From bc06a9e0874239cb6d4eebcb0ecd1a91ad9272db Mon Sep 17 00:00:00 2001 From: Shanker Donthineni Date: Fri, 19 May 2023 08:49:00 -0500 Subject: genirq: Use hlist for managing resend handlers The current implementation utilizes a bitmap for managing interrupt resend handlers, which is allocated based on the SPARSE_IRQ/NR_IRQS macros. However, this method may not efficiently utilize memory during runtime, particularly when IRQ_BITMAP_BITS is large. Address this issue by using an hlist to manage interrupt resend handlers instead of relying on a static bitmap memory allocation. Additionally, a new function, clear_irq_resend(), is introduced and called from irq_shutdown to ensure a graceful teardown of the interrupt. Signed-off-by: Shanker Donthineni Signed-off-by: Thomas Gleixner Link: https://lore.kernel.org/r/20230519134902.1495562-2-sdonthineni@nvidia.com --- include/linux/irqdesc.h | 3 +++ 1 file changed, 3 insertions(+) (limited to 'include/linux') diff --git a/include/linux/irqdesc.h b/include/linux/irqdesc.h index 844a8e30e6de..d9451d456a73 100644 --- a/include/linux/irqdesc.h +++ b/include/linux/irqdesc.h @@ -102,6 +102,9 @@ struct irq_desc { int parent_irq; struct module *owner; const char *name; +#ifdef CONFIG_HARDIRQS_SW_RESEND + struct hlist_node resend_node; +#endif } ____cacheline_internodealigned_in_smp; #ifdef CONFIG_SPARSE_IRQ -- cgit v1.2.3 From 0cfb4a1af386427cdaba98f18f501eb074985cfd Mon Sep 17 00:00:00 2001 From: Marc Zyngier Date: Thu, 15 Jun 2023 14:58:52 +0100 Subject: genirq: Use BIT() for the IRQD_* state flags As we're about to use the last bit available in the IRQD_* state flags, rewrite these flags with BIT(), which ensures that these constant do not represent a signed value. Signed-off-by: Marc Zyngier --- include/linux/irq.h | 46 +++++++++++++++++++++++----------------------- 1 file changed, 23 insertions(+), 23 deletions(-) (limited to 'include/linux') diff --git a/include/linux/irq.h b/include/linux/irq.h index b1b28affb32a..d9c86db69982 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -226,29 +226,29 @@ struct irq_data { */ enum { IRQD_TRIGGER_MASK = 0xf, - IRQD_SETAFFINITY_PENDING = (1 << 8), - IRQD_ACTIVATED = (1 << 9), - IRQD_NO_BALANCING = (1 << 10), - IRQD_PER_CPU = (1 << 11), - IRQD_AFFINITY_SET = (1 << 12), - IRQD_LEVEL = (1 << 13), - IRQD_WAKEUP_STATE = (1 << 14), - IRQD_MOVE_PCNTXT = (1 << 15), - IRQD_IRQ_DISABLED = (1 << 16), - IRQD_IRQ_MASKED = (1 << 17), - IRQD_IRQ_INPROGRESS = (1 << 18), - IRQD_WAKEUP_ARMED = (1 << 19), - IRQD_FORWARDED_TO_VCPU = (1 << 20), - IRQD_AFFINITY_MANAGED = (1 << 21), - IRQD_IRQ_STARTED = (1 << 22), - IRQD_MANAGED_SHUTDOWN = (1 << 23), - IRQD_SINGLE_TARGET = (1 << 24), - IRQD_DEFAULT_TRIGGER_SET = (1 << 25), - IRQD_CAN_RESERVE = (1 << 26), - IRQD_MSI_NOMASK_QUIRK = (1 << 27), - IRQD_HANDLE_ENFORCE_IRQCTX = (1 << 28), - IRQD_AFFINITY_ON_ACTIVATE = (1 << 29), - IRQD_IRQ_ENABLED_ON_SUSPEND = (1 << 30), + IRQD_SETAFFINITY_PENDING = BIT(8), + IRQD_ACTIVATED = BIT(9), + IRQD_NO_BALANCING = BIT(10), + IRQD_PER_CPU = BIT(11), + IRQD_AFFINITY_SET = BIT(12), + IRQD_LEVEL = BIT(13), + IRQD_WAKEUP_STATE = BIT(14), + IRQD_MOVE_PCNTXT = BIT(15), + IRQD_IRQ_DISABLED = BIT(16), + IRQD_IRQ_MASKED = BIT(17), + IRQD_IRQ_INPROGRESS = BIT(18), + IRQD_WAKEUP_ARMED = BIT(19), + IRQD_FORWARDED_TO_VCPU = BIT(20), + IRQD_AFFINITY_MANAGED = BIT(21), + IRQD_IRQ_STARTED = BIT(22), + IRQD_MANAGED_SHUTDOWN = BIT(23), + IRQD_SINGLE_TARGET = BIT(24), + IRQD_DEFAULT_TRIGGER_SET = BIT(25), + IRQD_CAN_RESERVE = BIT(26), + IRQD_MSI_NOMASK_QUIRK = BIT(27), + IRQD_HANDLE_ENFORCE_IRQCTX = BIT(28), + IRQD_AFFINITY_ON_ACTIVATE = BIT(29), + IRQD_IRQ_ENABLED_ON_SUSPEND = BIT(30), }; #define __irqd_to_state(d) ACCESS_PRIVATE((d)->common, state_use_accessors) -- cgit v1.2.3 From 9c15eeb5362c48dd27d51bd72e8873341fa9383c Mon Sep 17 00:00:00 2001 From: James Gowans Date: Thu, 8 Jun 2023 14:00:20 +0200 Subject: genirq: Allow fasteoi handler to resend interrupts on concurrent handling There is a class of interrupt controllers out there that, once they have signalled a given interrupt number, will still signal incoming instances of the *same* interrupt despite the original interrupt not having been EOIed yet. As long as the new interrupt reaches the *same* CPU, nothing bad happens, as that CPU still has its interrupts globally disabled, and we will only take the new interrupt once the interrupt has been EOIed. However, things become more "interesting" if an affinity change comes in while the interrupt is being handled. More specifically, while the per-irq lock is being dropped. This results in the affinity change taking place immediately. At this point, there is nothing that prevents the interrupt from firing on the new target CPU. We end-up with the interrupt running concurrently on two CPUs, which isn't a good thing. And that's where things become worse: the new CPU notices that the interrupt handling is in progress (irq_may_run() return false), and *drops the interrupt on the floor*. The whole race looks like this: CPU 0 | CPU 1 -----------------------------|----------------------------- interrupt start | handle_fasteoi_irq | set_affinity(CPU 1) handler | ... | interrupt start ... | handle_fasteoi_irq -> early out handle_fasteoi_irq return | interrupt end interrupt end | If the interrupt was an edge, too bad. The interrupt is lost, and the system will eventually die one way or another. Not great. A way to avoid this situation is to detect this problem at the point we handle the interrupt on the new target. Instead of dropping the interrupt, use the resend mechanism to force it to be replayed. Also, in order to limit the impact of this workaround to the pathetic architectures that require it, gate it behind a new irq flag aptly named IRQD_RESEND_WHEN_IN_PROGRESS. Suggested-by: Marc Zyngier Signed-off-by: James Gowans Cc: Thomas Gleixner Cc: Marc Zyngier Cc: KarimAllah Raslan Cc: Yipeng Zou Cc: Zhang Jianhua [maz: reworded commit mesage] Signed-off-by: Marc Zyngier Link: https://lore.kernel.org/r/20230608120021.3273400-3-jgowans@amazon.com --- include/linux/irq.h | 13 +++++++++++++ kernel/irq/chip.c | 16 +++++++++++++++- kernel/irq/debugfs.c | 2 ++ 3 files changed, 30 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/irq.h b/include/linux/irq.h index d9c86db69982..d8a6fdce9373 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -223,6 +223,8 @@ struct irq_data { * irq_chip::irq_set_affinity() when deactivated. * IRQD_IRQ_ENABLED_ON_SUSPEND - Interrupt is enabled on suspend by irq pm if * irqchip have flag IRQCHIP_ENABLE_WAKEUP_ON_SUSPEND set. + * IRQD_RESEND_WHEN_IN_PROGRESS - Interrupt may fire when already in progress in which + * case it must be resent at the next available opportunity. */ enum { IRQD_TRIGGER_MASK = 0xf, @@ -249,6 +251,7 @@ enum { IRQD_HANDLE_ENFORCE_IRQCTX = BIT(28), IRQD_AFFINITY_ON_ACTIVATE = BIT(29), IRQD_IRQ_ENABLED_ON_SUSPEND = BIT(30), + IRQD_RESEND_WHEN_IN_PROGRESS = BIT(31), }; #define __irqd_to_state(d) ACCESS_PRIVATE((d)->common, state_use_accessors) @@ -448,6 +451,16 @@ static inline bool irqd_affinity_on_activate(struct irq_data *d) return __irqd_to_state(d) & IRQD_AFFINITY_ON_ACTIVATE; } +static inline void irqd_set_resend_when_in_progress(struct irq_data *d) +{ + __irqd_to_state(d) |= IRQD_RESEND_WHEN_IN_PROGRESS; +} + +static inline bool irqd_needs_resend_when_in_progress(struct irq_data *d) +{ + return __irqd_to_state(d) & IRQD_RESEND_WHEN_IN_PROGRESS; +} + #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 49e7bc871fec..57cd8f475302 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -692,8 +692,16 @@ void handle_fasteoi_irq(struct irq_desc *desc) raw_spin_lock(&desc->lock); - if (!irq_may_run(desc)) + /* + * When an affinity change races with IRQ handling, the next interrupt + * can arrive on the new CPU before the original CPU has completed + * handling the previous one - it may need to be resent. + */ + if (!irq_may_run(desc)) { + if (irqd_needs_resend_when_in_progress(&desc->irq_data)) + desc->istate |= IRQS_PENDING; goto out; + } desc->istate &= ~(IRQS_REPLAY | IRQS_WAITING); @@ -715,6 +723,12 @@ void handle_fasteoi_irq(struct irq_desc *desc) cond_unmask_eoi_irq(desc, chip); + /* + * When the race described above happens this will resend the interrupt. + */ + if (unlikely(desc->istate & IRQS_PENDING)) + check_irq_resend(desc, false); + raw_spin_unlock(&desc->lock); return; out: diff --git a/kernel/irq/debugfs.c b/kernel/irq/debugfs.c index bbcaac64038e..5971a66be034 100644 --- a/kernel/irq/debugfs.c +++ b/kernel/irq/debugfs.c @@ -133,6 +133,8 @@ static const struct irq_bit_descr irqdata_states[] = { BIT_MASK_DESCR(IRQD_HANDLE_ENFORCE_IRQCTX), BIT_MASK_DESCR(IRQD_IRQ_ENABLED_ON_SUSPEND), + + BIT_MASK_DESCR(IRQD_RESEND_WHEN_IN_PROGRESS), }; static const struct irq_bit_descr irqdesc_states[] = { -- cgit v1.2.3 From f1771b85e3086c9506c3de81e993330bca568ba5 Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Tue, 16 May 2023 22:05:05 +0200 Subject: irqchip/mmp: Remove non-DT codepath Building with "W=1" warns about missing declarations for two functions in the mmp irqchip driver: drivers/irqchip/irq-mmp.c:248:13: error: no previous prototype for 'icu_init_irq' drivers/irqchip/irq-mmp.c:271:13: error: no previous prototype for 'mmp2_init_icu' The declarations are present in an unused header, but since there is no caller, it's best to just remove the functions and the header completely, making the driver DT-only to match the state of the platform. Fixes: 77acc85ce797 ("ARM: mmp: remove device definitions") Signed-off-by: Arnd Bergmann Signed-off-by: Marc Zyngier Link: https://lore.kernel.org/r/20230516200516.554663-2-arnd@kernel.org --- drivers/irqchip/irq-mmp.c | 127 -------------------------------------------- include/linux/irqchip/mmp.h | 10 ---- 2 files changed, 137 deletions(-) delete mode 100644 include/linux/irqchip/mmp.h (limited to 'include/linux') diff --git a/drivers/irqchip/irq-mmp.c b/drivers/irqchip/irq-mmp.c index 83455ca72439..25cf4f80e767 100644 --- a/drivers/irqchip/irq-mmp.c +++ b/drivers/irqchip/irq-mmp.c @@ -244,132 +244,6 @@ static void __exception_irq_entry mmp2_handle_irq(struct pt_regs *regs) generic_handle_domain_irq(icu_data[0].domain, hwirq); } -/* MMP (ARMv5) */ -void __init icu_init_irq(void) -{ - int