linux/drivers/irqchip/irq-mips-gic.c

1013 lines
25 KiB
C
Raw Normal View History

/*
* This file is subject to the terms and conditions of the GNU General Public
* License. See the file "COPYING" in the main directory of this archive
* for more details.
*
* Copyright (C) 2008 Ralf Baechle (ralf@linux-mips.org)
* Copyright (C) 2012 MIPS Technologies, Inc. All rights reserved.
*/
#define pr_fmt(fmt) "irq-mips-gic: " fmt
#include <linux/bitfield.h>
#include <linux/bitmap.h>
#include <linux/clocksource.h>
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
#include <linux/cpuhotplug.h>
#include <linux/init.h>
#include <linux/interrupt.h>
#include <linux/irq.h>
#include <linux/irqchip.h>
#include <linux/irqdomain.h>
#include <linux/of_address.h>
#include <linux/percpu.h>
#include <linux/sched.h>
#include <linux/smp.h>
#include <asm/mips-cps.h>
#include <asm/setup.h>
#include <asm/traps.h>
#include <dt-bindings/interrupt-controller/mips-gic.h>
#define GIC_MAX_INTRS 256
#define GIC_MAX_LONGS BITS_TO_LONGS(GIC_MAX_INTRS)
/* Add 2 to convert GIC CPU pin to core interrupt */
#define GIC_CPU_PIN_OFFSET 2
/* Mapped interrupt to pin X, then GIC will generate the vector (X+1). */
#define GIC_PIN_TO_VEC_OFFSET 1
/* Convert between local/shared IRQ number and GIC HW IRQ number. */
#define GIC_LOCAL_HWIRQ_BASE 0
#define GIC_LOCAL_TO_HWIRQ(x) (GIC_LOCAL_HWIRQ_BASE + (x))
#define GIC_HWIRQ_TO_LOCAL(x) ((x) - GIC_LOCAL_HWIRQ_BASE)
#define GIC_SHARED_HWIRQ_BASE GIC_NUM_LOCAL_INTRS
#define GIC_SHARED_TO_HWIRQ(x) (GIC_SHARED_HWIRQ_BASE + (x))
#define GIC_HWIRQ_TO_SHARED(x) ((x) - GIC_SHARED_HWIRQ_BASE)
MIPS: GIC: Introduce asm/mips-gic.h with accessor functions This patch introduces a new header providing accessor functions for the MIPS Global Interrupt Controller (GIC) mirroring those provided for the other 2 components of the MIPS Coherent Processing System (CPS) - the Coherence Manager (CM) & Cluster Power Controller (CPC). This header makes use of the new standardised CPS accessor macros where possible, but does require some custom accessors for cases where we have either a bit or a register per interrupt. A major advantage of this over the existing include/linux/irqchip/mips-gic.h definitions is that code performing accesses can become much simpler, for example this: gic_update_bits(GIC_REG(SHARED, GIC_SH_SET_TRIGGER) + GIC_INTR_OFS(intr), 1ul << GIC_INTR_BIT(intr), (unsigned long)trig << GIC_INTR_BIT(intr)); ...can become simply: change_gic_trig(intr, trig); The accessors handle 32 vs 64 bit in the same way as for CM & CPC code, which means that GIC code will also not need to worry about the access size in most cases. They are also accessible outside of drivers/irqchip/irq-mips-gic.c which will allow for simplification in the use of the non-interrupt portions of the GIC (eg. counters) which currently require the interrupt controller driver to expose helper functions for access. This patch doesn't change any existing code over to use the new accessors yet, since a wholesale change would be invasive & difficult to review. Instead follow-on patches will convert code piecemeal to use this new header. The one change to existing code is to rename gic_base to mips_gic_base & make it global, in order to fit in with the naming expected by the standardised CPS accessor macros. Signed-off-by: Paul Burton <paul.burton@imgtec.com> Acked-by: Marc Zyngier <marc.zyngier@arm.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Cc: linux-mips@linux-mips.org Patchwork: https://patchwork.linux-mips.org/patch/17020/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2017-08-13 04:36:10 +00:00
void __iomem *mips_gic_base;
static DEFINE_PER_CPU_READ_MOSTLY(unsigned long[GIC_MAX_LONGS], pcpu_masks);
static DEFINE_RAW_SPINLOCK(gic_lock);
static struct irq_domain *gic_irq_domain;
static int gic_shared_intrs;
static unsigned int gic_cpu_pin;
static struct irq_chip gic_level_irq_controller, gic_edge_irq_controller;
#ifdef CONFIG_GENERIC_IRQ_IPI
static DECLARE_BITMAP(ipi_resrv, GIC_MAX_INTRS);
static DECLARE_BITMAP(ipi_available, GIC_MAX_INTRS);
#endif /* CONFIG_GENERIC_IRQ_IPI */
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
static struct gic_all_vpes_chip_data {
u32 map;
bool mask;
} gic_all_vpes_chip_data[GIC_NUM_LOCAL_INTRS];
static int __gic_with_next_online_cpu(int prev)
{
unsigned int cpu;
/* Discover the next online CPU */
cpu = cpumask_next(prev, cpu_online_mask);
/* If there isn't one, we're done */
if (cpu >= nr_cpu_ids)
return cpu;
/*
* Move the access lock to the next CPU's GIC local register block.
