From 1e12c4a9393b75a744aada2c8115434572698bc3 Mon Sep 17 00:00:00 2001 From: Marc Zyngier Date: Thu, 11 Aug 2016 14:19:42 +0100 Subject: genirq: Correctly configure the trigger on chained interrupts Commit 1e2a7d78499e ("irqdomain: Don't set type when mapping an IRQ") moved the trigger configuration call from the irqdomain mapping to the interrupt being actually requested. This patch failed to handle the case where we configure a chained interrupt, which doesn't get requested through the usual path. In order to solve this, let's call __irq_set_trigger just before starting the cascade interrupt. Special care must be taken to make the flow handler stick, as the .irq_set_type method could have reset it (it doesn't know we're dealing with a chained interrupt). Based on an initial patch by Jon Hunter. Fixes: 1e2a7d78499e ("irqdomain: Don't set type when mapping an IRQ") Reported-by: John Stultz Reported-by: Linus Walleij Tested-by: John Stultz Acked-by: Jon Hunter Signed-off-by: Marc Zyngier --- kernel/irq/chip.c | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index b4c1bc7c9ca2..637389088b3f 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -820,6 +820,17 @@ __irq_do_set_handler(struct irq_desc *desc, irq_flow_handler_t handle, desc->name = name; if (handle != handle_bad_irq && is_chained) { + /* + * We're about to start this interrupt immediately, + * hence the need to set the trigger configuration. + * But the .set_type callback may have overridden the + * flow handler, ignoring that we're dealing with a + * chained interrupt. Reset it immediately because we + * do know better. + */ + __irq_set_trigger(desc, irqd_get_trigger_type(&desc->irq_data)); + desc->handle_irq = handle; + irq_settings_set_noprobe(desc); irq_settings_set_norequest(desc); irq_settings_set_nothread(desc); -- cgit v1.2.3 From 059e232089e45b0befc9933d31209c225e08b426 Mon Sep 17 00:00:00 2001 From: Marc Zyngier Date: Tue, 9 Aug 2016 07:50:44 +0100 Subject: irqchip/gic: Allow self-SGIs for SMP on UP configurations On systems where a single CPU is present, the GIC may not support having SGIs delivered to a target list. In that case, we use the self-SGI mechanism to allow the interrupt to be delivered locally. Tested-by: Fabio Estevam Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-gic.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/drivers/irqchip/irq-gic.c b/drivers/irqchip/irq-gic.c index c2cab572c511..390fac59c6bc 100644 --- a/drivers/irqchip/irq-gic.c +++ b/drivers/irqchip/irq-gic.c @@ -769,6 +769,13 @@ static void gic_raise_softirq(const struct cpumask *mask, unsigned int irq) int cpu; unsigned long flags, map = 0; + if (unlikely(nr_cpu_ids == 1)) { + /* Only one CPU? let's do a self-IPI... */ + writel_relaxed(2 << 24 | irq, + gic_data_dist_base(&gic_data[0]) + GIC_DIST_SOFTINT); + return; + } + raw_spin_lock_irqsave(&irq_controller_lock, flags); /* Convert our logical CPU mask into a physical one. */ -- cgit v1.2.3 From ccd9432a5c85f35df7b491a1b701560d247466a5 Mon Sep 17 00:00:00 2001 From: Sudeep Holla Date: Wed, 17 Aug 2016 13:49:19 +0100 Subject: irqchip/gicv3: Remove disabling redistributor and group1 non-secure interrupts As per the GICv3 specification, to power down a processor using GICv3 and allow automatic power-on if an interrupt must be sent to a processor, software must set Enable to zero for all interrupt groups(by writing to GICC_CTLR or ICC_IGRPEN{0,1}_EL1/3 as appropriate. When commit 3708d52fc6bb ("irqchip: gic-v3: Implement CPU