irq; - - max_icu_nr = 1; - mmp_icu_base = ioremap(0xd4282000, 0x1000); - icu_data[0].conf_enable = mmp_conf.conf_enable; - icu_data[0].conf_disable = mmp_conf.conf_disable; - icu_data[0].conf_mask = mmp_conf.conf_mask; - icu_data[0].nr_irqs = 64; - icu_data[0].virq_base = 0; - icu_data[0].domain = irq_domain_add_legacy(NULL, 64, 0, 0, - &irq_domain_simple_ops, - &icu_data[0]); - for (irq = 0; irq < 64; irq++) { - icu_mask_irq(irq_get_irq_data(irq)); - irq_set_chip_and_handler(irq, &icu_irq_chip, handle_level_irq); - } - irq_set_default_host(icu_data[0].domain); - set_handle_irq(mmp_handle_irq); -} - -/* MMP2 (ARMv7) */ -void __init mmp2_init_icu(void) -{ - int irq, end; - - max_icu_nr = 8; - mmp_icu_base = ioremap(0xd4282000, 0x1000); - icu_data[0].conf_enable = mmp2_conf.conf_enable; - icu_data[0].conf_disable = mmp2_conf.conf_disable; - icu_data[0].conf_mask = mmp2_conf.conf_mask; - icu_data[0].nr_irqs = 64; - icu_data[0].virq_base = 0; - icu_data[0].domain = irq_domain_add_legacy(NULL, 64, 0, 0, - &irq_domain_simple_ops, - &icu_data[0]); - icu_data[1].reg_status = mmp_icu_base + 0x150; - icu_data[1].reg_mask = mmp_icu_base + 0x168; - icu_data[1].clr_mfp_irq_base = icu_data[0].virq_base + - icu_data[0].nr_irqs; - icu_data[1].clr_mfp_hwirq = 1; /* offset to IRQ_MMP2_PMIC_BASE */ - icu_data[1].nr_irqs = 2; - icu_data[1].cascade_irq = 4; - icu_data[1].virq_base = icu_data[0].virq_base + icu_data[0].nr_irqs; - icu_data[1].domain = irq_domain_add_legacy(NULL, icu_data[1].nr_irqs, - icu_data[1].virq_base, 0, - &irq_domain_simple_ops, - &icu_data[1]); - icu_data[2].reg_status = mmp_icu_base + 0x154; - icu_data[2].reg_mask = mmp_icu_base + 0x16c; - icu_data[2].nr_irqs = 2; - icu_data[2].cascade_irq = 5; - icu_data[2].virq_base = icu_data[1].virq_base + icu_data[1].nr_irqs; - icu_data[2].domain = irq_domain_add_legacy(NULL, icu_data[2].nr_irqs, - icu_data[2].virq_base, 0, - &irq_domain_simple_ops, - &icu_data[2]); - icu_data[3].reg_status = mmp_icu_base + 0x180; - icu_data[3].reg_mask = mmp_icu_base + 0x17c; - icu_data[3].nr_irqs = 3; - icu_data[3].cascade_irq = 9; - icu_data[3].virq_base = icu_data[2].virq_base + icu_data[2].nr_irqs; - icu_data[3].domain = irq_domain_add_legacy(NULL, icu_data[3].nr_irqs, - icu_data[3].virq_base, 0, - &irq_domain_simple_ops, - &icu_data[3]); - icu_data[4].reg_status = mmp_icu_base + 0x158; - icu_data[4].reg_mask = mmp_icu_base + 0x170; - icu_data[4].nr_irqs = 5; - icu_data[4].cascade_irq = 17; - icu_data[4].virq_base = icu_data[3].virq_base + icu_data[3].nr_irqs; - icu_data[4].domain = irq_domain_add_legacy(NULL, icu_data[4].nr_irqs, - icu_data[4].virq_base, 0, - &irq_domain_simple_ops, - &icu_data[4]); - icu_data[5].reg_status = mmp_icu_base + 0x15c; - icu_data[5].reg_mask = mmp_icu_base + 0x174; - icu_data[5].nr_irqs = 15; - icu_data[5].cascade_irq = 35; - icu_data[5].virq_base = icu_data[4].virq_base + icu_data[4].nr_irqs; - icu_data[5].domain = irq_domain_add_legacy(NULL, icu_data[5].nr_irqs, - icu_data[5].virq_base, 0, - &irq_domain_simple_ops, - &icu_data[5]); - icu_data[6].reg_status = mmp_icu_base + 0x160; - icu_data[6].reg_mask = mmp_icu_base + 0x178; - icu_data[6].nr_irqs = 2; - icu_data[6].cascade_irq = 51; - icu_data[6].virq_base = icu_data[5].virq_base + icu_data[5].nr_irqs; - icu_data[6].domain = irq_domain_add_legacy(NULL, icu_data[6].nr_irqs, - icu_data[6].virq_base, 0, - &irq_domain_simple_ops, - &icu_data[6]); - icu_data[7].reg_status = mmp_icu_base + 0x188; - icu_data[7].reg_mask = mmp_icu_base + 0x184; - icu_data[7].nr_irqs = 2; - icu_data[7].cascade_irq = 55; - icu_data[7].virq_base = icu_data[6].virq_base + icu_data[6].nr_irqs; - icu_data[7].domain = irq_domain_add_legacy(NULL, icu_data[7].nr_irqs, - icu_data[7].virq_base, 0, - &irq_domain_simple_ops, - &icu_data[7]); - end = icu_data[7].virq_base + icu_data[7].nr_irqs; - for (irq = 0; irq < end; irq++) { - icu_mask_irq(irq_get_irq_data(irq)); - if (irq == icu_data[1].cascade_irq || - irq == icu_data[2].cascade_irq || - irq == icu_data[3].cascade_irq || - irq == icu_data[4].cascade_irq || - irq == icu_data[5].cascade_irq || - irq == icu_data[6].cascade_irq || - irq == icu_data[7].cascade_irq) { - irq_set_chip(irq, &icu_irq_chip); - irq_set_chained_handler(irq, icu_mux_irq_demux); - } else { - irq_set_chip_and_handler(irq, &icu_irq_chip, - handle_level_irq); - } - } - irq_set_default_host(icu_data[0].domain); - set_handle_irq(mmp2_handle_irq); -} - -#ifdef CONFIG_OF static int __init mmp_init_bases(struct device_node *node) { int ret, nr_irqs, irq, i = 0; @@ -548,4 +422,3 @@ err: return -EINVAL; } IRQCHIP_DECLARE(mmp2_mux_intc, "mrvl,mmp2-mux-intc", mmp2_mux_of_init); -#endif diff --git a/include/linux/irqchip/mmp.h b/include/linux/irqchip/mmp.h deleted file mode 100644 index aa1813749a4f..000000000000 --- a/include/linux/irqchip/mmp.h +++ /dev/null @@ -1,10 +0,0 @@ -/* SPDX-License-Identifier: GPL-2.0 */ -#ifndef __IRQCHIP_MMP_H -#define __IRQCHIP_MMP_H - -extern struct irq_chip icu_irq_chip; - -extern void icu_init_irq(void); -extern void mmp2_init_icu(void); - -#endif /* __IRQCHIP_MMP_H */ -- cgit v1.2.3 From 415e84294798d1cb041c902168393054cc4ad211 Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Tue, 16 May 2023 22:05:08 +0200 Subject: irqchip/gicv3: Add a iort_pmsi_get_dev_id() prototype iort_pmsi_get_dev_id() has a __weak definition in the driver, and an override in arm64 specific code, but the declaration is conditional and not always seen when the copy in the driver gets built: drivers/irqchip/irq-gic-v3-its-platform-msi.c:41:12: error: no previous prototype for 'iort_pmsi_get_dev_id' [-Werror=missing-prototypes] Move the existing declaration out of the #ifdef block to ensure it can be seen in all configurations. Signed-off-by: Arnd Bergmann Reviewed-by: Hanjun Guo Signed-off-by: Marc Zyngier Link: https://lore.kernel.org/r/20230516200516.554663-5-arnd@kernel.org --- include/linux/acpi_iort.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/acpi_iort.h b/include/linux/acpi_iort.h index b43be0987b19..6b70d02bc5f9 100644 --- a/include/linux/acpi_iort.h +++ b/include/linux/acpi_iort.h @@ -26,13 +26,14 @@ int iort_register_domain_token(int trans_id, phys_addr_t base, struct fwnode_handle *fw_node); void iort_deregister_domain_token(int trans_id); struct fwnode_handle *iort_find_domain_token(int trans_id); +int iort_pmsi_get_dev_id(struct device *dev, u32 *dev_id); + #ifdef CONFIG_ACPI_IORT void acpi_iort_init(void); u32 iort_msi_map_id(struct device *dev, u32 id); struct irq_domain *iort_get_device_domain(struct device *dev, u32 id, enum irq_domain_bus_token bus_token); void acpi_configure_pmsi_domain(struct device *dev); -int iort_pmsi_get_dev_id(struct device *dev, u32 *dev_id); void iort_get_rmr_sids(struct fwnode_handle *iommu_fwnode, struct list_head *head); void iort_put_rmr_sids(struct fwnode_handle *iommu_fwnode, -- cgit v1.2.3