic.txt new file mode 100644 index 000000000000..033cc82e5684 --- /dev/null +++ b/Documentation/devicetree/bindings/interrupt-controller/aspeed,ast2400-i2c-ic.txt @@ -0,0 +1,25 @@ +Device tree configuration for the I2C Interrupt Controller on the AST24XX and +AST25XX SoCs. + +Required Properties: +- #address-cells : should be 1 +- #size-cells : should be 1 +- #interrupt-cells : should be 1 +- compatible : should be "aspeed,ast2400-i2c-ic" + or "aspeed,ast2500-i2c-ic" +- reg : address start and range of controller +- interrupts : interrupt number +- interrupt-controller : denotes that the controller receives and fires + new interrupts for child busses + +Example: + +i2c_ic: interrupt-controller@0 { + #address-cells = <1>; + #size-cells = <1>; + #interrupt-cells = <1>; + compatible = "aspeed,ast2400-i2c-ic"; + reg = <0x0 0x40>; + interrupts = <12>; + interrupt-controller; +}; From f48e699ddf7056f83bb8e2dbe3c2ae8d1ff1a31a Mon Sep 17 00:00:00 2001 From: Brendan Higgins Date: Fri, 2 Jun 2017 18:29:49 -0700 Subject: [PATCH 034/112] irqchip/aspeed-i2c-ic: Add I2C IRQ controller for Aspeed The Aspeed 24XX/25XX chips share a single hardware interrupt across 14 separate I2C busses. This adds a dummy irqchip which maps the single hardware interrupt to software interrupts for each of the busses. Signed-off-by: Brendan Higgins Signed-off-by: Marc Zyngier --- drivers/irqchip/Makefile | 2 +- drivers/irqchip/irq-aspeed-i2c-ic.c | 115 ++++++++++++++++++++++++++++ 2 files changed, 116 insertions(+), 1 deletion(-) create mode 100644 drivers/irqchip/irq-aspeed-i2c-ic.c diff --git a/drivers/irqchip/Makefile b/drivers/irqchip/Makefile index b64c59b838a0..e067f9839b33 100644 --- a/drivers/irqchip/Makefile +++ b/drivers/irqchip/Makefile @@ -73,6 +73,6 @@ obj-$(CONFIG_MVEBU_ODMI) += irq-mvebu-odmi.o obj-$(CONFIG_MVEBU_PIC) += irq-mvebu-pic.o obj-$(CONFIG_LS_SCFG_MSI) += irq-ls-scfg-msi.o obj-$(CONFIG_EZNPS_GIC) += irq-eznps.o -obj-$(CONFIG_ARCH_ASPEED) += irq-aspeed-vic.o +obj-$(CONFIG_ARCH_ASPEED) += irq-aspeed-vic.o irq-aspeed-i2c-ic.o obj-$(CONFIG_STM32_EXTI) += irq-stm32-exti.o obj-$(CONFIG_QCOM_IRQ_COMBINER) += qcom-irq-combiner.o diff --git a/drivers/irqchip/irq-aspeed-i2c-ic.c b/drivers/irqchip/irq-aspeed-i2c-ic.c new file mode 100644 index 000000000000..815b88dd18f2 --- /dev/null +++ b/drivers/irqchip/irq-aspeed-i2c-ic.c @@ -0,0 +1,115 @@ +/* + * Aspeed 24XX/25XX I2C Interrupt Controller. + * + * Copyright (C) 2012-2017 ASPEED Technology Inc. + * Copyright 2017 IBM Corporation + * Copyright 2017 Google, Inc. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#include +#include +#include +#include +#include +#include +#include + + +#define ASPEED_I2C_IC_NUM_BUS 14 + +struct aspeed_i2c_ic { + void __iomem *base; + int parent_irq; + struct irq_domain *irq_domain; +}; + +/* + * The aspeed chip provides a single hardware interrupt for all of the I2C + * busses, so we use a dummy interrupt chip to translate this single interrupt + * into multiple interrupts, each associated with a single I2C bus. + */ +static void aspeed_i2c_ic_irq_handler(struct irq_desc *desc) +{ + struct aspeed_i2c_ic *i2c_ic = irq_desc_get_handler_data(desc); + struct irq_chip *chip = irq_desc_get_chip(desc); + unsigned long bit, status; + unsigned int bus_irq; + + chained_irq_enter(chip, desc); + status = readl(i2c_ic->base); + for_each_set_bit(bit, &status, ASPEED_I2C_IC_NUM_BUS) { + bus_irq = irq_find_mapping(i2c_ic->irq_domain, bit); + generic_handle_irq(bus_irq); + } + chained_irq_exit(chip, desc); +} + +/* + * Set simple handler and mark IRQ as valid. Nothing interesting to do here + * since we are using a dummy interrupt chip. + */ +static int aspeed_i2c_ic_map_irq_domain(struct irq_domain *domain, + unsigned int irq, irq_hw_number_t hwirq) +{ + irq_set_chip_and_handler(irq, &dummy_irq_chip, handle_simple_irq); + irq_set_chip_data(irq, domain->host_data); + + return 0; +} + +static const struct irq_domain_ops aspeed_i2c_ic_irq_domain_ops = { + .map = aspeed_i2c_ic_map_irq_domain, +}; + +static int __init aspeed_i2c_ic_of_init(struct device_node *node, + struct device_node *parent) +{ + struct aspeed_i2c_ic *i2c_ic; + int ret = 0; + + i2c_ic = kzalloc(sizeof(*i2c_ic), GFP_KERNEL); + if (!i2c_ic) + return -ENOMEM; + + i2c_ic->base = of_iomap(node, 0); + if (IS_ERR(i2c_ic->base)) { + ret = PTR_ERR(i2c_ic->base); + goto err_free_ic; + } + + i2c_ic->parent_irq = irq_of_parse_and_map(node, 0); + if (i2c_ic->parent_irq < 0) { + ret = i2c_ic->parent_irq; + goto err_iounmap; + } + + i2c_ic->irq_domain = irq_domain_add_linear(node, ASPEED_I2C_IC_NUM_BUS, + &aspeed_i2c_ic_irq_domain_ops, + NULL); + if (!i2c_ic->irq_domain) { + ret = -ENOMEM; + goto err_iounmap; + } + + i2c_ic->irq_domain->name = "aspeed-i2c-domain"; + + irq_set_chained_handler_and_data(i2c_ic->parent_irq, + aspeed_i2c_ic_irq_handler, i2c_ic); + + pr_info("i2c controller registered, irq %d\n", i2c_ic->parent_irq); + + return 0; + +err_iounmap: + iounmap(i2c_ic->base); +err_free_ic: + kfree(i2c_ic); + return ret; +} + +IRQCHIP_DECLARE(ast2400_i2c_ic, "aspeed,ast2400-i2c-ic", aspeed_i2c_ic_of_init); +IRQCHIP_DECLARE(ast2500_i2c_ic, "aspeed,ast2500-i2c-ic", aspeed_i2c_ic_of_init); From a1628366227aff3433f646cec2beb36d1f018f86 Mon Sep 17 00:00:00 2001 From: Wei Yongjun Date: Tue, 25 Apr 2017 16:12:23 +0000 Subject: [PATCH 035/112] irqchip/qcom: Use builtin_platform_driver to simplify the code Use the builtin_platform_driver() macro to make the code simpler. Signed-off-by: Wei Yongjun Signed-off-by: Marc Zyngier --- drivers/irqchip/qcom-irq-combiner.c | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-) diff --git a/drivers/irqchip/qcom-irq-combiner.c b/drivers/irqchip/qcom-irq-combiner.c index 226558698344..6aa3ea479214 100644 --- a/drivers/irqchip/qcom-irq-combiner.c +++ b/drivers/irqchip/qcom-irq-combiner.c @@ -288,9 +288,4 @@ static struct platform_driver qcom_irq_combiner_probe = { }, .probe = combiner_probe, }; - -static int __init register_qcom_irq_combiner(void) -{ - return platform_driver_register(&qcom_irq_combiner_probe); -} -device_initcall(register_qcom_irq_combiner); +builtin_platform_driver(qcom_irq_combiner_probe); From 11f69da0dada3e4472b5b1319fae9b1df194e3d6 Mon Sep 17 00:00:00 2001 From: Thomas Petazzoni Date: Wed, 21 Jun 2017 15:29:12 +0200 Subject: [PATCH 036/112] dt-bindings/interrupt-controller: Add DT binding for the Marvell GICP This commit adds the Device Tree binding documentation for the Marvell GICP, an extension to the GIC that allows to trigger GIC SPI interrupts using memory transactions. It is used by the ICU unit in the Marvell CP110 block to turn wired interrupts inside the CP into SPI interrupts at the GIC level in the AP. Signed-off-by: Thomas Petazzoni Signed-off-by: Marc Zyngier --- .../interrupt-controller/marvell,gicp.txt | 27 +++++++++++++++++++ 1 file changed, 27 insertions(+) create mode 100644 Documentation/devicetree/bindings/interrupt-controller/marvell,gicp.txt diff --git a/Documentation/devicetree/bindings/interrupt-controller/marvell,gicp.txt b/Documentation/devicetree/bindings/interrupt-controller/marvell,gicp.txt new file mode 100644 index 000000000000..64a00ceb7da4 --- /dev/null +++ b/Documentation/devicetree/bindings/interrupt-controller/marvell,gicp.txt @@ -0,0 +1,27 @@ +Marvell GICP Controller +----------------------- + +GICP is a Marvell extension of the GIC that allows to trigger GIC SPI +interrupts by doing a memory transaction. It is used by the ICU +located in the Marvell CP110 to turn wired interrupts inside the CP +into GIC SPI interrupts. + +Required properties: + +- compatible: Must be "marvell,ap806-gicp" + +- reg: Must be the address and size of the GICP SPI registers + +- marvell,spi-ranges: tuples of GIC SPI interrupts ranges available + for this GICP + +- msi-controller: indicates that this is an MSI controller + +Example: + +gicp_spi: gicp-spi@3f0040 { + compatible = "marvell,ap806-gicp"; + reg = <0x3f0040 0x10>; + marvell,spi-ranges = <64 64>, <288 64>; + msi-controller; +}; From 8947dfb257eb91d7487e06b7d2a069d82e7c19a2 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:01 +0200 Subject: [PATCH 037/112] x86/apic: Add name to irq chip Add the missing name, so debugging will work proper. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235443.266561988@linutronix.de Signed-off-by: Thomas Gleixner --- arch/x86/kernel/apic/vector.c | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/x86/kernel/apic/vector.c b/arch/x86/kernel/apic/vector.c index f3557a1eb562..6b21b9eeba2f 100644 --- a/arch/x86/kernel/apic/vector.c +++ b/arch/x86/kernel/apic/vector.c @@ -534,6 +534,7 @@ static int apic_set_affinity(struct irq_data *irq_data, } static struct irq_chip lapic_controller = { + .name = "APIC", .irq_ack = apic_ack_edge, .irq_set_affinity = apic_set_affinity, .irq_retrigger = apic_retrigger_irq, From 290be194ba9d489e1857cc45d0dd24bf3429156b Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:02 +0200 Subject: [PATCH 038/112] iommu/amd: Add name to irq chip Add the missing name, so debugging will work proper. Signed-off-by: Thomas Gleixner Acked-by: Joerg Roedel Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: iommu@lists.linux-foundation.org Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235443.343236995@linutronix.de --- drivers/iommu/amd_iommu.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/drivers/iommu/amd_iommu.c b/drivers/iommu/amd_iommu.c index 63cacf5d6cf2..590e1e8288a1 100644 --- a/drivers/iommu/amd_iommu.c +++ b/drivers/iommu/amd_iommu.c @@ -4386,10 +4386,11 @@ static void ir_compose_msi_msg(struct irq_data *irq_data, struct msi_msg *msg) } static struct irq_chip amd_ir_chip = { - .irq_ack = ir_ack_apic_edge, - .irq_set_affinity = amd_ir_set_affinity, - .irq_set_vcpu_affinity = amd_ir_set_vcpu_affinity, - .irq_compose_msi_msg = ir_compose_msi_msg, + .name = "AMD-IR", + .irq_ack = ir_ack_apic_edge, + .irq_set_affinity = amd_ir_set_affinity, + .irq_set_vcpu_affinity = amd_ir_set_vcpu_affinity, + .irq_compose_msi_msg = ir_compose_msi_msg, }; int amd_iommu_create_irq_domain(struct amd_iommu *iommu) From 1bb3a5a76386ba2886ee44b903eeff5765bd71d4 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:03 +0200 Subject: [PATCH 039/112] iommu/vt-d: Add name to irq chip Add the missing name, so debugging will work proper. Signed-off-by: Thomas Gleixner Acked-by: Joerg Roedel Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: iommu@lists.linux-foundation.org Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235443.431939968@linutronix.de --- drivers/iommu/intel_irq_remapping.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/drivers/iommu/intel_irq_remapping.c b/drivers/iommu/intel_irq_remapping.c index a190cbd76ef7..ba5b5800e249 100644 --- a/drivers/iommu/intel_irq_remapping.c +++ b/drivers/iommu/intel_irq_remapping.c @@ -1205,10 +1205,11 @@ static int intel_ir_set_vcpu_affinity(struct irq_data *data, void *info) } static struct irq_chip intel_ir_chip = { - .irq_ack = ir_ack_apic_edge, - .irq_set_affinity = intel_ir_set_affinity, - .irq_compose_msi_msg = intel_ir_compose_msi_msg, - .irq_set_vcpu_affinity = intel_ir_set_vcpu_affinity, + .name = "INTEL-IR", + .irq_ack = ir_ack_apic_edge, + .irq_set_affinity = intel_ir_set_affinity, + .irq_compose_msi_msg = intel_ir_compose_msi_msg, + .irq_set_vcpu_affinity = intel_ir_set_vcpu_affinity, }; static void intel_irq_remapping_prepare_irte(struct intel_ir_data *data, From 0165308a2f994939d2e1b36624f5a8f57746bc88 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:04 +0200 Subject: [PATCH 040/112] genirq/msi: Prevent overwriting domain name Prevent overwriting an already assigned domain name. Remove the extra check for chip->name, because if domain->name is NULL overwriting it with NULL is not a problem. Signed-off-by: Thomas Gleixner Acked-by: Marc Zyngier Cc: Jens Axboe Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235443.510684976@linutronix.de --- kernel/irq/msi.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/kernel/irq/msi.c b/kernel/irq/msi.c index fe4d48ec5bc4..9e3f1857c6bd 100644 --- a/kernel/irq/msi.c +++ b/kernel/irq/msi.c @@ -274,7 +274,8 @@ struct irq_domain *msi_create_irq_domain(struct fwnode_handle *fwnode, domain = irq_domain_create_hierarchy(parent, IRQ_DOMAIN_FLAG_MSI, 0, fwnode, &msi_domain_ops, info); - if (domain && info->chip && info->chip->name) + + if (domain && !domain->name && info->chip) domain->name = info->chip->name; return domain; From d59f6617eef0f76e34f7a9993f5645c5ef467e42 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:05 +0200 Subject: [PATCH 041/112] genirq: Allow fwnode to carry name information only In order to provide proper debug interface it's required to have domain names available when the domain is added. Non fwnode based architectures like x86 have no way to do so. It's not possible to use domain ops or host data for this as domain ops might be the same for several instances, but the names have to be unique. Extend the irqchip fwnode to allow transporting the domain name. If no node is supplied, create a 'unknown-N' placeholder. Warn if an invalid node is supplied and treat it like no node. This happens e.g. with i2 devices on x86 which hand in an ACPI type node which has no interface for retrieving the name. [ Folded a fix from Marc to make DT name parsing work ] Signed-off-by: Thomas Gleixner Acked-by: Marc Zyngier Cc: Jens Axboe Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235443.588784933@linutronix.de --- include/linux/irqdomain.h | 31 ++++++++++- kernel/irq/irqdomain.c | 105 +++++++++++++++++++++++++++++++++----- 2 files changed, 122 insertions(+), 14 deletions(-) diff --git a/include/linux/irqdomain.h b/include/linux/irqdomain.h index 9f3616085423..9cf32a2fbe69 100644 --- a/include/linux/irqdomain.h +++ b/include/linux/irqdomain.h @@ -189,6 +189,9 @@ enum { /* Irq domain implements MSI remapping */ IRQ_DOMAIN_FLAG_MSI_REMAP = (1 << 5), + /* Irq domain name was allocated in __irq_domain_add() */ + IRQ_DOMAIN_NAME_ALLOCATED = (1 << 6), + /* * Flags starting from IRQ_DOMAIN_FLAG_NONCORE are reserved * for implementation specific purposes and ignored by the @@ -203,7 +206,33 @@ static inline struct device_node *irq_domain_get_of_node(struct irq_domain *d) } #ifdef CONFIG_IRQ_DOMAIN -struct fwnode_handle *irq_domain_alloc_fwnode(void *data); +struct fwnode_handle *__irq_domain_alloc_fwnode(unsigned int type, int id, + const char *name, void *data); + +enum { + IRQCHIP_FWNODE_REAL, + IRQCHIP_FWNODE_NAMED, + IRQCHIP_FWNODE_NAMED_ID, +}; + +static inline +struct fwnode_handle *irq_domain_alloc_named_fwnode(const char *name) +{ + return __irq_domain_alloc_fwnode(IRQCHIP_FWNODE_NAMED, 0, name, NULL); +} + +static inline +struct fwnode_handle *irq_domain_alloc_named_id_fwnode(const char *name, int id) +{ + return __irq_domain_alloc_fwnode(IRQCHIP_FWNODE_NAMED_ID, id, name, + NULL); +} + +static inline struct fwnode_handle *irq_domain_alloc_fwnode(void *data) +{ + return __irq_domain_alloc_fwnode(IRQCHIP_FWNODE_REAL, 0, NULL, data); +} + void irq_domain_free_fwnode(struct fwnode_handle *fwnode); struct irq_domain *__irq_domain_add(struct fwnode_handle *fwnode, int size, irq_hw_number_t hwirq_max, int direct_max, diff --git a/kernel/irq/irqdomain.c b/kernel/irq/irqdomain.c index 70b9da72018b..e1b925bea205 100644 --- a/kernel/irq/irqdomain.c +++ b/kernel/irq/irqdomain.c @@ -26,39 +26,61 @@ static struct irq_domain *irq_default_domain; static void irq_domain_check_hierarchy(struct irq_domain *domain); struct irqchip_fwid { - struct fwnode_handle fwnode; - char *name; - void *data; + struct fwnode_handle fwnode; + unsigned int type; + char *name; + void *data; }; /** * irq_domain_alloc_fwnode - Allocate a fwnode_handle suitable for * identifying an irq domain - * @data: optional user-provided data + * @type: Type of irqchip_fwnode. See linux/irqdomain.h + * @name: Optional user provided domain name + * @id: Optional user provided id if name != NULL + * @data: Optional user-provided data * - * Allocate a struct device_node, and return a poiner to the embedded + * Allocate a struct irqchip_fwid, and return a poiner to the embedded * fwnode_handle (or NULL on failure). + * + * Note: The types IRQCHIP_FWNODE_NAMED and IRQCHIP_FWNODE_NAMED_ID are + * solely to transport name information to irqdomain creation code. The + * node is not stored. For other types the pointer is kept in the irq + * domain struct. */ -struct fwnode_handle *irq_domain_alloc_fwnode(void *data) +struct fwnode_handle *__irq_domain_alloc_fwnode(unsigned int type, int id, + const char *name, void *data) { struct irqchip_fwid *fwid; - char *name; + char *n; fwid = kzalloc(sizeof(*fwid), GFP_KERNEL); - name = kasprintf(GFP_KERNEL, "irqchip@%p", data); - if (!fwid || !name) { + switch (type) { + case IRQCHIP_FWNODE_NAMED: + n = kasprintf(GFP_KERNEL, "%s", name); + break; + case IRQCHIP_FWNODE_NAMED_ID: + n = kasprintf(GFP_KERNEL, "%s-%d", name, id); + break; + default: + n = kasprintf(GFP_KERNEL, "irqchip@%p", data); + break; + } + + if (!fwid || !n) { kfree(fwid); - kfree(name); + kfree(n); return NULL; } - fwid->name = name; + fwid->type = type; + fwid->name = n; fwid->data = data; fwid->fwnode.type = FWNODE_IRQCHIP; return &fwid->fwnode; } -EXPORT_SYMBOL_GPL(irq_domain_alloc_fwnode); +EXPORT_SYMBOL_GPL(__irq_domain_alloc_fwnode); /** * irq_domain_free_fwnode - Free a non-OF-backed fwnode_handle @@ -97,20 +119,75 @@ struct irq_domain *__irq_domain_add(struct fwnode_handle *fwnode, int size, void *host_data) { struct device_node *of_node = to_of_node(fwnode); + struct irqchip_fwid *fwid; struct irq_domain *domain; + static atomic_t unknown_domains; + domain = kzalloc_node(sizeof(*domain) + (sizeof(unsigned int) * size), GFP_KERNEL, of_node_to_nid(of_node)); if (WARN_ON(!domain)) return NULL; + if (fwnode && is_fwnode_irqchip(fwnode)) { + fwid = container_of(fwnode, struct irqchip_fwid, fwnode); + + switch (fwid->type) { + case IRQCHIP_FWNODE_NAMED: + case IRQCHIP_FWNODE_NAMED_ID: + domain->name = kstrdup(fwid->name, GFP_KERNEL); + if (!domain->name) { + kfree(domain); + return NULL; + } + domain->flags |= IRQ_DOMAIN_NAME_ALLOCATED; + break; + default: + domain->fwnode = fwnode; + domain->name = fwid->name; + break; + } + } else if (of_node) { + char *name; + + /* + * DT paths contain '/', which debugfs is legitimately + * unhappy about. Replace them with ':', which does + * the trick and is not as offensive as '\'... + */ + name = kstrdup(of_node_full_name(of_node), GFP_KERNEL); + if (!name) { + kfree(domain); + return NULL; + } + + strreplace(name, '/', ':'); + + domain->name = name; + domain->fwnode = fwnode; + domain->flags |= IRQ_DOMAIN_NAME_ALLOCATED; + } + + if (!domain->name) { + if (fwnode) { + pr_err("Invalid fwnode type (%d) for irqdomain\n", + fwnode->type); + } + domain->name = kasprintf(GFP_KERNEL, "unknown-%d", + atomic_inc_return(&unknown_domains)); + if (!domain->name) { + kfree(domain); + return NULL; + } + domain->flags |= IRQ_DOMAIN_NAME_ALLOCATED; + } + of_node_get(of_node); /* Fill structure */ INIT_RADIX_TREE(&domain->revmap_tree, GFP_KERNEL); domain->ops = ops; domain->host_data = host_data; - domain->fwnode = fwnode; domain->hwirq_max = hwirq_max; domain->revmap_size = size; domain->revmap_direct_max_irq = direct_max; @@ -152,6 +229,8 @@ void irq_domain_remove(struct irq_domain *domain) pr_debug("Removed domain %s\n", domain->name); of_node_put(irq_domain_get_of_node(domain)); + if (domain->flags & IRQ_DOMAIN_NAME_ALLOCATED) + kfree(domain->name); kfree(domain); } EXPORT_SYMBOL_GPL(irq_domain_remove); From 9d35f859590efa48be51b8ccded6550e0440e2c7 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:06 +0200 Subject: [PATCH 042/112] x86/vector: Create named irq domain Use the fwnode to create a named domain so diagnosis works. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235443.673635238@linutronix.de Signed-off-by: Thomas Gleixner --- arch/x86/kernel/apic/vector.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/arch/x86/kernel/apic/vector.c b/arch/x86/kernel/apic/vector.c index 6b21b9eeba2f..47c5d019fb7e 100644 --- a/arch/x86/kernel/apic/vector.c +++ b/arch/x86/kernel/apic/vector.c @@ -429,11 +429,16 @@ static void init_legacy_irqs(void) { } int __init arch_early_irq_init(void) { + struct fwnode_handle *fn; + init_legacy_irqs(); - x86_vector_domain = irq_domain_add_tree(NULL, &x86_vector_domain_ops, - NULL); + fn = irq_domain_alloc_named_fwnode("VECTOR"); + BUG_ON(!fn); + x86_vector_domain = irq_domain_create_tree(fn, &x86_vector_domain_ops, + NULL); BUG_ON(x86_vector_domain == NULL); + irq_domain_free_fwnode(fn); irq_set_default_host(x86_vector_domain); arch_init_msi_domain(x86_vector_domain); From 1b604745c8474c76e5fd1682ea5b7da0a1c6d440 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:07 +0200 Subject: [PATCH 043/112] x86/ioapic: Create named irq domain Use the fwnode to create a named domain so diagnosis works, but only when the the ioapic is not device tree based. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235443.752782603@linutronix.de Signed-off-by: Thomas Gleixner --- arch/x86/kernel/apic/io_apic.c | 22 ++++++++++++++++++++-- 1 file changed, 20 insertions(+), 2 deletions(-) diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c index 347bb9f65737..444ae9281216 100644 --- a/arch/x86/kernel/apic/io_apic.c +++ b/arch/x86/kernel/apic/io_apic.c @@ -2223,6 +2223,8 @@ static int mp_irqdomain_create(int ioapic) struct ioapic *ip = &ioapics[ioapic]; struct ioapic_domain_cfg *cfg = &ip->irqdomain_cfg; struct mp_ioapic_gsi *gsi_cfg = mp_ioapic_gsi_routing(ioapic); + struct fwnode_handle *fn; + char *name = "IO-APIC"; if (cfg->type == IOAPIC_DOMAIN_INVALID) return 0; @@ -2233,9 +2235,25 @@ static int mp_irqdomain_create(int ioapic) parent = irq_remapping_get_ir_irq_domain(&info); if (!parent) parent = x86_vector_domain; + else + name = "IO-APIC-IR"; + + /* Handle device tree enumerated APICs proper */ + if (cfg->dev) { + fn = of_node_to_fwnode(cfg->dev); + } else { + fn = irq_domain_alloc_named_id_fwnode(name, ioapic); + if (!fn) + return -ENOMEM; + } + + ip->irqdomain = irq_domain_create_linear(fn, hwirqs, cfg->ops, + (void *)(long)ioapic); + + /* Release fw handle if it was allocated above */ + if (!cfg->dev) + irq_domain_free_fwnode(fn); - ip->irqdomain = irq_domain_add_linear(cfg->dev, hwirqs, cfg->ops, - (void *)(long)ioapic); if (!ip->irqdomain) return -ENOMEM; From 5f432711ba94400fb39e9be81913ced81c141758 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:08 +0200 Subject: [PATCH 044/112] x86/htirq: Create named domain Use the fwnode to create a named domain so diagnosis works. Mark the init function __init while at it. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235443.829047007@linutronix.de Signed-off-by: Thomas Gleixner --- arch/x86/kernel/apic/htirq.c | 21 ++++++++++++++++----- 1 file changed, 16 insertions(+), 5 deletions(-) diff --git a/arch/x86/kernel/apic/htirq.c b/arch/x86/kernel/apic/htirq.c index ae50d3454d78..56ccf9346b08 100644 --- a/arch/x86/kernel/apic/htirq.c +++ b/arch/x86/kernel/apic/htirq.c @@ -150,16 +150,27 @@ static const struct irq_domain_ops htirq_domain_ops = { .deactivate = htirq_domain_deactivate, }; -void arch_init_htirq_domain(struct irq_domain *parent) +void __init arch_init_htirq_domain(struct irq_domain *parent) { + struct fwnode_handle *fn; + if (disable_apic) return; - htirq_domain = irq_domain_add_tree(NULL, &htirq_domain_ops, NULL); + fn = irq_domain_alloc_named_fwnode("PCI-HT"); + if (!fn) + goto warn; + + htirq_domain = irq_domain_create_tree(fn, &htirq_domain_ops, NULL); + irq_domain_free_fwnode(fn); if (!htirq_domain) - pr_warn("failed to initialize irqdomain for HTIRQ.\n"); - else - htirq_domain->parent = parent; + goto warn; + + htirq_domain->parent = parent; + return; + +warn: + pr_warn("Failed to initialize irqdomain for HTIRQ.\n"); } int arch_setup_ht_irq(int idx, int pos, struct pci_dev *dev, From f8409a6a4bf86e2d90ec8460df2874e4e19ebb27 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:09 +0200 Subject: [PATCH 045/112] x86/uv: Create named irq domain Use the fwnode to create a named domain so diagnosis works. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235443.907511074@linutronix.de Signed-off-by: Thomas Gleixner --- arch/x86/platform/uv/uv_irq.c | 18 +++++++++++++----- 1 file changed, 13 insertions(+), 5 deletions(-) diff --git a/arch/x86/platform/uv/uv_irq.c b/arch/x86/platform/uv/uv_irq.c index 776c6592136c..03fc397335b7 100644 --- a/arch/x86/platform/uv/uv_irq.c +++ b/arch/x86/platform/uv/uv_irq.c @@ -160,13 +160,21 @@ static struct irq_domain *uv_get_irq_domain(void) { static struct irq_domain *uv_domain; static DEFINE_MUTEX(uv_lock); + struct fwnode_handle *fn; mutex_lock(&uv_lock); - if (uv_domain == NULL) { - uv_domain = irq_domain_add_tree(NULL, &uv_domain_ops, NULL); - if (uv_domain) - uv_domain->parent = x86_vector_domain; - } + if (uv_domain) + goto out; + + fn = irq_domain_alloc_named_fwnode("UV-CORE"); + if (!fn) + goto out; + + uv_domain = irq_domain_create_tree(fn, &uv_domain_ops, NULL); + irq_domain_free_fwnode(fn); + if (uv_domain) + uv_domain->parent = x86_vector_domain; +out: mutex_unlock(&uv_lock); return uv_domain; From 667724c5a3109675cf3bfe7d75795b8608d1bcbe Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:10 +0200 Subject: [PATCH 046/112] x86/msi: Provide new iommu irqdomain interface Provide a new interface for creating the iommu remapping domains, so that the caller can supply a name and a id in order to create named irqdomains. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Joerg Roedel Cc: Keith Busch Cc: Peter Zijlstra Cc: iommu@lists.linux-foundation.org Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235443.986661206@linutronix.de Signed-off-by: Thomas Gleixner --- arch/x86/include/asm/irq_remapping.h | 2 ++ arch/x86/kernel/apic/msi.c | 15 +++++++++++++++ 2 files changed, 17 insertions(+) diff --git a/arch/x86/include/asm/irq_remapping.h b/arch/x86/include/asm/irq_remapping.h index a210eba2727c..039867558ed4 100644 --- a/arch/x86/include/asm/irq_remapping.h +++ b/arch/x86/include/asm/irq_remapping.h @@ -56,6 +56,8 @@ irq_remapping_get_irq_domain(struct irq_alloc_info *info); /* Create PCI MSI/MSIx irqdomain, use @parent as the parent irqdomain. */ extern struct irq_domain *arch_create_msi_irq_domain(struct irq_domain *parent); +extern struct irq_domain * +arch_create_remap_msi_irq_domain(struct irq_domain *par, const char *n, int id); /* Get parent irqdomain for interrupt remapping irqdomain */ static inline struct irq_domain *arch_get_ir_parent_domain(void) diff --git a/arch/x86/kernel/apic/msi.c b/arch/x86/kernel/apic/msi.c index c61aec7e65f4..0e6618e2e589 100644 --- a/arch/x86/kernel/apic/msi.c +++ b/arch/x86/kernel/apic/msi.c @@ -167,10 +167,25 @@ static struct msi_domain_info pci_msi_ir_domain_info = { .handler_name = "edge", }; +struct irq_domain *arch_create_remap_msi_irq_domain(struct irq_domain *parent, + const char *name, int id) +{ + struct fwnode_handle *fn; + struct irq_domain *d; + + fn = irq_domain_alloc_named_id_fwnode(name, id); + if (!fn) + return NULL; + d = pci_msi_create_irq_domain(fn, &pci_msi_ir_domain_info, parent); + irq_domain_free_fwnode(fn); + return d; +} + struct irq_domain *arch_create_msi_irq_domain(struct irq_domain *parent) { return pci_msi_create_irq_domain(NULL, &pci_msi_ir_domain_info, parent); } + #endif #ifdef CONFIG_DMAR_TABLE From cea29b656a5e5f1a7b7de42795c3ae6fc417ab0b Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:11 +0200 Subject: [PATCH 047/112] iommu/vt-d: Use named irq domain interface Signed-off-by: Thomas Gleixner Acked-by: Joerg Roedel Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: iommu@lists.linux-foundation.org Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235444.063083997@linutronix.de --- drivers/iommu/intel_irq_remapping.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) diff --git a/drivers/iommu/intel_irq_remapping.c b/drivers/iommu/intel_irq_remapping.c index ba5b5800e249..8fc641ea2e41 100644 --- a/drivers/iommu/intel_irq_remapping.c +++ b/drivers/iommu/intel_irq_remapping.c @@ -500,8 +500,9 @@ static void iommu_enable_irq_remapping(struct intel_iommu *iommu) static int intel_setup_irq_remapping(struct intel_iommu *iommu) { struct ir_table *ir_table; - struct page *pages; + struct fwnode_handle *fn; unsigned long *bitmap; + struct page *pages; if (iommu->ir_table) return 0; @@ -525,15 +526,24 @@ static int intel_setup_irq_remapping(struct intel_iommu *iommu) goto out_free_pages; } - iommu->ir_domain = irq_domain_add_hierarchy(arch_get_ir_parent_domain(), - 0, INTR_REMAP_TABLE_ENTRIES, - NULL, &intel_ir_domain_ops, - iommu); + fn = irq_domain_alloc_named_id_fwnode("INTEL-IR", iommu->seq_id); + if (!fn) + goto out_free_bitmap; + + iommu->ir_domain = + irq_domain_create_hierarchy(arch_get_ir_parent_domain(), + 0, INTR_REMAP_TABLE_ENTRIES, + fn, &intel_ir_domain_ops, + iommu); + irq_domain_free_fwnode(fn); if (!iommu->ir_domain) { pr_err("IR%d: failed to allocate irqdomain\n", iommu->seq_id); goto out_free_bitmap; } - iommu->ir_msi_domain = arch_create_msi_irq_domain(iommu->ir_domain); + iommu->ir_msi_domain = + arch_create_remap_msi_irq_domain(iommu->ir_domain, + "INTEL-IR-MSI", + iommu->seq_id); ir_table->base = page_address(pages); ir_table->bitmap = bitmap; From 3e49a8182277ea57736285aede5f43bfa6aa11b1 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:12 +0200 Subject: [PATCH 048/112] iommu/amd: Use named irq domain interface Signed-off-by: Thomas Gleixner Acked-by: Joerg Roedel Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: iommu@lists.linux-foundation.org Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235444.142270582@linutronix.de --- drivers/iommu/amd_iommu.c | 13 ++++++++++--- 1 file changed, 10 insertions(+), 3 deletions(-) diff --git a/drivers/iommu/amd_iommu.c b/drivers/iommu/amd_iommu.c index 590e1e8288a1..503849d186c3 100644 --- a/drivers/iommu/amd_iommu.c +++ b/drivers/iommu/amd_iommu.c @@ -4395,13 +4395,20 @@ static struct irq_chip amd_ir_chip = { int amd_iommu_create_irq_domain(struct amd_iommu *iommu) { - iommu->ir_domain = irq_domain_add_tree(NULL, &amd_ir_domain_ops, iommu); + struct fwnode_handle *fn; + + fn = irq_domain_alloc_named_id_fwnode("AMD-IR", iommu->index); + if (!fn) + return -ENOMEM; + iommu->ir_domain = irq_domain_create_tree(fn, &amd_ir_domain_ops, iommu); + irq_domain_free_fwnode(fn); if (!iommu->ir_domain) return -ENOMEM; iommu->ir_domain->parent = arch_get_ir_parent_domain(); - iommu->msi_domain = arch_create_msi_irq_domain(iommu->ir_domain); - + iommu->msi_domain = arch_create_remap_msi_irq_domain(iommu->ir_domain, + "AMD-IR-MSI", + iommu->index); return 0; } From 0323b9690448e1d1ada91dac9d8fa62f7285751a Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:13 +0200 Subject: [PATCH 049/112] x86/msi: Remove unused remap irq domain interface Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235444.221049665@linutronix.de --- arch/x86/include/asm/irq_remapping.h | 1 - arch/x86/kernel/apic/msi.c | 6 ------ 2 files changed, 7 deletions(-) diff --git a/arch/x86/include/asm/irq_remapping.h b/arch/x86/include/asm/irq_remapping.h index 039867558ed4..023b4a9fc846 100644 --- a/arch/x86/include/asm/irq_remapping.h +++ b/arch/x86/include/asm/irq_remapping.h @@ -55,7 +55,6 @@ extern struct irq_domain * irq_remapping_get_irq_domain(struct irq_alloc_info *info); /* Create PCI MSI/MSIx irqdomain, use @parent as the parent irqdomain. */ -extern struct irq_domain *arch_create_msi_irq_domain(struct irq_domain *parent); extern struct irq_domain * arch_create_remap_msi_irq_domain(struct irq_domain *par, const char *n, int id); diff --git a/arch/x86/kernel/apic/msi.c b/arch/x86/kernel/apic/msi.c index 0e6618e2e589..d79dc2afa536 100644 --- a/arch/x86/kernel/apic/msi.c +++ b/arch/x86/kernel/apic/msi.c @@ -180,12 +180,6 @@ struct irq_domain *arch_create_remap_msi_irq_domain(struct irq_domain *parent, irq_domain_free_fwnode(fn); return d; } - -struct irq_domain *arch_create_msi_irq_domain(struct irq_domain *parent) -{ - return pci_msi_create_irq_domain(NULL, &pci_msi_ir_domain_info, parent); -} - #endif #ifdef CONFIG_DMAR_TABLE From f8f37ca78915b51a73bf240409fcda30d811b76b Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:14 +0200 Subject: [PATCH 050/112] x86/msi: Create named irq domains Use the fwnode to create named irq domains so diagnosis works. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235444.299024560@linutronix.de --- arch/x86/kernel/apic/msi.c | 44 +++++++++++++++++++++++++++++--------- 1 file changed, 34 insertions(+), 10 deletions(-) diff --git a/arch/x86/kernel/apic/msi.c b/arch/x86/kernel/apic/msi.c index d79dc2afa536..9b18be764422 100644 --- a/arch/x86/kernel/apic/msi.c +++ b/arch/x86/kernel/apic/msi.c @@ -136,13 +136,20 @@ static struct msi_domain_info pci_msi_domain_info = { .handler_name = "edge", }; -void arch_init_msi_domain(struct irq_domain *parent) +void __init arch_init_msi_domain(struct irq_domain *parent) { + struct fwnode_handle *fn; + if (disable_apic) return; - msi_default_domain = pci_msi_create_irq_domain(NULL, - &pci_msi_domain_info, parent); + fn = irq_domain_alloc_named_fwnode("PCI-MSI"); + if (fn) { + msi_default_domain = + pci_msi_create_irq_domain(fn, &pci_msi_domain_info, + parent); + irq_domain_free_fwnode(fn); + } if (!msi_default_domain) pr_warn("failed to initialize irqdomain for MSI/MSI-x.\n"); } @@ -230,13 +237,20 @@ static struct irq_domain *dmar_get_irq_domain(void) { static struct irq_domain *dmar_domain; static DEFINE_MUTEX(dmar_lock); + struct fwnode_handle *fn; mutex_lock(&dmar_lock); - if (dmar_domain == NULL) - dmar_domain = msi_create_irq_domain(NULL, &dmar_msi_domain_info, - x86_vector_domain); - mutex_unlock(&dmar_lock); + if (dmar_domain) + goto out; + fn = irq_domain_alloc_named_fwnode("DMAR-MSI"); + if (fn) { + dmar_domain = msi_create_irq_domain(fn, &dmar_msi_domain_info, + x86_vector_domain); + irq_domain_free_fwnode(fn); + } +out: + mutex_unlock(&dmar_lock); return dmar_domain; } @@ -326,9 +340,10 @@ static struct msi_domain_info hpet_msi_domain_info = { struct irq_domain *hpet_create_irq_domain(int hpet_id) { - struct irq_domain *parent; - struct irq_alloc_info info; struct msi_domain_info *domain_info; + struct irq_domain *parent, *d; + struct irq_alloc_info info; + struct fwnode_handle *fn; if (x86_vector_domain == NULL) return NULL; @@ -349,7 +364,16 @@ struct irq_domain *hpet_create_irq_domain(int hpet_id) else hpet_msi_controller.name = "IR-HPET-MSI"; - return msi_create_irq_domain(NULL, domain_info, parent); + fn = irq_domain_alloc_named_id_fwnode(hpet_msi_controller.name, + hpet_id); + if (!fn) { + kfree(domain_info); + return NULL; + } + + d = msi_create_irq_domain(fn, domain_info, parent); + irq_domain_free_fwnode(fn); + return d; } int hpet_assign_irq(struct irq_domain *domain, struct hpet_dev *dev, From ae904cafd59d7120ef2afb97b252eadeba45e95f Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:15 +0200 Subject: [PATCH 051/112] PCI/vmd: Create named irq domain Use the fwnode to create a named domain so diagnosis works. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: linux-pci@vger.kernel.org Cc: Bjorn Helgaas Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235444.379861978@linutronix.de --- drivers/pci/host/vmd.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/drivers/pci/host/vmd.c b/drivers/pci/host/vmd.c index e27ad2a3bd33..31203d69616d 100644 --- a/drivers/pci/host/vmd.c +++ b/drivers/pci/host/vmd.c @@ -554,6 +554,7 @@ static int vmd_find_free_domain(void) static int vmd_enable_domain(struct vmd_dev *vmd) { struct pci_sysdata *sd = &vmd->sysdata; + struct fwnode_handle *fn; struct resource *res; u32 upper_bits; unsigned long flags; @@ -617,8 +618,13 @@ static int vmd_enable_domain(struct vmd_dev *vmd) sd->node = pcibus_to_node(vmd->dev->bus); - vmd->irq_domain = pci_msi_create_irq_domain(NULL, &vmd_msi_domain_info, + fn = irq_domain_alloc_named_id_fwnode("VMD-MSI", vmd->sysdata.domain); + if (!fn) + return -ENODEV; + + vmd->irq_domain = pci_msi_create_irq_domain(fn, &vmd_msi_domain_info, x86_vector_domain); + irq_domain_free_fwnode(fn); if (!vmd->irq_domain) return -ENODEV; From 9dc6be3d419398eae9a19cd09b7969ceff8eaf10 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:16 +0200 Subject: [PATCH 052/112] genirq/irqdomain: Add map counter Add a map counter instead of counting radix tree entries for diagnosis. That also gives correct information for linear domains. Signed-off-by: Thomas Gleixner Acked-by: Marc Zyngier Cc: Jens Axboe Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235444.459397746@linutronix.de --- include/linux/irqdomain.h | 2 ++ kernel/irq/irqdomain.c | 4 ++++ 2 files changed, 6 insertions(+) diff --git a/include/linux/irqdomain.h b/include/linux/irqdomain.h index 9cf32a2fbe69..17ccd54d936d 100644 --- a/include/linux/irqdomain.h +++ b/include/linux/irqdomain.h @@ -130,6 +130,7 @@ struct irq_domain_chip_generic; * @host_data: private data pointer for use by owner. Not touched by irq_domain * core code. * @flags: host per irq_domain flags + * @mapcount: The number of mapped interrupts * * Optional elements * @of_node: Pointer to device tree nodes associated with the irq_domain. Used @@ -152,6 +153,7 @@ struct irq_domain { const struct irq_domain_ops *ops; void *host_data; unsigned int flags; + unsigned int mapcount; /* Optional data */ struct fwnode_handle *fwnode; diff --git a/kernel/irq/irqdomain.c b/kernel/irq/irqdomain.c index e1b925bea205..8d5805c655b6 100644 --- a/kernel/irq/irqdomain.c +++ b/kernel/irq/irqdomain.c @@ -423,6 +423,7 @@ void irq_domain_disassociate(struct irq_domain *domain, unsigned int irq) irq_data->domain = NULL; irq_data->hwirq = 0; + domain->mapcount--; /* Clear reverse map for this hwirq */ if (hwirq < domain->revmap_size) { @@ -474,6 +475,7 @@ int irq_domain_associate(struct irq_domain *domain, unsigned int virq, domain->name = irq_data->chip->name; } + domain->mapcount++; if (hwirq < domain->revmap_size) { domain->linear_revmap[hwirq] = virq; } else { @@ -1081,6 +1083,7 @@ static void irq_domain_insert_irq(int virq) struct irq_domain *domain = data->domain; irq_hw_number_t hwirq = data->hwirq; + domain->mapcount++; if (hwirq < domain->revmap_size) { domain->linear_revmap[hwirq] = virq; } else { @@ -1110,6 +1113,7 @@ static void irq_domain_remove_irq(int virq) struct irq_domain *domain = data->domain; irq_hw_number_t hwirq = data->hwirq; + domain->mapcount--; if (hwirq < domain->revmap_size) { domain->linear_revmap[hwirq] = 0; } else { From 087cdfb662ae50e3826e7cd2e54b6519d07b60f0 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:17 +0200 Subject: [PATCH 053/112] genirq/debugfs: Add proper debugfs interface Debugging (hierarchical) interupt domains is tedious as there is no information about the hierarchy and no information about states of interrupts in the various domain levels. Add a debugfs directory 'irq' and subdirectories 'domains' and 'irqs'. The domains directory contains the domain files. The content is information about the domain. If the domain is part of a hierarchy then the parent domains are printed as well. # ls /sys/kernel/debug/irq/domains/ default INTEL-IR-2 INTEL-IR-MSI-2 IO-APIC-IR-2 PCI-MSI DMAR-MSI INTEL-IR-3 INTEL-IR-MSI-3 IO-APIC-IR-3 unknown-1 INTEL-IR-0 INTEL-IR-MSI-0 IO-APIC-IR-0 IO-APIC-IR-4 VECTOR INTEL-IR-1 INTEL-IR-MSI-1 IO-APIC-IR-1 PCI-HT # cat /sys/kernel/debug/irq/domains/VECTOR name: VECTOR size: 0 mapped: 216 flags: 0x00000041 # cat /sys/kernel/debug/irq/domains/IO-APIC-IR-0 name: IO-APIC-IR-0 size: 24 mapped: 19 flags: 0x00000041 parent: INTEL-IR-3 name: INTEL-IR-3 size: 65536 mapped: 167 flags: 0x00000041 parent: VECTOR name: VECTOR size: 0 mapped: 216 flags: 0x00000041 Unfortunately there is no per cpu information about the VECTOR domain (yet). The irqs directory contains detailed information about mapped interrupts. # cat /sys/kernel/debug/irq/irqs/3 handler: handle_edge_irq status: 0x00004000 istate: 0x00000000 ddepth: 1 wdepth: 0 dstate: 0x01018000 IRQD_IRQ_DISABLED IRQD_SINGLE_TARGET IRQD_MOVE_PCNTXT node: 0 affinity: 0-143 effectiv: 0 pending: domain: IO-APIC-IR-0 hwirq: 0x3 chip: IR-IO-APIC flags: 0x10 IRQCHIP_SKIP_SET_WAKE parent: domain: INTEL-IR-3 hwirq: 0x20000 chip: INTEL-IR flags: 0x0 parent: domain: VECTOR hwirq: 0x3 chip: APIC flags: 0x0 This was developed to simplify the debugging of the managed affinity changes. Signed-off-by: Thomas Gleixner Acked-by: Marc Zyngier Cc: Jens Axboe Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235444.537566163@linutronix.de Signed-off-by: Thomas Gleixner --- include/linux/irqdesc.h | 4 + include/linux/irqdomain.h | 4 + kernel/irq/Kconfig | 11 ++ kernel/irq/Makefile | 1 + kernel/irq/debugfs.c | 215 ++++++++++++++++++++++++++++++++++++++ kernel/irq/internals.h | 22 ++++ kernel/irq/irqdesc.c | 1 + kernel/irq/irqdomain.c | 87 ++++++++++++++- kernel/irq/manage.c | 1 + 9 files changed, 345 insertions(+), 1 deletion(-) create mode 100644 kernel/irq/debugfs.c diff --git a/include/linux/irqdesc.h b/include/linux/irqdesc.h index c9be57931b58..d425a3a09722 100644 --- a/include/linux/irqdesc.h +++ b/include/linux/irqdesc.h @@ -46,6 +46,7 @@ struct pt_regs; * @rcu: rcu head for delayed free * @kobj: kobject used to represent this struct in sysfs * @dir: /proc/irq/ procfs entry + * @debugfs_file: dentry for the debugfs file * @name: flow handler name for /proc/interrupts output */ struct irq_desc { @@ -88,6 +89,9 @@ struct irq_desc { #ifdef CONFIG_PROC_FS struct proc_dir_entry *dir; #endif +#ifdef CONFIG_GENERIC_IRQ_DEBUGFS + struct dentry *debugfs_file; +#endif #ifdef CONFIG_SPARSE_IRQ struct rcu_head rcu; struct kobject kobj; diff --git a/include/linux/irqdomain.h b/include/linux/irqdomain.h index 17ccd54d936d..914b0c31d233 100644 --- a/include/linux/irqdomain.h +++ b/include/linux/irqdomain.h @@ -139,6 +139,7 @@ struct irq_domain_chip_generic; * setting up one or more generic chips for interrupt controllers * drivers using the generic chip library which uses this pointer. * @parent: Pointer to parent irq_domain to support hierarchy irq_domains + * @debugfs_file: dentry for the domain debugfs file * * Revmap data, used internally by irq_domain * @revmap_direct_max_irq: The largest hwirq that can be set for controllers that @@ -162,6 +163,9 @@ struct irq_domain { #ifdef CONFIG_IRQ_DOMAIN_HIERARCHY struct irq_domain *parent; #endif +#ifdef CONFIG_GENERIC_IRQ_DEBUGFS + struct dentry *debugfs_file; +#endif /* reverse map data. The linear map gets appended to the irq_domain */ irq_hw_number_t hwirq_max; diff --git a/kernel/irq/Kconfig b/kernel/irq/Kconfig index 3bbfd6a9c475..8d9498e51585 100644 --- a/kernel/irq/Kconfig +++ b/kernel/irq/Kconfig @@ -108,4 +108,15 @@ config SPARSE_IRQ If you don't know what to do here, say N. +config GENERIC_IRQ_DEBUGFS + bool "Expose irq internals in debugfs" + depends on DEBUG_FS + default n + ---help--- + + Exposes internal state information through debugfs. Mostly for + developers and debugging of hard to diagnose interrupt problems. + + If you don't know what to do here, say N. + endmenu diff --git a/kernel/irq/Makefile b/kernel/irq/Makefile index 1d3ee3169202..c61fc9c2d1f7 100644 --- a/kernel/irq/Makefile +++ b/kernel/irq/Makefile @@ -10,3 +10,4 @@ obj-$(CONFIG_PM_SLEEP) += pm.o obj-$(CONFIG_GENERIC_MSI_IRQ) += msi.o obj-$(CONFIG_GENERIC_IRQ_IPI) += ipi.o obj-$(CONFIG_SMP) += affinity.o +obj-$(CONFIG_GENERIC_IRQ_DEBUGFS) += debugfs.o diff --git a/kernel/irq/debugfs.c b/kernel/irq/debugfs.c new file mode 100644 index 000000000000..50ee2f6593e8 --- /dev/null +++ b/kernel/irq/debugfs.c @@ -0,0 +1,215 @@ +/* + * Copyright 2017 Thomas Gleixner + * + * This file is licensed under the GPL V2. + */ +#include +#include +#include + +#include "internals.h" + +static struct dentry *irq_dir; + +struct irq_bit_descr { + unsigned int mask; + char *name; +}; +#define BIT_MASK_DESCR(m) { .mask = m, .name = #m } + +static void irq_debug_show_bits(struct seq_file *m, int ind, unsigned int state, + const struct irq_bit_descr *sd, int size) +{ + int i; + + for (i = 0; i < size; i++, sd++) { + if (state & sd->mask) + seq_printf(m, "%*s%s\n", ind + 12, "", sd->name); + } +} + +#ifdef CONFIG_SMP +static void irq_debug_show_masks(struct seq_file *m, struct irq_desc *desc) +{ + struct irq_data *data = irq_desc_get_irq_data(desc); + struct cpumask *msk; + + msk = irq_data_get_affinity_mask(data); + seq_printf(m, "affinity: %*pbl\n", cpumask_pr_args(msk)); +#ifdef CONFIG_GENERIC_PENDING_IRQ + msk = desc->pending_mask; + seq_printf(m, "pending: %*pbl\n", cpumask_pr_args(msk)); +#endif +} +#else +static void irq_debug_show_masks(struct seq_file *m, struct irq_desc *desc) { } +#endif + +static const struct irq_bit_descr irqchip_flags[] = { + BIT_MASK_DESCR(IRQCHIP_SET_TYPE_MASKED), + BIT_MASK_DESCR(IRQCHIP_EOI_IF_HANDLED), + BIT_MASK_DESCR(IRQCHIP_MASK_ON_SUSPEND), + BIT_MASK_DESCR(IRQCHIP_ONOFFLINE_ENABLED), + BIT_MASK_DESCR(IRQCHIP_SKIP_SET_WAKE), + BIT_MASK_DESCR(IRQCHIP_ONESHOT_SAFE), + BIT_MASK_DESCR(IRQCHIP_EOI_THREADED), +}; + +static void +irq_debug_show_chip(struct seq_file *m, struct irq_data *data, int ind) +{ + struct irq_chip *chip = data->chip; + + if (!chip) { + seq_printf(m, "chip: None\n"); + return; + } + seq_printf(m, "%*schip: %s\n", ind, "", chip->name); + seq_printf(m, "%*sflags: 0x%lx\n", ind + 1, "", chip->flags); + irq_debug_show_bits(m, ind, chip->flags, irqchip_flags, + ARRAY_SIZE(irqchip_flags)); +} + +static void +irq_debug_show_data(struct seq_file *m, struct irq_data *data, int ind) +{ + seq_printf(m, "%*sdomain: %s\n", ind, "", + data->domain ? data->domain->name : ""); + seq_printf(m, "%*shwirq: 0x%lx\n", ind + 1, "", data->hwirq); + irq_debug_show_chip(m, data, ind + 1); +#ifdef CONFIG_IRQ_DOMAIN_HIERARCHY + if (!data->parent_data) + return; + seq_printf(m, "%*sparent:\n", ind + 1, ""); + irq_debug_show_data(m, data->parent_data, ind + 4); +#endif +} + +static const struct irq_bit_descr irqdata_states[] = { + BIT_MASK_DESCR(IRQ_TYPE_EDGE_RISING), + BIT_MASK_DESCR(IRQ_TYPE_EDGE_FALLING), + BIT_MASK_DESCR(IRQ_TYPE_LEVEL_HIGH), + BIT_MASK_DESCR(IRQ_TYPE_LEVEL_LOW), + BIT_MASK_DESCR(IRQD_LEVEL), + + BIT_MASK_DESCR(IRQD_ACTIVATED), + BIT_MASK_DESCR(IRQD_IRQ_STARTED), + BIT_MASK_DESCR(IRQD_IRQ_DISABLED), + BIT_MASK_DESCR(IRQD_IRQ_MASKED), + BIT_MASK_DESCR(IRQD_IRQ_INPROGRESS), + + BIT_MASK_DESCR(IRQD_PER_CPU), + BIT_MASK_DESCR(IRQD_NO_BALANCING), + + BIT_MASK_DESCR(IRQD_MOVE_PCNTXT), + BIT_MASK_DESCR(IRQD_AFFINITY_SET), + BIT_MASK_DESCR(IRQD_SETAFFINITY_PENDING), + BIT_MASK_DESCR(IRQD_AFFINITY_MANAGED), + BIT_MASK_DESCR(IRQD_MANAGED_SHUTDOWN), + + BIT_MASK_DESCR(IRQD_FORWARDED_TO_VCPU), + + BIT_MASK_DESCR(IRQD_WAKEUP_STATE), + BIT_MASK_DESCR(IRQD_WAKEUP_ARMED), +}; + +static const struct irq_bit_descr irqdesc_states[] = { + BIT_MASK_DESCR(_IRQ_NOPROBE), + BIT_MASK_DESCR(_IRQ_NOREQUEST), + BIT_MASK_DESCR(_IRQ_NOTHREAD), + BIT_MASK_DESCR(_IRQ_NOAUTOEN), + BIT_MASK_DESCR(_IRQ_NESTED_THREAD), + BIT_MASK_DESCR(_IRQ_PER_CPU_DEVID), + BIT_MASK_DESCR(_IRQ_IS_POLLED), + BIT_MASK_DESCR(_IRQ_DISABLE_UNLAZY), +}; + +static const struct irq_bit_descr irqdesc_istates[] = { + BIT_MASK_DESCR(IRQS_AUTODETECT), + BIT_MASK_DESCR(IRQS_SPURIOUS_DISABLED), + BIT_MASK_DESCR(IRQS_POLL_INPROGRESS), + BIT_MASK_DESCR(IRQS_ONESHOT), + BIT_MASK_DESCR(IRQS_REPLAY), + BIT_MASK_DESCR(IRQS_WAITING), + BIT_MASK_DESCR(IRQS_PENDING), + BIT_MASK_DESCR(IRQS_SUSPENDED), +}; + + +static int irq_debug_show(struct seq_file *m, void *p) +{ + struct irq_desc *desc = m->private; + struct irq_data *data; + + raw_spin_lock_irq(&desc->lock); + data = irq_desc_get_irq_data(desc); + seq_printf(m, "handler: %pf\n", desc->handle_irq); + seq_printf(m, "status: 0x%08x\n", desc->status_use_accessors); + irq_debug_show_bits(m, 0, desc->status_use_accessors, irqdesc_states, + ARRAY_SIZE(irqdesc_states)); + seq_printf(m, "istate: 0x%08x\n", desc->istate); + irq_debug_show_bits(m, 0, desc->istate, irqdesc_istates, + ARRAY_SIZE(irqdesc_istates)); + seq_printf(m, "ddepth: %u\n", desc->depth); + seq_printf(m, "wdepth: %u\n", desc->wake_depth); + seq_printf(m, "dstate: 0x%08x\n", irqd_get(data)); + irq_debug_show_bits(m, 0, irqd_get(data), irqdata_states, + ARRAY_SIZE(irqdata_states)); + seq_printf(m, "node: %d\n", irq_data_get_node(data)); + irq_debug_show_masks(m, desc); + irq_debug_show_data(m, data, 0); + raw_spin_unlock_irq(&desc->lock); + return 0; +} + +static int irq_debug_open(struct inode *inode, struct file *file) +{ + return single_open(file, irq_debug_show, inode->i_private); +} + +static const struct file_operations dfs_irq_ops = { + .open = irq_debug_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +void irq_add_debugfs_entry(unsigned int irq, struct irq_desc *desc) +{ + char name [10]; + + if (!irq_dir || !desc || desc->debugfs_file) + return; + + sprintf(name, "%d", irq); + desc->debugfs_file = debugfs_create_file(name, 0444, irq_dir, desc, + &dfs_irq_ops); +} + +void irq_remove_debugfs_entry(struct irq_desc *desc) +{ + if (desc->debugfs_file) + debugfs_remove(desc->debugfs_file); +} + +static int __init irq_debugfs_init(void) +{ + struct dentry *root_dir; + int irq; + + root_dir = debugfs_create_dir("irq", NULL); + if (!root_dir) + return -ENOMEM; + + irq_domain_debugfs_init(root_dir); + + irq_dir = debugfs_create_dir("irqs", root_dir); + + irq_lock_sparse(); + for_each_active_irq(irq) + irq_add_debugfs_entry(irq, irq_to_desc(irq)); + irq_unlock_sparse(); + + return 0; +} +__initcall(irq_debugfs_init); diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index 921a2419720c..094db5bfb83f 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -169,6 +169,11 @@ irq_put_desc_unlock(struct irq_desc *desc, unsigned long flags) #define __irqd_to_state(d) ACCESS_PRIVATE((d)->common, state_use_accessors) +static inline unsigned int irqd_get(struct irq_data *d) +{ + return __irqd_to_state(d); +} + /* * Manipulation functions for irq_data.state */ @@ -237,3 +242,20 @@ irq_init_generic_chip(struct irq_chip_generic *gc, const char *name, int num_ct, unsigned int irq_base, void __iomem *reg_base, irq_flow_handler_t handler) { } #endif /* CONFIG_GENERIC_IRQ_CHIP */ + +#ifdef CONFIG_GENERIC_IRQ_DEBUGFS +void irq_add_debugfs_entry(unsigned int irq, struct irq_desc *desc); +void irq_remove_debugfs_entry(struct irq_desc *desc); +# ifdef CONFIG_IRQ_DOMAIN +void irq_domain_debugfs_init(struct dentry *root); +# else +static inline void irq_domain_debugfs_init(struct dentry *root); +# endif +#else /* CONFIG_GENERIC_IRQ_DEBUGFS */ +static inline void irq_add_debugfs_entry(unsigned int irq, struct irq_desc *d) +{ +} +static inline void irq_remove_debugfs_entry(struct irq_desc *d) +{ +} +#endif /* CONFIG_GENERIC_IRQ_DEBUGFS */ diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 09abce2ea8f0..feade536b6d1 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -394,6 +394,7 @@ static void free_desc(unsigned int irq) { struct irq_desc *desc = irq_to_desc(irq); + irq_remove_debugfs_entry(desc); unregister_irq_proc(irq, desc); /* diff --git a/kernel/irq/irqdomain.c b/kernel/irq/irqdomain.c index 8d5805c655b6..75e1f0851c33 100644 --- a/kernel/irq/irqdomain.c +++ b/kernel/irq/irqdomain.c @@ -29,9 +29,17 @@ struct irqchip_fwid { struct fwnode_handle fwnode; unsigned int type; char *name; - void *data; + void *data; }; +#ifdef CONFIG_GENERIC_IRQ_DEBUGFS +static void debugfs_add_domain_dir(struct irq_domain *d); +static void debugfs_remove_domain_dir(struct irq_domain *d); +#else +static inline void debugfs_add_domain_dir(struct irq_domain *d) { } +static inline void debugfs_remove_domain_dir(struct irq_domain *d) { } +#endif + /** * irq_domain_alloc_fwnode - Allocate a fwnode_handle suitable for * identifying an irq domain @@ -194,6 +202,7 @@ struct irq_domain *__irq_domain_add(struct fwnode_handle *fwnode, int size, irq_domain_check_hierarchy(domain); mutex_lock(&irq_domain_mutex); + debugfs_add_domain_dir(domain); list_add(&domain->link, &irq_domain_list); mutex_unlock(&irq_domain_mutex); @@ -213,6 +222,7 @@ EXPORT_SYMBOL_GPL(__irq_domain_add); void irq_domain_remove(struct irq_domain *domain) { mutex_lock(&irq_domain_mutex); + debugfs_remove_domain_dir(domain); WARN_ON(!radix_tree_empty(&domain->revmap_tree)); @@ -1599,3 +1609,78 @@ static void irq_domain_check_hierarchy(struct irq_domain *domain) { } #endif /* CONFIG_IRQ_DOMAIN_HIERARCHY */ + +#ifdef CONFIG_GENERIC_IRQ_DEBUGFS +static struct dentry *domain_dir; + +static void +irq_domain_debug_show_one(struct seq_file *m, struct irq_domain *d, int ind) +{ + seq_printf(m, "%*sname: %s\n", ind, "", d->name); + seq_printf(m, "%*ssize: %u\n", ind + 1, "", + d->revmap_size + d->revmap_direct_max_irq); + seq_printf(m, "%*smapped: %u\n", ind + 1, "", d->mapcount); + seq_printf(m, "%*sflags: 0x%08x\n", ind +1 , "", d->flags); +#ifdef CONFIG_IRQ_DOMAIN_HIERARCHY + if (!d->parent) + return; + seq_printf(m, "%*sparent: %s\n", ind + 1, "", d->parent->name); + irq_domain_debug_show_one(m, d->parent, ind + 4); +#endif +} + +static int irq_domain_debug_show(struct seq_file *m, void *p) +{ + struct irq_domain *d = m->private; + + /* Default domain? Might be NULL */ + if (!d) { + if (!irq_default_domain) + return 0; + d = irq_default_domain; + } + irq_domain_debug_show_one(m, d, 0); + return 0; +} + +static int irq_domain_debug_open(struct inode *inode, struct file *file) +{ + return single_open(file, irq_domain_debug_show, inode->i_private); +} + +static const struct file_operations dfs_domain_ops = { + .open = irq_domain_debug_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +static void debugfs_add_domain_dir(struct irq_domain *d) +{ + if (!d->name || !domain_dir || d->debugfs_file) + return; + d->debugfs_file = debugfs_create_file(d->name, 0444, domain_dir, d, + &dfs_domain_ops); +} + +static void debugfs_remove_domain_dir(struct irq_domain *d) +{ + if (d->debugfs_file) + debugfs_remove(d->debugfs_file); +} + +void __init irq_domain_debugfs_init(struct dentry *root) +{ + struct irq_domain *d; + + domain_dir = debugfs_create_dir("domains", root); + if (!domain_dir) + return; + + debugfs_create_file("default", 0444, domain_dir, NULL, &dfs_domain_ops); + mutex_lock(&irq_domain_mutex); + list_for_each_entry(d, &irq_domain_list, link) + debugfs_add_domain_dir(d); + mutex_unlock(&irq_domain_mutex); +} +#endif diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 4c34696ca575..284f4eb1ffbe 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -1398,6 +1398,7 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) wake_up_process(new->secondary->thread); register_irq_proc(irq, desc); + irq_add_debugfs_entry(irq, desc); new->dir = NULL; register_handler_proc(irq, new); free_cpumask_var(mask); From 1bb0401680da156ce1549e915e711bf5b2534cc5 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:18 +0200 Subject: [PATCH 054/112] genirq: Add missing comment for IRQD_STARTED Signed-off-by: Thomas Gleixner Cc: Marc Zyngier Cc: Jens Axboe Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235444.614913014@linutronix.de --- include/linux/irq.h | 1 + 1 file changed, 1 insertion(+) diff --git a/include/linux/irq.h b/include/linux/irq.h index d996314b6522..7e62e10e5856 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -200,6 +200,7 @@ struct irq_data { * IRQD_WAKEUP_ARMED - Wakeup mode armed * IRQD_FORWARDED_TO_VCPU - The interrupt is forwarded to a VCPU * IRQD_AFFINITY_MANAGED - Affinity is auto-managed by the kernel + * IRQD_IRQ_STARTED - Startup state of the interrupt */ enum { IRQD_TRIGGER_MASK = 0xf, From cdd16365b0bd7c0cd19e2cc768b6bdc8021f32c3 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:19 +0200 Subject: [PATCH 055/112] genirq: Provide irq_fixup_move_pending() If an CPU goes offline, the interrupts are migrated away, but a eventually pending interrupt move, which has not yet been made effective is kept pending even if the outgoing CPU is the sole target of the pending affinity mask. What's worse is, that the pending affinity mask is discarded even if it would contain a valid subset of the online CPUs. Implement a helper function which allows to avoid these issues. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235444.691345468@linutronix.de --- include/linux/irq.h | 5 +++++ kernel/irq/migration.c | 30 ++++++++++++++++++++++++++++++ 2 files changed, 35 insertions(+) diff --git a/include/linux/irq.h b/include/linux/irq.h index 7e62e10e5856..d008065e2f4d 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -491,9 +491,14 @@ extern void irq_migrate_all_off_this_cpu(void); #if defined(CONFIG_SMP) && defined(CONFIG_GENERIC_PENDING_IRQ) void irq_move_irq(struct irq_data *data); void irq_move_masked_irq(struct irq_data *data); +bool irq_fixup_move_pending(struct irq_desc *desc, bool force_clear); #else static inline void irq_move_irq(struct irq_data *data) { } static inline void irq_move_masked_irq(struct irq_data *data) { } +static inline bool irq_fixup_move_pending(struct irq_desc *desc, bool fclear) +{ + return false; +} #endif extern int no_irq_affinity; diff --git a/kernel/irq/migration.c b/kernel/irq/migration.c index 37ddb7bda651..6ca054a3f91d 100644 --- a/kernel/irq/migration.c +++ b/kernel/irq/migration.c @@ -4,6 +4,36 @@ #include "internals.h" +/** + * irq_fixup_move_pending - Cleanup irq move pending from a dying CPU + * @desc: Interrupt descpriptor to clean up + * @force_clear: If set clear the move pending bit unconditionally. + * If not set, clear it only when the dying CPU is the + * last one in the pending mask. + * + * Returns true if the pending bit was set and the pending mask contains an + * online CPU other than the dying CPU. + */ +bool irq_fixup_move_pending(struct irq_desc *desc, bool force_clear) +{ + struct irq_data *data = irq_desc_get_irq_data(desc); + + if (!irqd_is_setaffinity_pending(data)) + return false; + + /* + * The outgoing CPU might be the last online target in a pending + * interrupt move. If that's the case clear the pending move bit. + */ + if (cpumask_any_and(desc->pending_mask, cpu_online_mask) >= nr_cpu_ids) { + irqd_clr_move_pending(data); + return false; + } + if (force_clear) + irqd_clr_move_pending(data); + return true; +} + void irq_move_masked_irq(struct irq_data *idata) { struct irq_desc *desc = irq_data_to_desc(idata); From 8e7b632237df8b17526411d1d98f838580bb6aa3 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:20 +0200 Subject: [PATCH 056/112] x86/irq: Cleanup pending irq move in fixup_irqs() If an CPU goes offline, the interrupts are migrated away, but a eventually pending interrupt move, which has not yet been made effective is kept pending even if the outgoing CPU is the sole target of the pending affinity mask. What's worse is, that the pending affinity mask is discarded even if it would contain a valid subset of the online CPUs. Use the newly introduced helper to: - Discard a pending move when the outgoing CPU is the only target in the pending mask. - Use the pending mask instead of the affinity mask to find a valid target for the CPU if the pending mask intersects with the online CPUs. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235444.774068557@linutronix.de --- arch/x86/kernel/irq.c | 25 +++++++++++++++++++++---- 1 file changed, 21 insertions(+), 4 deletions(-) diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c index f34fe7444836..9696007df67b 100644 --- a/arch/x86/kernel/irq.c +++ b/arch/x86/kernel/irq.c @@ -440,9 +440,9 @@ void fixup_irqs(void) int ret; for_each_irq_desc(irq, desc) { + const struct cpumask *affinity; int break_affinity = 0; int set_affinity = 1; - const struct cpumask *affinity; if (!desc) continue; @@ -454,19 +454,36 @@ void fixup_irqs(void) data = irq_desc_get_irq_data(desc); affinity = irq_data_get_affinity_mask(data); + if (!irq_has_action(irq) || irqd_is_per_cpu(data) || cpumask_subset(affinity, cpu_online_mask)) { + irq_fixup_move_pending(desc, false); raw_spin_unlock(&desc->lock); continue; } /* - * Complete the irq move. This cpu is going down and for - * non intr-remapping case, we can't wait till this interrupt - * arrives at this cpu before completing the irq move. + * Complete an eventually pending irq move cleanup. If this + * interrupt was moved in hard irq context, then the + * vectors need to be cleaned up. It can't wait until this + * interrupt actually happens and this CPU was involved. */ irq_force_complete_move(desc); + /* + * If there is a setaffinity pending, then try to reuse the + * pending mask, so the last change of the affinity does + * not get lost. If there is no move pending or the pending + * mask does not contain any online CPU, use the current + * affinity mask. + */ + if (irq_fixup_move_pending(desc, true)) + affinity = desc->pending_mask; + + /* + * If the mask does not contain an offline CPU, break + * affinity and use cpu_online_mask as fall back. + */ if (cpumask_any_and(affinity, cpu_online_mask) >= nr_cpu_ids) { break_affinity = 1; affinity = cpu_online_mask; From cba4235e6031e9318d68186f6d765c531cbea4e1 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:21 +0200 Subject: [PATCH 057/112] genirq: Remove mask argument from setup_affinity() No point to have this alloc/free dance of cpumasks. Provide a static mask for setup_affinity() and protect it proper. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235444.851571573@linutronix.de --- kernel/irq/internals.h | 2 +- kernel/irq/manage.c | 53 ++++++++++++++++++------------------------ kernel/irq/proc.c | 8 ++++--- 3 files changed, 29 insertions(+), 34 deletions(-) diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index 094db5bfb83f..33ca83816b8c 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -109,7 +109,7 @@ static inline void unregister_handler_proc(unsigned int irq, extern bool irq_can_set_affinity_usr(unsigned int irq); -extern int irq_select_affinity_usr(unsigned int irq, struct cpumask *mask); +extern int irq_select_affinity_usr(unsigned int irq); extern void irq_set_thread_affinity(struct irq_desc *desc); diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 284f4eb1ffbe..e2f20d553d60 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -345,15 +345,18 @@ EXPORT_SYMBOL_GPL(irq_set_affinity_notifier); /* * Generic version of the affinity autoselector. */ -static int setup_affinity(struct irq_desc *desc, struct cpumask *mask) +static int irq_setup_affinity(struct irq_desc *desc) { struct cpumask *set = irq_default_affinity; - int node = irq_desc_get_node(desc); + int ret, node = irq_desc_get_node(desc); + static DEFINE_RAW_SPINLOCK(mask_lock); + static struct cpumask mask; /* Excludes PER_CPU and NO_BALANCE interrupts */ if (!