PM notifier") was introduced there were no firmware implementations(in particular PSCI) handling this. Linux kernel may not be aware of the CPU power state details and might fail to identify the power states that require quiescing the CPU interface. Even if it can be aware of those details, it can't determine which CPU power state have been triggered at the platform level and how the power control is implemented. This patch make disabling redistributor and group1 non-secure interrupts in the power down path and re-enabling of redistributor in the power-up path conditional. It will be handled in the kernel if and only if the non-secure accesses are permitted to access and modify control registers. It is left to the platform implementation otherwise. Cc: Marc Zyngier Cc: Lorenzo Pieralisi Cc: Thomas Gleixner Cc: Jason Cooper Tested-by: Christopher Covington Signed-off-by: Sudeep Holla Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-gic-v3.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c index 6fc56c3466b0..ede5672ab34d 100644 --- a/drivers/irqchip/irq-gic-v3.c +++ b/drivers/irqchip/irq-gic-v3.c @@ -667,13 +667,20 @@ static int gic_set_affinity(struct irq_data *d, const struct cpumask *mask_val, #endif #ifdef CONFIG_CPU_PM +/* Check whether it's single security state view */ +static bool gic_dist_security_disabled(void) +{ + return readl_relaxed(gic_data.dist_base + GICD_CTLR) & GICD_CTLR_DS; +} + static int gic_cpu_pm_notifier(struct notifier_block *self, unsigned long cmd, void *v) { if (cmd == CPU_PM_EXIT) { - gic_enable_redist(true); + if (gic_dist_security_disabled()) + gic_enable_redist(true); gic_cpu_sys_reg_init(); - } else if (cmd == CPU_PM_ENTER) { + } else if (cmd == CPU_PM_ENTER && gic_dist_security_disabled()) { gic_write_grpen1(0); gic_enable_redist(false); } -- cgit v1.2.3 From 7611da865c1060b2a7c87a15de663a59035747f8 Mon Sep 17 00:00:00 2001 From: David Daney Date: Thu, 18 Aug 2016 15:41:58 -0700 Subject: irqchip/gicv3-its: Disable the ITS before initializing it When starting a kexec/kdump kernel, the GIC ITS will already have been enabled. According to the ARM Generic Interrupt Controller Architecture Specification (GIC architecture Version 3.0 and version 4.0), writing to GITS_BASER or GITS_CBASER is "UNPREDICTABLE" when the ITS is enabled. On Cavium Thunder systems, this prevents the ITS from being initializing in the kexec/kdump kernel, resulting in failure to register/enable interrupts for all devices. The fix is to disable the ITS if it is not already in the disabled state. This allows the ITS to be properly initialized and then re-enabled in the kexec/kdump kernel. Acked-by: Marc Zyngier Signed-off-by: David Daney Signed-off-by: Marc Zyngier --- drivers/irqchip/irq-gic-v3-its.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c index 7ceaba81efb4..36b9c28a5c91 100644 --- a/drivers/irqchip/irq-gic-v3-its.c +++ b/drivers/irqchip/irq-gic-v3-its.c @@ -1545,7 +1545,12 @@ static int its_force_quiescent(void __iomem *base) u32 val; val = readl_relaxed(base + GITS_CTLR); - if (val & GITS_CTLR_QUIESCENT) + /* + * GIC architecture specification requires the ITS to be both + * disabled and quiescent for writes to GITS_BASER or + * GITS_CBASER to not have UNPREDICTABLE results. + */ + if ((val & GITS_CTLR_QUIESCENT) && !