*
* Set GIC_VL_OTHER. Since the caller holds gic_lock nothing can
* clobber the written value.
*/
write_gic_vl_other(mips_cm_vp_id(cpu));
return cpu;
}
static inline void gic_unlock_cluster(void)
{
if (mips_cps_multicluster_cpus())
mips_cm_unlock_other();
}
/**
* for_each_online_cpu_gic() - Iterate over online CPUs, access local registers
* @cpu: An integer variable to hold the current CPU number
* @gic_lock: A pointer to raw spin lock used as a guard
*
* Iterate over online CPUs & configure the other/redirect register region to
* access each CPUs GIC local register block, which can be accessed from the
* loop body using read_gic_vo_*() or write_gic_vo_*() accessor functions or
* their derivatives.
*/
#define for_each_online_cpu_gic(cpu, gic_lock) \
guard(raw_spinlock_irqsave)(gic_lock); \
for ((cpu) = __gic_with_next_online_cpu(-1); \
(cpu) < nr_cpu_ids; \
gic_unlock_cluster(), \
(cpu) = __gic_with_next_online_cpu(cpu))
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
/**
* gic_irq_lock_cluster() - Lock redirect block access to IRQ's cluster
* @d: struct irq_data corresponding to the interrupt we're interested in
*
* Locks redirect register block access to the global register block of the GIC
* within the remote cluster that the IRQ corresponding to @d is affine to,
* returning true when this redirect block setup & locking has been performed.
*
* If @d is affine to the local cluster then no locking is performed and this
* function will return false, indicating to the caller that it should access
* the local clusters registers without the overhead of indirection through the
* redirect block.
*
* In summary, if this function returns true then the caller should access GIC
* registers using redirect register block accessors & then call
* mips_cm_unlock_other() when done. If this function returns false then the
* caller should trivially access GIC registers in the local cluster.
*
* Returns true if locking performed, else false.
*/
static bool gic_irq_lock_cluster(struct irq_data *d)
{
unsigned int cpu, cl;
cpu = cpumask_first(irq_data_get_effective_affinity_mask(d));
BUG_ON(cpu >= NR_CPUS);
cl = cpu_cluster(&cpu_data[cpu]);
if (cl == cpu_cluster(&current_cpu_data))
return false;
if (mips_cps_numcores(cl) == 0)
return false;
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
mips_cm_lock_other(cl, 0, 0, CM_GCR_Cx_OTHER_BLOCK_GLOBAL);
return true;
}
static void gic_clear_pcpu_masks(unsigned int intr)
{
unsigned int i;
/* Clear the interrupt's bit in all pcpu_masks */
for_each_possible_cpu(i)
clear_bit(intr, per_cpu_ptr(pcpu_masks, i));
}
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
static bool gic_local_irq_is_routable(int intr)
{
u32 vpe_ctl;
/* All local interrupts are routable in EIC mode. */
if (cpu_has_veic)
return true;
vpe_ctl = read_gic_vl_ctl();
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
switch (intr) {
case GIC_LOCAL_INT_TIMER:
return vpe_ctl & GIC_VX_CTL_TIMER_ROUTABLE;
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
case GIC_LOCAL_INT_PERFCTR:
return vpe_ctl & GIC_VX_CTL_PERFCNT_ROUTABLE;
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
case GIC_LOCAL_INT_FDC:
return vpe_ctl & GIC_VX_CTL_FDC_ROUTABLE;
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
case GIC_LOCAL_INT_SWINT0:
case GIC_LOCAL_INT_SWINT1:
return vpe_ctl & GIC_VX_CTL_SWINT_ROUTABLE;