__irq_can_set_affinity(desc)) return 0; + raw_spin_lock(&mask_lock); /* * Preserve the managed affinity setting and a userspace affinity * setup, but make sure that one of the targets is online. @@ -367,43 +370,42 @@ static int setup_affinity(struct irq_desc *desc, struct cpumask *mask) irqd_clear(&desc->irq_data, IRQD_AFFINITY_SET); } - cpumask_and(mask, cpu_online_mask, set); + cpumask_and(&mask, cpu_online_mask, set); if (node != NUMA_NO_NODE) { const struct cpumask *nodemask = cpumask_of_node(node); /* make sure at least one of the cpus in nodemask is online */ - if (cpumask_intersects(mask, nodemask)) - cpumask_and(mask, mask, nodemask); + if (cpumask_intersects(&mask, nodemask)) + cpumask_and(&mask, &mask, nodemask); } - irq_do_set_affinity(&desc->irq_data, mask, false); - return 0; + ret = irq_do_set_affinity(&desc->irq_data, &mask, false); + raw_spin_unlock(&mask_lock); + return ret; } #else /* Wrapper for ALPHA specific affinity selector magic */ -static inline int setup_affinity(struct irq_desc *d, struct cpumask *mask) +int irq_setup_affinity(struct irq_desc *desc) { - return irq_select_affinity(irq_desc_get_irq(d)); + return irq_select_affinity(irq_desc_get_irq(desc)); } #endif /* - * Called when affinity is set via /proc/irq + * Called when a bogus affinity is set via /proc/irq */ -int irq_select_affinity_usr(unsigned int irq, struct cpumask *mask) +int irq_select_affinity_usr(unsigned int irq) { struct irq_desc *desc = irq_to_desc(irq); unsigned long flags; int ret; raw_spin_lock_irqsave(&desc->lock, flags); - ret = setup_affinity(desc, mask); + ret = irq_setup_affinity(desc); raw_spin_unlock_irqrestore(&desc->lock, flags); return ret; } - #else -static inline int -setup_affinity(struct irq_desc *desc, struct cpumask *mask) +static inline int setup_affinity(struct irq_desc *desc) { return 0; } @@ -1128,7 +1130,6 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) struct irqaction *old, **old_ptr; unsigned long flags, thread_mask = 0; int ret, nested, shared = 0; - cpumask_var_t mask; if (!desc) return -EINVAL; @@ -1187,11 +1188,6 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) } } - if (!alloc_cpumask_var(&mask, GFP_KERNEL)) { - ret = -ENOMEM; - goto out_thread; - } - /* * Drivers are often written to work w/o knowledge about the * underlying irq chip implementation, so a request for a @@ -1256,7 +1252,7 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) */ if (thread_mask == ~0UL) { ret = -EBUSY; - goto out_mask; + goto out_unlock; } /* * The thread_mask for the action is or'ed to @@ -1300,7 +1296,7 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) pr_err("Threaded irq requested with handler=NULL and !ONESHOT for irq %d\n", irq); ret = -EINVAL; - goto out_mask; + goto out_unlock; } if (!shared) { @@ -1308,7 +1304,7 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) if (ret) { pr_err("Failed to request resources for %s (irq %d) on irqchip %s\n", new->name, irq, desc->irq_data.chip->name); - goto out_mask; + goto out_unlock; } init_waitqueue_head(&desc->wait_for_threads); @@ -1320,7 +1316,7 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) if (ret) { irq_release_resources(desc); - goto out_mask; + goto out_unlock; } } @@ -1357,7 +1353,7 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) } /* Set default affinity mask once everything is setup */ - setup_affinity(desc, mask); + irq_setup_affinity(desc); } else if (new->flags & IRQF_TRIGGER_MASK) { unsigned int nmsk = new->flags & IRQF_TRIGGER_MASK; @@ -1401,8 +1397,6 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) irq_add_debugfs_entry(irq, desc); new->dir = NULL; register_handler_proc(irq, new); - free_cpumask_var(mask); - return 0; mismatch: @@ -1415,9 +1409,8 @@ mismatch: } ret = -EBUSY; -out_mask: +out_unlock: raw_spin_unlock_irqrestore(&desc->lock, flags); - free_cpumask_var(mask); out_thread: if (new->thread) { diff --git a/kernel/irq/proc.c b/kernel/irq/proc.c index c53edad7b459..d35bb8d4c317 100644 --- a/kernel/irq/proc.c +++ b/kernel/irq/proc.c @@ -120,9 +120,11 @@ static ssize_t write_irq_affinity(int type, struct file *file, * one online CPU still has to be targeted. */ if (!cpumask_intersects(new_value, cpu_online_mask)) { - /* Special case for empty set - allow the architecture - code to set default SMP affinity. */ - err = irq_select_affinity_usr(irq, new_value) ? -EINVAL : count; + /* + * Special case for empty set - allow the architecture code + * to set default SMP affinity. + */ + err = irq_select_affinity_usr(irq) ? -EINVAL : count; } else { irq_set_affinity(irq, new_value); err = count; From 43564bd97d0e6182bbd43b51b33254c728832551 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:22 +0200 Subject: [PATCH 058/112] genirq: Rename setup_affinity() to irq_setup_affinity() Rename it with a proper irq_ prefix and make it available for other files in the core code. Preparatory patch for moving the irq affinity setup around. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235444.928501004@linutronix.de --- kernel/irq/internals.h | 6 ++++++ kernel/irq/manage.c | 7 +------ 2 files changed, 7 insertions(+), 6 deletions(-) diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index 33ca83816b8c..2d7927d9fb57 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -116,6 +116,12 @@ extern void irq_set_thread_affinity(struct irq_desc *desc); extern int irq_do_set_affinity(struct irq_data *data, const struct cpumask *dest, bool force); +#ifdef CONFIG_SMP +extern int irq_setup_affinity(struct irq_desc *desc); +#else +static inline int irq_setup_affinity(struct irq_desc *desc) { return 0; } +#endif + /* Inline functions for support of irq chips on slow busses */ static inline void chip_bus_lock(struct irq_desc *desc) { diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index e2f20d553d60..907fb791ff63 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -345,7 +345,7 @@ EXPORT_SYMBOL_GPL(irq_set_affinity_notifier); /* * Generic version of the affinity autoselector. */ -static int irq_setup_affinity(struct irq_desc *desc) +int irq_setup_affinity(struct irq_desc *desc) { struct cpumask *set = irq_default_affinity; int ret, node = irq_desc_get_node(desc); @@ -404,11 +404,6 @@ int irq_select_affinity_usr(unsigned int irq) raw_spin_unlock_irqrestore(&desc->lock, flags); return ret; } -#else -static inline int setup_affinity(struct irq_desc *desc) -{ - return 0; -} #endif /** From 2e051552df69af6d134c2592d0d6f1ac80f01190 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:23 +0200 Subject: [PATCH 059/112] genirq: Move initial affinity setup to irq_startup() The startup vs. setaffinity ordering of interrupts depends on the IRQF_NOAUTOEN flag. Chained interrupts are not getting any affinity assignment at all. A regular interrupt is started up and then the affinity is set. A IRQF_NOAUTOEN marked interrupt is not started up, but the affinity is set nevertheless. Move the affinity setup to startup_irq() so the ordering is always the same and chained interrupts get the proper default affinity assigned as well. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235445.020534783@linutronix.de --- kernel/irq/chip.c | 2 ++ kernel/irq/manage.c | 15 ++++++--------- 2 files changed, 8 insertions(+), 9 deletions(-) diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index bc1331f84fb5..e290d73b88e2 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -213,6 +213,8 @@ int irq_startup(struct irq_desc *desc, bool resend) irq_enable(desc); } irq_state_set_started(desc); + /* Set default affinity mask once everything is setup */ + irq_setup_affinity(desc); } if (resend) diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 907fb791ff63..1e283073cecc 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -1327,6 +1327,12 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) if (new->flags & IRQF_ONESHOT) desc->istate |= IRQS_ONESHOT; + /* Exclude IRQ from balancing if requested */ + if (new->flags & IRQF_NOBALANCING) { + irq_settings_set_no_balancing(desc); + irqd_set(&desc->irq_data, IRQD_NO_BALANCING); + } + if (irq_settings_can_autoenable(desc)) { irq_startup(desc, true); } else { @@ -1341,15 +1347,6 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) desc->depth = 1; } - /* Exclude IRQ from balancing if requested */ - if (new->flags & IRQF_NOBALANCING) { - irq_settings_set_no_balancing(desc); - irqd_set(&desc->irq_data, IRQD_NO_BALANCING); - } - - /* Set default affinity mask once everything is setup */ - irq_setup_affinity(desc); - } else if (new->flags & IRQF_TRIGGER_MASK) { unsigned int nmsk = new->flags & IRQF_TRIGGER_MASK; unsigned int omsk = irqd_get_trigger_type(&desc->irq_data); From 137221df69c6f8a7002f82dc3d95052d34f5667e Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 20 Jun 2017 01:37:24 +0200 Subject: [PATCH 060/112] genirq: Move pending helpers to internal.h So that the affinity code can reuse them. Signed-off-by: Christoph Hellwig Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/20170619235445.109426284@linutronix.de --- kernel/irq/internals.h | 38 ++++++++++++++++++++++++++++++++++++++ kernel/irq/manage.c | 28 ---------------------------- 2 files changed, 38 insertions(+), 28 deletions(-) diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index 2d7927d9fb57..20b197f0a7b5 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -249,6 +249,44 @@ irq_init_generic_chip(struct irq_chip_generic *gc, const char *name, void __iomem *reg_base, irq_flow_handler_t handler) { } #endif /* CONFIG_GENERIC_IRQ_CHIP */ +#ifdef CONFIG_GENERIC_PENDING_IRQ +static inline bool irq_can_move_pcntxt(struct irq_data *data) +{ + return irqd_can_move_in_process_context(data); +} +static inline bool irq_move_pending(struct irq_data *data) +{ + return irqd_is_setaffinity_pending(data); +} +static inline void +irq_copy_pending(struct irq_desc *desc, const struct cpumask *mask) +{ + cpumask_copy(desc->pending_mask, mask); +} +static inline void +irq_get_pending(struct cpumask *mask, struct irq_desc *desc) +{ + cpumask_copy(mask, desc->pending_mask); +} +#else /* CONFIG_GENERIC_PENDING_IRQ */ +static inline bool irq_can_move_pcntxt(struct irq_data *data) +{ + return true; +} +static inline bool irq_move_pending(struct irq_data *data) +{ + return false; +} +static inline void +irq_copy_pending(struct irq_desc *desc, const struct cpumask *mask) +{ +} +static inline void +irq_get_pending(struct cpumask *mask, struct irq_desc *desc) +{ +} +#endif /* CONFIG_GENERIC_PENDING_IRQ */ + #ifdef CONFIG_GENERIC_IRQ_DEBUGFS void irq_add_debugfs_entry(unsigned int irq, struct irq_desc *desc); void irq_remove_debugfs_entry(struct irq_desc *desc); diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 1e283073cecc..7dcf19397c39 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -168,34 +168,6 @@ void irq_set_thread_affinity(struct irq_desc *desc) set_bit(IRQTF_AFFINITY, &action->thread_flags); } -#ifdef CONFIG_GENERIC_PENDING_IRQ -static inline bool irq_can_move_pcntxt(struct irq_data *data) -{ - return irqd_can_move_in_process_context(data); -} -static inline bool irq_move_pending(struct irq_data *data) -{ - return irqd_is_setaffinity_pending(data); -} -static inline void -irq_copy_pending(struct irq_desc *desc, const struct cpumask *mask) -{ - cpumask_copy(desc->pending_mask, mask); -} -static inline void -irq_get_pending(struct cpumask *mask, struct irq_desc *desc) -{ - cpumask_copy(mask, desc->pending_mask); -} -#else -static inline bool irq_can_move_pcntxt(struct irq_data *data) { return true; } -static inline bool irq_move_pending(struct irq_data *data) { return false; } -static inline void -irq_copy_pending(struct irq_desc *desc, const struct cpumask *mask) { } -static inline void -irq_get_pending(struct cpumask *mask, struct irq_desc *desc) { } -#endif - int irq_do_set_affinity(struct irq_data *data, const struct cpumask *mask, bool force) { From 0dd945ff4647a1f29c6ae8f4f9a69c8f37c994cf Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:25 +0200 Subject: [PATCH 061/112] genirq/cpuhotplug: Remove irq disabling logic This is called from stop_machine() with interrupts disabled. No point in disabling them some more. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235445.198042748@linutronix.de --- kernel/irq/cpuhotplug.c | 12 ++++-------- 1 file changed, 4 insertions(+), 8 deletions(-) diff --git a/kernel/irq/cpuhotplug.c b/kernel/irq/cpuhotplug.c index 011f8c4c63da..705139831590 100644 --- a/kernel/irq/cpuhotplug.c +++ b/kernel/irq/cpuhotplug.c @@ -59,11 +59,8 @@ static bool migrate_one_irq(struct irq_desc *desc) */ void irq_migrate_all_off_this_cpu(void) { - unsigned int irq; struct irq_desc *desc; - unsigned long flags; - - local_irq_save(flags); + unsigned int irq; for_each_active_irq(irq) { bool affinity_broken; @@ -73,10 +70,9 @@ void irq_migrate_all_off_this_cpu(void) affinity_broken = migrate_one_irq(desc); raw_spin_unlock(&desc->lock); - if (affinity_broken) - pr_warn_ratelimited("IRQ%u no longer affine to CPU%u\n", + if (affinity_broken) { + pr_warn_ratelimited("IRQ %u: no longer affine to CPU%u\n", irq, smp_processor_id()); + } } - - local_irq_restore(flags); } From 735c09524d3e7c92315e8e2699a1b9acb4fb415c Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:26 +0200 Subject: [PATCH 062/112] genirq/cpuhotplug: Dont claim success on error In case the affinity of an interrupt was broken, a printk is emitted. But if the affinity cannot be set at all due to a missing irq_set_affinity() callback or due to a failing callback, the message is still printed preceeded by a warning/error. That makes no sense whatsoever. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235445.274852976@linutronix.de --- kernel/irq/cpuhotplug.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/kernel/irq/cpuhotplug.c b/kernel/irq/cpuhotplug.c index 705139831590..9c5521b247d5 100644 --- a/kernel/irq/cpuhotplug.c +++ b/kernel/irq/cpuhotplug.c @@ -37,11 +37,14 @@ static bool migrate_one_irq(struct irq_desc *desc) c = irq_data_get_irq_chip(d); if (!c->irq_set_affinity) { pr_debug("IRQ%u: unable to set affinity\n", d->irq); + ret = false; } else { int r = irq_do_set_affinity(d, affinity, false); - if (r) + if (r) { pr_warn_ratelimited("IRQ%u: set affinity failed(%d).\n", d->irq, r); + ret = false; + } } return ret; From e8a7035039306c90bcc99129ffc18e0be052bbb9 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:27 +0200 Subject: [PATCH 063/112] genirq/cpuhotplug: Reorder check logic Move the checks for a valid irq chip and the irq_set_affinity() callback right in front of the whole migration logic. No point in doing a gazillion of other things when the interrupt cannot be migrated at all. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235445.354181630@linutronix.de --- kernel/irq/cpuhotplug.c | 36 ++++++++++++++++++++---------------- 1 file changed, 20 insertions(+), 16 deletions(-) diff --git a/kernel/irq/cpuhotplug.c b/kernel/irq/cpuhotplug.c index 9c5521b247d5..41fe1e04d5d9 100644 --- a/kernel/irq/cpuhotplug.c +++ b/kernel/irq/cpuhotplug.c @@ -17,9 +17,20 @@ static bool migrate_one_irq(struct irq_desc *desc) { struct irq_data *d = irq_desc_get_irq_data(desc); + struct irq_chip *chip = irq_data_get_irq_chip(d); const struct cpumask *affinity = d->common->affinity; - struct irq_chip *c; - bool ret = false; + bool brokeaff = false; + int err; + + /* + * IRQ chip might be already torn down, but the irq descriptor is + * still in the radix tree. Also if the chip has no affinity setter, + * nothing can be done here. + */ + if (!chip || !chip->irq_set_affinity) { + pr_debug("IRQ %u: Unable to migrate away\n", d->irq); + return false; + } /* * If this is a per-CPU interrupt, or the affinity does not @@ -31,23 +42,16 @@ static bool migrate_one_irq(struct irq_desc *desc) if (cpumask_any_and(affinity, cpu_online_mask) >= nr_cpu_ids) { affinity = cpu_online_mask; - ret = true; + brokeaff = true; } - c = irq_data_get_irq_chip(d); - if (!c->irq_set_affinity) { - pr_debug("IRQ%u: unable to set affinity\n", d->irq); - ret = false; - } else { - int r = irq_do_set_affinity(d, affinity, false); - if (r) { - pr_warn_ratelimited("IRQ%u: set affinity failed(%d).\n", - d->irq, r); - ret = false; - } + err = irq_do_set_affinity(d, affinity, false); + if (err) { + pr_warn_ratelimited("IRQ%u: set affinity failed(%d).\n", + d->irq, err); + return false; } - - return ret; + return brokeaff; } /** From 91f26cb4cd3c22bd656ab46c49329aacaaab5504 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:28 +0200 Subject: [PATCH 064/112] genirq/cpuhotplug: Do not migrated shutdown irqs Interrupts, which are shut down are tried to be migrated as well. That's pointless because the interrupt cannot fire and the next startup will move it to the proper place anyway. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235445.447550992@linutronix.de --- kernel/irq/cpuhotplug.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/kernel/irq/cpuhotplug.c b/kernel/irq/cpuhotplug.c index 41fe1e04d5d9..09b20e127aee 100644 --- a/kernel/irq/cpuhotplug.c +++ b/kernel/irq/cpuhotplug.c @@ -33,10 +33,15 @@ static bool migrate_one_irq(struct irq_desc *desc) } /* - * If this is a per-CPU interrupt, or the affinity does not - * include this CPU, then we have nothing to do. + * No move required, if: + * - Interrupt is per cpu + * - Interrupt is not started + * - Affinity mask does not include this CPU. + * + * Note: Do not check desc->action as this might be a chained + * interrupt. */ - if (irqd_is_per_cpu(d) || + if (irqd_is_per_cpu(d) || !irqd_is_started(d) || !cpumask_test_cpu(smp_processor_id(), affinity)) return false; From f0383c24b4855f6a4b5a358c7b2d2c16e0437e9b Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:29 +0200 Subject: [PATCH 065/112] genirq/cpuhotplug: Add support for cleaning up move in progress In order to move x86 to the generic hotplug migration code, add support for cleaning up move in progress bits. On architectures which have this x86 specific (mis)feature not enabled, this is optimized out by the compiler. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235445.525817311@linutronix.de --- arch/x86/include/asm/irq.h | 1 - include/linux/irq.h | 2 ++ kernel/irq/cpuhotplug.c | 28 ++++++++++++++++++++++++++-- kernel/irq/internals.h | 10 +++++++++- 4 files changed, 37 insertions(+), 4 deletions(-) diff --git a/arch/x86/include/asm/irq.h b/arch/x86/include/asm/irq.h index 16d3fa211962..668cca540025 100644 --- a/arch/x86/include/asm/irq.h +++ b/arch/x86/include/asm/irq.h @@ -29,7 +29,6 @@ struct irq_desc; #include extern int check_irq_vectors_for_cpu_disable(void); extern void fixup_irqs(void); -extern void irq_force_complete_move(struct irq_desc *desc); #endif #ifdef CONFIG_HAVE_KVM diff --git a/include/linux/irq.h b/include/linux/irq.h index d008065e2f4d..299271a4953c 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -491,10 +491,12 @@ extern void irq_migrate_all_off_this_cpu(void); #if defined(CONFIG_SMP) && defined(CONFIG_GENERIC_PENDING_IRQ) void irq_move_irq(struct irq_data *data); void irq_move_masked_irq(struct irq_data *data); +void irq_force_complete_move(struct irq_desc *desc); bool irq_fixup_move_pending(struct irq_desc *desc, bool force_clear); #else static inline void irq_move_irq(struct irq_data *data) { } static inline void irq_move_masked_irq(struct irq_data *data) { } +static inline void irq_force_complete_move(struct irq_desc *desc) { } static inline bool irq_fixup_move_pending(struct irq_desc *desc, bool fclear) { return false; diff --git a/kernel/irq/cpuhotplug.c b/kernel/irq/cpuhotplug.c index 09b20e127aee..4be4bd669d81 100644 --- a/kernel/irq/cpuhotplug.c +++ b/kernel/irq/cpuhotplug.c @@ -18,7 +18,7 @@ static bool migrate_one_irq(struct irq_desc *desc) { struct irq_data *d = irq_desc_get_irq_data(desc); struct irq_chip *chip = irq_data_get_irq_chip(d); - const struct cpumask *affinity = d->common->affinity; + const struct cpumask *affinity; bool brokeaff = false; int err; @@ -41,9 +41,33 @@ static bool migrate_one_irq(struct irq_desc *desc) * Note: Do not check desc->action as this might be a chained * interrupt. */ + affinity = irq_data_get_affinity_mask(d); if (irqd_is_per_cpu(d) || !irqd_is_started(d) || - !cpumask_test_cpu(smp_processor_id(), affinity)) + !cpumask_test_cpu(smp_processor_id(), affinity)) { + /* + * If an irq move is pending, abort it if the dying CPU is + * the sole target. + */ + irq_fixup_move_pending(desc, false); return false; + } + + /* + * Complete an eventually pending irq move cleanup. If this + * interrupt was moved in hard irq context, then the vectors need + * to be cleaned up. It can't wait until this interrupt actually + * happens and this CPU was involved. + */ + irq_force_complete_move(desc); + + /* + * If there is a setaffinity pending, then try to reuse the pending + * mask, so the last change of the affinity does not get lost. If + * there is no move pending or the pending mask does not contain + * any online CPU, use the current affinity mask. + */ + if (irq_fixup_move_pending(desc, true)) + affinity = irq_desc_get_pending_mask(desc); if (cpumask_any_and(affinity, cpu_online_mask) >= nr_cpu_ids) { affinity = cpu_online_mask; diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index 20b197f0a7b5..fd4fa8382b8f 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -268,6 +268,10 @@ irq_get_pending(struct cpumask *mask, struct irq_desc *desc) { cpumask_copy(mask, desc->pending_mask); } +static inline struct cpumask *irq_desc_get_pending_mask(struct irq_desc *desc) +{ + return desc->pending_mask; +} #else /* CONFIG_GENERIC_PENDING_IRQ */ static inline bool irq_can_move_pcntxt(struct irq_data *data) { @@ -285,7 +289,11 @@ static inline void irq_get_pending(struct cpumask *mask, struct irq_desc *desc) { } -#endif /* CONFIG_GENERIC_PENDING_IRQ */ +static inline struct cpumask *irq_desc_get_pending_mask(struct irq_desc *desc) +{ + return NULL; +} +#endif /* !CONFIG_GENERIC_PENDING_IRQ */ #ifdef CONFIG_GENERIC_IRQ_DEBUGFS void irq_add_debugfs_entry(unsigned int irq, struct irq_desc *desc); From 47a06d3a783217acae02976f15ca07ddc1ac024f Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:30 +0200 Subject: [PATCH 066/112] genirq/cpuhotplug: Add support for conditional masking Interrupts which cannot be migrated in process context, need to be masked before the affinity is changed forcefully. Add support for that. Will be compiled out for architectures which do not have this x86 specific issue. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235445.604565591@linutronix.de --- kernel/irq/cpuhotplug.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-) diff --git a/kernel/irq/cpuhotplug.c b/kernel/irq/cpuhotplug.c index 4be4bd669d81..6f46587a9ce5 100644 --- a/kernel/irq/cpuhotplug.c +++ b/kernel/irq/cpuhotplug.c @@ -18,6 +18,7 @@ static bool migrate_one_irq(struct irq_desc *desc) { struct irq_data *d = irq_desc_get_irq_data(desc); struct irq_chip *chip = irq_data_get_irq_chip(d); + bool maskchip = !irq_can_move_pcntxt(d) && !irqd_irq_masked(d); const struct cpumask *affinity; bool brokeaff = false; int err; @@ -69,6 +70,10 @@ static bool migrate_one_irq(struct irq_desc *desc) if (irq_fixup_move_pending(desc, true)) affinity = irq_desc_get_pending_mask(desc); + /* Mask the chip for interrupts which cannot move in process context */ + if (maskchip && chip->irq_mask) + chip->irq_mask(d); + if (cpumask_any_and(affinity, cpu_online_mask) >= nr_cpu_ids) { affinity = cpu_online_mask; brokeaff = true; @@ -78,8 +83,12 @@ static bool migrate_one_irq(struct irq_desc *desc) if (err) { pr_warn_ratelimited("IRQ%u: set affinity failed(%d).\n", d->irq, err); - return false; + brokeaff = false; } + + if (maskchip && chip->irq_unmask) + chip->irq_unmask(d); + return brokeaff; } From 77f85e66aa8be563ae5804eebf74a78ec6ef5555 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:31 +0200 Subject: [PATCH 067/112] genirq/cpuhotplug: Set force affinity flag on hotplug migration Set the force migration flag when migrating interrupts away from an outgoing CPU. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235445.681874648@linutronix.de --- kernel/irq/cpuhotplug.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/irq/cpuhotplug.c b/kernel/irq/cpuhotplug.c index 6f46587a9ce5..e09cb91a7c8b 100644 --- a/kernel/irq/cpuhotplug.c +++ b/kernel/irq/cpuhotplug.c @@ -79,7 +79,7 @@ static bool migrate_one_irq(struct irq_desc *desc) brokeaff = true; } - err = irq_do_set_affinity(d, affinity, false); + err = irq_do_set_affinity(d, affinity, true); if (err) { pr_warn_ratelimited("IRQ%u: set affinity failed(%d).\n", d->irq, err); From 654abd0a7baf144998147787121da0f9422dafc8 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:32 +0200 Subject: [PATCH 068/112] x86/irq: Restructure fixup_irqs() Reorder fixup_irqs() so it matches the flow in the generic migration code. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235445.774272454@linutronix.de --- arch/x86/kernel/irq.c | 46 +++++++++++++++++++------------------------ 1 file changed, 20 insertions(+), 26 deletions(-) diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c index 9696007df67b..78bd2b84eac6 100644 --- a/arch/x86/kernel/irq.c +++ b/arch/x86/kernel/irq.c @@ -433,7 +433,6 @@ int check_irq_vectors_for_cpu_disable(void) void fixup_irqs(void) { unsigned int irq, vector; - static int warned; struct irq_desc *desc; struct irq_data *data; struct irq_chip *chip; @@ -441,18 +440,27 @@ void fixup_irqs(void) for_each_irq_desc(irq, desc) { const struct cpumask *affinity; - int break_affinity = 0; - int set_affinity = 1; + bool break_affinity = false; if (!desc) continue; - if (irq == 2) - continue; /* interrupt's are disabled at this point */ raw_spin_lock(&desc->lock); data = irq_desc_get_irq_data(desc); + chip = irq_data_get_irq_chip(data); + /* + * The interrupt descriptor might have been cleaned up + * already, but it is not yet removed from the radix + * tree. If the chip does not have an affinity setter, + * nothing to do here. + */ + if (!chip && !chip->irq_set_affinity) { + raw_spin_unlock(&desc->lock); + continue; + } + affinity = irq_data_get_affinity_mask(data); if (!irq_has_action(irq) || irqd_is_per_cpu(data) || @@ -485,30 +493,18 @@ void fixup_irqs(void) * affinity and use cpu_online_mask as fall back. */ if (cpumask_any_and(affinity, cpu_online_mask) >= nr_cpu_ids) { - break_affinity = 1; + broke_affinity = true; affinity = cpu_online_mask; } - chip = irq_data_get_irq_chip(data); - /* - * The interrupt descriptor might have been cleaned up - * already, but it is not yet removed from the radix tree - */ - if (!chip) { - raw_spin_unlock(&desc->lock); - continue; - } - if (!irqd_can_move_in_process_context(data) && chip->irq_mask) chip->irq_mask(data); - if (chip->irq_set_affinity) { - ret = chip->irq_set_affinity(data, affinity, true); - if (ret == -ENOSPC) - pr_crit("IRQ %d set affinity failed because there are no available vectors. The device assigned to this IRQ is unstable.\n", irq); - } else { - if (!(warned++)) - set_affinity = 0; + ret = chip->irq_set_affinity(data, affinity, true); + if (ret) { + pr_crit("IRQ %u: Force affinity failed (%d)\n", + d->irq, ret); + broke_affinity = false; } /* @@ -522,10 +518,8 @@ void fixup_irqs(void) raw_spin_unlock(&desc->lock); - if (break_affinity && set_affinity) + if (broke_affinity) pr_notice("Broke affinity for irq %i\n", irq); - else if (!set_affinity) - pr_notice("Cannot set affinity for irq %i\n", irq); } /* From ad7a929fa4bb1143357aa83043a149d5c27c68fd Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:33 +0200 Subject: [PATCH 069/112] x86/irq: Use irq_migrate_all_off_this_cpu() The generic migration code supports all the required features already. Remove the x86 specific implementation and use the generic one. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235445.851311033@linutronix.de --- arch/x86/Kconfig | 1 + arch/x86/kernel/irq.c | 89 +------------------------------------------ 2 files changed, 3 insertions(+), 87 deletions(-) diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 0efb4c9497bc..fcf1dad582f5 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -87,6 +87,7 @@ config X86 select GENERIC_EARLY_IOREMAP select GENERIC_FIND_FIRST_BIT select GENERIC_IOMAP + select GENERIC_IRQ_MIGRATION if SMP select GENERIC_IRQ_PROBE select GENERIC_IRQ_SHOW select GENERIC_PENDING_IRQ if SMP diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c index 78bd2b84eac6..4aa03c5a14c9 100644 --- a/arch/x86/kernel/irq.c +++ b/arch/x86/kernel/irq.c @@ -432,95 +432,12 @@ int check_irq_vectors_for_cpu_disable(void) /* A cpu has been removed from cpu_online_mask. Reset irq affinities. */ void fixup_irqs(void) { - unsigned int irq, vector; + unsigned int irr, vector; struct irq_desc *desc; struct irq_data *data; struct irq_chip *chip; - int ret; - for_each_irq_desc(irq, desc) { - const struct cpumask *affinity; - bool break_affinity = false; - - if (!desc) - continue; - - /* interrupt's are disabled at this point */ - raw_spin_lock(&desc->lock); - - data = irq_desc_get_irq_data(desc); - chip = irq_data_get_irq_chip(data); - /* - * The interrupt descriptor might have been cleaned up - * already, but it is not yet removed from the radix - * tree. If the chip does not have an affinity setter, - * nothing to do here. - */ - if (!chip && !chip->irq_set_affinity) { - raw_spin_unlock(&desc->lock); - continue; - } - - affinity = irq_data_get_affinity_mask(data); - - if (!irq_has_action(irq) || irqd_is_per_cpu(data) || - cpumask_subset(affinity, cpu_online_mask)) { - irq_fixup_move_pending(desc, false); - raw_spin_unlock(&desc->lock); - continue; - } - - /* - * Complete an eventually pending irq move cleanup. If this - * interrupt was moved in hard irq context, then the - * vectors need to be cleaned up. It can't wait until this - * interrupt actually happens and this CPU was involved. - */ - irq_force_complete_move(desc); - - /* - * If there is a setaffinity pending, then try to reuse the - * pending mask, so the last change of the affinity does - * not get lost. If there is no move pending or the pending - * mask does not contain any online CPU, use the current - * affinity mask. - */ - if (irq_fixup_move_pending(desc, true)) - affinity = desc->pending_mask; - - /* - * If the mask does not contain an offline CPU, break - * affinity and use cpu_online_mask as fall back. - */ - if (cpumask_any_and(affinity, cpu_online_mask) >= nr_cpu_ids) { - broke_affinity = true; - affinity = cpu_online_mask; - } - - if (!irqd_can_move_in_process_context(data) && chip->irq_mask) - chip->irq_mask(data); - - ret = chip->irq_set_affinity(data, affinity, true); - if (ret) { - pr_crit("IRQ %u: Force affinity failed (%d)\n", - d->irq, ret); - broke_affinity = false; - } - - /* - * We unmask if the irq was not marked masked by the - * core code. That respects the lazy irq disable - * behaviour. - */ - if (!irqd_can_move_in_process_context(data) && - !irqd_irq_masked(data) && chip->irq_unmask) - chip->irq_unmask(data); - - raw_spin_unlock(&desc->lock); - - if (broke_affinity) - pr_notice("Broke affinity for irq %i\n", irq); - } + irq_migrate_all_off_this_cpu(); /* * We can remove mdelay() and then send spuriuous interrupts to @@ -539,8 +456,6 @@ void fixup_irqs(void) * nothing else will touch it. */ for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; vector++) { - unsigned int irr; - if (IS_ERR_OR_NULL(__this_cpu_read(vector_irq[vector]))) continue; From 36d84fb45140f151fa4e145381dbce5e5ffed24d Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:34 +0200 Subject: [PATCH 070/112] genirq: Move irq_fixup_move_pending() to core Now that x86 uses the generic code, the function declaration and inline stub can move to the core internal header. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235445.928156166@linutronix.de --- include/linux/irq.h | 5 ----- kernel/irq/internals.h | 5 +++++ 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/include/linux/irq.h b/include/linux/irq.h index 299271a4953c..2b7e5a70d05f 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -492,15 +492,10 @@ extern void irq_migrate_all_off_this_cpu(void); void irq_move_irq(struct irq_data *data); void irq_move_masked_irq(struct irq_data *data); void irq_force_complete_move(struct irq_desc *desc); -bool irq_fixup_move_pending(struct irq_desc *desc, bool force_clear); #else static inline void irq_move_irq(struct irq_data *data) { } static inline void irq_move_masked_irq(struct irq_data *data) { } static inline void irq_force_complete_move(struct irq_desc *desc) { } -static inline bool irq_fixup_move_pending(struct irq_desc *desc, bool fclear) -{ - return false; -} #endif extern int no_irq_affinity; diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index fd4fa8382b8f..040806f1124c 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -272,6 +272,7 @@ static inline struct cpumask *irq_desc_get_pending_mask(struct irq_desc *desc) { return desc->pending_mask; } +bool irq_fixup_move_pending(struct irq_desc *desc, bool force_clear); #else /* CONFIG_GENERIC_PENDING_IRQ */ static inline bool irq_can_move_pcntxt(struct irq_data *data) { @@ -293,6 +294,10 @@ static inline struct cpumask *irq_desc_get_pending_mask(struct irq_desc *desc) { return NULL; } +static inline bool irq_fixup_move_pending(struct irq_desc *desc, bool fclear) +{ + return false; +} #endif /* !CONFIG_GENERIC_PENDING_IRQ */ #ifdef CONFIG_GENERIC_IRQ_DEBUGFS From 047dc6331de58da51818582c0db0dbfcb837e614 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:35 +0200 Subject: [PATCH 071/112] genirq: Remove pointless arg from show_irq_affinity The third argument of the internal helper function is unused. Remove it. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235446.004958600@linutronix.de --- kernel/irq/proc.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/kernel/irq/proc.c b/kernel/irq/proc.c index d35bb8d4c317..eff7c0c8f9b9 100644 --- a/kernel/irq/proc.c +++ b/kernel/irq/proc.c @@ -37,7 +37,7 @@ static struct proc_dir_entry *root_irq_dir; #ifdef CONFIG_SMP -static int show_irq_affinity(int type, struct seq_file *m, void *v) +static int show_irq_affinity(int type, struct seq_file *m) { struct irq_desc *desc = irq_to_desc((long)m->private); const struct cpumask *mask = desc->irq_common_data.affinity; @@ -80,12 +80,12 @@ static int irq_affinity_hint_proc_show(struct seq_file *m, void *v) int no_irq_affinity; static int irq_affinity_proc_show(struct seq_file *m, void *v) { - return show_irq_affinity(0, m, v); + return show_irq_affinity(0, m); } static int irq_affinity_list_proc_show(struct seq_file *m, void *v) { - return show_irq_affinity(1, m, v); + return show_irq_affinity(1, m); } From 4ab764c336123157690ee0000a1dcf81851c58d1 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:36 +0200 Subject: [PATCH 072/112] genirq: Remove pointless gfp argument All callers hand in GPF_KERNEL. No point to have an extra argument for that. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235446.082544752@linutronix.de --- kernel/irq/irqdesc.c | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-) diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index feade536b6d1..48d4f0365e52 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -54,14 +54,14 @@ static void __init init_irq_default_affinity(void) #endif #ifdef CONFIG_SMP -static int alloc_masks(struct irq_desc *desc, gfp_t gfp, int node) +static int alloc_masks(struct irq_desc *desc, int node) { if (!zalloc_cpumask_var_node(&desc->irq_common_data.affinity, - gfp, node)) + GFP_KERNEL, node)) return -ENOMEM; #ifdef CONFIG_GENERIC_PENDING_IRQ - if (!zalloc_cpumask_var_node(&desc->pending_mask, gfp, node)) { + if (!zalloc_cpumask_var_node(&desc->pending_mask, GFP_KERNEL, node)) { free_cpumask_var(desc->irq_common_data.affinity); return -ENOMEM; } @@ -86,7 +86,7 @@ static void desc_smp_init(struct irq_desc *desc, int node, #else static inline int -alloc_masks(struct irq_desc *desc, gfp_t gfp, int node) { return 0; } +alloc_masks(struct irq_desc *desc, int node) { return 0; } static inline void desc_smp_init(struct irq_desc *desc, int node, const struct cpumask *affinity) { } #endif @@ -344,9 +344,8 @@ static struct irq_desc *alloc_desc(int irq, int node, unsigned int flags, struct module *owner) { struct irq_desc *desc; - gfp_t gfp = GFP_KERNEL; - desc = kzalloc_node(sizeof(*desc), gfp, node); + desc = kzalloc_node(sizeof(*desc), GFP_KERNEL, node); if (!desc) return NULL; /* allocate based on nr_cpu_ids */ @@ -354,7 +353,7 @@ static struct irq_desc *alloc_desc(int irq, int node, unsigned int flags, if (!desc->kstat_irqs) goto err_desc; - if (alloc_masks(desc, gfp, node)) + if (alloc_masks(desc, node)) goto err_kstat; raw_spin_lock_init(&desc->lock); @@ -525,7 +524,7 @@ int __init early_irq_init(void) for (i = 0; i < count; i++) { desc[i].kstat_irqs = alloc_percpu(unsigned int); - alloc_masks(&desc[i], GFP_KERNEL, node); + alloc_masks(&desc[i], node); raw_spin_lock_init(&desc[i].lock); lockdep_set_class(&desc[i].lock, &irq_desc_lock_class); desc_set_defaults(i, &desc[i], node, NULL, NULL); From c1a80386965e9fa3c2f8d1d57966216fe02c9124 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:37 +0200 Subject: [PATCH 073/112] genirq/proc: Replace ever repeating type cast The proc file setup repeats the same ugly type cast for the irq number over and over. Do it once and hand in the local void pointer. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235446.160866358@linutronix.de --- kernel/irq/proc.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/kernel/irq/proc.c b/kernel/irq/proc.c index eff7c0c8f9b9..cbc4c5e377ec 100644 --- a/kernel/irq/proc.c +++ b/kernel/irq/proc.c @@ -326,6 +326,7 @@ void register_handler_proc(unsigned int irq, struct irqaction *action) void register_irq_proc(unsigned int irq, struct irq_desc *desc) { static DEFINE_MUTEX(register_lock); + void __maybe_unused *irqp = (void *)(unsigned long) irq; char name [MAX_NAMELEN]; if (!root_irq_dir || (desc->irq_data.chip == &no_irq_chip)) @@ -351,20 +352,19 @@ void register_irq_proc(unsigned int irq, struct irq_desc *desc) #ifdef CONFIG_SMP /* create /proc/irq//smp_affinity */ proc_create_data("smp_affinity", 0644, desc->dir, - &irq_affinity_proc_fops, (void *)(long)irq); + &irq_affinity_proc_fops, irqp); /* create /proc/irq//affinity_hint */ proc_create_data("affinity_hint", 0444, desc->dir, - &irq_affinity_hint_proc_fops, (void *)(long)irq); + &irq_affinity_hint_proc_fops, irqp); /* create /proc/irq//smp_affinity_list */ proc_create_data("smp_affinity_list", 0644, desc->dir, - &irq_affinity_list_proc_fops, (void *)(long)irq); + &irq_affinity_list_proc_fops, irqp); proc_create_data("node", 0444, desc->dir, - &irq_node_proc_fops, (void *)(long)irq); + &irq_node_proc_fops, irqp); #endif - proc_create_data("spurious", 0444, desc->dir, &irq_spurious_proc_fops, (void *)(long)irq); From 0d3f54257dc300f2db480d6a46b34bdb87f18c1b Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:38 +0200 Subject: [PATCH 074/112] genirq: Introduce effective affinity mask There is currently no way to evaluate the effective affinity mask of a given interrupt. Many irq chips allow only a single target CPU or a subset of CPUs in the affinity mask. Updating the mask at the time of setting the affinity to the subset would be counterproductive because information for cpu hotplug about assigned interrupt affinities gets lost. On CPU hotplug it's also pointless to force migrate an interrupt, which is not targeted at the CPU effectively. But currently the information is not available. Provide a seperate mask to be updated by the irq_chip->irq_set_affinity() implementations. Implement the read only proc files so the user can see the effective mask as well w/o trying to deduce it from /proc/interrupts. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235446.247834245@linutronix.de --- include/linux/irq.h | 29 ++++++++++++++ kernel/irq/Kconfig | 4 ++ kernel/irq/debugfs.c | 4 ++ kernel/irq/irqdesc.c | 14 +++++++ kernel/irq/proc.c | 90 ++++++++++++++++++++++++++++++++++++++++---- 5 files changed, 134 insertions(+), 7 deletions(-) diff --git a/include/linux/irq.h b/include/linux/irq.h index 2b7e5a70d05f..4087ef268ba9 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -137,6 +137,9 @@ struct irq_domain; * @affinity: IRQ affinity on SMP. If this is an IPI * related irq, then this is the mask of the * CPUs to which an IPI can be sent. + * @effective_affinity: The effective IRQ affinity on SMP as some irq + * chips do not allow multi CPU destinations. + * A subset of @affinity. * @msi_desc: MSI descriptor * @ipi_offset: Offset of first IPI target cpu in @affinity. Optional. */ @@ -148,6 +151,9 @@ struct irq_common_data { void *handler_data; struct msi_desc *msi_desc; cpumask_var_t affinity; +#ifdef CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK + cpumask_var_t effective_affinity; +#endif #ifdef CONFIG_GENERIC_IRQ_IPI unsigned int ipi_offset; #endif @@ -737,6 +743,29 @@ static inline struct cpumask *irq_data_get_affinity_mask(struct irq_data *d) return d->common->affinity; } +#ifdef CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK +static inline +struct cpumask *irq_data_get_effective_affinity_mask(struct irq_data *d) +{ + return d->common->effective_affinity; +} +static inline void irq_data_update_effective_affinity(struct irq_data *d, + const struct cpumask *m) +{ + cpumask_copy(d->common->effective_affinity, m); +} +#else +static inline void irq_data_update_effective_affinity(struct irq_data *d, + const struct cpumask *m) +{ +} +static inline +struct cpumask *irq_data_get_effective_affinity_mask(struct irq_data *d) +{ + return d->common->affinity; +} +#endif + unsigned int arch_dynirq_lower_bound(unsigned int from); int __irq_alloc_descs(int irq, unsigned int from, unsigned int cnt, int node, diff --git a/kernel/irq/Kconfig b/kernel/irq/Kconfig index 8d9498e51585..fcbb1d6d51cb 100644 --- a/kernel/irq/Kconfig +++ b/kernel/irq/Kconfig @@ -21,6 +21,10 @@ config GENERIC_IRQ_SHOW config GENERIC_IRQ_SHOW_LEVEL bool +# Supports effective affinity mask +config GENERIC_IRQ_EFFECTIVE_AFF_MASK + bool + # Facility to allocate a hardware interrupt. This is legacy support # and should not be used in new code. Use irq domains instead. config GENERIC_IRQ_LEGACY_ALLOC_HWIRQ diff --git a/kernel/irq/debugfs.c b/kernel/irq/debugfs.c index 50ee2f6593e8..edbef252d0c4 100644 --- a/kernel/irq/debugfs.c +++ b/kernel/irq/debugfs.c @@ -36,6 +36,10 @@ static void irq_debug_show_masks(struct seq_file *m, struct irq_desc *desc) msk = irq_data_get_affinity_mask(data); seq_printf(m, "affinity: %*pbl\n", cpumask_pr_args(msk)); +#ifdef CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK + msk = irq_data_get_effective_affinity_mask(data); + seq_printf(m, "effectiv: %*pbl\n", cpumask_pr_args(msk)); +#endif #ifdef CONFIG_GENERIC_PENDING_IRQ msk = desc->pending_mask; seq_printf(m, "pending: %*pbl\n", cpumask_pr_args(msk)); diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 48d4f0365e52..35a95fadcfda 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -60,8 +60,19 @@ static int alloc_masks(struct irq_desc *desc, int node) GFP_KERNEL, node)) return -ENOMEM; +#ifdef CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK + if (!zalloc_cpumask_var_node(&desc->irq_common_data.effective_affinity, + GFP_KERNEL, node)) { + free_cpumask_var(desc->irq_common_data.affinity); + return -ENOMEM; + } +#endif + #ifdef CONFIG_GENERIC_PENDING_IRQ if (!zalloc_cpumask_var_node(&desc->pending_mask, GFP_KERNEL, node)) { +#ifdef CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK + free_cpumask_var(desc->irq_common_data.effective_affinity); +#endif free_cpumask_var(desc->irq_common_data.affinity); return -ENOMEM; } @@ -324,6 +335,9 @@ static void free_masks(struct irq_desc *desc) free_cpumask_var(desc->pending_mask); #endif free_cpumask_var(desc->irq_common_data.affinity); +#ifdef CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK + free_cpumask_var(desc->irq_common_data.effective_affinity); +#endif } #else static inline void free_masks(struct irq_desc *desc) { } diff --git a/kernel/irq/proc.c b/kernel/irq/proc.c index cbc4c5e377ec..7f9642a1e267 100644 --- a/kernel/irq/proc.c +++ b/kernel/irq/proc.c @@ -37,19 +37,47 @@ static struct proc_dir_entry *root_irq_dir; #ifdef CONFIG_SMP +enum { + AFFINITY, + AFFINITY_LIST, + EFFECTIVE, + EFFECTIVE_LIST, +}; + static int show_irq_affinity(int type, struct seq_file *m) { struct irq_desc *desc = irq_to_desc((long)m->private); - const struct cpumask *mask = desc->irq_common_data.affinity; + const struct cpumask *mask; + switch (type) { + case AFFINITY: + case AFFINITY_LIST: + mask = desc->irq_common_data.affinity; #ifdef CONFIG_GENERIC_PENDING_IRQ - if (irqd_is_setaffinity_pending(&desc->irq_data)) - mask = desc->pending_mask; + if (irqd_is_setaffinity_pending(&desc->irq_data)) + mask = desc->pending_mask; #endif - if (type) + break; + case EFFECTIVE: + case EFFECTIVE_LIST: +#ifdef CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK + mask = desc->irq_common_data.effective_affinity; + break; +#else + return -EINVAL; +#endif + }; + + switch (type) { + case AFFINITY_LIST: + case EFFECTIVE_LIST: seq_printf(m, "%*pbl\n", cpumask_pr_args(mask)); - else + break; + case AFFINITY: + case EFFECTIVE: seq_printf(m, "%*pb\n", cpumask_pr_args(mask)); + break; + } return 0; } @@ -80,12 +108,12 @@ static int irq_affinity_hint_proc_show(struct seq_file *m, void *v) int no_irq_affinity; static int irq_affinity_proc_show(struct seq_file *m, void *v) { - return show_irq_affinity(0, m); + return show_irq_affinity(AFFINITY, m); } static int irq_affinity_list_proc_show(struct seq_file *m, void *v) { - return show_irq_affinity(1, m); + return show_irq_affinity(AFFINITY_LIST, m); } @@ -185,6 +213,44 @@ static const struct file_operations irq_affinity_list_proc_fops = { .write = irq_affinity_list_proc_write, }; +#ifdef CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK +static int irq_effective_aff_proc_show(struct seq_file *m, void *v) +{ + return show_irq_affinity(EFFECTIVE, m); +} + +static int irq_effective_aff_list_proc_show(struct seq_file *m, void *v) +{ + return show_irq_affinity(EFFECTIVE_LIST, m); +} + +static int irq_effective_aff_proc_open(struct inode *inode, struct file *file) +{ + return single_open(file, irq_effective_aff_proc_show, PDE_DATA(inode)); +} + +static int irq_effective_aff_list_proc_open(struct inode *inode, + struct file *file) +{ + return single_open(file, irq_effective_aff_list_proc_show, + PDE_DATA(inode)); +} + +static const struct file_operations irq_effective_aff_proc_fops = { + .open = irq_effective_aff_proc_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +static const struct file_operations irq_effective_aff_list_proc_fops = { + .open = irq_effective_aff_list_proc_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; +#endif + static int default_affinity_show(struct seq_file *m, void *v) { seq_printf(m, "%*pb\n", cpumask_pr_args(irq_default_affinity)); @@ -364,6 +430,12 @@ void register_irq_proc(unsigned int irq, struct irq_desc *desc) proc_create_data("node", 0444, desc->dir, &irq_node_proc_fops, irqp); +# ifdef CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK + proc_create_data("effective_affinity", 0444, desc->dir, + &irq_effective_aff_proc_fops, irqp); + proc_create_data("effective_affinity_list", 0444, desc->dir, + &irq_effective_aff_list_proc_fops, irqp); +# endif #endif proc_create_data("spurious", 0444, desc->dir, &irq_spurious_proc_fops, (void *)(long)irq); @@ -383,6 +455,10 @@ void unregister_irq_proc(unsigned int irq, struct irq_desc *desc) remove_proc_entry("affinity_hint", desc->dir); remove_proc_entry("smp_affinity_list", desc->dir); remove_proc_entry("node", desc->dir); +# ifdef CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK + remove_proc_entry("effective_affinity", desc->dir); + remove_proc_entry("effective_affinity_list", desc->dir); +# endif #endif remove_proc_entry("spurious", desc->dir); From 415fcf1a2293046e0c1f4ab8558a87bad66652b1 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:39 +0200 Subject: [PATCH 075/112] genirq/cpuhotplug: Use effective affinity mask If the architecture supports the effective affinity mask, migrating interrupts away which are not targeted by the effective mask is pointless. They can stay in the user or system supplied affinity mask, but won't be targetted at any given point as the affinity setter functions need to validate against the online cpu mask anyway. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235446.328488490@linutronix.de --- kernel/irq/cpuhotplug.c | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-) diff --git a/kernel/irq/cpuhotplug.c b/kernel/irq/cpuhotplug.c index e09cb91a7c8b..0b093db3336b 100644 --- a/kernel/irq/cpuhotplug.c +++ b/kernel/irq/cpuhotplug.c @@ -14,6 +14,14 @@ #include "internals.h" +/* For !GENERIC_IRQ_EFFECTIVE_AFF_MASK this looks at general affinity mask */ +static inline bool irq_needs_fixup(struct irq_data *d) +{ + const struct cpumask *m = irq_data_get_effective_affinity_mask(d); + + return cpumask_test_cpu(smp_processor_id(), m); +} + static bool migrate_one_irq(struct irq_desc *desc) { struct irq_data *d = irq_desc_get_irq_data(desc); @@ -42,9 +50,7 @@ static bool migrate_one_irq(struct irq_desc *desc) * Note: Do not check desc->action as this might be a chained * interrupt. */ - affinity = irq_data_get_affinity_mask(d); - if (irqd_is_per_cpu(d) || !irqd_is_started(d) || - !cpumask_test_cpu(smp_processor_id(), affinity)) { + if (irqd_is_per_cpu(d) || !irqd_is_started(d) || !irq_needs_fixup(d)) { /* * If an irq move is pending, abort it if the dying CPU is * the sole target. @@ -69,6 +75,8 @@ static bool migrate_one_irq(struct irq_desc *desc) */ if (irq_fixup_move_pending(desc, true)) affinity = irq_desc_get_pending_mask(desc); + else + affinity = irq_data_get_affinity_mask(d); /* Mask the chip for interrupts which cannot move in process context */ if (maskchip && chip->irq_mask) From ad95212ee6e0b62f38b287b40c9ab6a1ba3e892b Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:40 +0200 Subject: [PATCH 076/112] x86/apic: Move flat_cpu_mask_to_apicid_and() into C source No point in having inlines assigned to function pointers at multiple places. Just bloats the text. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235446.405975721@linutronix.de --- arch/x86/include/asm/apic.h | 28 ++++++---------------------- arch/x86/kernel/apic/apic.c | 16 ++++++++++++++++ 2 files changed, 22 insertions(+), 22 deletions(-) diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h index bdffcd9eab2b..a86be0a082d0 100644 --- a/arch/x86/include/asm/apic.h +++ b/arch/x86/include/asm/apic.h @@ -540,28 +540,12 @@ static inline int default_phys_pkg_id(int cpuid_apic, int index_msb) #endif -static inline int -flat_cpu_mask_to_apicid_and(const struct cpumask *cpumask, - const struct cpumask *andmask, - unsigned int *apicid) -{ - unsigned long cpu_mask = cpumask_bits(cpumask)[0] & - cpumask_bits(andmask)[0] & - cpumask_bits(cpu_online_mask)[0] & - APIC_ALL_CPUS; - - if (likely(cpu_mask)) { - *apicid = (unsigned int)cpu_mask; - return 0; - } else { - return -EINVAL; - } -} - -extern int -default_cpu_mask_to_apicid_and(const struct cpumask *cpumask, - const struct cpumask *andmask, - unsigned int *apicid); +extern int flat_cpu_mask_to_apicid_and(const struct cpumask *cpumask, + const struct cpumask *andmask, + unsigned int *apicid); +extern int default_cpu_mask_to_apicid_and(const struct cpumask *cpumask, + const struct cpumask *andmask, + unsigned int *apicid); static inline void flat_vector_allocation_domain(int cpu, struct cpumask *retmask, diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 2d75faf743f2..e9b322f05051 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -2220,6 +2220,22 @@ int default_cpu_mask_to_apicid_and(const struct cpumask *cpumask, return -EINVAL; } +int flat_cpu_mask_to_apicid_and(const struct cpumask *cpumask, + const struct cpumask *andmask, + unsigned int *apicid) +{ + unsigned long cpu_mask = cpumask_bits(cpumask)[0] & + cpumask_bits(andmask)[0] & + cpumask_bits(cpu_online_mask)[0] & + APIC_ALL_CPUS; + + if (likely(cpu_mask)) { + *apicid = (unsigned int)cpu_mask; + return 0; + } + return -EINVAL; +} + /* * Override the generic EOI implementation with an optimized version. * Only called during early boot when only one CPU is active and with From bbcf9574bc6fb85d22f2718d48da7f98830a7870 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:41 +0200 Subject: [PATCH 077/112] x86/uv: Use default_cpu_mask_to_apicid_and() Same functionality except the extra bits ored on the apicid. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235446.482841015@linutronix.de --- arch/x86/kernel/apic/x2apic_uv_x.c | 19 ++++--------------- 1 file changed, 4 insertions(+), 15 deletions(-) diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c index b487b3a01615..fd5bb20c8430 100644 --- a/arch/x86/kernel/apic/x2apic_uv_x.c +++ b/arch/x86/kernel/apic/x2apic_uv_x.c @@ -530,23 +530,12 @@ uv_cpu_mask_to_apicid_and(const struct cpumask *cpumask, const struct cpumask *andmask, unsigned int *apicid) { - int unsigned cpu; + int ret = default_cpu_mask_to_apicid_and(cpumask, andmask, apicid); - /* - * We're using fixed IRQ delivery, can only return one phys APIC ID. - * May as well be the first. - */ - for_each_cpu_and(cpu, cpumask, andmask) { - if (cpumask_test_cpu(cpu, cpu_online_mask)) - break; - } + if (!ret) + *apicid |= uv_apicid_hibits; - if (likely(cpu < nr_cpu_ids)) { - *apicid = per_cpu(x86_cpu_to_apicid, cpu) | uv_apicid_hibits; - return 0; - } - - return -EINVAL; + return ret; } static unsigned int x2apic_get_apic_id(unsigned long x) From 52b166af40faec9813cd5ac26d6ba9adec2e3a9d Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:42 +0200 Subject: [PATCH 078/112] x86/apic: Move online masking to core code All implementations of apic->cpu_mask_to_apicid_and() mask out the offline cpus. The callsite already has a mask available, which has the offline CPUs removed. Use that and remove the extra bits. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235446.560868224@linutronix.de --- arch/x86/kernel/apic/apic.c | 27 +++++++++------------------ arch/x86/kernel/apic/vector.c | 5 ++++- arch/x86/kernel/apic/x2apic_cluster.c | 25 +++++++++---------------- 3 files changed, 22 insertions(+), 35 deletions(-) diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index e9b322f05051..8a0bde3fc488 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -2205,19 +2205,12 @@ int default_cpu_mask_to_apicid_and(const struct cpumask *cpumask, const struct cpumask *andmask, unsigned int *apicid) { - unsigned int cpu; + unsigned int cpu = cpumask_first_and(cpumask, andmask); - for_each_cpu_and(cpu, cpumask, andmask) { - if (cpumask_test_cpu(cpu, cpu_online_mask)) - break; - } - - if (likely(cpu < nr_cpu_ids)) { - *apicid = per_cpu(x86_cpu_to_apicid, cpu); - return 0; - } - - return -EINVAL; + if (cpu >= nr_cpu_ids) + return -EINVAL; + *apicid = per_cpu(x86_cpu_to_apicid, cpu); + return 0; } int flat_cpu_mask_to_apicid_and(const struct cpumask *cpumask, @@ -2226,14 +2219,12 @@ int flat_cpu_mask_to_apicid_and(const struct cpumask *cpumask, { unsigned long cpu_mask = cpumask_bits(cpumask)[0] & cpumask_bits(andmask)[0] & - cpumask_bits(cpu_online_mask)[0] & APIC_ALL_CPUS; - if (likely(cpu_mask)) { - *apicid = (unsigned int)cpu_mask; - return 0; - } - return -EINVAL; + if (!cpu_mask) + return -EINVAL; + *apicid = (unsigned int)cpu_mask; + return 0; } /* diff --git a/arch/x86/kernel/apic/vector.c b/arch/x86/kernel/apic/vector.c index 47c5d019fb7e..0f94ddbb6beb 100644 --- a/arch/x86/kernel/apic/vector.c +++ b/arch/x86/kernel/apic/vector.c @@ -221,8 +221,11 @@ success: * Cache destination APIC IDs into cfg->dest_apicid. This cannot fail * as we already established, that mask & d->domain & cpu_online_mask * is not empty. + * + * vector_searchmask is a subset of d->domain and has the offline + * cpus masked out. */ - BUG_ON(apic->cpu_mask_to_apicid_and(mask, d->domain, + BUG_ON(apic->cpu_mask_to_apicid_and(mask, vector_searchmask, &d->cfg.dest_apicid)); return 0; } diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c index 5a35f208ed95..d73baa8c1a17 100644 --- a/arch/x86/kernel/apic/x2apic_cluster.c +++ b/arch/x86/kernel/apic/x2apic_cluster.c @@ -108,31 +108,24 @@ x2apic_cpu_mask_to_apicid_and(const struct cpumask *cpumask, const struct cpumask *andmask, unsigned int *apicid) { + unsigned int cpu; u32 dest = 0; u16 cluster; - int i; - for_each_cpu_and(i, cpumask, andmask) { - if (!cpumask_test_cpu(i, cpu_online_mask)) - continue; - dest = per_cpu(x86_cpu_to_logical_apicid, i); - cluster = x2apic_cluster(i); - break; - } - - if (!dest) + cpu = cpumask_first_and(cpumask, andmask); + if (cpu >= nr_cpu_ids) return -EINVAL; - for_each_cpu_and(i, cpumask, andmask) { - if (!cpumask_test_cpu(i, cpu_online_mask)) + dest = per_cpu(x86_cpu_to_logical_apicid, cpu); + cluster = x2apic_cluster(cpu); + + for_each_cpu_and(cpu, cpumask, andmask) { + if (cluster != x2apic_cluster(cpu)) continue; - if (cluster != x2apic_cluster(i)) - continue; - dest |= per_cpu(x86_cpu_to_logical_apicid, i); + dest |= per_cpu(x86_cpu_to_logical_apicid, cpu); } *apicid = dest; - return 0; } From 91cd9cb7ee1c081304d0e61f09e9faccb33d3df7 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:43 +0200 Subject: [PATCH 079/112] x86/apic: Move cpumask and to core code All implementations of apic->cpu_mask_to_apicid_and() and the two incoming cpumasks to search for the target. Move that operation to the call site and rename it to cpu_mask_to_apicid() Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235446.641575516@linutronix.de --- arch/x86/include/asm/apic.h | 15 ++++++--------- arch/x86/kernel/apic/apic.c | 14 ++++---------- arch/x86/kernel/apic/apic_flat_64.c | 4 ++-- arch/x86/kernel/apic/apic_noop.c | 2 +- arch/x86/kernel/apic/apic_numachip.c | 4 ++-- arch/x86/kernel/apic/bigsmp_32.c | 2 +- arch/x86/kernel/apic/probe_32.c | 2 +- arch/x86/kernel/apic/vector.c | 6 +++--- arch/x86/kernel/apic/x2apic_cluster.c | 10 ++++------ arch/x86/kernel/apic/x2apic_phys.c | 2 +- arch/x86/kernel/apic/x2apic_uv_x.c | 8 +++----- arch/x86/xen/apic.c | 2 +- 12 files changed, 29 insertions(+), 42 deletions(-) diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h index a86be0a082d0..3e64e99edc36 100644 --- a/arch/x86/include/asm/apic.h +++ b/arch/x86/include/asm/apic.h @@ -296,9 +296,8 @@ struct apic { /* Can't be NULL on 64-bit */ unsigned long (*set_apic_id)(unsigned int id); - int (*cpu_mask_to_apicid_and)(const struct cpumask *cpumask, - const struct cpumask *andmask, - unsigned int *apicid); + int (*cpu_mask_to_apicid)(const struct cpumask *cpumask, + unsigned int *apicid); /* ipi */ void (*send_IPI)(int cpu, int vector); @@ -540,12 +539,10 @@ static inline int default_phys_pkg_id(int cpuid_apic, int index_msb) #endif -extern int flat_cpu_mask_to_apicid_and(const struct cpumask *cpumask, - const struct cpumask *andmask, - unsigned int *apicid); -extern int default_cpu_mask_to_apicid_and(const struct cpumask *cpumask, - const struct cpumask *andmask, - unsigned int *apicid); +extern int flat_cpu_mask_to_apicid(const struct cpumask *cpumask, + unsigned int *apicid); +extern int default_cpu_mask_to_apicid(const struct cpumask *cpumask, + unsigned int *apicid); static inline void flat_vector_allocation_domain(int cpu, struct cpumask *retmask, diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 8a0bde3fc488..169dd422dc30 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -2201,11 +2201,9 @@ void default_init_apic_ldr(void) apic_write(APIC_LDR, val); } -int default_cpu_mask_to_apicid_and(const struct cpumask *cpumask, - const struct cpumask *andmask, - unsigned int *apicid) +int default_cpu_mask_to_apicid(const struct cpumask *mask, unsigned int *apicid) { - unsigned int cpu = cpumask_first_and(cpumask, andmask); + unsigned int cpu = cpumask_first(mask); if (cpu >= nr_cpu_ids) return -EINVAL; @@ -2213,13 +2211,9 @@ int default_cpu_mask_to_apicid_and(const struct cpumask *cpumask, return 0; } -int flat_cpu_mask_to_apicid_and(const struct cpumask *cpumask, - const struct cpumask *andmask, - unsigned int *apicid) +int flat_cpu_mask_to_apicid(const struct cpumask *mask, unsigned int *apicid) { - unsigned long cpu_mask = cpumask_bits(cpumask)[0] & - cpumask_bits(andmask)[0] & - APIC_ALL_CPUS; + unsigned long cpu_mask = cpumask_bits(mask)[0] & APIC_ALL_CPUS; if (!cpu_mask) return -EINVAL; diff --git a/arch/x86/kernel/apic/apic_flat_64.c b/arch/x86/kernel/apic/apic_flat_64.c index a4d7ff20ed22..dedd5a41ba48 100644 --- a/arch/x86/kernel/apic/apic_flat_64.c +++ b/arch/x86/kernel/apic/apic_flat_64.c @@ -172,7 +172,7 @@ static struct apic apic_flat __ro_after_init = { .get_apic_id = flat_get_apic_id, .set_apic_id = set_apic_id, - .cpu_mask_to_apicid_and = flat_cpu_mask_to_apicid_and, + .cpu_mask_to_apicid = flat_cpu_mask_to_apicid, .send_IPI = default_send_IPI_single, .send_IPI_mask = flat_send_IPI_mask, @@ -268,7 +268,7 @@ static struct apic apic_physflat __ro_after_init = { .get_apic_id = flat_get_apic_id, .set_apic_id = set_apic_id, - .cpu_mask_to_apicid_and = default_cpu_mask_to_apicid_and, + .cpu_mask_to_apicid = default_cpu_mask_to_apicid, .send_IPI = default_send_IPI_single_phys, .send_IPI_mask = default_send_IPI_mask_sequence_phys, diff --git a/arch/x86/kernel/apic/apic_noop.c b/arch/x86/kernel/apic/apic_noop.c index 2262eb6df796..6599f437b4ab 100644 --- a/arch/x86/kernel/apic/apic_noop.c +++ b/arch/x86/kernel/apic/apic_noop.c @@ -141,7 +141,7 @@ struct apic apic_noop __ro_after_init = { .get_apic_id = noop_get_apic_id, .set_apic_id = NULL, - .cpu_mask_to_apicid_and = flat_cpu_mask_to_apicid_and, + .cpu_mask_to_apicid = flat_cpu_mask_to_apicid, .send_IPI = noop_send_IPI, .send_IPI_mask = noop_send_IPI_mask, diff --git a/arch/x86/kernel/apic/apic_numachip.c b/arch/x86/kernel/apic/apic_numachip.c index e08fe2c8dd8c..2fda912219a6 100644 --- a/arch/x86/kernel/apic/apic_numachip.c +++ b/arch/x86/kernel/apic/apic_numachip.c @@ -267,7 +267,7 @@ static const struct apic apic_numachip1 __refconst = { .get_apic_id = numachip1_get_apic_id, .set_apic_id = numachip1_set_apic_id, - .cpu_mask_to_apicid_and = default_cpu_mask_to_apicid_and, + .cpu_mask_to_apicid = default_cpu_mask_to_apicid, .send_IPI = numachip_send_IPI_one, .send_IPI_mask = numachip_send_IPI_mask, @@ -318,7 +318,7 @@ static const struct apic apic_numachip2 __refconst = { .get_apic_id = numachip2_get_apic_id, .set_apic_id = numachip2_set_apic_id, - .cpu_mask_to_apicid_and = default_cpu_mask_to_apicid_and, + .cpu_mask_to_apicid = default_cpu_mask_to_apicid, .send_IPI = numachip_send_IPI_one, .send_IPI_mask = numachip_send_IPI_mask, diff --git a/arch/x86/kernel/apic/bigsmp_32.c b/arch/x86/kernel/apic/bigsmp_32.c index 56012010332c..456e45e8bf84 100644 --- a/arch/x86/kernel/apic/bigsmp_32.c +++ b/arch/x86/kernel/apic/bigsmp_32.c @@ -172,7 +172,7 @@ static struct apic apic_bigsmp __ro_after_init = { .get_apic_id = bigsmp_get_apic_id, .set_apic_id = NULL, - .cpu_mask_to_apicid_and = default_cpu_mask_to_apicid_and, + .cpu_mask_to_apicid = default_cpu_mask_to_apicid, .send_IPI = default_send_IPI_single_phys, .send_IPI_mask = default_send_IPI_mask_sequence_phys, diff --git a/arch/x86/kernel/apic/probe_32.c b/arch/x86/kernel/apic/probe_32.c index 2e8f7f048f4f..63287659adb6 100644 --- a/arch/x86/kernel/apic/probe_32.c +++ b/arch/x86/kernel/apic/probe_32.c @@ -102,7 +102,7 @@ static struct apic apic_default __ro_after_init = { .get_apic_id = default_get_apic_id, .set_apic_id = NULL, - .cpu_mask_to_apicid_and = flat_cpu_mask_to_apicid_and, + .cpu_mask_to_apicid = flat_cpu_mask_to_apicid, .send_IPI = default_send_IPI_single, .send_IPI_mask = default_send_IPI_mask_logical, diff --git a/arch/x86/kernel/apic/vector.c b/arch/x86/kernel/apic/vector.c index 0f94ddbb6beb..1f57f5a08c44 100644 --- a/arch/x86/kernel/apic/vector.c +++ b/arch/x86/kernel/apic/vector.c @@ -141,7 +141,7 @@ static int __assign_irq_vector(int irq, struct apic_chip_data *d, /* * Clear the offline cpus from @vector_cpumask for searching * and verify whether the result overlaps with @mask. If true, - * then the call to apic->cpu_mask_to_apicid_and() will + * then the call to apic->cpu_mask_to_apicid() will * succeed as well. If not, no point in trying to find a * vector in this mask. */ @@ -225,8 +225,8 @@ success: * vector_searchmask is a subset of d->domain and has the offline * cpus masked out. */ - BUG_ON(apic->cpu_mask_to_apicid_and(mask, vector_searchmask, - &d->cfg.dest_apicid)); + cpumask_and(vector_searchmask, vector_searchmask, mask); + BUG_ON(apic->cpu_mask_to_apicid(vector_searchmask, &d->cfg.dest_apicid)); return 0; } diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c index d73baa8c1a17..61474259bf3f 100644 --- a/arch/x86/kernel/apic/x2apic_cluster.c +++ b/arch/x86/kernel/apic/x2apic_cluster.c @@ -104,22 +104,20 @@ static void x2apic_send_IPI_all(int vector) } static int -x2apic_cpu_mask_to_apicid_and(const struct cpumask *cpumask, - const struct cpumask *andmask, - unsigned int *apicid) +x2apic_cpu_mask_to_apicid(const struct cpumask *mask, unsigned int *apicid) { unsigned int cpu; u32 dest = 0; u16 cluster; - cpu = cpumask_first_and(cpumask, andmask); + cpu = cpumask_first(mask); if (cpu >= nr_cpu_ids) return -EINVAL; dest = per_cpu(x86_cpu_to_logical_apicid, cpu); cluster = x2apic_cluster(cpu); - for_each_cpu_and(cpu, cpumask, andmask) { + for_each_cpu(cpu, mask) { if (cluster != x2apic_cluster(cpu)) continue; dest |= per_cpu(x86_cpu_to_logical_apicid, cpu); @@ -249,7 +247,7 @@ static struct apic apic_x2apic_cluster __ro_after_init = { .get_apic_id = x2apic_get_apic_id, .set_apic_id = x2apic_set_apic_id, - .cpu_mask_to_apicid_and = x2apic_cpu_mask_to_apicid_and, + .cpu_mask_to_apicid = x2apic_cpu_mask_to_apicid, .send_IPI = x2apic_send_IPI, .send_IPI_mask = x2apic_send_IPI_mask, diff --git a/arch/x86/kernel/apic/x2apic_phys.c b/arch/x86/kernel/apic/x2apic_phys.c index ff111f05a314..3baf0c3dc875 100644 --- a/arch/x86/kernel/apic/x2apic_phys.c +++ b/arch/x86/kernel/apic/x2apic_phys.c @@ -127,7 +127,7 @@ static struct apic apic_x2apic_phys __ro_after_init = { .get_apic_id = x2apic_get_apic_id, .set_apic_id = x2apic_set_apic_id, - .cpu_mask_to_apicid_and = default_cpu_mask_to_apicid_and, + .cpu_mask_to_apicid = default_cpu_mask_to_apicid, .send_IPI = x2apic_send_IPI, .send_IPI_mask = x2apic_send_IPI_mask, diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c index fd5bb20c8430..ad0223f013f6 100644 --- a/arch/x86/kernel/apic/x2apic_uv_x.c +++ b/arch/x86/kernel/apic/x2apic_uv_x.c @@ -526,11 +526,9 @@ static void uv_init_apic_ldr(void) } static int -uv_cpu_mask_to_apicid_and(const struct cpumask *cpumask, - const struct cpumask *andmask, - unsigned int *apicid) +uv_cpu_mask_to_apicid(const struct cpumask *mask, unsigned int *apicid) { - int ret = default_cpu_mask_to_apicid_and(cpumask, andmask, apicid); + int ret = default_cpu_mask_to_apicid(mask, apicid); if (!ret) *apicid |= uv_apicid_hibits; @@ -603,7 +601,7 @@ static struct apic apic_x2apic_uv_x __ro_after_init = { .get_apic_id = x2apic_get_apic_id, .set_apic_id = set_apic_id, - .cpu_mask_to_apicid_and = uv_cpu_mask_to_apicid_and, + .cpu_mask_to_apicid = uv_cpu_mask_to_apicid, .send_IPI = uv_send_IPI_one, .send_IPI_mask = uv_send_IPI_mask, diff --git a/arch/x86/xen/apic.c b/arch/x86/xen/apic.c index bcea81f36fc5..b5e48da7fbff 100644 --- a/arch/x86/xen/apic.c +++ b/arch/x86/xen/apic.c @@ -178,7 +178,7 @@ static struct apic xen_pv_apic = { .get_apic_id = xen_get_apic_id, .set_apic_id = xen_set_apic_id, /* Can be NULL on 32-bit. */ - .cpu_mask_to_apicid_and = flat_cpu_mask_to_apicid_and, + .cpu_mask_to_apicid = flat_cpu_mask_to_apicid, #ifdef CONFIG_SMP .send_IPI_mask = xen_send_IPI_mask, From 0e24f7c9f67e218546ad44160d2a12d9d8be0171 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:44 +0200 Subject: [PATCH 080/112] x86/apic: Add irq_data argument to apic->cpu_mask_to_apicid() The decision to which CPUs an interrupt is effectively routed happens in the various apic->cpu_mask_to_apicid() implementations To support effective affinity masks this information needs to be updated in irq_data. Add a pointer to irq_data to the callbacks and feed it through the call chain. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235446.720739075@linutronix.de --- arch/x86/include/asm/apic.h | 5 +++++ arch/x86/kernel/apic/apic.c | 9 +++++++-- arch/x86/kernel/apic/vector.c | 25 +++++++++++++++---------- arch/x86/kernel/apic/x2apic_cluster.c | 3 ++- arch/x86/kernel/apic/x2apic_uv_x.c | 5 +++-- 5 files changed, 32 insertions(+), 15 deletions(-) diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h index 3e64e99edc36..5f01671c68f2 100644 --- a/arch/x86/include/asm/apic.h +++ b/arch/x86/include/asm/apic.h @@ -252,6 +252,8 @@ static inline int x2apic_enabled(void) { return 0; } #define x2apic_supported() (0) #endif /* !CONFIG_X86_X2APIC */ +struct irq_data; + /* * Copyright 2004 James Cleverdon, IBM. * Subject to the GNU Public License, v.2 @@ -297,6 +299,7 @@ struct apic { unsigned long (*set_apic_id)(unsigned int id); int (*cpu_mask_to_apicid)(const struct cpumask *cpumask, + struct irq_data *irqdata, unsigned int *apicid); /* ipi */ @@ -540,8 +543,10 @@ static inline int default_phys_pkg_id(int cpuid_apic, int index_msb) #endif extern int flat_cpu_mask_to_apicid(const struct cpumask *cpumask, + struct irq_data *irqdata, unsigned int *apicid); extern int default_cpu_mask_to_apicid(const struct cpumask *cpumask, + struct irq_data *irqdata, unsigned int *apicid); static inline void diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 169dd422dc30..14e5a47192b7 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -2201,7 +2201,9 @@ void default_init_apic_ldr(void) apic_write(APIC_LDR, val); } -int default_cpu_mask_to_apicid(const struct cpumask *mask, unsigned int *apicid) +int default_cpu_mask_to_apicid(const struct cpumask *mask, + struct irq_data *irqdata, + unsigned int *apicid) { unsigned int cpu = cpumask_first(mask); @@ -2211,7 +2213,10 @@ int default_cpu_mask_to_apicid(const struct cpumask *mask, unsigned int *apicid) return 0; } -int flat_cpu_mask_to_apicid(const struct cpumask *mask, unsigned int *apicid) +int flat_cpu_mask_to_apicid(const struct cpumask *mask, + struct irq_data *irqdata, + unsigned int *apicid) + { unsigned long cpu_mask = cpumask_bits(mask)[0] & APIC_ALL_CPUS; diff --git a/arch/x86/kernel/apic/vector.c b/arch/x86/kernel/apic/vector.c index 1f57f5a08c44..b270a76ee640 100644 --- a/arch/x86/kernel/apic/vector.c +++ b/arch/x86/kernel/apic/vector.c @@ -103,7 +103,8 @@ static void free_apic_chip_data(struct apic_chip_data *data) } static int __assign_irq_vector(int irq, struct apic_chip_data *d, - const struct cpumask *mask) + const struct cpumask *mask, + struct irq_data *irqdata) { /* * NOTE! The local APIC isn't very good at handling @@ -226,32 +227,35 @@ success: * cpus masked out. */ cpumask_and(vector_searchmask, vector_searchmask, mask); - BUG_ON(apic->cpu_mask_to_apicid(vector_searchmask, &d->cfg.dest_apicid)); + BUG_ON(apic->cpu_mask_to_apicid(vector_searchmask, irqdata, + &d->cfg.dest_apicid)); return 0; } static int assign_irq_vector(int irq, struct apic_chip_data *data, - const struct cpumask *mask) + const struct cpumask *mask, + struct irq_data *irqdata) { int err; unsigned long flags; raw_spin_lock_irqsave(&vector_lock, flags); - err = __assign_irq_vector(irq, data, mask); + err = __assign_irq_vector(irq, data, mask, irqdata); raw_spin_unlock_irqrestore(&vector_lock, flags); return err; } static int assign_irq_vector_policy(int irq, int node, struct apic_chip_data *data, - struct irq_alloc_info *info) + struct irq_alloc_info *info, + struct irq_data *irqdata) { if (info && info->mask) - return assign_irq_vector(irq, data, info->mask); + return assign_irq_vector(irq, data, info->mask, irqdata); if (node != NUMA_NO_NODE && - assign_irq_vector(irq, data, cpumask_of_node(node)) == 0) + assign_irq_vector(irq, data, cpumask_of_node(node), irqdata) == 0) return 0; - return assign_irq_vector(irq, data, apic->target_cpus()); + return assign_irq_vector(irq, data, apic->target_cpus(), irqdata); } static void clear_irq_vector(int irq, struct apic_chip_data *data) @@ -363,7 +367,8 @@ static int x86_vector_alloc_irqs(struct irq_domain *domain, unsigned int virq, irq_data->chip = &lapic_controller; irq_data->chip_data = data; irq_data->hwirq = virq + i; - err = assign_irq_vector_policy(virq + i, node, data, info); + err = assign_irq_vector_policy(virq + i, node, data, info, + irq_data); if (err) goto error; } @@ -537,7 +542,7 @@ static int apic_set_affinity(struct irq_data *irq_data, if (!cpumask_intersects(dest, cpu_online_mask)) return -EINVAL; - err = assign_irq_vector(irq, data, dest); + err = assign_irq_vector(irq, data, dest, irq_data); return err ? err : IRQ_SET_MASK_OK; } diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c index 61474259bf3f..305031e0a228 100644 --- a/arch/x86/kernel/apic/x2apic_cluster.c +++ b/arch/x86/kernel/apic/x2apic_cluster.c @@ -104,7 +104,8 @@ static void x2apic_send_IPI_all(int vector) } static int -x2apic_cpu_mask_to_apicid(const struct cpumask *mask, unsigned int *apicid) +x2apic_cpu_mask_to_apicid(const struct cpumask *mask, struct irq_data *irqdata, + unsigned int *apicid) { unsigned int cpu; u32 dest = 0; diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c index ad0223f013f6..0d57bb9079c9 100644 --- a/arch/x86/kernel/apic/x2apic_uv_x.c +++ b/arch/x86/kernel/apic/x2apic_uv_x.c @@ -526,9 +526,10 @@ static void uv_init_apic_ldr(void) } static int -uv_cpu_mask_to_apicid(const struct cpumask *mask, unsigned int *apicid) +uv_cpu_mask_to_apicid(const struct cpumask *mask, struct irq_data *irqdata, + unsigned int *apicid) { - int ret = default_cpu_mask_to_apicid(mask, apicid); + int ret = default_cpu_mask_to_apicid(mask, irqdata, apicid); if (!ret) *apicid |= uv_apicid_hibits; From ef1c2cc88531a967fa97d1ac1f3f8a64ee6910b4 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:45 +0200 Subject: [PATCH 081/112] xen/events: Add support for effective affinity mask Update the effective affinity mask when an interrupt was successfully targeted to a CPU. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235446.799944725@linutronix.de --- drivers/xen/events/events_base.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/xen/events/events_base.c b/drivers/xen/events/events_base.c index b52852f38cff..2e567d8433b3 100644 --- a/drivers/xen/events/events_base.c +++ b/drivers/xen/events/events_base.c @@ -1343,8 +1343,12 @@ static int set_affinity_irq(struct irq_data *data, const struct cpumask *dest, bool force) { unsigned tcpu = cpumask_first_and(dest, cpu_online_mask); + int ret = rebind_irq_to_cpu(data->irq, tcpu); - return rebind_irq_to_cpu(data->irq, tcpu); + if (!ret) + irq_data_update_effective_affinity(data, cpumask_of(tcpu)); + + return ret; } static void enable_dynirq(struct irq_data *data) From c7d6c9dd871f42c4e0ce5563d2f684e78ea673cf Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:46 +0200 Subject: [PATCH 082/112] x86/apic: Implement effective irq mask update Add the effective irq mask update to the apic implementations and enable effective irq masks for x86. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235446.878370703@linutronix.de --- arch/x86/Kconfig | 1 + arch/x86/kernel/apic/apic.c | 3 +++ arch/x86/kernel/apic/x2apic_cluster.c | 4 ++++ 3 files changed, 8 insertions(+) diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index fcf1dad582f5..0172c0b844d4 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -87,6 +87,7 @@ config X86 select GENERIC_EARLY_IOREMAP select GENERIC_FIND_FIRST_BIT select GENERIC_IOMAP + select GENERIC_IRQ_EFFECTIVE_AFF_MASK if SMP select GENERIC_IRQ_MIGRATION if SMP select GENERIC_IRQ_PROBE select GENERIC_IRQ_SHOW diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 14e5a47192b7..e7409468efc6 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -2210,6 +2210,7 @@ int default_cpu_mask_to_apicid(const struct cpumask *mask, if (cpu >= nr_cpu_ids) return -EINVAL; *apicid = per_cpu(x86_cpu_to_apicid, cpu); + irq_data_update_effective_affinity(irqdata, cpumask_of(cpu)); return 0; } @@ -2218,11 +2219,13 @@ int flat_cpu_mask_to_apicid(const struct cpumask *mask, unsigned int *apicid) { + struct cpumask *effmsk = irq_data_get_effective_affinity_mask(irqdata); unsigned long cpu_mask = cpumask_bits(mask)[0] & APIC_ALL_CPUS; if (!cpu_mask) return -EINVAL; *apicid = (unsigned int)cpu_mask; + cpumask_bits(effmsk)[0] = cpu_mask; return 0; } diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c index 305031e0a228..481237cb1544 100644 --- a/arch/x86/kernel/apic/x2apic_cluster.c +++ b/arch/x86/kernel/apic/x2apic_cluster.c @@ -4,6 +4,7 @@ #include #include #include +#include #include #include @@ -107,6 +108,7 @@ static int x2apic_cpu_mask_to_apicid(const struct cpumask *mask, struct irq_data *irqdata, unsigned int *apicid) { + struct cpumask *effmsk = irq_data_get_effective_affinity_mask(irqdata); unsigned int cpu; u32 dest = 0; u16 cluster; @@ -118,10 +120,12 @@ x2apic_cpu_mask_to_apicid(const struct cpumask *mask, struct irq_data *irqdata, dest = per_cpu(x86_cpu_to_logical_apicid, cpu); cluster = x2apic_cluster(cpu); + cpumask_clear(effmsk); for_each_cpu(cpu, mask) { if (cluster != x2apic_cluster(cpu)) continue; dest |= per_cpu(x86_cpu_to_logical_apicid, cpu); + cpumask_set_cpu(cpu, effmsk); } *apicid = dest; From 54fdf6a0875ca380647ac1cc9b5b8f2dbbbfa131 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:47 +0200 Subject: [PATCH 083/112] genirq: Introduce IRQD_MANAGED_SHUTDOWN Affinity managed interrupts should keep their assigned affinity accross CPU hotplug. To avoid magic hackery in device drivers, the core code shall manage them transparently. This will set these interrupts into a managed shutdown state when the last CPU of the assigned affinity mask goes offline. The interrupt will be restarted when one of the CPUs in the assigned affinity mask comes back online. Introduce the necessary state flag and the accessor functions. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235446.954523476@linutronix.de --- include/linux/irq.h | 8 ++++++++ kernel/irq/internals.h | 10 ++++++++++ 2 files changed, 18 insertions(+) diff --git a/include/linux/irq.h b/include/linux/irq.h index 4087ef268ba9..0e37276c5315 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -207,6 +207,8 @@ struct irq_data { * IRQD_FORWARDED_TO_VCPU - The interrupt is forwarded to a VCPU * IRQD_AFFINITY_MANAGED - Affinity is auto-managed by the kernel * IRQD_IRQ_STARTED - Startup state of the interrupt + * IRQD_MANAGED_SHUTDOWN - Interrupt was shutdown due to empty affinity + * mask. Applies only to affinity managed irqs. */ enum { IRQD_TRIGGER_MASK = 0xf, @@ -225,6 +227,7 @@ enum { IRQD_FORWARDED_TO_VCPU = (1 << 20), IRQD_AFFINITY_MANAGED = (1 << 21), IRQD_IRQ_STARTED = (1 << 22), + IRQD_MANAGED_SHUTDOWN = (1 << 23), }; #define __irqd_to_state(d) ACCESS_PRIVATE((d)->common, state_use_accessors) @@ -343,6 +346,11 @@ static inline bool irqd_is_started(struct irq_data *d) return __irqd_to_state(d) & IRQD_IRQ_STARTED; } +static inline bool irqd_is_managed_shutdown(struct irq_data *d) +{ + return __irqd_to_state(d) & IRQD_MANAGED_SHUTDOWN; +} + #undef __irqd_to_state static inline irq_hw_number_t irqd_to_hwirq(struct irq_data *d) diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index 040806f1124c..ca4666b4cd39 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -193,6 +193,16 @@ static inline void irqd_clr_move_pending(struct irq_data *d) __irqd_to_state(d) &= ~IRQD_SETAFFINITY_PENDING; } +static inline void irqd_set_managed_shutdown(struct irq_data *d) +{ + __irqd_to_state(d) |= IRQD_MANAGED_SHUTDOWN; +} + +static inline void irqd_clr_managed_shutdown(struct irq_data *d) +{ + __irqd_to_state(d) &= ~IRQD_MANAGED_SHUTDOWN; +} + static inline void irqd_clear(struct irq_data *d, unsigned int mask) { __irqd_to_state(d) &= ~mask; From 708d174b6c32bffc5d73793bc7a267bcafeb6558 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:48 +0200 Subject: [PATCH 084/112] genirq: Split out irq_startup() code Split out the inner workings of irq_startup() so it can be reused to handle managed interrupts gracefully. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235447.033235144@linutronix.de --- kernel/irq/chip.c | 29 ++++++++++++++++++----------- 1 file changed, 18 insertions(+), 11 deletions(-) diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index e290d73b88e2..1163089aa245 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -195,6 +195,23 @@ static void irq_state_set_started(struct irq_desc *desc) irqd_set(&desc->irq_data, IRQD_IRQ_STARTED); } +static int __irq_startup(struct irq_desc *desc) +{ + struct irq_data *d = irq_desc_get_irq_data(desc); + int ret = 0; + + irq_domain_activate_irq(d); + if (d->chip->irq_startup) { + ret = d->chip->irq_startup(d); + irq_state_clr_disabled(desc); + irq_state_clr_masked(desc); + } else { + irq_enable(desc); + } + irq_state_set_started(desc); + return ret; +} + int irq_startup(struct irq_desc *desc, bool resend) { int ret = 0; @@ -204,19 +221,9 @@ int irq_startup(struct irq_desc *desc, bool resend) if (irqd_is_started(&desc->irq_data)) { irq_enable(desc); } else { - irq_domain_activate_irq(&desc->irq_data); - if (desc->irq_data.chip->irq_startup) { - ret = desc->irq_data.chip->irq_startup(&desc->irq_data); - irq_state_clr_disabled(desc); - irq_state_clr_masked(desc); - } else { - irq_enable(desc); - } - irq_state_set_started(desc); - /* Set default affinity mask once everything is setup */ + ret = __irq_startup(desc); irq_setup_affinity(desc); } - if (resend) check_irq_resend(desc); From 4cde9c6b826834b861a2b58653ab33150f562064 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:49 +0200 Subject: [PATCH 085/112] genirq: Add force argument to irq_startup() In order to handle managed interrupts gracefully on irq_startup() so they won't lose their assigned affinity, it's necessary to allow startups which keep the interrupts in managed shutdown state, if none of the assigend CPUs is online. This allows drivers to request interrupts w/o the CPUs being online, which avoid online/offline churn in drivers. Add a force argument which can override that decision and let only request_irq() and enable_irq() allow the managed shutdown handling. enable_irq() is required, because the interrupt might be requested with IRQF_NOAUTOEN and enable_irq() invokes irq_startup() which would then wreckage the assignment again. All other callers force startup and potentially break the assigned affinity. No functional change as this only adds the function argument. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235447.112094565@linutronix.de --- kernel/irq/autoprobe.c | 4 ++-- kernel/irq/chip.c | 4 ++-- kernel/irq/internals.h | 9 ++++++++- kernel/irq/manage.c | 4 ++-- 4 files changed, 14 insertions(+), 7 deletions(-) diff --git a/kernel/irq/autoprobe.c b/kernel/irq/autoprobe.c index 0119b9d467ae..d30a0dd5cc02 100644 --- a/kernel/irq/autoprobe.c +++ b/kernel/irq/autoprobe.c @@ -53,7 +53,7 @@ unsigned long probe_irq_on(void) if (desc->irq_data.chip->irq_set_type) desc->irq_data.chip->irq_set_type(&desc->irq_data, IRQ_TYPE_PROBE); - irq_startup(desc, false); + irq_startup(desc, IRQ_NORESEND, IRQ_START_FORCE); } raw_spin_unlock_irq(&desc->lock); } @@ -70,7 +70,7 @@ unsigned long probe_irq_on(void) raw_spin_lock_irq(&desc->lock); if (!desc->action && irq_settings_can_probe(desc)) { desc->istate |= IRQS_AUTODETECT | IRQS_WAITING; - if (irq_startup(desc, false)) + if (irq_startup(desc, IRQ_NORESEND, IRQ_START_FORCE)) desc->istate |= IRQS_PENDING; } raw_spin_unlock_irq(&desc->lock); diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index 1163089aa245..b7599e952d3b 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -212,7 +212,7 @@ static int __irq_startup(struct irq_desc *desc) return ret; } -int irq_startup(struct irq_desc *desc, bool resend) +int irq_startup(struct irq_desc *desc, bool resend, bool force) { int ret = 0; @@ -892,7 +892,7 @@ __irq_do_set_handler(struct irq_desc *desc, irq_flow_handler_t handle, irq_settings_set_norequest(desc); irq_settings_set_nothread(desc); desc->action = &chained_action; - irq_startup(desc, true); + irq_startup(desc, IRQ_RESEND, IRQ_START_FORCE); } } diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index ca4666b4cd39..5fd105e252c3 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -66,7 +66,14 @@ extern int __irq_set_trigger(struct irq_desc *desc, unsigned long flags); extern void __disable_irq(struct irq_desc *desc); extern void __enable_irq(struct irq_desc *desc); -extern int irq_startup(struct irq_desc *desc, bool resend); +#define IRQ_RESEND true +#define IRQ_NORESEND false + +#define IRQ_START_FORCE true +#define IRQ_START_COND false + +extern int irq_startup(struct irq_desc *desc, bool resend, bool force); + extern void irq_shutdown(struct irq_desc *desc); extern void irq_enable(struct irq_desc *desc); extern void irq_disable(struct irq_desc *desc); diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 7dcf19397c39..3577c091ac7b 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -509,7 +509,7 @@ void __enable_irq(struct irq_desc *desc) * time. If it was already started up, then irq_startup() * will invoke irq_enable() under the hood. */ - irq_startup(desc, true); + irq_startup(desc, IRQ_RESEND, IRQ_START_COND); break; } default: @@ -1306,7 +1306,7 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) } if (irq_settings_can_autoenable(desc)) { - irq_startup(desc, true); + irq_startup(desc, IRQ_RESEND, IRQ_START_COND); } else { /* * Shared interrupts do not go well with disabling From 761ea388e8c4e3ac883a94e16bcc8c51fa419d4f Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:50 +0200 Subject: [PATCH 086/112] genirq: Handle managed irqs gracefully in irq_startup() Affinity managed interrupts should keep their assigned affinity accross CPU hotplug. To avoid magic hackery in device drivers, the core code shall manage them transparently and set these interrupts into a managed shutdown state when the last CPU of the assigned affinity mask goes offline. The interrupt will be restarted when one of the CPUs in the assigned affinity mask comes back online. Add the necessary logic to irq_startup(). If an interrupt is requested and started up, the code checks whether it is affinity managed and if so, it checks whether a CPU in the interrupts affinity mask is online. If not, it puts the interrupt into managed shutdown state. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235447.189851170@linutronix.de --- include/linux/irq.h | 2 +- kernel/irq/chip.c | 64 ++++++++++++++++++++++++++++++++++++++++++--- 2 files changed, 62 insertions(+), 4 deletions(-) diff --git a/include/linux/irq.h b/include/linux/irq.h index 0e37276c5315..807042b46af1 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -346,7 +346,7 @@ static inline bool irqd_is_started(struct irq_data *d) return __irqd_to_state(d) & IRQD_IRQ_STARTED; } -static inline bool irqd_is_managed_shutdown(struct irq_data *d) +static inline bool irqd_is_managed_and_shutdown(struct irq_data *d) { return __irqd_to_state(d) & IRQD_MANAGED_SHUTDOWN; } diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index b7599e952d3b..fc89eeb8a6b4 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -195,6 +195,52 @@ static void irq_state_set_started(struct irq_desc *desc) irqd_set(&desc->irq_data, IRQD_IRQ_STARTED); } +enum { + IRQ_STARTUP_NORMAL, + IRQ_STARTUP_MANAGED, + IRQ_STARTUP_ABORT, +}; + +#ifdef CONFIG_SMP +static int +__irq_startup_managed(struct irq_desc *desc, struct cpumask *aff, bool force) +{ + struct irq_data *d = irq_desc_get_irq_data(desc); + + if (!irqd_affinity_is_managed(d)) + return IRQ_STARTUP_NORMAL; + + irqd_clr_managed_shutdown(d); + + if (cpumask_any_and(aff, cpu_online_mask) > nr_cpu_ids) { + /* + * Catch code which fiddles with enable_irq() on a managed + * and potentially shutdown IRQ. Chained interrupt + * installment or irq auto probing should not happen on + * managed irqs either. Emit a warning, break the affinity + * and start it up as a normal interrupt. + */ + if (WARN_ON_ONCE(force)) + return IRQ_STARTUP_NORMAL; + /* + * The interrupt was requested, but there is no online CPU + * in it's affinity mask. Put it into managed shutdown + * state and let the cpu hotplug mechanism start it up once + * a CPU in the mask becomes available. + */ + irqd_set_managed_shutdown(d); + return IRQ_STARTUP_ABORT; + } + return IRQ_STARTUP_MANAGED; +} +#else +static int +__irq_startup_managed(struct irq_desc *desc, struct cpumask *aff, bool force) +{ + return IRQ_STARTUP_NORMAL; +} +#endif + static int __irq_startup(struct irq_desc *desc) { struct irq_data *d = irq_desc_get_irq_data(desc); @@ -214,15 +260,27 @@ static int __irq_startup(struct irq_desc *desc) int irq_startup(struct irq_desc *desc, bool resend, bool force) { + struct irq_data *d = irq_desc_get_irq_data(desc); + struct cpumask *aff = irq_data_get_affinity_mask(d); int ret = 0; desc->depth = 0; - if (irqd_is_started(&desc->irq_data)) { + if (irqd_is_started(d)) { irq_enable(desc); } else { - ret = __irq_startup(desc); - irq_setup_affinity(desc); + switch (__irq_startup_managed(desc, aff, force)) { + case IRQ_STARTUP_NORMAL: + ret = __irq_startup(desc); + irq_setup_affinity(desc); + break; + case IRQ_STARTUP_MANAGED: + ret = __irq_startup(desc); + irq_set_affinity_locked(d, aff, false); + break; + case IRQ_STARTUP_ABORT: + return 0; + } } if (resend) check_irq_resend(desc); From c5cb83bb337c25caae995d992d1cdf9b317f83de Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:51 +0200 Subject: [PATCH 087/112] genirq/cpuhotplug: Handle managed IRQs on CPU hotplug If a CPU goes offline, interrupts affine to the CPU are moved away. If the outgoing CPU is the last CPU in the affinity mask the migration code breaks the affinity and sets it it all online cpus. This is a problem for affinity managed interrupts as CPU hotplug is often used for power management purposes. If the affinity is broken, the interrupt is not longer affine to the CPUs to which it was allocated. The affinity spreading allows to lay out multi queue devices in a way that they are assigned to a single CPU or a group of CPUs. If the last CPU goes offline, then the queue is not longer used, so the interrupt can be shutdown gracefully and parked until one of the assigned CPUs comes online again. Add a graceful shutdown mechanism into the irq affinity breaking code path, mark the irq as MANAGED_SHUTDOWN and leave the affinity mask unmodified. In the online path, scan the active interrupts for managed interrupts and if the interrupt is functional and the newly online CPU is part of the affinity mask, restart the interrupt if it is marked MANAGED_SHUTDOWN or if the interrupts is started up, try to add the CPU back to the effective affinity mask. Originally-by: Christoph Hellwig Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/20170619235447.273417334@linutronix.de --- include/linux/cpuhotplug.h | 1 + include/linux/irq.h | 5 +++++ kernel/cpu.c | 5 +++++ kernel/irq/cpuhotplug.c | 45 ++++++++++++++++++++++++++++++++++++++ 4 files changed, 56 insertions(+) diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h index 0f2a80377520..c15f22c54535 100644 --- a/include/linux/cpuhotplug.h +++ b/include/linux/cpuhotplug.h @@ -124,6 +124,7 @@ enum cpuhp_state { CPUHP_AP_ONLINE_IDLE, CPUHP_AP_SMPBOOT_THREADS, CPUHP_AP_X86_VDSO_VMA_ONLINE, + CPUHP_AP_IRQ_AFFINITY_ONLINE, CPUHP_AP_PERF_ONLINE, CPUHP_AP_PERF_X86_ONLINE, CPUHP_AP_PERF_X86_UNCORE_ONLINE, diff --git a/include/linux/irq.h b/include/linux/irq.h index 807042b46af1..19cea6326599 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -500,7 +500,12 @@ extern int irq_set_affinity_locked(struct irq_data *data, const struct cpumask *cpumask, bool force); extern int irq_set_vcpu_affinity(unsigned int irq, void *vcpu_info); +#if defined(CONFIG_SMP) && defined(CONFIG_GENERIC_IRQ_MIGRATION) extern void irq_migrate_all_off_this_cpu(void); +extern int irq_affinity_online_cpu(unsigned int cpu); +#else +# define irq_affinity_online_cpu NULL +#endif #if defined(CONFIG_SMP) && defined(CONFIG_GENERIC_PENDING_IRQ) void irq_move_irq(struct irq_data *data); diff --git a/kernel/cpu.c b/kernel/cpu.c index cb5103413bd8..b86b32ebb3b2 100644 --- a/kernel/cpu.c +++ b/kernel/cpu.c @@ -1252,6 +1252,11 @@ static struct cpuhp_step cpuhp_ap_states[] = { .startup.single = smpboot_unpark_threads, .teardown.single = NULL, }, + [CPUHP_AP_IRQ_AFFINITY_ONLINE] = { + .name = "irq/affinity:online", + .startup.single = irq_affinity_online_cpu, + .teardown.single = NULL, + }, [CPUHP_AP_PERF_ONLINE] = { .name = "perf:online", .startup.single = perf_event_init_cpu, diff --git a/kernel/irq/cpuhotplug.c b/kernel/irq/cpuhotplug.c index 0b093db3336b..b7964e72ded7 100644 --- a/kernel/irq/cpuhotplug.c +++ b/kernel/irq/cpuhotplug.c @@ -83,6 +83,15 @@ static bool migrate_one_irq(struct irq_desc *desc) chip->irq_mask(d); if (cpumask_any_and(affinity, cpu_online_mask) >= nr_cpu_ids) { + /* + * If the interrupt is managed, then shut it down and leave + * the affinity untouched. + */ + if (irqd_affinity_is_managed(d)) { + irqd_set_managed_shutdown(d); + irq_shutdown(desc); + return false; + } affinity = cpu_online_mask; brokeaff = true; } @@ -129,3 +138,39 @@ void irq_migrate_all_off_this_cpu(void) } } } + +static void irq_restore_affinity_of_irq(struct irq_desc *desc, unsigned int cpu) +{ + struct irq_data *data = irq_desc_get_irq_data(desc); + const struct cpumask *affinity = irq_data_get_affinity_mask(data); + + if (!irqd_affinity_is_managed(data) || !desc->action || + !irq_data_get_irq_chip(data) || !cpumask_test_cpu(cpu, affinity)) + return; + + if (irqd_is_managed_and_shutdown(data)) + irq_startup(desc, IRQ_RESEND, IRQ_START_COND); + else + irq_set_affinity_locked(data, affinity, false); +} + +/** + * irq_affinity_online_cpu - Restore affinity for managed interrupts + * @cpu: Upcoming CPU for which interrupts should be restored + */ +int irq_affinity_online_cpu(unsigned int cpu) +{ + struct irq_desc *desc; + unsigned int irq; + + irq_lock_sparse(); + for_each_active_irq(irq) { + desc = irq_to_desc(irq); + raw_spin_lock_irq(&desc->lock); + irq_restore_affinity_of_irq(desc, cpu); + raw_spin_unlock_irq(&desc->lock); + } + irq_unlock_sparse(); + + return 0; +} From d52dd44175bd27ad9d8e34a994fb80877c1f6d61 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:52 +0200 Subject: [PATCH 088/112] genirq: Introduce IRQD_SINGLE_TARGET flag Many interrupt chips allow only a single CPU as interrupt target. The core code has no knowledge about that. That's unfortunate as it could avoid trying to readd a newly online CPU to the effective affinity mask. Add the status flag and the necessary accessors. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235447.352343969@linutronix.de --- include/linux/irq.h | 16 ++++++++++++++++ kernel/irq/debugfs.c | 1 + 2 files changed, 17 insertions(+) diff --git a/include/linux/irq.h b/include/linux/irq.h index 19cea6326599..00db35b61e9e 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -209,6 +209,7 @@ struct irq_data { * IRQD_IRQ_STARTED - Startup state of the interrupt * IRQD_MANAGED_SHUTDOWN - Interrupt was shutdown due to empty affinity * mask. Applies only to affinity managed irqs. + * IRQD_SINGLE_TARGET - IRQ allows only a single affinity target */ enum { IRQD_TRIGGER_MASK = 0xf, @@ -228,6 +229,7 @@ enum { IRQD_AFFINITY_MANAGED = (1 << 21), IRQD_IRQ_STARTED = (1 << 22), IRQD_MANAGED_SHUTDOWN = (1 << 23), + IRQD_SINGLE_TARGET = (1 << 24), }; #define __irqd_to_state(d) ACCESS_PRIVATE((d)->common, state_use_accessors) @@ -276,6 +278,20 @@ static inline bool irqd_is_level_type(struct irq_data *d) return __irqd_to_state(d) & IRQD_LEVEL; } +/* + * Must only be called of irqchip.irq_set_affinity() or low level + * hieararchy domain allocation functions. + */ +static inline void irqd_set_single_target(struct irq_data *d) +{ + __irqd_to_state(d) |= IRQD_SINGLE_TARGET; +} + +static inline bool irqd_is_single_target(struct irq_data *d) +{ + return __irqd_to_state(d) & IRQD_SINGLE_TARGET; +} + static inline bool irqd_is_wakeup_set(struct irq_data *d) { return __irqd_to_state(d) & IRQD_WAKEUP_STATE; diff --git a/kernel/irq/debugfs.c b/kernel/irq/debugfs.c index edbef252d0c4..dbd6e78db213 100644 --- a/kernel/irq/debugfs.c +++ b/kernel/irq/debugfs.c @@ -105,6 +105,7 @@ static const struct irq_bit_descr irqdata_states[] = { BIT_MASK_DESCR(IRQD_PER_CPU), BIT_MASK_DESCR(IRQD_NO_BALANCING), + BIT_MASK_DESCR(IRQD_SINGLE_TARGET), BIT_MASK_DESCR(IRQD_MOVE_PCNTXT), BIT_MASK_DESCR(IRQD_AFFINITY_SET), BIT_MASK_DESCR(IRQD_SETAFFINITY_PENDING), From 8f31a9845db348f5781df47ce04c79e4cfe90016 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:53 +0200 Subject: [PATCH 089/112] genirq/cpuhotplug: Avoid irq affinity setting for single targets Avoid trying to add a newly online CPU to the effective affinity mask of an started up interrupt. That interrupt will either stay on the already online CPU or move around for no value. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235447.431321047@linutronix.de --- kernel/irq/cpuhotplug.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) diff --git a/kernel/irq/cpuhotplug.c b/kernel/irq/cpuhotplug.c index b7964e72ded7..aee8f7ec40af 100644 --- a/kernel/irq/cpuhotplug.c +++ b/kernel/irq/cpuhotplug.c @@ -148,9 +148,17 @@ static void irq_restore_affinity_of_irq(struct irq_desc *desc, unsigned int cpu) !irq_data_get_irq_chip(data) || !cpumask_test_cpu(cpu, affinity)) return; - if (irqd_is_managed_and_shutdown(data)) + if (irqd_is_managed_and_shutdown(data)) { irq_startup(desc, IRQ_RESEND, IRQ_START_COND); - else + return; + } + + /* + * If the interrupt can only be directed to a single target + * CPU then it is already assigned to a CPU in the affinity + * mask. No point in trying to move it around. + */ + if (!irqd_is_single_target(data)) irq_set_affinity_locked(data, affinity, false); } From 3ca57222c36ba31b80aa25de313f3c8ab26a8102 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 20 Jun 2017 01:37:54 +0200 Subject: [PATCH 090/112] x86/apic: Mark single target interrupts If the interrupt destination mode of the APIC is physical then the effective affinity is restricted to a single CPU. Mark the interrupt accordingly in the domain allocation code, so the core code can avoid pointless affinity setting attempts. Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: Marc Zyngier Cc: Michael Ellerman Cc: Keith Busch Cc: Peter Zijlstra Cc: Christoph Hellwig Link: http://lkml.kernel.org/r/20170619235447.508846202@linutronix.de --- arch/x86/kernel/apic/vector.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/arch/x86/kernel/apic/vector.c b/arch/x86/kernel/apic/vector.c index b270a76ee640..2567dc012200 100644 --- a/arch/x86/kernel/apic/vector.c +++ b/arch/x86/kernel/apic/vector.c @@ -371,6 +371,13 @@ static int x86_vector_alloc_irqs(struct irq_domain *domain, unsigned int virq, irq_data); if (err) goto error; + /* + * If the apic destination mode is physical, then the + * effective affinity is restricted to a single target + * CPU. Mark the interrupt accordingly. + */ + if (!apic->irq_dest_mode) + irqd_set_single_target(irq_data); } return 0; From 9a0ef98e186d86fb3c1ff3ec267a76f067005f74 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 20 Jun 2017 01:37:55 +0200 Subject: [PATCH 091/112] genirq/affinity: Assign vectors to all present CPUs Currently the irq vector spread algorithm is restricted to online CPUs, which ties the IRQ mapping to the currently online devices and doesn't deal nicely with the fact that CPUs could come and go rapidly due to e.g. power management. Instead assign vectors to all present CPUs to avoid this churn. Build a map of all possible CPUs for a given node, as the architectures only provide a map of all onlines CPUs. Do this dynamically on each call for the vector assingments, which is a bit suboptimal and could be optimized in the future by provinding a mapping from the arch code. Signed-off-by: Christoph Hellwig Signed-off-by: Thomas Gleixner Cc: Jens Axboe Cc: linux-block@vger.kernel.org Cc: Sagi Grimberg Cc: Marc Zyngier Cc: Michael Ellerman Cc: linux-nvme@lists.infradead.org Cc: Keith Busch Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/20170603140403.27379-5-hch@lst.de --- kernel/irq/affinity.c | 76 +++++++++++++++++++++++++++++++++++-------- 1 file changed, 63 insertions(+), 13 deletions(-) diff --git a/kernel/irq/affinity.c b/kernel/irq/affinity.c index e2d356dd7581..d2747f9c5707 100644 --- a/kernel/irq/affinity.c +++ b/kernel/irq/affinity.c @@ -1,4 +1,7 @@ - +/* + * Copyright (C) 2016 Thomas Gleixner. + * Copyright (C) 2016-2017 Christoph Hellwig. + */ #include #include #include @@ -35,13 +38,54 @@ static void irq_spread_init_one(struct cpumask *irqmsk, struct cpumask *nmsk, } } -static int get_nodes_in_cpumask(const struct cpumask *mask, nodemask_t *nodemsk) +static cpumask_var_t *alloc_node_to_present_cpumask(void) +{ + cpumask_var_t *masks; + int node; + + masks = kcalloc(nr_node_ids, sizeof(cpumask_var_t), GFP_KERNEL); + if (!masks) + return NULL; + + for (node = 0; node < nr_node_ids; node++) { + if (!zalloc_cpumask_var(&masks[node], GFP_KERNEL)) + goto out_unwind; + } + + return masks; + +out_unwind: + while (--node >= 0) + free_cpumask_var(masks[node]); + kfree(masks); + return NULL; +} + +static void free_node_to_present_cpumask(cpumask_var_t *masks) +{ + int node; + + for (node = 0; node < nr_node_ids; node++) + free_cpumask_var(masks[node]); + kfree(masks); +} + +static void build_node_to_present_cpumask(cpumask_var_t *masks) +{ + int cpu; + + for_each_present_cpu(cpu) + cpumask_set_cpu(cpu, masks[cpu_to_node(cpu)]); +} + +static int get_nodes_in_cpumask(cpumask_var_t *node_to_present_cpumask, + const struct cpumask *mask, nodemask_t *nodemsk) { int n, nodes = 0; /* Calculate the number of nodes in the supplied affinity mask */ - for_each_online_node(n) { - if (cpumask_intersects(mask, cpumask_of_node(n))) { + for_each_node(n) { + if (cpumask_intersects(mask, node_to_present_cpumask[n])) { node_set(n, *nodemsk); nodes++; } @@ -64,7 +108,7 @@ irq_create_affinity_masks(int nvecs, const struct irq_affinity *affd) int last_affv = affv + affd->pre_vectors; nodemask_t nodemsk = NODE_MASK_NONE; struct cpumask *masks; - cpumask_var_t nmsk; + cpumask_var_t nmsk, *node_to_present_cpumask; if (!zalloc_cpumask_var(&nmsk, GFP_KERNEL)) return NULL; @@ -73,13 +117,19 @@ irq_create_affinity_masks(int nvecs, const struct irq_affinity *affd) if (!masks) goto out; + node_to_present_cpumask = alloc_node_to_present_cpumask(); + if (!node_to_present_cpumask) + goto out; + /* Fill out vectors at the beginning that don't need affinity */ for (curvec = 0; curvec < affd->pre_vectors; curvec++) cpumask_copy(masks + curvec, irq_default_affinity); /* Stabilize the cpumasks */ get_online_cpus(); - nodes = get_nodes_in_cpumask(cpu_online_mask, &nodemsk); + build_node_to_present_cpumask(node_to_present_cpumask); + nodes = get_nodes_in_cpumask(node_to_present_cpumask, cpu_present_mask, + &nodemsk); /* * If the number of nodes in the mask is greater than or equal the @@ -87,7 +137,8 @@ irq_create_affinity_masks(int nvecs, const struct irq_affinity *affd) */ if (affv <= nodes) { for_each_node_mask(n, nodemsk) { - cpumask_copy(masks + curvec, cpumask_of_node(n)); + cpumask_copy(masks + curvec, + node_to_present_cpumask[n]); if (++curvec == last_affv) break; } @@ -101,7 +152,7 @@ irq_create_affinity_masks(int nvecs, const struct irq_affinity *affd) vecs_per_node = (affv - (curvec - affd->pre_vectors)) / nodes; /* Get the cpus on this node which are in the mask */ - cpumask_and(nmsk, cpu_online_mask, cpumask_of_node(n)); + cpumask_and(nmsk, cpu_present_mask, node_to_present_cpumask[n]); /* Calculate the number of cpus per vector */ ncpus = cpumask_weight(nmsk); @@ -133,6 +184,7 @@ done: /* Fill out vectors at the end that don't need affinity */ for (; curvec < nvecs; curvec++) cpumask_copy(masks + curvec, irq_default_affinity); + free_node_to_present_cpumask(node_to_present_cpumask); out: free_cpumask_var(nmsk); return masks; @@ -147,12 +199,10 @@ int irq_calc_affinity_vectors(int maxvec, const struct irq_affinity *affd) { int resv = affd->pre_vectors + affd->post_vectors; int vecs = maxvec - resv; - int cpus; + int ret; - /* Stabilize the cpumasks */ get_online_cpus(); - cpus = cpumask_weight(cpu_online_mask); + ret = min_t(int, cpumask_weight(cpu_present_mask), vecs) + resv; put_online_cpus(); - - return min(cpus, vecs) + resv; + return ret; } From 61d0a000b7746665c7cfcff766532f6f2a922a61 Mon Sep 17 00:00:00 2001 From: Marc Zyngier Date: Thu, 22 Jun 2017 11:34:57 +0100 Subject: [PATCH 092/112] genirq/irqdomain: Add irq_domain_update_bus_token helper We can have irq domains that are identified by the same fwnode (because they are serviced by the same HW), and yet have different functionnality (because they serve different busses, for example). This is what we use the bus_token field. Since we don't use this field when generating the domain name, all the aliasing domains will get the same name, and the debugfs file creation fails. Also, bus_token is updated by individual drivers, and the core code is unaware of that update. In order to sort this mess, let's introduce a helper that takes care of updating bus_token, and regenerate the debugfs file. A separate patch will update all the individual users. Signed-off-by: Marc Zyngier Signed-off-by: Thomas Gleixner --- include/linux/irqdomain.h | 3 +++ kernel/irq/irqdomain.c | 31 +++++++++++++++++++++++++++++++ 2 files changed, 34 insertions(+) diff --git a/include/linux/irqdomain.h b/include/linux/irqdomain.h index 914b0c31d233..222f47af12f4 100644 --- a/include/linux/irqdomain.h +++ b/include/linux/irqdomain.h @@ -273,6 +273,9 @@ static inline bool is_fwnode_irqchip(struct fwnode_handle *fwnode) return fwnode && fwnode->type == FWNODE_IRQCHIP; } +extern void irq_domain_update_bus_token(struct irq_domain *domain, + enum irq_domain_bus_token bus_token); + static inline struct irq_domain *irq_find_matching_fwnode(struct fwnode_handle *fwnode, enum irq_domain_bus_token bus_token) diff --git a/kernel/irq/irqdomain.c b/kernel/irq/irqdomain.c index 75e1f0851c33..f6adeaeb4c16 100644 --- a/kernel/irq/irqdomain.c +++ b/kernel/irq/irqdomain.c @@ -245,6 +245,37 @@ void irq_domain_remove(struct irq_domain *domain) } EXPORT_SYMBOL_GPL(irq_domain_remove); +void irq_domain_update_bus_token(struct irq_domain *domain, + enum irq_domain_bus_token bus_token) +{ + char *name; + + if (domain->bus_token == bus_token) + return; + + mutex_lock(&irq_domain_mutex); + + domain->bus_token = bus_token; + + name = kasprintf(GFP_KERNEL, "%s-%d", domain->name, bus_token); + if (!name) { + mutex_unlock(&irq_domain_mutex); + return; + } + + debugfs_remove_domain_dir(domain); + + if (domain->flags & IRQ_DOMAIN_NAME_ALLOCATED) + kfree(domain->name); + else + domain->flags |= IRQ_DOMAIN_NAME_ALLOCATED; + + domain->name = name; + debugfs_add_domain_dir(domain); + + mutex_unlock(&irq_domain_mutex); +} + /** * irq_domain_add_simple() - Register an irq_domain and optionally map a range of irqs * @of_node: pointer to interrupt controller's device tree node. From 96f0d93a487e13233c95397aa3ecaf9cc8ff04bf Mon Sep 17 00:00:00 2001 From: Marc Zyngier Date: Thu, 22 Jun 2017 11:42:50 +0100 Subject: [PATCH 093/112] irqchip/MSI: Use irq_domain_update_bus_token instead of an open coded access Now that we have irq_domain_update_bus_token(), switch everyone over to it. The debugfs code thanks you for your continued support. Signed-off-by: Marc Zyngier Signed-off-by: Thomas Gleixner --- drivers/base/platform-msi.c | 2 +- drivers/irqchip/irq-armada-370-xp.c | 2 +- drivers/irqchip/irq-gic-v2m.c | 2 +- drivers/irqchip/irq-gic-v3-its.c | 2 +- drivers/irqchip/irq-mips-cpu.c | 2 +- drivers/irqchip/irq-mips-gic.c | 2 +- drivers/pci/msi.c | 2 +- drivers/staging/fsl-mc/bus/fsl-mc-msi.c | 2 +- 8 files changed, 8 insertions(+), 8 deletions(-) diff --git a/drivers/base/platform-msi.c b/drivers/base/platform-msi.c index d35e9a20caf7..e5473525e7b2 100644 --- a/drivers/base/platform-msi.c +++ b/drivers/base/platform-msi.c @@ -195,7 +195,7 @@ struct irq_domain *platform_msi_create_irq_domain(struct fwnode_handle *fwnode, domain = msi_create_irq_domain(fwnode, info, parent); if (domain) - domain->bus_token = DOMAIN_BUS_PLATFORM_MSI; + irq_domain_update_bus_token(domain, DOMAIN_BUS_PLATFORM_MSI); return domain; } diff --git a/drivers/irqchip/irq-armada-370-xp.c b/drivers/irqchip/irq-armada-370-xp.c index eb0d4d41b156..5e16d042f281 100644 --- a/drivers/irqchip/irq-armada-370-xp.c +++ b/drivers/irqchip/irq-armada-370-xp.c @@ -563,7 +563,7 @@ static int __init armada_370_xp_mpic_of_init(struct device_node *node, irq_domain_add_linear(node, nr_irqs, &armada_370_xp_mpic_irq_ops, NULL); BUG_ON(!armada_370_xp_mpic_domain); - armada_370_xp_mpic_domain->bus_token = DOMAIN_BUS_WIRED; + irq_domain_update_bus_token(armada_370_xp_mpic_domain, DOMAIN_BUS_WIRED); /* Setup for the boot CPU */ armada_xp_mpic_perf_init(); diff --git a/drivers/irqchip/irq-gic-v2m.c b/drivers/irqchip/irq-gic-v2m.c index 863e073c6f7f..993a8426a453 100644 --- a/drivers/irqchip/irq-gic-v2m.c +++ b/drivers/irqchip/irq-gic-v2m.c @@ -280,7 +280,7 @@ static int gicv2m_allocate_domains(struct irq_domain *parent) return -ENOMEM; } - inner_domain->bus_token = DOMAIN_BUS_NEXUS; + irq_domain_update_bus_token(inner_domain, DOMAIN_BUS_NEXUS); inner_domain->parent = parent; pci_domain = pci_msi_create_irq_domain(v2m->fwnode, &gicv2m_msi_domain_info, diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c index 45ea193325d2..059016541277 100644 --- a/drivers/irqchip/irq-gic-v3-its.c +++ b/drivers/irqchip/irq-gic-v3-its.c @@ -1661,7 +1661,7 @@ static int its_init_domain(struct fwnode_handle *handle, struct its_node *its) } inner_domain->parent = its_parent; - inner_domain->bus_token = DOMAIN_BUS_NEXUS; + irq_domain_update_bus_token(inner_domain, DOMAIN_BUS_NEXUS); inner_domain->flags |= IRQ_DOMAIN_FLAG_MSI_REMAP; info->ops = &its_msi_domain_ops; info->data = its; diff --git a/drivers/irqchip/irq-mips-cpu.c b/drivers/irqchip/irq-mips-cpu.c index b247f3c743ac..0a8ed1c05518 100644 --- a/drivers/irqchip/irq-mips-cpu.c +++ b/drivers/irqchip/irq-mips-cpu.c @@ -240,7 +240,7 @@ static void mips_cpu_register_ipi_domain(struct device_node *of_node) ipi_domain_state); if (!ipi_domain) panic("Failed to add MIPS CPU IPI domain"); - ipi_domain->bus_token = DOMAIN_BUS_IPI; + irq_domain_update_bus_token(ipi_domain, DOMAIN_BUS_IPI); } #else /* !CONFIG_GENERIC_IRQ_IPI */ diff --git a/drivers/irqchip/irq-mips-gic.c b/drivers/irqchip/irq-mips-gic.c index eb7fbe159963..d3a6dc800e3c 100644 --- a/drivers/irqchip/irq-mips-gic.c +++ b/drivers/irqchip/irq-mips-gic.c @@ -960,7 +960,7 @@ static void __init __gic_init(unsigned long gic_base_addr, panic("Failed to add GIC IPI domain"); gic_ipi_domain->name = "mips-gic-ipi"; - gic_ipi_domain->bus_token = DOMAIN_BUS_IPI; + irq_domain_update_bus_token(gic_ipi_domain, DOMAIN_BUS_IPI); if (node && !of_property_read_u32_array(node, "mti,reserved-ipi-vectors", v, 2)) { diff --git a/drivers/pci/msi.c b/drivers/pci/msi.c index ba44fdfda66b..fbad5dca3219 100644 --- a/drivers/pci/msi.c +++ b/drivers/pci/msi.c @@ -1463,7 +1463,7 @@ struct irq_domain *pci_msi_create_irq_domain(struct fwnode_handle *fwnode, if (!domain) return NULL; - domain->bus_token = DOMAIN_BUS_PCI_MSI; + irq_domain_update_bus_token(domain, DOMAIN_BUS_PCI_MSI); return domain; } EXPORT_SYMBOL_GPL(pci_msi_create_irq_domain); diff --git a/drivers/staging/fsl-mc/bus/fsl-mc-msi.c b/drivers/staging/fsl-mc/bus/fsl-mc-msi.c index b8b2c86e63d4..17d5cf9f91d4 100644 --- a/drivers/staging/fsl-mc/bus/fsl-mc-msi.c +++ b/drivers/staging/fsl-mc/bus/fsl-mc-msi.c @@ -170,7 +170,7 @@ struct irq_domain *fsl_mc_msi_create_irq_domain(struct fwnode_handle *fwnode, domain = msi_create_irq_domain(fwnode, info, parent); if (domain) - domain->bus_token = DOMAIN_BUS_FSL_MC_MSI; + irq_domain_update_bus_token(domain, DOMAIN_BUS_FSL_MC_MSI); return domain; } From 6a6544e520abecd484ab8b67fb50d1fc003f3275 Mon Sep 17 00:00:00 2001 From: Marc Zyngier Date: Tue, 20 Jun 2017 22:17:44 +0100 Subject: [PATCH 094/112] genirq/irqdomain: Remove auto-recursive hierarchy support It did seem like a good idea at the time, but it never really caught on, and auto-recursive domains remain unused 3 years after having been introduced. Oh well, time for a late spring cleanup. Signed-off-by: Marc Zyngier Signed-off-by: Thomas Gleixner --- include/linux/irqdomain.h | 9 +++---- kernel/irq/irqdomain.c | 55 +++++++++------------------------------ kernel/irq/msi.c | 2 +- 3 files changed, 17 insertions(+), 49 deletions(-) diff --git a/include/linux/irqdomain.h b/include/linux/irqdomain.h index 222f47af12f4..cac77a5c5555 100644 --- a/include/linux/irqdomain.h +++ b/include/linux/irqdomain.h @@ -180,8 +180,8 @@ enum { /* Irq domain is hierarchical */ IRQ_DOMAIN_FLAG_HIERARCHY = (1 << 0), - /* Core calls alloc/free recursive through the domain hierarchy. */ - IRQ_DOMAIN_FLAG_AUTO_RECURSIVE = (1 << 1), + /* Irq domain name was allocated in __irq_domain_add() */ + IRQ_DOMAIN_NAME_ALLOCATED = (1 << 6), /* Irq domain is an IPI domain with virq per cpu */ IRQ_DOMAIN_FLAG_IPI_PER_CPU = (1 << 2), @@ -195,9 +195,6 @@ enum { /* Irq domain implements MSI remapping */ IRQ_DOMAIN_FLAG_MSI_REMAP = (1 << 5), - /* Irq domain name was allocated in __irq_domain_add() */ - IRQ_DOMAIN_NAME_ALLOCATED = (1 << 6), - /* * Flags starting from IRQ_DOMAIN_FLAG_NONCORE are reserved * for implementation specific purposes and ignored by the @@ -448,7 +445,7 @@ static inline int irq_domain_alloc_irqs(struct irq_domain *domain, NULL); } -extern int irq_domain_alloc_irqs_recursive(struct irq_domain *domain, +extern int irq_domain_alloc_irqs_hierarchy(struct irq_domain *domain, unsigned int irq_base, unsigned int nr_irqs, void *arg); extern int irq_domain_set_hwirq_and_chip(struct irq_domain *domain, diff --git a/kernel/irq/irqdomain.c b/kernel/irq/irqdomain.c index f6adeaeb4c16..14fe862aa2e3 100644 --- a/kernel/irq/irqdomain.c +++ b/kernel/irq/irqdomain.c @@ -1342,43 +1342,18 @@ void irq_domain_free_irqs_top(struct irq_domain *domain, unsigned int virq, irq_domain_free_irqs_common(domain, virq, nr_irqs); } -static bool irq_domain_is_auto_recursive(struct irq_domain *domain) -{ - return domain->flags & IRQ_DOMAIN_FLAG_AUTO_RECURSIVE; -} - -static void irq_domain_free_irqs_recursive(struct irq_domain *domain, +static void irq_domain_free_irqs_hierarchy(struct irq_domain *domain, unsigned int irq_base, unsigned int nr_irqs) { domain->ops->free(domain, irq_base, nr_irqs); - if (irq_domain_is_auto_recursive(domain)) { - BUG_ON(!domain->parent); - irq_domain_free_irqs_recursive(domain->parent, irq_base, - nr_irqs); - } } -int irq_domain_alloc_irqs_recursive(struct irq_domain *domain, +int irq_domain_alloc_irqs_hierarchy(struct irq_domain *domain, unsigned int irq_base, unsigned int nr_irqs, void *arg) { - int ret = 0; - struct irq_domain *parent = domain->parent; - bool recursive = irq_domain_is_auto_recursive(domain); - - BUG_ON(recursive && !parent); - if (recursive) - ret = irq_domain_alloc_irqs_recursive(parent, irq_base, - nr_irqs, arg); - if (ret < 0) - return ret; - - ret = domain->ops->alloc(domain, irq_base, nr_irqs, arg); - if (ret < 0 && recursive) - irq_domain_free_irqs_recursive(parent, irq_base, nr_irqs); - - return ret; + return domain->ops->alloc(domain, irq_base, nr_irqs, arg); } /** @@ -1439,7 +1414,7 @@ int __irq_domain_alloc_irqs(struct irq_domain *domain, int irq_base, } mutex_lock(&irq_domain_mutex); - ret = irq_domain_alloc_irqs_recursive(domain, virq, nr_irqs, arg); + ret = irq_domain_alloc_irqs_hierarchy(domain, virq, nr_irqs, arg); if (ret < 0) { mutex_unlock(&irq_domain_mutex); goto out_free_irq_data; @@ -1474,7 +1449,7 @@ void irq_domain_free_irqs(unsigned int virq, unsigned int nr_irqs) mutex_lock(&irq_domain_mutex); for (i = 0; i < nr_irqs; i++) irq_domain_remove_irq(virq + i); - irq_domain_free_irqs_recursive(data->domain, virq, nr_irqs); + irq_domain_free_irqs_hierarchy(data->domain, virq, nr_irqs); mutex_unlock(&irq_domain_mutex); irq_domain_free_irq_data(virq, nr_irqs); @@ -1494,15 +1469,11 @@ int irq_domain_alloc_irqs_parent(struct irq_domain *domain, unsigned int irq_base, unsigned int nr_irqs, void *arg) { - /* irq_domain_alloc_irqs_recursive() has called parent's alloc() */ - if (irq_domain_is_auto_recursive(domain)) - return 0; + if (!domain->parent) + return -ENOSYS; - domain = domain->parent; - if (domain) - return irq_domain_alloc_irqs_recursive(domain, irq_base, - nr_irqs, arg); - return -ENOSYS; + return irq_domain_alloc_irqs_hierarchy(domain->parent, irq_base, + nr_irqs, arg); } EXPORT_SYMBOL_GPL(irq_domain_alloc_irqs_parent); @@ -1517,10 +1488,10 @@ EXPORT_SYMBOL_GPL(irq_domain_alloc_irqs_parent); void irq_domain_free_irqs_parent(struct irq_domain *domain, unsigned int irq_base, unsigned int nr_irqs) { - /* irq_domain_free_irqs_recursive() will call parent's free */ - if (!irq_domain_is_auto_recursive(domain) && domain->parent) - irq_domain_free_irqs_recursive(domain->parent, irq_base, - nr_irqs); + if (!domain->parent) + return; + + irq_domain_free_irqs_hierarchy(domain->parent, irq_base, nr_irqs); } EXPORT_SYMBOL_GPL(irq_domain_free_irqs_parent); diff --git a/kernel/irq/msi.c b/kernel/irq/msi.c index 9e3f1857c6bd..48eadf416c24 100644 --- a/kernel/irq/msi.c +++ b/kernel/irq/msi.c @@ -315,7 +315,7 @@ int msi_domain_populate_irqs(struct irq_domain *domain, struct device *dev, ops->set_desc(arg, desc); /* Assumes the domain mutex is held! */ - ret = irq_domain_alloc_irqs_recursive(domain, virq, 1, arg); + ret = irq_domain_alloc_irqs_hierarchy(domain, virq, 1, arg); if (ret) break; From f39a29bb5ca3f1c54fa677d9de17a09079b7e898 Mon Sep 17 00:00:00 2001 From: Thomas Petazzoni Date: Wed, 21 Jun 2017 15:29:13 +0200 Subject: [PATCH 095/112] dt-bindings/interrupt-controller: Add DT binding for the Marvell ICU This commit adds the Device Tree binding documentation for the Marvell ICU interrupt controller, which collects wired interrupts from the devices located into the CP110 hardware block of Marvell Armada 7K/8K, and converts them into SPI interrupts in the GIC located in the AP hardware block, using the GICP extension. Acked-by: Rob Herring Signed-off-by: Thomas Petazzoni Signed-off-by: Marc Zyngier --- .../interrupt-controller/marvell,icu.txt | 51 +++++++++++++++++++ 1 file changed, 51 insertions(+) create mode 100644 Documentation/devicetree/bindings/interrupt-controller/marvell,icu.txt diff --git a/Documentation/devicetree/bindings/interrupt-controller/marvell,icu.txt b/Documentation/devicetree/bindings/interrupt-controller/marvell,icu.txt new file mode 100644 index 000000000000..aa8bf2ec8905 --- /dev/null +++ b/Documentation/devicetree/bindings/interrupt-controller/marvell,icu.txt @@ -0,0 +1,51 @@ +Marvell ICU Interrupt Controller +-------------------------------- + +The Marvell ICU (Interrupt Consolidation Unit) controller is +responsible for collecting all wired-interrupt sources in the CP and +communicating them to the GIC in the AP, the unit translates interrupt +requests on input wires to MSG memory mapped transactions to the GIC. + +Required properties: + +- compatible: Should be "marvell,cp110-icu" + +- reg: Should contain ICU registers location and length. + +- #interrupt-cells: Specifies the number of cells needed to encode an + interrupt source. The value shall be 3. + + The 1st cell is the group type of the ICU interrupt. Possible group + types are: + + ICU_GRP_NSR (0x0) : Shared peripheral interrupt, non-secure + ICU_GRP_SR (0x1) : Shared peripheral interrupt, secure + ICU_GRP_SEI (0x4) : System error interrupt + ICU_GRP_REI (0x5) : RAM error interrupt + + The 2nd cell is the index of the interrupt in the ICU unit. + + The 3rd cell is the type of the interrupt. See arm,gic.txt for + details. + +- interrupt-controller: Identifies the node as an interrupt + controller. + +- msi-parent: Should point to the GICP controller, the GIC extension + that allows to trigger interrupts using MSG memory mapped + transactions. + +Example: + +icu: interrupt-controller@1e0000 { + compatible = "marvell,cp110-icu"; + reg = <0x1e0000 0x10>; + #interrupt-cells = <3>; + interrupt-controller; + msi-parent = <&gicp>; +}; + +usb3h0: usb3@500000 { + interrupt-parent = <&icu>; + interrupts = ; +}; From a68a63cb4dfc30e8a79b444aabc7747bb7621acf Mon Sep 17 00:00:00 2001 From: Thomas Petazzoni Date: Wed, 21 Jun 2017 15:29:14 +0200 Subject: [PATCH 096/112] irqchip/irq-mvebu-gicp: Add new driver for Marvell GICP This commit adds a simple driver for the Marvell GICP, a hardware unit that converts memory writes into GIC SPI interrupts. The driver provides a number of functions to the ICU driver to allocate GICP interrupts, and get the physical addresses that the ICUs should write to to set/clear interrupts. Signed-off-by: Thomas Petazzoni Reviewed-by: Marc Zyngier Signed-off-by: Marc Zyngier --- drivers/irqchip/Kconfig | 3 + drivers/irqchip/Makefile | 1 + drivers/irqchip/irq-mvebu-gicp.c | 279 +++++++++++++++++++++++++++++++ drivers/irqchip/irq-mvebu-gicp.h | 11 ++ 4 files changed, 294 insertions(+) create mode 100644 drivers/irqchip/irq-mvebu-gicp.c create mode 100644 drivers/irqchip/irq-mvebu-gicp.h diff --git a/drivers/irqchip/Kconfig b/drivers/irqchip/Kconfig index 478f8ace2664..e527ee596d54 100644 --- a/drivers/irqchip/Kconfig +++ b/drivers/irqchip/Kconfig @@ -268,6 +268,9 @@ config IRQ_MXS select IRQ_DOMAIN select STMP_DEVICE +config MVEBU_GICP + bool + config MVEBU_ODMI bool select GENERIC_MSI_IRQ_DOMAIN diff --git a/drivers/irqchip/Makefile b/drivers/irqchip/Makefile index e067f9839b33..f214f4573307 100644 --- a/drivers/irqchip/Makefile +++ b/drivers/irqchip/Makefile @@ -69,6 +69,7 @@ obj-$(CONFIG_ARCH_SA1100) += irq-sa11x0.o obj-$(CONFIG_INGENIC_IRQ) += irq-ingenic.o obj-$(CONFIG_IMX_GPCV2) += irq-imx-gpcv2.o obj-$(CONFIG_PIC32_EVIC) += irq-pic32-evic.o +obj-$(CONFIG_MVEBU_GICP) += irq-mvebu-gicp.o obj-$(CONFIG_MVEBU_ODMI) += irq-mvebu-odmi.o obj-$(CONFIG_MVEBU_PIC) += irq-mvebu-pic.o obj-$(CONFIG_LS_SCFG_MSI) += irq-ls-scfg-msi.o diff --git a/drivers/irqchip/irq-mvebu-gicp.c b/drivers/irqchip/irq-mvebu-gicp.c new file mode 100644 index 000000000000..45358ac9bb1d --- /dev/null +++ b/drivers/irqchip/irq-mvebu-gicp.c @@ -0,0 +1,279 @@ +/* + * Copyright (C) 2017 Marvell + * + * Thomas Petazzoni + * + * This file is licensed under the terms of the GNU General Public + * License version 2. This program is licensed "as is" without any + * warranty of any kind, whether express or implied. + */ + +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +#include "irq-mvebu-gicp.h" + +#define GICP_SETSPI_NSR_OFFSET 0x0 +#define GICP_CLRSPI_NSR_OFFSET 0x8 + +struct mvebu_gicp_spi_range { + unsigned int start; + unsigned int count; +}; + +struct mvebu_gicp { + struct mvebu_gicp_spi_range *spi_ranges; + unsigned int spi_ranges_cnt; + unsigned int spi_cnt; + unsigned long *spi_bitmap; + spinlock_t spi_lock; + struct resource *res; + struct device *dev; +}; + +static int gicp_idx_to_spi(struct mvebu_gicp *gicp, int idx) +{ + int i; + + for (i = 0; i < gicp->spi_ranges_cnt; i++) { + struct mvebu_gicp_spi_range *r = &gicp->spi_ranges[i]; + + if (idx < r->count) + return r->start + idx; + + idx -= r->count; + } + + return -EINVAL; +} + +int mvebu_gicp_get_doorbells(struct device_node *dn, phys_addr_t *setspi, + phys_addr_t *clrspi) +{ + struct platform_device *pdev; + struct mvebu_gicp *gicp; + + pdev = of_find_device_by_node(dn); + if (!pdev) + return -ENODEV; + + gicp = platform_get_drvdata(pdev); + if (!gicp) + return -ENODEV; + + *setspi = gicp->res->start + GICP_SETSPI_NSR_OFFSET; + *clrspi = gicp->res->start + GICP_CLRSPI_NSR_OFFSET; + + return 0; +} + +static void gicp_compose_msi_msg(struct irq_data *data, struct msi_msg *msg) +{ + struct mvebu_gicp *gicp = data->chip_data; + phys_addr_t setspi = gicp->res->start + GICP_SETSPI_NSR_OFFSET; + + msg->data = data->hwirq; + msg->address_lo = lower_32_bits(setspi); + msg->address_hi = upper_32_bits(setspi); +} + +static struct irq_chip gicp_irq_chip = { + .name = "GICP", + .irq_mask = irq_chip_mask_parent, + .irq_unmask = irq_chip_unmask_parent, + .irq_eoi = irq_chip_eoi_parent, + .irq_set_affinity = irq_chip_set_affinity_parent, + .irq_set_type = irq_chip_set_type_parent, + .irq_compose_msi_msg = gicp_compose_msi_msg, +}; + +static int gicp_irq_domain_alloc(struct irq_domain *domain, unsigned int virq, + unsigned int nr_irqs, void *args) +{ + struct mvebu_gicp *gicp = domain->host_data; + struct irq_fwspec fwspec; + unsigned int hwirq; + int ret; + + spin_lock(&gicp->spi_lock); + hwirq = find_first_zero_bit(gicp->spi_bitmap, gicp->spi_cnt); + if (hwirq == gicp->spi_cnt) { + spin_unlock(&gicp->spi_lock); + return -ENOSPC; + } + __set_bit(hwirq, gicp->spi_bitmap); + spin_unlock(&gicp->spi_lock); + + fwspec.fwnode = domain->parent->fwnode; + fwspec.param_count = 3; + fwspec.param[0] = GIC_SPI; + fwspec.param[1] = gicp_idx_to_spi(gicp, hwirq) - 32; + /* + * Assume edge rising for now, it will be properly set when + * ->set_type() is called + */ + fwspec.param[2] = IRQ_TYPE_EDGE_RISING; + + ret = irq_domain_alloc_irqs_parent(domain, virq, 1, &fwspec); + if (ret) { + dev_err(gicp->dev, "Cannot allocate parent IRQ\n"); + goto free_hwirq; + } + + ret = irq_domain_set_hwirq_and_chip(domain, virq, hwirq, + &gicp_irq_chip, gicp); + if (ret) + goto free_irqs_parent; + + return 0; + +free_irqs_parent: + irq_domain_free_irqs_parent(domain, virq, nr_irqs); +free_hwirq: + spin_lock(&gicp->spi_lock); + __clear_bit(hwirq, gicp->spi_bitmap); + spin_unlock(&gicp->spi_lock); + return ret; +} + +static void gicp_irq_domain_free(struct irq_domain *domain, + unsigned int virq, unsigned int nr_irqs) +{ + struct mvebu_gicp *gicp = domain->host_data; + struct irq_data *d = irq_domain_get_irq_data(domain, virq); + + if (d->hwirq >= gicp->spi_cnt) { + dev_err(gicp->dev, "Invalid hwirq %lu\n", d->hwirq); + return; + } + + irq_domain_free_irqs_parent(domain, virq, nr_irqs); + + spin_lock(&gicp->spi_lock); + __clear_bit(d->hwirq, gicp->spi_bitmap); + spin_unlock(&gicp->spi_lock); +} + +static const struct irq_domain_ops gicp_domain_ops = { + .alloc = gicp_irq_domain_alloc, + .free = gicp_irq_domain_free, +}; + +static struct irq_chip gicp_msi_irq_chip = { + .name = "GICP", + .irq_set_type = irq_chip_set_type_parent, +}; + +static struct msi_domain_ops gicp_msi_ops = { +}; + +static struct msi_domain_info gicp_msi_domain_info = { + .flags = (MSI_FLAG_USE_DEF_DOM_OPS | MSI_FLAG_USE_DEF_CHIP_OPS), + .ops = &gicp_msi_ops, + .chip = &gicp_msi_irq_chip, +}; + +static int mvebu_gicp_probe(struct platform_device *pdev) +{ + struct mvebu_gicp *gicp; + struct irq_domain *inner_domain, *plat_domain, *parent_domain; + struct device_node *node = pdev->dev.of_node; + struct device_node *irq_parent_dn; + int ret, i; + + gicp = devm_kzalloc(&pdev->dev, sizeof(*gicp), GFP_KERNEL); + if (!gicp) + return -ENOMEM; + + gicp->dev = &pdev->dev; + + gicp->res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + if (!gicp->res) + return -ENODEV; + + ret = of_property_count_u32_elems(node, "marvell,spi-ranges"); + if (ret < 0) + return ret; + + gicp->spi_ranges_cnt = ret / 2; + + gicp->spi_ranges = + devm_kzalloc(&pdev->dev, + gicp->spi_ranges_cnt * + sizeof(struct mvebu_gicp_spi_range), + GFP_KERNEL); + if (!gicp->spi_ranges) + return -ENOMEM; + + for (i = 0; i < gicp->spi_ranges_cnt; i++) { + of_property_read_u32_index(node, "marvell,spi-ranges", + i * 2, + &gicp->spi_ranges[i].start); + + of_property_read_u32_index(node, "marvell,spi-ranges", + i * 2 + 1, + &gicp->spi_ranges[i].count); + + gicp->spi_cnt += gicp->spi_ranges[i].count; + } + + gicp->spi_bitmap = devm_kzalloc(&pdev->dev, + BITS_TO_LONGS(gicp->spi_cnt), + GFP_KERNEL); + if (!gicp->spi_bitmap) + return -ENOMEM; + + irq_parent_dn = of_irq_find_parent(node); + if (!irq_parent_dn) { + dev_err(&pdev->dev, "failed to find parent IRQ node\n"); + return -ENODEV; + } + + parent_domain = irq_find_host(irq_parent_dn); + if (!parent_domain) { + dev_err(&pdev->dev, "failed to find parent IRQ domain\n"); + return -ENODEV; + } + + inner_domain = irq_domain_create_hierarchy(parent_domain, 0, + gicp->spi_cnt, + of_node_to_fwnode(node), + &gicp_domain_ops, gicp); + if (!inner_domain) + return -ENOMEM; + + + plat_domain = platform_msi_create_irq_domain(of_node_to_fwnode(node), + &gicp_msi_domain_info, + inner_domain); + if (!plat_domain) { + irq_domain_remove(inner_domain); + return -ENOMEM; + } + + platform_set_drvdata(pdev, gicp); + + return 0; +} + +static const struct of_device_id mvebu_gicp_of_match[] = { + { .compatible = "marvell,ap806-gicp", }, + {}, +}; + +static struct platform_driver mvebu_gicp_driver = { + .probe = mvebu_gicp_probe, + .driver = { + .name = "mvebu-gicp", + .of_match_table = mvebu_gicp_of_match, + }, +}; +builtin_platform_driver(mvebu_gicp_driver); diff --git a/drivers/irqchip/irq-mvebu-gicp.h b/drivers/irqchip/irq-mvebu-gicp.h new file mode 100644 index 000000000000..98535e886ea5 --- /dev/null +++ b/drivers/irqchip/irq-mvebu-gicp.h @@ -0,0 +1,11 @@ +#ifndef __MVEBU_GICP_H__ +#define __MVEBU_GICP_H__ + +#include + +struct device_node; + +int mvebu_gicp_get_doorbells(struct device_node *dn, phys_addr_t *setspi, + phys_addr_t *clrspi); + +#endif /* __MVEBU_GICP_H__ */ From e0de91a977bfe4c199138fa256a211a648ef6999 Mon Sep 17 00:00:00 2001 From: Thomas Petazzoni Date: Wed, 21 Jun 2017 15:29:15 +0200 Subject: [PATCH 097/112] irqchip/irq-mvebu-icu: Add new driver for Marvell ICU The Marvell ICU unit is found in the CP110 block of the Marvell Armada 7K and 8K SoCs. It collects the wired interrupts of the devices located in the CP110 and turns them into SPI interrupts in the GIC located in the AP806 side of the SoC, by using a memory transaction. Until now, the ICU was configured in a static fashion by the firmware, and Linux was relying on this static configuration. By having Linux configure the ICU, we are more flexible, and we can allocate dynamically the GIC SPI interrupts only for devices that are actually in use. The driver was initially written by Hanna Hawa . Reviewed-by: Marc Zyngier Signed-off-by: Thomas Petazzoni Signed-off-by: Marc Zyngier --- drivers/irqchip/Kconfig | 3 + drivers/irqchip/Makefile | 1 + drivers/irqchip/irq-mvebu-icu.c | 289 ++++++++++++++++++ .../interrupt-controller/mvebu-icu.h | 15 + 4 files changed, 308 insertions(+) create mode 100644 drivers/irqchip/irq-mvebu-icu.c create mode 100644 include/dt-bindings/interrupt-controller/mvebu-icu.h diff --git a/drivers/irqchip/Kconfig b/drivers/irqchip/Kconfig index e527ee596d54..676232a94f9f 100644 --- a/drivers/irqchip/Kconfig +++ b/drivers/irqchip/Kconfig @@ -271,6 +271,9 @@ config IRQ_MXS config MVEBU_GICP bool +config MVEBU_ICU + bool + config MVEBU_ODMI bool select GENERIC_MSI_IRQ_DOMAIN diff --git a/drivers/irqchip/Makefile b/drivers/irqchip/Makefile index f214f4573307..e88d856cc09c 100644 --- a/drivers/irqchip/Makefile +++ b/drivers/irqchip/Makefile @@ -70,6 +70,7 @@ obj-$(CONFIG_INGENIC_IRQ) += irq-ingenic.o obj-$(CONFIG_IMX_GPCV2) += irq-imx-gpcv2.o obj-$(CONFIG_PIC32_EVIC) += irq-pic32-evic.o obj-$(CONFIG_MVEBU_GICP) += irq-mvebu-gicp.o +obj-$(CONFIG_MVEBU_ICU) += irq-mvebu-icu.o obj-$(CONFIG_MVEBU_ODMI) += irq-mvebu-odmi.o obj-$(CONFIG_MVEBU_PIC) += irq-mvebu-pic.o obj-$(CONFIG_LS_SCFG_MSI) += irq-ls-scfg-msi.o diff --git a/drivers/irqchip/irq-mvebu-icu.c b/drivers/irqchip/irq-mvebu-icu.c new file mode 100644 index 000000000000..e18c48d3a92e --- /dev/null +++ b/drivers/irqchip/irq-mvebu-icu.c @@ -0,0 +1,289 @@ +/* + * Copyright (C) 2017 Marvell + * + * Hanna Hawa + * Thomas Petazzoni + * + * This file is licensed under the terms of the GNU General Public + * License version 2. This program is licensed "as is" without any + * warranty of any kind, whether express or implied. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +#include "irq-mvebu-gicp.h" + +/* ICU registers */ +#define ICU_SETSPI_NSR_AL 0x10 +#define ICU_SETSPI_NSR_AH 0x14 +#define ICU_CLRSPI_NSR_AL 0x18 +#define ICU_CLRSPI_NSR_AH 0x1c +#define ICU_INT_CFG(x) (0x100 + 4 * (x)) +#define ICU_INT_ENABLE BIT(24) +#define ICU_IS_EDGE BIT(28) +#define ICU_GROUP_SHIFT 29 + +/* ICU definitions */ +#define ICU_MAX_IRQS 207 +#define ICU_SATA0_ICU_ID 109 +#define ICU_SATA1_ICU_ID 107 + +struct mvebu_icu { + struct irq_chip irq_chip; + void __iomem *base; + struct irq_domain *domain; + struct device *dev; +}; + +struct mvebu_icu_irq_data { + struct mvebu_icu *icu; + unsigned int icu_group; + unsigned int type; +}; + +static void mvebu_icu_write_msg(struct msi_desc *desc, struct msi_msg *msg) +{ + struct irq_data *d = irq_get_irq_data(desc->irq); + struct mvebu_icu_irq_data *icu_irqd = d->chip_data; + struct mvebu_icu *icu = icu_irqd->icu; + unsigned int icu_int; + + if (msg->address_lo || msg->address_hi) { + /* Configure the ICU with irq number & type */ + icu_int = msg->data | ICU_INT_ENABLE; + if (icu_irqd->type & IRQ_TYPE_EDGE_RISING) + icu_int |= ICU_IS_EDGE; + icu_int |= icu_irqd->icu_group << ICU_GROUP_SHIFT; + } else { + /* De-configure the ICU */ + icu_int = 0; + } + + writel_relaxed(icu_int, icu->base + ICU_INT_CFG(d->hwirq)); + + /* + * The SATA unit has 2 ports, and a dedicated ICU entry per + * port. The ahci sata driver supports only one irq interrupt + * per SATA unit. To solve this conflict, we configure the 2 + * SATA wired interrupts in the south bridge into 1 GIC + * interrupt in the north bridge. Even if only a single port + * is enabled, if sata node is enabled, both interrupts are + * configured (regardless of which port is actually in use). + */ + if (d->hwirq == ICU_SATA0_ICU_ID || d->hwirq == ICU_SATA1_ICU_ID) { + writel_relaxed(icu_int, + icu->base + ICU_INT_CFG(ICU_SATA0_ICU_ID)); + writel_relaxed(icu_int, + icu->base + ICU_INT_CFG(ICU_SATA1_ICU_ID)); + } +} + +static int +mvebu_icu_irq_domain_translate(struct irq_domain *d, struct irq_fwspec *fwspec, + unsigned long *hwirq, unsigned int *type) +{ + struct mvebu_icu *icu = d->host_data; + unsigned int icu_group; + + /* Check the count of the parameters in dt */ + if (WARN_ON(fwspec->param_count < 3)) { + dev_err(icu->dev, "wrong ICU parameter count %d\n", + fwspec->param_count); + return -EINVAL; + } + + /* Only ICU group type is handled */ + icu_group = fwspec->param[0]; + if (icu_group != ICU_GRP_NSR && icu_group != ICU_GRP_SR && + icu_group != ICU_GRP_SEI && icu_group != ICU_GRP_REI) { + dev_err(icu->dev, "wrong ICU group type %x\n", icu_group); + return -EINVAL; + } + + *hwirq = fwspec->param[1]; + if (*hwirq >= ICU_MAX_IRQS) { + dev_err(icu->dev, "invalid interrupt number %ld\n", *hwirq); + return -EINVAL; + } + + /* Mask the type to prevent wrong DT configuration */ + *type = fwspec->param[2] & IRQ_TYPE_SENSE_MASK; + + return 0; +} + +static int +mvebu_icu_irq_domain_alloc(struct irq_domain *domain, unsigned int virq, + unsigned int nr_irqs, void *args) +{ + int err; + unsigned long hwirq; + struct irq_fwspec *fwspec = args; + struct mvebu_icu *icu = platform_msi_get_host_data(domain); + struct mvebu_icu_irq_data *icu_irqd; + + icu_irqd = kmalloc(sizeof(*icu_irqd), GFP_KERNEL); + if (!icu_irqd) + return -ENOMEM; + + err = mvebu_icu_irq_domain_translate(domain, fwspec, &hwirq, + &icu_irqd->type); + if (err) { + dev_err(icu->dev, "failed to translate ICU parameters\n"); + goto free_irqd; + } + + icu_irqd->icu_group = fwspec->param[0]; + icu_irqd->icu = icu; + + err = platform_msi_domain_alloc(domain, virq, nr_irqs); + if (err) { + dev_err(icu->dev, "failed to allocate ICU interrupt in parent domain\n"); + goto free_irqd; + } + + /* Make sure there is no interrupt left pending by the firmware */ + err = irq_set_irqchip_state(virq, IRQCHIP_STATE_PENDING, false); + if (err) + goto free_msi; + + err = irq_domain_set_hwirq_and_chip(domain, virq, hwirq, + &icu->irq_chip, icu_irqd); + if (err) { + dev_err(icu->dev, "failed to set the data to IRQ domain\n"); + goto free_msi; + } + + return 0; + +free_msi: + platform_msi_domain_free(domain, virq, nr_irqs); +free_irqd: + kfree(icu_irqd); + return err; +} + +static void +mvebu_icu_irq_domain_free(struct irq_domain *domain, unsigned int virq, + unsigned int nr_irqs) +{ + struct irq_data *d = irq_get_irq_data(virq); + struct mvebu_icu_irq_data *icu_irqd = d->chip_data; + + kfree(icu_irqd); + + platform_msi_domain_free(domain, virq, nr_irqs); +} + +static const struct irq_domain_ops mvebu_icu_domain_ops = { + .translate = mvebu_icu_irq_domain_translate, + .alloc = mvebu_icu_irq_domain_alloc, + .free = mvebu_icu_irq_domain_free, +}; + +static int mvebu_icu_probe(struct platform_device *pdev) +{ + struct mvebu_icu *icu; + struct device_node *node = pdev->dev.of_node; + struct device_node *gicp_dn; + struct resource *res; + phys_addr_t setspi, clrspi; + u32 i, icu_int; + int ret; + + icu = devm_kzalloc(&pdev->dev, sizeof(struct mvebu_icu), + GFP_KERNEL); + if (!icu) + return -ENOMEM; + + icu->dev = &pdev->dev; + + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + icu->base = devm_ioremap_resource(&pdev->dev, res); + if (IS_ERR(icu->base)) { + dev_err(&pdev->dev, "Failed to map icu base address.\n"); + return PTR_ERR(icu->base); + } + + icu->irq_chip.name = devm_kasprintf(&pdev->dev, GFP_KERNEL, + "ICU.%x", + (unsigned int)res->start); + if (!icu->irq_chip.name) + return -ENOMEM; + + icu->irq_chip.irq_mask = irq_chip_mask_parent; + icu->irq_chip.irq_unmask = irq_chip_unmask_parent; + icu->irq_chip.irq_eoi = irq_chip_eoi_parent; + icu->irq_chip.irq_set_type = irq_chip_set_type_parent; +#ifdef CONFIG_SMP + icu->irq_chip.irq_set_affinity = irq_chip_set_affinity_parent; +#endif + + /* + * We're probed after MSI domains have been resolved, so force + * resolution here. + */ + pdev->dev.msi_domain = of_msi_get_domain(&pdev->dev, node, + DOMAIN_BUS_PLATFORM_MSI); + if (!pdev->dev.msi_domain) + return -EPROBE_DEFER; + + gicp_dn = irq_domain_get_of_node(pdev->dev.msi_domain); + if (!gicp_dn) + return -ENODEV; + + ret = mvebu_gicp_get_doorbells(gicp_dn, &setspi, &clrspi); + if (ret) + return ret; + + /* Set Clear/Set ICU SPI message address in AP */ + writel_relaxed(upper_32_bits(setspi), icu->base + ICU_SETSPI_NSR_AH); + writel_relaxed(lower_32_bits(setspi), icu->base + ICU_SETSPI_NSR_AL); + writel_relaxed(upper_32_bits(clrspi), icu->base + ICU_CLRSPI_NSR_AH); + writel_relaxed(lower_32_bits(clrspi), icu->base + ICU_CLRSPI_NSR_AL); + + /* + * Clean all ICU interrupts with type SPI_NSR, required to + * avoid unpredictable SPI assignments done by firmware. + */ + for (i = 0 ; i < ICU_MAX_IRQS ; i++) { + icu_int = readl(icu->base + ICU_INT_CFG(i)); + if ((icu_int >> ICU_GROUP_SHIFT) == ICU_GRP_NSR) + writel_relaxed(0x0, icu->base + ICU_INT_CFG(i)); + } + + icu->domain = + platform_msi_create_device_domain(&pdev->dev, ICU_MAX_IRQS, + mvebu_icu_write_msg, + &mvebu_icu_domain_ops, icu); + if (!icu->domain) { + dev_err(&pdev->dev, "Failed to create ICU domain\n"); + return -ENOMEM; + } + + return 0; +} + +static const struct of_device_id mvebu_icu_of_match[] = { + { .compatible = "marvell,cp110-icu", }, + {}, +}; + +static struct platform_driver mvebu_icu_driver = { + .probe = mvebu_icu_probe, + .driver = { + .name = "mvebu-icu", + .of_match_table = mvebu_icu_of_match, + }, +}; +builtin_platform_driver(mvebu_icu_driver); diff --git a/include/dt-bindings/interrupt-controller/mvebu-icu.h b/include/dt-bindings/interrupt-controller/mvebu-icu.h new file mode 100644 index 000000000000..8249558545c7 --- /dev/null +++ b/include/dt-bindings/interrupt-controller/mvebu-icu.h @@ -0,0 +1,15 @@ +/* + * This header provides constants for the MVEBU ICU driver. + */ + +#ifndef _DT_BINDINGS_INTERRUPT_CONTROLLER_MVEBU_ICU_H +#define _DT_BINDINGS_INTERRUPT_CONTROLLER_MVEBU_ICU_H + +/* interrupt specifier cell 0 */ + +#define ICU_GRP_NSR 0x0 +#define ICU_GRP_SR 0x1 +#define ICU_GRP_SEI 0x4 +#define ICU_GRP_REI 0x5 + +#endif From 935bba7ccf0246fae8247aa8dab7aa85c43de2d8 Mon Sep 17 00:00:00 2001 From: Arvind Yadav Date: Thu, 22 Jun 2017 16:05:30 +0530 Subject: [PATCH 098/112] irqchip/gic-v3-its: Make of_device_ids const of_device_ids are not supposed to change at runtime. All functions working with of_device_ids provided by work with const of_device_ids. So mark the non-const structs as const. Signed-off-by: Arvind Yadav Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-gic-v3-its.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c index b33528026c5c..fee7d13a5587 100644 --- a/drivers/irqchip/irq-gic-v3-its.c +++ b/drivers/irqchip/irq-gic-v3-its.c @@ -1804,7 +1804,7 @@ int its_cpu_init(void) return 0; } -static struct of_device_id its_device_id[] = { +static const struct of_device_id its_device_id[] = { { .compatible = "arm,gic-v3-its", }, {}, }; From 5b8670612d4014f33d9f130693bfba5327726411 Mon Sep 17 00:00:00 2001 From: Arvind Yadav Date: Thu, 22 Jun 2017 16:11:13 +0530 Subject: [PATCH 099/112] irqchip/gic-v3-its-platform-msi: Make of_device_ids const of_device_ids are not supposed to change at runtime. All functions working with of_device_ids provided by work with const of_device_ids. So mark the non-const structs as const. Signed-off-by: Arvind Yadav Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-gic-v3-its-platform-msi.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/irqchip/irq-gic-v3-its-platform-msi.c b/drivers/irqchip/irq-gic-v3-its-platform-msi.c index 9e9dda33eb17..249240d9a425 100644 --- a/drivers/irqchip/irq-gic-v3-its-platform-msi.c +++ b/drivers/irqchip/irq-gic-v3-its-platform-msi.c @@ -86,7 +86,7 @@ static struct msi_domain_info its_pmsi_domain_info = { .chip = &its_pmsi_irq_chip, }; -static struct of_device_id its_device_id[] = { +static const struct of_device_id its_device_id[] = { { .compatible = "arm,gic-v3-its", }, {}, }; From dbd2b826723319eb6e4bee5214f8901e336c06be Mon Sep 17 00:00:00 2001 From: Ganapatrao Kulkarni Date: Thu, 22 Jun 2017 11:40:12 +0530 Subject: [PATCH 100/112] irqchip/gic-v3-its: Add ACPI NUMA node mapping Add code to parse SRAT ITS Affinity sub table as defined in ACPI 6.2. Later in per device probe, ITS devices are mapped to numa node using ITS Id to proximity domain mapping. [maz: fix dependency on ACPICA, fixed structure name, minor cleanups] Reviewed-by: Lorenzo Pieralisi Signed-off-by: Ganapatrao Kulkarni Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-gic-v3-its.c | 76 +++++++++++++++++++++++++++++++- 1 file changed, 75 insertions(+), 1 deletion(-) diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c index fee7d13a5587..63cd0f2b8707 100644 --- a/drivers/irqchip/irq-gic-v3-its.c +++ b/drivers/irqchip/irq-gic-v3-its.c @@ -1836,6 +1836,78 @@ static int __init its_of_probe(struct device_node *node) #define ACPI_GICV3_ITS_MEM_SIZE (SZ_128K) +#if defined(CONFIG_ACPI_NUMA) && (ACPI_CA_VERSION >= 0x20170531) +struct its_srat_map { + /* numa node id */ + u32 numa_node; + /* GIC ITS ID */ + u32 its_id; +}; + +static struct its_srat_map its_srat_maps[MAX_NUMNODES] __initdata; +static int its_in_srat __initdata; + +static int __init acpi_get_its_numa_node(u32 its_id) +{ + int i; + + for (i = 0; i < its_in_srat; i++) { + if (its_id == its_srat_maps[i].its_id) + return its_srat_maps[i].numa_node; + } + return NUMA_NO_NODE; +} + +static int __init gic_acpi_parse_srat_its(struct acpi_subtable_header *header, + const unsigned long end) +{ + int node; + struct acpi_srat_gic_its_affinity *its_affinity; + + its_affinity = (struct acpi_srat_gic_its_affinity *)header; + if (!its_affinity) + return -EINVAL; + + if (its_affinity->header.length < sizeof(*its_affinity)) { + pr_err("SRAT: Invalid header length %d in ITS affinity\n", + its_affinity->header.length); + return -EINVAL; + } + + if (its_in_srat >= MAX_NUMNODES) { + pr_err("SRAT: ITS affinity exceeding max count[%d]\n", + MAX_NUMNODES); + return -EINVAL; + } + + node = acpi_map_pxm_to_node(its_affinity->proximity_domain); + + if (node == NUMA_NO_NODE || node >= MAX_NUMNODES) { + pr_err("SRAT: Invalid NUMA node %d in ITS affinity\n", node); + return 0; + } + + its_srat_maps[its_in_srat].numa_node = node; + its_srat_maps[its_in_srat].its_id = its_affinity->its_id; + its_in_srat++; + pr_info("SRAT: PXM %d -> ITS %d -> Node %d\n", + its_affinity->proximity_domain, its_affinity->its_id, node); + + return 0; +} + +static void __init acpi_table_parse_srat_its(void) +{ + acpi_table_parse_entries(ACPI_SIG_SRAT, + sizeof(struct acpi_table_srat), + ACPI_SRAT_TYPE_GIC_ITS_AFFINITY, + gic_acpi_parse_srat_its, 0); +} +#else +static void __init acpi_table_parse_srat_its(void) { } +static int __init acpi_get_its_numa_node(u32 its_id) { return NUMA_NO_NODE; } +#endif + static int __init gic_acpi_parse_madt_its(struct acpi_subtable_header *header, const unsigned long end) { @@ -1864,7 +1936,8 @@ static int __init gic_acpi_parse_madt_its(struct acpi_subtable_header *header, goto dom_err; } - err = its_probe_one(&res, dom_handle, NUMA_NO_NODE); + err = its_probe_one(&res, dom_handle, + acpi_get_its_numa_node(its_entry->translation_id)); if (!err) return 0; @@ -1876,6 +1949,7 @@ dom_err: static void __init its_acpi_probe(void) { + acpi_table_parse_srat_its(); acpi_table_parse_madt(ACPI_MADT_TYPE_GENERIC_TRANSLATOR, gic_acpi_parse_madt_its, 0); } From 6c31e123dc4c59eeaae6ac1cd08b929e8b6f7651 Mon Sep 17 00:00:00 2001 From: Shanker Donthineni Date: Thu, 22 Jun 2017 18:19:14 -0500 Subject: [PATCH 101/112] irqchip/gic-v3-its: Don't assume GICv3 hardware supports 16bit INTID The current ITS driver is assuming every ITS hardware implementation supports minimum of 16bit INTID. But this is not true, as per GICv3 specification, INTID field is IMPLEMENTATION DEFINED in the range of 14-24 bits. We might see an unpredictable system behavior on systems where hardware support less than 16bits and software tries to use 64K LPI interrupts. On Qualcomm Datacenter Technologies QDF2400 platform, boot log shows confusing information about number of LPI chunks as shown below. The QDF2400 ITS hardware supports 24bit INTID. This patch allocates the memory resources for PEND/PROP tables based on discoverable value which is specified in GITS_TYPER.IDbits. Also it fixes the log message that reflects the correct number of LPI chunks were allocated. ITS@0xff7efe0000: allocated 524288 Devices @3c0400000 (indirect, esz 8, psz 64K, shr 1) ITS@0xff7efe0000: allocated 8192 Interrupt Collections @3c0130000 (flat, esz 8, psz 64K, shr 1) ITS@0xff7efe0000: allocated 8192 Virtual CPUs @3c0140000 (flat, esz 8, psz 64K, shr 1) ITS: Allocated 524032 chunks for LPIs PCI/MSI: ITS@0xff7efe0000 domain created Platform MSI: ITS@0xff7efe0000 domain created Signed-off-by: Shanker Donthineni Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-gic-v3-its.c | 24 ++++++++++-------------- 1 file changed, 10 insertions(+), 14 deletions(-) diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c index 63cd0f2b8707..fed99c55e2f4 100644 --- a/drivers/irqchip/irq-gic-v3-its.c +++ b/drivers/irqchip/irq-gic-v3-its.c @@ -691,9 +691,11 @@ static struct irq_chip its_irq_chip = { */ #define IRQS_PER_CHUNK_SHIFT 5 #define IRQS_PER_CHUNK (1 << IRQS_PER_CHUNK_SHIFT) +#define ITS_MAX_LPI_NRBITS 16 /* 64K LPIs */ static unsigned long *lpi_bitmap; static u32 lpi_chunks; +static u32 lpi_id_bits; static DEFINE_SPINLOCK(lpi_lock); static int its_lpi_to_chunk(int lpi) @@ -789,17 +791,13 @@ static void its_lpi_free(struct event_lpi_map *map) } /* - * We allocate 64kB for PROPBASE. That gives us at most 64K LPIs to + * We allocate memory for PROPBASE to cover 2 ^ lpi_id_bits LPIs to * deal with (one configuration byte per interrupt). PENDBASE has to * be 64kB aligned (one bit per LPI, plus 8192 bits for SPI/PPI/SGI). */ -#define LPI_PROPBASE_SZ SZ_64K -#define LPI_PENDBASE_SZ (LPI_PROPBASE_SZ / 8 + SZ_1K) - -/* - * This is how many bits of ID we need, including the useless ones. - */ -#define LPI_NRBITS ilog2(LPI_PROPBASE_SZ + SZ_8K) +#define LPI_NRBITS lpi_id_bits +#define LPI_PROPBASE_SZ ALIGN(BIT(LPI_NRBITS), SZ_64K) +#define LPI_PENDBASE_SZ ALIGN(BIT(LPI_NRBITS) / 8, SZ_64K) #define LPI_PROP_DEFAULT_PRIO 0xa0 @@ -807,6 +805,7 @@ static int __init its_alloc_lpi_tables(void) { phys_addr_t paddr; + lpi_id_bits = min_t(u32, gic_rdists->id_bits, ITS_MAX_LPI_NRBITS); gic_rdists->prop_page = alloc_pages(GFP_NOWAIT, get_order(LPI_PROPBASE_SZ)); if (!gic_rdists->prop_page) { @@ -825,7 +824,7 @@ static int __init its_alloc_lpi_tables(void) /* Make sure the GIC will observe the written configuration */ gic_flush_dcache_to_poc(page_address(gic_rdists->prop_page), LPI_PROPBASE_SZ); - return 0; + return its_lpi_init(lpi_id_bits); } static const char *its_base_type_string[] = { @@ -1100,7 +1099,7 @@ static void its_cpu_init_lpis(void) * hence the 'max(LPI_PENDBASE_SZ, SZ_64K)' below. */ pend_page = alloc_pages(GFP_NOWAIT | __GFP_ZERO, - get_order(max(LPI_PENDBASE_SZ, SZ_64K))); + get_order(max_t(u32, LPI_PENDBASE_SZ, SZ_64K))); if (!pend_page) { pr_err("Failed to allocate PENDBASE for CPU%d\n", smp_processor_id()); @@ -1975,8 +1974,5 @@ int __init its_init(struct fwnode_handle *handle, struct rdists *rdists, } gic_rdists = rdists; - its_alloc_lpi_tables(); - its_lpi_init(rdists->id_bits); - - return 0; + return its_alloc_lpi_tables(); } From c2ce34c0a0e5187195ecade872be950d2611ba68 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Sat, 24 Jun 2017 11:05:59 +0200 Subject: [PATCH 102/112] genirq/debugfs: Remove pointless NULL pointer check debugfs_remove() has it's own NULL pointer check. Remove the conditional and make irq_remove_debugfs_entry() an inline helper Reported-by: kbuild test robot Signed-off-by: Thomas Gleixner --- kernel/irq/debugfs.c | 7 ------- kernel/irq/internals.h | 7 ++++++- 2 files changed, 6 insertions(+), 8 deletions(-) diff --git a/kernel/irq/debugfs.c b/kernel/irq/debugfs.c index dbd6e78db213..4d384edc0c64 100644 --- a/kernel/irq/debugfs.c +++ b/kernel/irq/debugfs.c @@ -3,7 +3,6 @@ * * This file is licensed under the GPL V2. */ -#include #include #include @@ -191,12 +190,6 @@ void irq_add_debugfs_entry(unsigned int irq, struct irq_desc *desc) &dfs_irq_ops); } -void irq_remove_debugfs_entry(struct irq_desc *desc) -{ - if (desc->debugfs_file) - debugfs_remove(desc->debugfs_file); -} - static int __init irq_debugfs_init(void) { struct dentry *root_dir; diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index 5fd105e252c3..a573e0771baf 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -318,8 +318,13 @@ static inline bool irq_fixup_move_pending(struct irq_desc *desc, bool fclear) #endif /* !CONFIG_GENERIC_PENDING_IRQ */ #ifdef CONFIG_GENERIC_IRQ_DEBUGFS +#include + void irq_add_debugfs_entry(unsigned int irq, struct irq_desc *desc); -void irq_remove_debugfs_entry(struct irq_desc *desc); +static inline void irq_remove_debugfs_entry(struct irq_desc *desc) +{ + debugfs_remove(desc->debugfs_file); +} # ifdef CONFIG_IRQ_DOMAIN void irq_domain_debugfs_init(struct dentry *root); # else From b2d3d61adb7b73cfe5f82404f7a130a76fc64232 Mon Sep 17 00:00:00 2001 From: Daniel Lezcano Date: Fri, 23 Jun 2017 16:11:07 +0200 Subject: [PATCH 103/112] genirq/timings: Add infrastructure to track the interrupt timings The interrupt framework gives a lot of information about each interrupt. It does not keep track of when those interrupts occur though, which is a prerequisite for estimating the next interrupt arrival for power management purposes. Add a mechanism to record the timestamp for each interrupt occurrences in a per-CPU circular buffer to help with the prediction of the next occurrence using a statistical model. Each CPU can store up to IRQ_TIMINGS_SIZE events , the current value of IRQ_TIMINGS_SIZE is 32. Each event is encoded into a single u64, where the high 48 bits are used for the timestamp and the low 16 bits are for the irq number. A static key is introduced so when the irq prediction is switched off at runtime, the overhead is near to zero. It results in most of the code in internals.h for inline reasons and a very few in the new file timings.c. The latter will contain more in the next patch which will provide the statistical model for the next event prediction. Signed-off-by: Daniel Lezcano Signed-off-by: Thomas Gleixner Acked-by: Nicolas Pitre Cc: Jens Axboe Cc: Hannes Reinecke Cc: Vincent Guittot Cc: "Rafael J . Wysocki" Cc: Peter Zijlstra Cc: Bjorn Helgaas Link: http://lkml.kernel.org/r/1498227072-5980-1-git-send-email-daniel.lezcano@linaro.org --- include/linux/interrupt.h | 5 +++ kernel/irq/Kconfig | 3 ++ kernel/irq/Makefile | 1 + kernel/irq/handle.c | 2 + kernel/irq/internals.h | 90 +++++++++++++++++++++++++++++++++++++++ kernel/irq/manage.c | 3 ++ kernel/irq/timings.c | 30 +++++++++++++ 7 files changed, 134 insertions(+) create mode 100644 kernel/irq/timings.c diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h index a6fba4804672..9f617238a2f7 100644 --- a/include/linux/interrupt.h +++ b/include/linux/interrupt.h @@ -703,6 +703,11 @@ static inline void init_irq_proc(void) } #endif +#ifdef CONFIG_IRQ_TIMINGS +void irq_timings_enable(void); +void irq_timings_disable(void); +#endif + struct seq_file; int show_interrupts(struct seq_file *p, void *v); int arch_show_interrupts(struct seq_file *p, int prec); diff --git a/kernel/irq/Kconfig b/kernel/irq/Kconfig index fcbb1d6d51cb..27c4e774071c 100644 --- a/kernel/irq/Kconfig +++ b/kernel/irq/Kconfig @@ -85,6 +85,9 @@ config GENERIC_MSI_IRQ_DOMAIN config HANDLE_DOMAIN_IRQ bool +config IRQ_TIMINGS + bool + config IRQ_DOMAIN_DEBUG bool "Expose hardware/virtual IRQ mapping via debugfs" depends on IRQ_DOMAIN && DEBUG_FS diff --git a/kernel/irq/Makefile b/kernel/irq/Makefile index c61fc9c2d1f7..e4aef7351f2b 100644 --- a/kernel/irq/Makefile +++ b/kernel/irq/Makefile @@ -1,5 +1,6 @@ obj-y := irqdesc.o handle.o manage.o spurious.o resend.o chip.o dummychip.o devres.o +obj-$(CONFIG_IRQ_TIMINGS) += timings.o obj-$(CONFIG_GENERIC_IRQ_CHIP) += generic-chip.o obj-$(CONFIG_GENERIC_IRQ_PROBE) += autoprobe.o obj-$(CONFIG_IRQ_DOMAIN) += irqdomain.o diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index d3f24905852c..eb4d3e8945b8 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -138,6 +138,8 @@ irqreturn_t __handle_irq_event_percpu(struct irq_desc *desc, unsigned int *flags unsigned int irq = desc->irq_data.irq; struct irqaction *action; + record_irq_time(desc); + for_each_action_of_desc(desc, action) { irqreturn_t res; diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index a573e0771baf..b95b74920433 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -8,6 +8,7 @@ #include #include #include +#include #ifdef CONFIG_SPARSE_IRQ # define IRQ_BITMAP_BITS (NR_IRQS + 8196) @@ -57,6 +58,7 @@ enum { IRQS_WAITING = 0x00000080, IRQS_PENDING = 0x00000200, IRQS_SUSPENDED = 0x00000800, + IRQS_TIMINGS = 0x00001000, }; #include "debug.h" @@ -255,6 +257,94 @@ static inline void irq_pm_remove_action(struct irq_desc *desc, struct irqaction *action) { } #endif +#ifdef CONFIG_IRQ_TIMINGS + +#define IRQ_TIMINGS_SHIFT 5 +#define IRQ_TIMINGS_SIZE (1 << IRQ_TIMINGS_SHIFT) +#define IRQ_TIMINGS_MASK (IRQ_TIMINGS_SIZE - 1) + +/** + * struct irq_timings - irq timings storing structure + * @values: a circular buffer of u64 encoded values + * @count: the number of elements in the array + */ +struct irq_timings { + u64 values[IRQ_TIMINGS_SIZE]; + int count; +}; + +DECLARE_PER_CPU(struct irq_timings, irq_timings); + +static inline void irq_remove_timings(struct irq_desc *desc) +{ + desc->istate &= ~IRQS_TIMINGS; +} + +static inline void irq_setup_timings(struct irq_desc *desc, struct irqaction *act) +{ + /* + * We don't need the measurement because the idle code already + * knows the next expiry event. + */ + if (act->flags & __IRQF_TIMER) + return; + + desc->istate |= IRQS_TIMINGS; +} + +extern void irq_timings_enable(void); +extern void irq_timings_disable(void); + +DECLARE_STATIC_KEY_FALSE(irq_timing_enabled); + +/* + * The interrupt number and the timestamp are encoded into a single + * u64 variable to optimize the size. + * 48 bit time stamp and 16 bit IRQ number is way sufficient. + * Who cares an IRQ after 78 hours of idle time? + */ +static inline u64 irq_timing_encode(u64 timestamp, int irq) +{ + return (timestamp << 16) | irq; +} + +static inline int irq_timing_decode(u64 value, u64 *timestamp) +{ + *timestamp = value >> 16; + return value & U16_MAX; +} + +/* + * The function record_irq_time is only called in one place in the + * interrupts handler. We want this function always inline so the code + * inside is embedded in the function and the static key branching + * code can act at the higher level. Without the explicit + * __always_inline we can end up with a function call and a small + * overhead in the hotpath for nothing. + */ +static __always_inline void record_irq_time(struct irq_desc *desc) +{ + if (!static_branch_likely(&irq_timing_enabled)) + return; + + if (desc->istate & IRQS_TIMINGS) { + struct irq_timings *timings = this_cpu_ptr(&irq_timings); + + timings->values[timings->count & IRQ_TIMINGS_MASK] = + irq_timing_encode(local_clock(), + irq_desc_get_irq(desc)); + + timings->count++; + } +} +#else +static inline void irq_remove_timings(struct irq_desc *desc) {} +static inline void irq_setup_timings(struct irq_desc *desc, + struct irqaction *act) {}; +static inline void record_irq_time(struct irq_desc *desc) {} +#endif /* CONFIG_IRQ_TIMINGS */ + + #ifdef CONFIG_GENERIC_IRQ_CHIP void irq_init_generic_chip(struct irq_chip_generic *gc, const char *name, int num_ct, unsigned int irq_base, diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 3577c091ac7b..5c11c1730ba5 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -1348,6 +1348,8 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) raw_spin_unlock_irqrestore(&desc->lock, flags); + irq_setup_timings(desc, new); + /* * Strictly no need to wake it up, but hung_task complains * when no hard interrupt wakes the thread up. @@ -1474,6 +1476,7 @@ static struct irqaction *__free_irq(unsigned int irq, void *dev_id) irq_settings_clr_disable_unlazy(desc); irq_shutdown(desc); irq_release_resources(desc); + irq_remove_timings(desc); } #ifdef CONFIG_SMP diff --git a/kernel/irq/timings.c b/kernel/irq/timings.c new file mode 100644 index 000000000000..56cf6870fa26 --- /dev/null +++ b/kernel/irq/timings.c @@ -0,0 +1,30 @@ +/* + * linux/kernel/irq/timings.c + * + * Copyright (C) 2016, Linaro Ltd - Daniel Lezcano + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + * + */ +#include +#include +#include +#include + +#include "internals.h" + +DEFINE_STATIC_KEY_FALSE(irq_timing_enabled); + +DEFINE_PER_CPU(struct irq_timings, irq_timings); + +void irq_timings_enable(void) +{ + static_branch_enable(&irq_timing_enabled); +} + +void irq_timings_disable(void) +{ + static_branch_disable(&irq_timing_enabled); +} From e1c921495534002d727b15a76a2f8c20b6b108b5 Mon Sep 17 00:00:00 2001 From: Daniel Lezcano Date: Fri, 23 Jun 2017 16:11:08 +0200 Subject: [PATCH 104/112] genirq/timings: Add infrastructure for estimating the next interrupt arrival time An interrupt behaves with a burst of activity with periodic interval of time followed by one or two peaks of longer interval. As the time intervals are periodic, statistically speaking they follow a normal distribution and each interrupts can be tracked individually. Add a mechanism to compute the statistics on all interrupts, except the timers which are deterministic from a prediction point of view, as their expiry time is known. The goal is to extract the periodicity for each interrupt, with the last timestamp and sum them, so the next event can be predicted to a certain extent. Taking the earliest prediction gives the expected wakeup on the system (assuming a timer won't expire before). Signed-off-by: Daniel Lezcano Signed-off-by: Thomas Gleixner Cc: Nicolas Pitre Cc: Jens Axboe Cc: Hannes Reinecke Cc: Vincent Guittot Cc: "Rafael J . Wysocki" Cc: Peter Zijlstra Cc: Bjorn Helgaas