(val & GITS_CTLR_ENABLE)) return 0; /* Disable the generation of all interrupts to this ITS */ -- cgit v1.2.3 From 4396f46c8c628329bd35ee4b84140b8b001a11eb Mon Sep 17 00:00:00 2001 From: Shawn Lin Date: Mon, 22 Aug 2016 16:21:52 +0800 Subject: genirq: Fix potential memleak when failing to get irq pm Obviously we should free action here if irq_chip_pm_get failed. Fixes: be45beb2df69: "genirq: Add runtime power management support for IRQ chips" Signed-off-by: Shawn Lin Cc: Jon Hunter Cc: Marc Zyngier Link: http://lkml.kernel.org/r/1471854112-13006-1-git-send-email-shawn.lin@rock-chips.com Signed-off-by: Thomas Gleixner --- kernel/irq/manage.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 73a2b786b5e9..9530fcd27704 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -1681,8 +1681,10 @@ int request_threaded_irq(unsigned int irq, irq_handler_t handler, action->dev_id = dev_id; retval = irq_chip_pm_get(&desc->irq_data); - if (retval < 0) + if (retval < 0) { + kfree(action); return retval; + } chip_bus_lock(desc); retval = __setup_irq(irq, desc, action); @@ -1985,8 +1987,10 @@ int request_percpu_irq(unsigned int irq, irq_handler_t handler, action->percpu_dev_id = dev_id; retval = irq_chip_pm_get(&desc->irq_data); - if (retval < 0) + if (retval < 0) { + kfree(action); return retval; + } chip_bus_lock(desc); retval = __setup_irq(irq, desc, action); -- cgit v1.2.3 From 3ee0ce2a54dff07d09440723594df89bc1a12e79 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 11 Aug 2016 07:06:45 -0700 Subject: genirq/affinity: Use get/put_online_cpus around cpumask operations Without locking out CPU mask operations we might end up with an inconsistent view of the cpumask in the function. Fixes: 5e385a6ef31f: "genirq: Add a helper to spread an affinity mask for MSI/MSI-X vectors" Signed-off-by: Christoph Hellwig Link: http://lkml.kernel.org/r/1470924405-25728-1-git-send-email-hch@lst.de Signed-off-by: Thomas Gleixner --- kernel/irq/affinity.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/kernel/irq/affinity.c b/kernel/irq/affinity.c index f68959341c0f..32f6cfcff212 100644 --- a/kernel/irq/affinity.c +++ b/kernel/irq/affinity.c @@ -39,6 +39,7 @@ struct cpumask *irq_create_affinity_mask(unsigned int *nr_vecs) return NULL; } + get_online_cpus(); if (max_vecs >= num_online_cpus()) { cpumask_copy(affinity_mask, cpu_online_mask); *nr_vecs = num_online_cpus(); @@ -56,6 +57,7 @@ struct cpumask *irq_create_affinity_mask(unsigned int *nr_vecs) } *nr_vecs = vecs; } + put_online_cpus(); return affinity_mask; } -- cgit v1.2.3 From 6a33fa2b87513fee44cb8f0cd17b1acd6316bc6b Mon Sep 17 00:00:00 2001 From: Paul Burton Date: Fri, 19 Aug 2016 18:07:14 +0100 Subject: irqchip/mips-gic: Cleanup chip and handler setup gic_shared_irq_domain_map() is called from gic_irq_domain_alloc() where the wrong chip has been set, and is then overwritten. Tidy this up by setting the correct chip the first time, and setting the handle_level_irq handler from gic_irq_domain_alloc() too. gic_shared_irq_domain_map() is also called from gic_irq_domain_map(), which now calls irq_set_chip_and_handler() to retain its previous behaviour. This patch prepares for a follow-on which will call gic_shared_irq_domain_map() from a callback where the lock on the struct irq_desc is held, which without this change would cause the call to irq_set_chip_and_handler() to lead to a deadlock. Fixes: c98c1822ee13 ("irqchip/mips-gic: Add device hierarchy domain") Signed-off-by: Paul Burton Cc: linux-mips@linux-mips.org Cc: Jason Cooper Cc: Marc Zyngier Cc: stable@vger.kernel.org Link: http://lkml.kernel.org/r/20160819170715.27820-1-paul.burton@imgtec.com Signed-off-by: Thomas Gleixner --- drivers/irqchip/irq-mips-gic.