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
default:
return true;
}
}
static void gic_bind_eic_interrupt(int irq, int set)
{
/* Convert irq vector # to hw int # */
irq -= GIC_PIN_TO_VEC_OFFSET;
/* Set irq to use shadow set */
write_gic_vl_eic_shadow_set(irq, set);
}
static void gic_send_ipi(struct irq_data *d, unsigned int cpu)
{
irq_hw_number_t hwirq = GIC_HWIRQ_TO_SHARED(irqd_to_hwirq(d));
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
if (gic_irq_lock_cluster(d)) {
write_gic_redir_wedge(GIC_WEDGE_RW | hwirq);
mips_cm_unlock_other();
} else {
write_gic_wedge(GIC_WEDGE_RW | hwirq);
}
}
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
int gic_get_c0_compare_int(void)
{
if (!gic_local_irq_is_routable(GIC_LOCAL_INT_TIMER))
return MIPS_CPU_IRQ_BASE + cp0_compare_irq;
return irq_create_mapping(gic_irq_domain,
GIC_LOCAL_TO_HWIRQ(GIC_LOCAL_INT_TIMER));
}
int gic_get_c0_perfcount_int(void)
{
if (!gic_local_irq_is_routable(GIC_LOCAL_INT_PERFCTR)) {
/* Is the performance counter shared with the timer? */
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
if (cp0_perfcount_irq < 0)
return -1;
return MIPS_CPU_IRQ_BASE + cp0_perfcount_irq;
}
return irq_create_mapping(gic_irq_domain,
GIC_LOCAL_TO_HWIRQ(GIC_LOCAL_INT_PERFCTR));
}
int gic_get_c0_fdc_int(void)
{
if (!gic_local_irq_is_routable(GIC_LOCAL_INT_FDC)) {
/* Is the FDC IRQ even present? */
if (cp0_fdc_irq < 0)
return -1;
return MIPS_CPU_IRQ_BASE + cp0_fdc_irq;
}
return irq_create_mapping(gic_irq_domain,
GIC_LOCAL_TO_HWIRQ(GIC_LOCAL_INT_FDC));
}
static void gic_handle_shared_int(bool chained)
{
unsigned int intr;
unsigned long *pcpu_mask;
DECLARE_BITMAP(pending, GIC_MAX_INTRS);
/* Get per-cpu bitmaps */
pcpu_mask = this_cpu_ptr(pcpu_masks);
if (mips_cm_is64)
__ioread64_copy(pending, addr_gic_pend(),
DIV_ROUND_UP(gic_shared_intrs, 64));
else
__ioread32_copy(pending, addr_gic_pend(),
DIV_ROUND_UP(gic_shared_intrs, 32));
bitmap_and(pending, pending, pcpu_mask, gic_shared_intrs);
for_each_set_bit(intr, pending, gic_shared_intrs) {
if (chained)
generic_handle_domain_irq(gic_irq_domain,
GIC_SHARED_TO_HWIRQ(intr));
else
irqchip/mips: Fix RCU violation when using irqdomain lookup on interrupt entry Since d4a45c68dc81 ("irqdomain: Protect the linear revmap with RCU"), any irqdomain lookup requires the RCU read lock to be held. This assumes that the architecture code will be structured such as irq_enter() will be called *before* the interrupt is looked up in the irq domain. However, this isn't the case for MIPS, and a number of drivers are structured to do it the other way around when handling an interrupt in their root irqchip (secondary irqchips are OK by construction). This results in a RCU splat on a lockdep-enabled kernel when the kernel takes an interrupt from idle, as reported by Guenter Roeck. Note that this could have fired previously if any driver had used tree-based irqdomain, which always had the RCU requirement. To solve this, provide a MIPS-specific helper (do_domain_IRQ()) as the pendent of do_IRQ() that will do thing in the right order (and maybe save some cycles in the process). Ideally, MIPS would be moved over to using handle_domain_irq(), but that's much more ambitious. Reported-by: Guenter Roeck <linux@roeck-us.net> Tested-by: Guenter Roeck <linux@roeck-us.net> [maz: add dependency on CONFIG_IRQ_DOMAIN after report from the kernelci bot] Signed-off-by: Marc Zyngier <maz@kernel.org> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Serge Semin <fancer.lancer@gmail.com> Link: https://lore.kernel.org/r/20210705172352.GA56304@roeck-us.net Link: https://lore.kernel.org/r/20210706110647.3979002-1-maz@kernel.org