Link: http://lkml.kernel.org/r/1498227072-5980-2-git-send-email-daniel.lezcano@linaro.org --- include/linux/interrupt.h | 1 + kernel/irq/internals.h | 19 +++ kernel/irq/timings.c | 339 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 359 insertions(+) diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h index 9f617238a2f7..37f8e354f564 100644 --- a/include/linux/interrupt.h +++ b/include/linux/interrupt.h @@ -706,6 +706,7 @@ static inline void init_irq_proc(void) #ifdef CONFIG_IRQ_TIMINGS void irq_timings_enable(void); void irq_timings_disable(void); +u64 irq_timings_next_event(u64 now); #endif struct seq_file; diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index b95b74920433..9da14d125df4 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -275,13 +275,21 @@ struct irq_timings { DECLARE_PER_CPU(struct irq_timings, irq_timings); +extern void irq_timings_free(int irq); +extern int irq_timings_alloc(int irq); + static inline void irq_remove_timings(struct irq_desc *desc) { desc->istate &= ~IRQS_TIMINGS; + + irq_timings_free(irq_desc_get_irq(desc)); } static inline void irq_setup_timings(struct irq_desc *desc, struct irqaction *act) { + int irq = irq_desc_get_irq(desc); + int ret; + /* * We don't need the measurement because the idle code already * knows the next expiry event. @@ -289,6 +297,17 @@ static inline void irq_setup_timings(struct irq_desc *desc, struct irqaction *ac if (act->flags & __IRQF_TIMER) return; + /* + * In case the timing allocation fails, we just want to warn, + * not fail, so letting the system boot anyway. + */ + ret = irq_timings_alloc(irq); + if (ret) { + pr_warn("Failed to allocate irq timing stats for irq%d (%d)", + irq, ret); + return; + } + desc->istate |= IRQS_TIMINGS; } diff --git a/kernel/irq/timings.c b/kernel/irq/timings.c index 56cf6870fa26..c8c1d073fbf1 100644 --- a/kernel/irq/timings.c +++ b/kernel/irq/timings.c @@ -8,10 +8,16 @@ * published by the Free Software Foundation. * */ +#include #include +#include #include #include +#include #include +#include + +#include #include "internals.h" @@ -19,6 +25,18 @@ DEFINE_STATIC_KEY_FALSE(irq_timing_enabled); DEFINE_PER_CPU(struct irq_timings, irq_timings); +struct irqt_stat { + u64 next_evt; + u64 last_ts; + u64 variance; + u32 avg; + u32 nr_samples; + int anomalies; + int valid; +}; + +static DEFINE_IDR(irqt_stats); + void irq_timings_enable(void) { static_branch_enable(&irq_timing_enabled); @@ -28,3 +46,324 @@ void irq_timings_disable(void) { static_branch_disable(&irq_timing_enabled); } + +/** + * irqs_update - update the irq timing statistics with a new timestamp + * + * @irqs: an irqt_stat struct pointer + * @ts: the new timestamp + * + * The statistics are computed online, in other words, the code is + * designed to compute the statistics on a stream of values rather + * than doing multiple passes on the values to compute the average, + * then the variance. The integer division introduces a loss of + * precision but with an acceptable error margin regarding the results + * we would have with the double floating precision: we are dealing + * with nanosec, so big numbers, consequently the mantisse is + * negligeable, especially when converting the time in usec + * afterwards. + * + * The computation happens at idle time. When the CPU is not idle, the + * interrupts' timestamps are stored in the circular buffer, when the + * CPU goes idle and this routine is called, all the buffer's values + * are injected in the statistical model continuying to extend the + * statistics from the previous busy-idle cycle. + * + * The observations showed a device will trigger a burst of periodic + * interrupts followed by one or two peaks of longer time, for + * instance when a SD card device flushes its cache, then the periodic + * intervals occur again. A one second inactivity period resets the + * stats, that gives us the certitude the statistical values won't + * exceed 1x10^9, thus the computation won't overflow. + * + * Basically, the purpose of the algorithm is to watch the periodic + * interrupts and eliminate the peaks. + * + * An interrupt is considered periodically stable if the interval of + * its occurences follow the normal distribution, thus the values + * comply with: + * + * avg - 3 x stddev < value < avg + 3 x stddev + * + * Which can be simplified to: + * + * -3 x stddev < value - avg < 3 x stddev + * + * abs(value - avg) < 3 x stddev + * + * In order to save a costly square root computation, we use the + * variance. For the record, stddev = sqrt(variance). The equation + * above becomes: + * + * abs(value - avg) < 3 x sqrt(variance) + * + * And finally we square it: + * + * (value - avg) ^ 2 < (3 x sqrt(variance)) ^ 2 + * + * (value - avg) x (value - avg) < 9 x variance + * + * Statistically speaking, any values out of this interval is + * considered as an anomaly and is discarded. However, a normal + * distribution appears when the number of samples is 30 (it is the + * rule of thumb in statistics, cf. "30 samples" on Internet). When + * there are three consecutive anomalies, the statistics are resetted. + * + */ +static void irqs_update(struct irqt_stat *irqs, u64 ts) +{ + u64 old_ts = irqs->last_ts; + u64 variance = 0; + u64 interval; + s64 diff; + + /* + * The timestamps are absolute time values, we need to compute + * the timing interval between two interrupts. + */ + irqs->last_ts = ts; + + /* + * The interval type is u64 in order to deal with the same + * type in our computation, that prevent mindfuck issues with + * overflow, sign and division. + */ + interval = ts - old_ts; + + /* + * The interrupt triggered more than one second apart, that + * ends the sequence as predictible for our purpose. In this + * case, assume we have the beginning of a sequence and the + * timestamp is the first value. As it is impossible to + * predict anything at this point, return. + * + * Note the first timestamp of the sequence will always fall + * in this test because the old_ts is zero. That is what we + * want as we need another timestamp to compute an interval. + */ + if (interval >= NSEC_PER_SEC) { + memset(irqs, 0, sizeof(*irqs)); + irqs->last_ts = ts; + return; + } + + /* + * Pre-compute the delta with the average as the result is + * used several times in this function. + */ + diff = interval - irqs->avg; + + /* + * Increment the number of samples. + */ + irqs->nr_samples++; + + /* + * Online variance divided by the number of elements if there + * is more than one sample. Normally the formula is division + * by nr_samples - 1 but we assume the number of element will be + * more than 32 and dividing by 32 instead of 31 is enough + * precise. + */ + if (likely(irqs->nr_samples > 1)) + variance = irqs->variance >> IRQ_TIMINGS_SHIFT; + + /* + * The rule of thumb in statistics for the normal distribution + * is having at least 30 samples in order to have the model to + * apply. Values outside the interval are considered as an + * anomaly. + */ + if ((irqs->nr_samples >= 30) && ((diff * diff) > (9 * variance))) { + /* + * After three consecutive anomalies, we reset the + * stats as it is no longer stable enough. + */ + if (irqs->anomalies++ >= 3) { + memset(irqs, 0, sizeof(*irqs)); + irqs->last_ts = ts; + return; + } + } else { + /* + * The anomalies must be consecutives, so at this + * point, we reset the anomalies counter. + */ + irqs->anomalies = 0; + } + + /* + * The interrupt is considered stable enough to try to predict + * the next event on it. + */ + irqs->valid = 1; + + /* + * Online average algorithm: + * + * new_average = average + ((value - average) / count) + * + * The variance computation depends on the new average + * to be computed here first. + * + */ + irqs->avg = irqs->avg + (diff >> IRQ_TIMINGS_SHIFT); + + /* + * Online variance algorithm: + * + * new_variance = variance + (value - average) x (value - new_average) + * + * Warning: irqs->avg is updated with the line above, hence + * 'interval - irqs->avg' is no longer equal to 'diff' + */ + irqs->variance = irqs->variance + (diff * (interval - irqs->avg)); + + /* + * Update the next event + */ + irqs->next_evt = ts + irqs->avg; +} + +/** + * irq_timings_next_event - Return when the next event is supposed to arrive + * + * During the last busy cycle, the number of interrupts is incremented + * and stored in the irq_timings structure. This information is + * necessary to: + * + * - know if the index in the table wrapped up: + * + * If more than the array size interrupts happened during the + * last busy/idle cycle, the index wrapped up and we have to + * begin with the next element in the array which is the last one + * in the sequence, otherwise it is a the index 0. + * + * - have an indication of the interrupts activity on this CPU + * (eg. irq/sec) + * + * The values are 'consumed' after inserting in the statistical model, + * thus the count is reinitialized. + * + * The array of values **must** be browsed in the time direction, the + * timestamp must increase between an element and the next one. + * + * Returns a nanosec time based estimation of the earliest interrupt, + * U64_MAX otherwise. + */ +u64 irq_timings_next_event(u64 now) +{ + struct irq_timings *irqts = this_cpu_ptr(&irq_timings); + struct irqt_stat *irqs; + struct irqt_stat __percpu *s; + u64 ts, next_evt = U64_MAX; + int i, irq = 0; + + /* + * This function must be called with the local irq disabled in + * order to prevent the timings circular buffer to be updated + * while we are reading it. + */ + WARN_ON_ONCE(!irqs_disabled()); + + /* + * Number of elements in the circular buffer: If it happens it + * was flushed before, then the number of elements could be + * smaller than IRQ_TIMINGS_SIZE, so the count is used, + * otherwise the array size is used as we wrapped. The index + * begins from zero when we did not wrap. That could be done + * in a nicer way with the proper circular array structure + * type but with the cost of extra computation in the + * interrupt handler hot path. We choose efficiency. + * + * Inject measured irq/timestamp to the statistical model + * while decrementing the counter because we consume the data + * from our circular buffer. + */ + for (i = irqts->count & IRQ_TIMINGS_MASK, + irqts->count = min(IRQ_TIMINGS_SIZE, irqts->count); + irqts->count > 0; irqts->count--, i = (i + 1) & IRQ_TIMINGS_MASK) { + + irq = irq_timing_decode(irqts->values[i], &ts); + + s = idr_find(&irqt_stats, irq); + if (s) { + irqs = this_cpu_ptr(s); + irqs_update(irqs, ts); + } + } + + /* + * Look in the list of interrupts' statistics, the earliest + * next event. + */ + idr_for_each_entry(&irqt_stats, s, i) { + + irqs = this_cpu_ptr(s); + + if (!irqs->valid) + continue; + + if (irqs->next_evt <= now) { + irq = i; + next_evt = now; + + /* + * This interrupt mustn't use in the future + * until new events occur and update the + * statistics. + */ + irqs->valid = 0; + break; + } + + if (irqs->next_evt < next_evt) { + irq = i; + next_evt = irqs->next_evt; + } + } + + return next_evt; +} + +void irq_timings_free(int irq) +{ + struct irqt_stat __percpu *s; + + s = idr_find(&irqt_stats, irq); + if (s) { + free_percpu(s); + idr_remove(&irqt_stats, irq); + } +} + +int irq_timings_alloc(int irq) +{ + struct irqt_stat __percpu *s; + int id; + + /* + * Some platforms can have the same private interrupt per cpu, + * so this function may be be called several times with the + * same interrupt number. Just bail out in case the per cpu + * stat structure is already allocated. + */ + s = idr_find(&irqt_stats, irq); + if (s) + return 0; + + s = alloc_percpu(*s); + if (!s) + return -ENOMEM; + + idr_preload(GFP_KERNEL); + id = idr_alloc(&irqt_stats, s, irq, irq + 1, GFP_NOWAIT); + idr_preload_end(); + + if (id < 0) { + free_percpu(s); + return id; + } + + return 0; +} From d829b8fb2431595422289cfc210f0a955a8bec74 Mon Sep 17 00:00:00 2001 From: Jeffy Chen Date: Mon, 26 Jun 2017 19:33:33 +0800 Subject: [PATCH 105/112] genirq: Set irq masked state when initializing irq_desc The irq default state is set to disabled when allocating irq desc, but the masked state flag is not set. This is inconsistent vs. the state tracking logic which is used to prevent unnecessary calls to hardware level irq chip functions. Set the masked state flag as well. Signed-off-by: Jeffy Chen Signed-off-by: Thomas Gleixner Cc: tfiga@chromium.org Cc: briannorris@chromium.org Cc: dianders@chromium.org Link: http://lkml.kernel.org/r/1498476814-12563-1-git-send-email-jeffy.chen@rock-chips.com --- kernel/irq/irqdesc.c | 1 + 1 file changed, 1 insertion(+) diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 35a95fadcfda..948b50e78549 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -116,6 +116,7 @@ static void desc_set_defaults(unsigned int irq, struct irq_desc *desc, int node, desc->irq_data.chip_data = NULL; irq_settings_clr_and_set(desc, ~0, _IRQ_DEFAULT_INIT_FLAGS); irqd_set(&desc->irq_data, IRQD_IRQ_DISABLED); + irqd_set(&desc->irq_data, IRQD_IRQ_MASKED); desc->handle_irq = handle_bad_irq; desc->depth = 1; desc->irq_count = 0; From bf22ff45bed664aefb5c4e43029057a199b7070c Mon Sep 17 00:00:00 2001 From: Jeffy Chen Date: Mon, 26 Jun 2017 19:33:34 +0800 Subject: [PATCH 106/112] genirq: Avoid unnecessary low level irq function calls Check irq state in enable/disable/unmask/mask_irq to avoid unnecessary low level irq function calls. This has two advantages: - Conditionals are faster than hardware access - Solves issues with the underlying refcounting of the pinctrl infrastructure Suggested-by: Thomas Gleixner Signed-off-by: Jeffy Chen Signed-off-by: Thomas Gleixner Cc: tfiga@chromium.org Cc: briannorris@chromium.org Cc: dianders@chromium.org Link: http://lkml.kernel.org/r/1498476814-12563-2-git-send-email-jeffy.chen@rock-chips.com --- kernel/irq/chip.c | 53 +++++++++++++++++++++++++++++------------------ 1 file changed, 33 insertions(+), 20 deletions(-) diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index fc89eeb8a6b4..2e30d925a40d 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -314,22 +314,32 @@ void irq_shutdown(struct irq_desc *desc) void irq_enable(struct irq_desc *desc) { - irq_state_clr_disabled(desc); - if (desc->irq_data.chip->irq_enable) - desc->irq_data.chip->irq_enable(&desc->irq_data); - else - desc->irq_data.chip->irq_unmask(&desc->irq_data); - irq_state_clr_masked(desc); + if (!irqd_irq_disabled(&desc->irq_data)) { + unmask_irq(desc); + } else { + irq_state_clr_disabled(desc); + if (desc->irq_data.chip->irq_enable) { + desc->irq_data.chip->irq_enable(&desc->irq_data); + irq_state_clr_masked(desc); + } else { + unmask_irq(desc); + } + } } static void __irq_disable(struct irq_desc *desc, bool mask) { - irq_state_set_disabled(desc); - if (desc->irq_data.chip->irq_disable) { - desc->irq_data.chip->irq_disable(&desc->irq_data); - irq_state_set_masked(desc); - } else if (mask) { - mask_irq(desc); + if (irqd_irq_disabled(&desc->irq_data)) { + if (mask) + mask_irq(desc); + } else { + irq_state_set_disabled(desc); + if (desc->irq_data.chip->irq_disable) { + desc->irq_data.chip->irq_disable(&desc->irq_data); + irq_state_set_masked(desc); + } else if (mask) { + mask_irq(desc); + } } } @@ -378,18 +388,21 @@ void irq_percpu_disable(struct irq_desc *desc, unsigned int cpu) static inline void mask_ack_irq(struct irq_desc *desc) { - if (desc->irq_data.chip->irq_mask_ack) + if (desc->irq_data.chip->irq_mask_ack) { desc->irq_data.chip->irq_mask_ack(&desc->irq_data); - else { - desc->irq_data.chip->irq_mask(&desc->irq_data); + irq_state_set_masked(desc); + } else { + mask_irq(desc); if (desc->irq_data.chip->irq_ack) desc->irq_data.chip->irq_ack(&desc->irq_data); } - irq_state_set_masked(desc); } void mask_irq(struct irq_desc *desc) { + if (irqd_irq_masked(&desc->irq_data)) + return; + if (desc->irq_data.chip->irq_mask) { desc->irq_data.chip->irq_mask(&desc->irq_data); irq_state_set_masked(desc); @@ -398,6 +411,9 @@ void mask_irq(struct irq_desc *desc) void unmask_irq(struct irq_desc *desc) { + if (!irqd_irq_masked(&desc->irq_data)) + return; + if (desc->irq_data.chip->irq_unmask) { desc->irq_data.chip->irq_unmask(&desc->irq_data); irq_state_clr_masked(desc); @@ -411,10 +427,7 @@ void unmask_threaded_irq(struct irq_desc *desc) if (chip->flags & IRQCHIP_EOI_THREADED) chip->irq_eoi(&desc->irq_data); - if (chip->irq_unmask) { - chip->irq_unmask(&desc->irq_data); - irq_state_clr_masked(desc); - } + unmask_irq(desc); } /* From 5f042e7cbd9ebd3580077dcdc21f35e68c2adf5f Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 26 Jun 2017 12:20:56 +0200 Subject: [PATCH 107/112] blk-mq: Include all present CPUs in the default queue mapping This way we get a nice distribution independent of the current cpu online / offline state. Signed-off-by: Christoph Hellwig Reviewed-by: Jens Axboe Cc: Keith Busch Cc: linux-block@vger.kernel.org Cc: linux-nvme@lists.infradead.org Link: http://lkml.kernel.org/r/20170626102058.10200-2-hch@lst.de Signed-off-by: Thomas Gleixner --- block/blk-mq-cpumap.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/block/blk-mq-cpumap.c b/block/blk-mq-cpumap.c index 8e61e8640e17..5eaecd40f701 100644 --- a/block/blk-mq-cpumap.c +++ b/block/blk-mq-cpumap.c @@ -35,7 +35,6 @@ int blk_mq_map_queues(struct blk_mq_tag_set *set) { unsigned int *map = set->mq_map; unsigned int nr_queues = set->nr_hw_queues; - const struct cpumask *online_mask = cpu_online_mask; unsigned int i, nr_cpus, nr_uniq_cpus, queue, first_sibling; cpumask_var_t cpus; @@ -44,7 +43,7 @@ int blk_mq_map_queues(struct blk_mq_tag_set *set) cpumask_clear(cpus); nr_cpus = nr_uniq_cpus = 0; - for_each_cpu(i, online_mask) { + for_each_present_cpu(i) { nr_cpus++; first_sibling = get_first_sibling(i); if (!cpumask_test_cpu(first_sibling, cpus)) @@ -54,7 +53,7 @@ int blk_mq_map_queues(struct blk_mq_tag_set *set) queue = 0; for_each_possible_cpu(i) { - if (!cpumask_test_cpu(i, online_mask)) { + if (!cpumask_test_cpu(i, cpu_present_mask)) { map[i] = 0; continue; } From 4b855ad37194f7bdbb200ce7a1c7051fecb56a08 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 26 Jun 2017 12:20:57 +0200 Subject: [PATCH 108/112] blk-mq: Create hctx for each present CPU Currently we only create hctx for online CPUs, which can lead to a lot of churn due to frequent soft offline / online operations. Instead allocate one for each present CPU to avoid this and dramatically simplify the code. Signed-off-by: Christoph Hellwig Reviewed-by: Jens Axboe Cc: Keith Busch Cc: linux-block@vger.kernel.org Cc: linux-nvme@lists.infradead.org Link: http://lkml.kernel.org/r/20170626102058.10200-3-hch@lst.de Signed-off-by: Thomas Gleixner --- block/blk-mq.c | 120 ++++--------------------------------- block/blk-mq.h | 5 -- include/linux/cpuhotplug.h | 1 - 3 files changed, 11 insertions(+), 115 deletions(-) diff --git a/block/blk-mq.c b/block/blk-mq.c index bb66c96850b1..dd390e27824d 100644 --- a/block/blk-mq.c +++ b/block/blk-mq.c @@ -37,9 +37,6 @@ #include "blk-wbt.h" #include "blk-mq-sched.h" -static DEFINE_MUTEX(all_q_mutex); -static LIST_HEAD(all_q_list); - static void blk_mq_poll_stats_start(struct request_queue *q); static void blk_mq_poll_stats_fn(struct blk_stat_callback *cb); static void __blk_mq_stop_hw_queues(struct request_queue *q, bool sync); @@ -1975,8 +1972,8 @@ static void blk_mq_init_cpu_queues(struct request_queue *q, INIT_LIST_HEAD(&__ctx->rq_list); __ctx->queue = q; - /* If the cpu isn't online, the cpu is mapped to first hctx */ - if (!cpu_online(i)) + /* If the cpu isn't present, the cpu is mapped to first hctx */ + if (!cpu_present(i)) continue; hctx = blk_mq_map_queue(q, i); @@ -2019,8 +2016,7 @@ static void blk_mq_free_map_and_requests(struct blk_mq_tag_set *set, } } -static void blk_mq_map_swqueue(struct request_queue *q, - const struct cpumask *online_mask) +static void blk_mq_map_swqueue(struct request_queue *q) { unsigned int i, hctx_idx; struct blk_mq_hw_ctx *hctx; @@ -2038,13 +2034,11 @@ static void blk_mq_map_swqueue(struct request_queue *q, } /* - * Map software to hardware queues + * Map software to hardware queues. + * + * If the cpu isn't present, the cpu is mapped to first hctx. */ - for_each_possible_cpu(i) { - /* If the cpu isn't online, the cpu is mapped to first hctx */ - if (!cpumask_test_cpu(i, online_mask)) - continue; - + for_each_present_cpu(i) { hctx_idx = q->mq_map[i]; /* unmapped hw queue can be remapped after CPU topo changed */ if (!set->tags[hctx_idx] && @@ -2330,16 +2324,8 @@ struct request_queue *blk_mq_init_allocated_queue(struct blk_mq_tag_set *set, blk_queue_softirq_done(q, set->ops->complete); blk_mq_init_cpu_queues(q, set->nr_hw_queues); - - get_online_cpus(); - mutex_lock(&all_q_mutex); - - list_add_tail(&q->all_q_node, &all_q_list); blk_mq_add_queue_tag_set(set, q); - blk_mq_map_swqueue(q, cpu_online_mask); - - mutex_unlock(&all_q_mutex); - put_online_cpus(); + blk_mq_map_swqueue(q); if (!(set->flags & BLK_MQ_F_NO_SCHED)) { int ret; @@ -2365,18 +2351,12 @@ void blk_mq_free_queue(struct request_queue *q) { struct blk_mq_tag_set *set = q->tag_set; - mutex_lock(&all_q_mutex); - list_del_init(&q->all_q_node); - mutex_unlock(&all_q_mutex); - blk_mq_del_queue_tag_set(q); - blk_mq_exit_hw_queues(q, set, set->nr_hw_queues); } /* Basically redo blk_mq_init_queue with queue frozen */ -static void blk_mq_queue_reinit(struct request_queue *q, - const struct cpumask *online_mask) +static void blk_mq_queue_reinit(struct request_queue *q) { WARN_ON_ONCE(!atomic_read(&q->mq_freeze_depth)); @@ -2389,76 +2369,12 @@ static void blk_mq_queue_reinit(struct request_queue *q, * involves free and re-allocate memory, worthy doing?) */ - blk_mq_map_swqueue(q, online_mask); + blk_mq_map_swqueue(q); blk_mq_sysfs_register(q); blk_mq_debugfs_register_hctxs(q); } -/* - * New online cpumask which is going to be set in this hotplug event. - * Declare this cpumasks as global as cpu-hotplug operation is invoked - * one-by-one and dynamically allocating this could result in a failure. - */ -static struct cpumask cpuhp_online_new; - -static void blk_mq_queue_reinit_work(void) -{ - struct request_queue *q; - - mutex_lock(&all_q_mutex); - /* - * We need to freeze and reinit all existing queues. Freezing - * involves synchronous wait for an RCU grace period and doing it - * one by one may take a long time. Start freezing all queues in - * one swoop and then wait for the completions so that freezing can - * take place in parallel. - */ - list_for_each_entry(q, &all_q_list, all_q_node) - blk_freeze_queue_start(q); - list_for_each_entry(q, &all_q_list, all_q_node) - blk_mq_freeze_queue_wait(q); - - list_for_each_entry(q, &all_q_list, all_q_node) - blk_mq_queue_reinit(q, &cpuhp_online_new); - - list_for_each_entry(q, &all_q_list, all_q_node) - blk_mq_unfreeze_queue(q); - - mutex_unlock(&all_q_mutex); -} - -static int blk_mq_queue_reinit_dead(unsigned int cpu) -{ - cpumask_copy(&cpuhp_online_new, cpu_online_mask); - blk_mq_queue_reinit_work(); - return 0; -} - -/* - * Before hotadded cpu starts handling requests, new mappings must be - * established. Otherwise, these requests in hw queue might never be - * dispatched. - * - * For example, there is a single hw queue (hctx) and two CPU queues (ctx0 - * for CPU0, and ctx1 for CPU1). - * - * Now CPU1 is just onlined and a request is inserted into ctx1->rq_list - * and set bit0 in pending bitmap as ctx1->index_hw is still zero. - * - * And then while running hw queue, blk_mq_flush_busy_ctxs() finds bit0 is set - * in pending bitmap and tries to retrieve requests in hctx->ctxs[0]->rq_list. - * But htx->ctxs[0] is a pointer to ctx0, so the request in ctx1->rq_list is - * ignored. - */ -static int blk_mq_queue_reinit_prepare(unsigned int cpu) -{ - cpumask_copy(&cpuhp_online_new, cpu_online_mask); - cpumask_set_cpu(cpu, &cpuhp_online_new); - blk_mq_queue_reinit_work(); - return 0; -} - static int __blk_mq_alloc_rq_maps(struct blk_mq_tag_set *set) { int i; @@ -2669,7 +2585,7 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, blk_mq_update_queue_map(set); list_for_each_entry(q, &set->tag_list, tag_set_list) { blk_mq_realloc_hw_ctxs(set, q); - blk_mq_queue_reinit(q, cpu_online_mask); + blk_mq_queue_reinit(q); } list_for_each_entry(q, &set->tag_list, tag_set_list) @@ -2885,24 +2801,10 @@ bool blk_mq_poll(struct request_queue *q, blk_qc_t cookie) } EXPORT_SYMBOL_GPL(blk_mq_poll); -void blk_mq_disable_hotplug(void) -{ - mutex_lock(&all_q_mutex); -} - -void blk_mq_enable_hotplug(void) -{ - mutex_unlock(&all_q_mutex); -} - static int __init blk_mq_init(void) { cpuhp_setup_state_multi(CPUHP_BLK_MQ_DEAD, "block/mq:dead", NULL, blk_mq_hctx_notify_dead); - - cpuhp_setup_state_nocalls(CPUHP_BLK_MQ_PREPARE, "block/mq:prepare", - blk_mq_queue_reinit_prepare, - blk_mq_queue_reinit_dead); return 0; } subsys_initcall(blk_mq_init); diff --git a/block/blk-mq.h b/block/blk-mq.h index cc67b48e3551..558df56544d2 100644 --- a/block/blk-mq.h +++ b/block/blk-mq.h @@ -56,11 +56,6 @@ void __blk_mq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq, bool at_head); void blk_mq_insert_requests(struct blk_mq_hw_ctx *hctx, struct blk_mq_ctx *ctx, struct list_head *list); -/* - * CPU hotplug helpers - */ -void blk_mq_enable_hotplug(void); -void blk_mq_disable_hotplug(void); /* * CPU -> queue mappings diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h index c15f22c54535..7f815d915977 100644 --- a/include/linux/cpuhotplug.h +++ b/include/linux/cpuhotplug.h @@ -58,7 +58,6 @@ enum cpuhp_state { CPUHP_XEN_EVTCHN_PREPARE, CPUHP_ARM_SHMOBILE_SCU_PREPARE, CPUHP_SH_SH3X_PREPARE, - CPUHP_BLK_MQ_PREPARE, CPUHP_NET_FLOW_PREPARE, CPUHP_TOPOLOGY_PREPARE, CPUHP_NET_IUCV_PREPARE, From 425a17cbfff933c4cca4eeef5caa5926d198dd85 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 26 Jun 2017 12:20:58 +0200 Subject: [PATCH 109/112] nvme: Allocate queues for all possible CPUs MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Unlike most drŅ–vers that simply pass the maximum possible vectors to pci_alloc_irq_vectors NVMe needs to configure the device before allocting the vectors, so it needs a manual update for the new scheme of using all present CPUs. Signed-off-by: Christoph Hellwig Reviewed-by: Jens Axboe Cc: Keith Busch Cc: linux-block@vger.kernel.org Cc: linux-nvme@lists.infradead.org Link: http://lkml.kernel.org/r/20170626102058.10200-4-hch@lst.de Signed-off-by: Thomas Gleixner --- drivers/nvme/host/pci.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c index 951042a375d6..b3dcd7abc6d7 100644 --- a/drivers/nvme/host/pci.c +++ b/drivers/nvme/host/pci.c @@ -1525,7 +1525,7 @@ static int nvme_setup_io_queues(struct nvme_dev *dev) struct pci_dev *pdev = to_pci_dev(dev->dev); int result, nr_io_queues, size; - nr_io_queues = num_online_cpus(); + nr_io_queues = num_present_cpus(); result = nvme_set_queue_count(&dev->ctrl, &nr_io_queues); if (result < 0) return result; From 866d7c1b0a3c70387646c4e455e727a58c5d465a Mon Sep 17 00:00:00 2001 From: Suzuki K Poulose Date: Fri, 30 Jun 2017 10:58:28 +0100 Subject: [PATCH 110/112] irqchip/gic-v3: Fix out-of-bound access in gic_set_affinity The GICv3 driver doesn't check if the target CPU for gic_set_affinity is valid before going ahead and making the changes. This triggers the following splat with KASAN: [ 141.189434] BUG: KASAN: global-out-of-bounds in gic_set_affinity+0x8c/0x140 [ 141.189704] Read of size 8 at addr ffff200009741d20 by task swapper/1/0 [ 141.189958] [ 141.190158] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 4.12.0-rc7 [ 141.190458] Hardware name: Foundation-v8A (DT) [ 141.190658] Call trace: [ 141.190908] [] dump_backtrace+0x0/0x328 [ 141.191224] [] show_stack+0x14/0x20 [ 141.191507] [] dump_stack+0xa4/0xc8 [ 141.191858] [] print_address_description+0x13c/0x250 [ 141.192219] [] kasan_report+0x210/0x300 [ 141.192547] [] __asan_load8+0x84/0x98 [ 141.192874] [] gic_set_affinity+0x8c/0x140 [ 141.193158] [] irq_do_set_affinity+0x54/0xb8 [ 141.193473] [] irq_set_affinity_locked+0x64/0xf0 [ 141.193828] [] __irq_set_affinity+0x48/0x78 [ 141.194158] [] arm_perf_starting_cpu+0x104/0x150 [ 141.194513] [] cpuhp_invoke_callback+0x17c/0x1f8 [ 141.194783] [] notify_cpu_starting+0x8c/0xb8 [ 141.195130] [] secondary_start_kernel+0x15c/0x200 [ 141.195390] [<0000000080db81b4>] 0x80db81b4 [ 141.195603] [ 141.195685] The buggy address belongs to the variable: [ 141.196012] __cpu_logical_map+0x200/0x220 [ 141.196176] [ 141.196315] Memory state around the buggy address: [ 141.196586] ffff200009741c00: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 [ 141.196913] ffff200009741c80: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 [ 141.197158] >ffff200009741d00: 00 00 00 00 fa fa fa fa 00 00 00 00 00 00 00 00 [ 141.197487] ^ [ 141.197758] ffff200009741d80: 00 00 00 00 00 00 00 00 fa fa fa fa 00 00 00 00 [ 141.198060] ffff200009741e00: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 [ 141.198358] ================================================================== [ 141.198609] Disabling lock debugging due to kernel taint [ 141.198961] CPU1: Booted secondary processor [410fd051] This patch adds the check to make sure the cpu is valid. Fixes: commit 021f653791ad17e03f98 ("irqchip: gic-v3: Initial support for GICv3") Cc: stable@vger.kernel.org Signed-off-by: Suzuki K Poulose Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-gic-v3.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c index c132f29322cc..dbffb7ab6203 100644 --- a/drivers/irqchip/irq-gic-v3.c +++ b/drivers/irqchip/irq-gic-v3.c @@ -645,6 +645,9 @@ static int gic_set_affinity(struct irq_data *d, const struct cpumask *mask_val, int enabled; u64 val; + if (cpu >= nr_cpu_ids) + return -EINVAL; + if (gic_irq_in_rdist(d)) return -EINVAL; From 478a2db82863db9f1c135111e78298e6ed5022f3 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Fri, 30 Jun 2017 11:00:49 +0300 Subject: [PATCH 111/112] irqchip/irq-mvebu-gicp: Allocate enough memory for spi_bitmap BITS_TO_LONGS() gives us the number of longs we need, but we want to allocate the number of bytes. Fixes: a68a63cb4dfc ("irqchip/irq-mvebu-gicp: Add new driver for Marvell GICP") Acked-by: Thomas Petazzoni Signed-off-by: Dan Carpenter Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-mvebu-gicp.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/irqchip/irq-mvebu-gicp.c b/drivers/irqchip/irq-mvebu-gicp.c index 45358ac9bb1d..b283fc90be1e 100644 --- a/drivers/irqchip/irq-mvebu-gicp.c +++ b/drivers/irqchip/irq-mvebu-gicp.c @@ -226,8 +226,8 @@ static int mvebu_gicp_probe(struct platform_device *pdev) } gicp->spi_bitmap = devm_kzalloc(&pdev->dev, - BITS_TO_LONGS(gicp->spi_cnt), - GFP_KERNEL); + BITS_TO_LONGS(gicp->spi_cnt) * sizeof(long), + GFP_KERNEL); if (!gicp->spi_bitmap) return -ENOMEM; From ca38701935138a4aba7dd9e4daf21e13a4663f43 Mon Sep 17 00:00:00 2001 From: "Pedro H. Penna" Date: Fri, 30 Jun 2017 08:33:56 -0300 Subject: [PATCH 112/112] irqchip/or1k-pic: Fix interrupt acknowledgement Usually, hardware implicitly acknowledges interrupts when reading them. However, if this is not the case, the IRQ gets fired over and over again in the current implementation. This patch uses the right mask acknowledge function to handle the aforementioned situation on or1k processors that interact with such kind of hardware. Acked-by: Stafford Horne Signed-off-by: Pedro H. Penna Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-or1k-pic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/irqchip/irq-or1k-pic.c b/drivers/irqchip/irq-or1k-pic.c index 6a9a3e79218b..dd9d5d12fea2 100644 --- a/drivers/irqchip/irq-or1k-pic.c +++ b/drivers/irqchip/irq-or1k-pic.c @@ -70,7 +70,7 @@ static struct or1k_pic_dev or1k_pic_level = { .name = "or1k-PIC-level", .irq_unmask = or1k_pic_unmask, .irq_mask = or1k_pic_mask, - .irq_mask_ack = or1k_pic_mask, + .irq_mask_ack = or1k_pic_mask_ack, }, .handle = handle_level_irq, .flags = IRQ_LEVEL | IRQ_NOPROBE,