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) diff --git a/drivers/irqchip/irq-mips-gic.c b/drivers/irqchip/irq-mips-gic.c index c5f33c3bd228..2e0f49992a98 100644 --- a/drivers/irqchip/irq-mips-gic.c +++ b/drivers/irqchip/irq-mips-gic.c @@ -713,9 +713,6 @@ static int gic_shared_irq_domain_map(struct irq_domain *d, unsigned int virq, unsigned long flags; int i; - irq_set_chip_and_handler(virq, &gic_level_irq_controller, - handle_level_irq); - spin_lock_irqsave(&gic_lock, flags); gic_map_to_pin(intr, gic_cpu_pin); gic_map_to_vpe(intr, mips_cm_vp_id(vpe)); @@ -732,6 +729,10 @@ static int gic_irq_domain_map(struct irq_domain *d, unsigned int virq, { if (GIC_HWIRQ_TO_LOCAL(hw) < GIC_NUM_LOCAL_INTRS) return gic_local_irq_domain_map(d, virq, hw); + + irq_set_chip_and_handler(virq, &gic_level_irq_controller, + handle_level_irq); + return gic_shared_irq_domain_map(d, virq, hw, 0); } @@ -771,11 +772,13 @@ static int gic_irq_domain_alloc(struct irq_domain *d, unsigned int virq, hwirq = GIC_SHARED_TO_HWIRQ(base_hwirq + i); ret = irq_domain_set_hwirq_and_chip(d, virq + i, hwirq, - &gic_edge_irq_controller, + &gic_level_irq_controller, NULL); if (ret) goto error; + irq_set_handler(virq + i, handle_level_irq); + ret = gic_shared_irq_domain_map(d, virq + i, hwirq, cpu); if (ret) goto error; -- cgit v1.2.3 From 2564970a381651865364974ea414384b569cb9c0 Mon Sep 17 00:00:00 2001 From: Paul Burton Date: Fri, 19 Aug 2016 18:07:15 +0100 Subject: irqchip/mips-gic: Implement activate op for device domain If an IRQ is setup using __setup_irq(), which is used by the request_irq() family of functions, and we are using an SMP kernel then the affinity of the IRQ will be set via setup_affinity() immediately after the IRQ is enabled. This call to gic_set_affinity() will lead to the interrupt being mapped to a VPE. However there are other ways to use IRQs which don't cause affinity to be set, for example if it is used to chain to another IRQ controller with irq_set_chained_handler_and_data(). The irq_set_chained_handler_and_data() code path will enable the IRQ, but will not trigger a call to gic_set_affinity() and in this case nothing will map the interrupt to a VPE, meaning that the interrupt is never received. Fix this by implementing the activate operation for the GIC device IRQ domain, using gic_shared_irq_domain_map() to map the interrupt to the correct pin of cpu 0. Fixes: c98c1822ee13 ("irqchip/mips-gic: Add device hierarchy domain") Signed-off-by: Paul Burton Cc: linux-mips@linux-mips.org Cc: Jason Cooper Cc: Marc Zyngier Cc: stable@vger.kernel.org Link: http://lkml.kernel.org/r/20160819170715.27820-2-paul.burton@imgtec.com Signed-off-by: Thomas Gleixner --- drivers/irqchip/irq-mips-gic.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/drivers/irqchip/irq-mips-gic.c b/drivers/irqchip/irq-mips-gic.c index 2e0f49992a98..83f498393a7f 100644 --- a/drivers/irqchip/irq-mips-gic.c +++ b/drivers/irqchip/irq-mips-gic.c @@ -893,10 +893,17 @@ void gic_dev_domain_free(struct irq_domain *d, unsigned int virq, return; } +static void gic_dev_domain_activate(struct irq_domain *domain, + struct irq_data *d) +{ + gic_shared_irq_domain_map(domain, d->irq, d->hwirq, 0); +} + static struct irq_domain_ops gic_dev_domain_ops = { .xlate = gic_dev_domain_xlate, .alloc = gic_dev_domain_alloc, .free = gic_dev_domain_free, + .activate = gic_dev_domain_activate, }; static int gic_ipi_domain_xlate(struct irq_domain *d, struct device_node *ctrlr, -- cgit v1.2.3