2021-07-06 10:38:59 +00:00
do_domain_IRQ(gic_irq_domain,
GIC_SHARED_TO_HWIRQ(intr));
}
}
static void gic_mask_irq(struct irq_data *d)
{
unsigned int intr = GIC_HWIRQ_TO_SHARED(d->hwirq);
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
if (gic_irq_lock_cluster(d)) {
write_gic_redir_rmask(intr);
mips_cm_unlock_other();
} else {
write_gic_rmask(intr);
}
gic_clear_pcpu_masks(intr);
}
static void gic_unmask_irq(struct irq_data *d)
{
unsigned int intr = GIC_HWIRQ_TO_SHARED(d->hwirq);
unsigned int cpu;
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
if (gic_irq_lock_cluster(d)) {
write_gic_redir_smask(intr);
mips_cm_unlock_other();
} else {
write_gic_smask(intr);
}
gic_clear_pcpu_masks(intr);
cpu = cpumask_first(irq_data_get_effective_affinity_mask(d));
set_bit(intr, per_cpu_ptr(pcpu_masks, cpu));
}
static void gic_ack_irq(struct irq_data *d)
{
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
unsigned int irq = GIC_HWIRQ_TO_SHARED(d->hwirq);
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
if (gic_irq_lock_cluster(d)) {
write_gic_redir_wedge(irq);
mips_cm_unlock_other();
} else {
write_gic_wedge(irq);
}
}
static int gic_set_type(struct irq_data *d, unsigned int type)
{
unsigned int irq, pol, trig, dual;
unsigned long flags;
irq = GIC_HWIRQ_TO_SHARED(d->hwirq);
raw_spin_lock_irqsave(&gic_lock, flags);
switch (type & IRQ_TYPE_SENSE_MASK) {
case IRQ_TYPE_EDGE_FALLING:
pol = GIC_POL_FALLING_EDGE;
trig = GIC_TRIG_EDGE;
dual = GIC_DUAL_SINGLE;
break;
case IRQ_TYPE_EDGE_RISING:
pol = GIC_POL_RISING_EDGE;
trig = GIC_TRIG_EDGE;
dual = GIC_DUAL_SINGLE;
break;
case IRQ_TYPE_EDGE_BOTH:
pol = 0; /* Doesn't matter */
trig = GIC_TRIG_EDGE;
dual = GIC_DUAL_DUAL;
break;
case IRQ_TYPE_LEVEL_LOW:
pol = GIC_POL_ACTIVE_LOW;
trig = GIC_TRIG_LEVEL;
dual = GIC_DUAL_SINGLE;
break;
case IRQ_TYPE_LEVEL_HIGH:
default:
pol = GIC_POL_ACTIVE_HIGH;
trig = GIC_TRIG_LEVEL;
dual = GIC_DUAL_SINGLE;
break;
}
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
if (gic_irq_lock_cluster(d)) {
change_gic_redir_pol(irq, pol);
change_gic_redir_trig(irq, trig);
change_gic_redir_dual(irq, dual);
mips_cm_unlock_other();
} else {
change_gic_pol(irq, pol);
change_gic_trig(irq, trig);
change_gic_dual(irq, dual);
}
if (trig == GIC_TRIG_EDGE)
irq_set_chip_handler_name_locked(d, &gic_edge_irq_controller,
handle_edge_irq, NULL);
else
irq_set_chip_handler_name_locked(d, &gic_level_irq_controller,
handle_level_irq, NULL);
raw_spin_unlock_irqrestore(&gic_lock, flags);
return 0;
}
#ifdef CONFIG_SMP
static int gic_set_affinity(struct irq_data *d, const struct cpumask *cpumask,
bool force)
{
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
unsigned int irq = GIC_HWIRQ_TO_SHARED(d->hwirq);
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
unsigned int cpu, cl, old_cpu, old_cl;
unsigned long flags;
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
/*
* The GIC specifies that we can only route an interrupt to one VP(E),
* ie. CPU in Linux parlance, at a time. Therefore we always route to
* the first online CPU in the mask.
*/
cpu = cpumask_first_and(cpumask, cpu_online_mask);
if (cpu >= NR_CPUS)
return -EINVAL;
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
old_cpu = cpumask_first(irq_data_get_effective_affinity_mask(d));
old_cl = cpu_cluster(&cpu_data[old_cpu]);
cl = cpu_cluster(&cpu_data[cpu]);
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
raw_spin_lock_irqsave(&gic_lock, flags);
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
/*
* If we're moving affinity between clusters, stop routing the
* interrupt to any VP(E) in the old cluster.
*/
if (cl != old_cl) {
if (gic_irq_lock_cluster(d)) {
write_gic_redir_map_vp(irq, 0);
mips_cm_unlock_other();
} else {
write_gic_map_vp(irq, 0);
}
}
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
/*
* Update effective affinity - after this gic_irq_lock_cluster() will
* begin operating on the new cluster.
*/
irq_data_update_effective_affinity(d, cpumask_of(cpu));
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
/*
* If we're moving affinity between clusters, configure the interrupt
* trigger type in the new cluster.
*/
if (cl != old_cl)
gic_set_type(d, irqd_get_trigger_type(d));
/* Route the interrupt to its new VP(E) */
if (gic_irq_lock_cluster(d)) {
write_gic_redir_map_pin(irq,
GIC_MAP_PIN_MAP_TO_PIN | gic_cpu_pin);
write_gic_redir_map_vp(irq, BIT(mips_cm_vp_id(cpu)));
/* Update the pcpu_masks */
gic_clear_pcpu_masks(irq);
if (read_gic_redir_mask(irq))
set_bit(irq, per_cpu_ptr(pcpu_masks, cpu));
mips_cm_unlock_other();
} else {
write_gic_map_pin(irq, GIC_MAP_PIN_MAP_TO_PIN | gic_cpu_pin);
write_gic_map_vp(irq, BIT(mips_cm_vp_id(cpu)));
/* Update the pcpu_masks */
gic_clear_pcpu_masks(irq);
if (read_gic_mask(irq))
set_bit(irq, per_cpu_ptr(pcpu_masks, cpu));
}
raw_spin_unlock_irqrestore(&gic_lock, flags);
return IRQ_SET_MASK_OK;
}
#endif
static struct irq_chip gic_level_irq_controller = {
.name = "MIPS GIC",
.irq_mask = gic_mask_irq,
.irq_unmask = gic_unmask_irq,
.irq_set_type = gic_set_type,
#ifdef CONFIG_SMP
.irq_set_affinity = gic_set_affinity,
#endif
};
static struct irq_chip gic_edge_irq_controller = {
.name = "MIPS GIC",
.irq_ack = gic_ack_irq,
.irq_mask = gic_mask_irq,
.irq_unmask = gic_unmask_irq,
.irq_set_type = gic_set_type,
#ifdef CONFIG_SMP
.irq_set_affinity = gic_set_affinity,
#endif
.ipi_send_single = gic_send_ipi,
};
static void gic_handle_local_int(bool chained)
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
{
unsigned long pending, masked;
unsigned int intr;
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
pending = read_gic_vl_pend();
masked = read_gic_vl_mask();
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
bitmap_and(&pending, &pending, &masked, GIC_NUM_LOCAL_INTRS);
for_each_set_bit(intr, &pending, GIC_NUM_LOCAL_INTRS) {
if (chained)
generic_handle_domain_irq(gic_irq_domain,
GIC_LOCAL_TO_HWIRQ(intr));
else
irqchip/mips: Fix RCU violation when using irqdomain lookup on interrupt entry Since d4a45c68dc81 ("irqdomain: Protect the linear revmap with RCU"), any irqdomain lookup requires the RCU read lock to be held. This assumes that the architecture code will be structured such as irq_enter() will be called *before* the interrupt is looked up in the irq domain. However, this isn't the case for MIPS, and a number of drivers are structured to do it the other way around when handling an interrupt in their root irqchip (secondary irqchips are OK by construction). This results in a RCU splat on a lockdep-enabled kernel when the kernel takes an interrupt from idle, as reported by Guenter Roeck. Note that this could have fired previously if any driver had used tree-based irqdomain, which always had the RCU requirement. To solve this, provide a MIPS-specific helper (do_domain_IRQ()) as the pendent of do_IRQ() that will do thing in the right order (and maybe save some cycles in the process). Ideally, MIPS would be moved over to using handle_domain_irq(), but that's much more ambitious. Reported-by: Guenter Roeck <linux@roeck-us.net> Tested-by: Guenter Roeck <linux@roeck-us.net> [maz: add dependency on CONFIG_IRQ_DOMAIN after report from the kernelci bot] Signed-off-by: Marc Zyngier <maz@kernel.org> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Serge Semin <fancer.lancer@gmail.com> Link: https://lore.kernel.org/r/20210705172352.GA56304@roeck-us.net Link: https://lore.kernel.org/r/20210706110647.3979002-1-maz@kernel.org
2021-07-06 10:38:59 +00:00
do_domain_IRQ(gic_irq_domain,
GIC_LOCAL_TO_HWIRQ(intr));
}
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
}
static void gic_mask_local_irq(struct irq_data *d)
{
int intr = GIC_HWIRQ_TO_LOCAL(d->hwirq);
write_gic_vl_rmask(BIT(intr));
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
}
static void gic_unmask_local_irq(struct irq_data *d)
{
int intr = GIC_HWIRQ_TO_LOCAL(d->hwirq);
write_gic_vl_smask(BIT(intr));
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
}
static struct irq_chip gic_local_irq_controller = {
.name = "MIPS GIC Local",
.irq_mask = gic_mask_local_irq,
.irq_unmask = gic_unmask_local_irq,
};
static void gic_mask_local_irq_all_vpes(struct irq_data *d)
{
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
struct gic_all_vpes_chip_data *cd;
int intr, cpu;
if (!mips_cps_multicluster_cpus())
return;
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
intr = GIC_HWIRQ_TO_LOCAL(d->hwirq);
cd = irq_data_get_irq_chip_data(d);
cd->mask = false;
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
for_each_online_cpu_gic(cpu, &gic_lock)
write_gic_vo_rmask(BIT(intr));
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
}
static void gic_unmask_local_irq_all_vpes(struct irq_data *d)
{
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
struct gic_all_vpes_chip_data *cd;
int intr, cpu;
if (!mips_cps_multicluster_cpus())
return;
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
intr = GIC_HWIRQ_TO_LOCAL(d->hwirq);
cd = irq_data_get_irq_chip_data(d);
cd->mask = true;
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
for_each_online_cpu_gic(cpu, &gic_lock)
write_gic_vo_smask(BIT(intr));
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
}
static void gic_all_vpes_irq_cpu_online(void)
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
{
static const unsigned int local_intrs[] = {
GIC_LOCAL_INT_TIMER,
GIC_LOCAL_INT_PERFCTR,
GIC_LOCAL_INT_FDC,
};
unsigned long flags;
int i;
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
raw_spin_lock_irqsave(&gic_lock, flags);
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
for (i = 0; i < ARRAY_SIZE(local_intrs); i++) {
unsigned int intr = local_intrs[i];
struct gic_all_vpes_chip_data *cd;
if (!gic_local_irq_is_routable(intr))
continue;
cd = &gic_all_vpes_chip_data[intr];
write_gic_vl_map(mips_gic_vx_map_reg(intr), cd->map);
if (cd->mask)
write_gic_vl_smask(BIT(intr));
}
raw_spin_unlock_irqrestore(&gic_lock, flags);
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
}
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
static struct irq_chip gic_all_vpes_local_irq_controller = {
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
.name = "MIPS GIC Local",
.irq_mask = gic_mask_local_irq_all_vpes,
.irq_unmask = gic_unmask_local_irq_all_vpes,
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
};
static void __gic_irq_dispatch(void)
{
gic_handle_local_int(false);
gic_handle_shared_int(false);
}
static void gic_irq_dispatch(struct irq_desc *desc)
{
gic_handle_local_int(true);
gic_handle_shared_int(true);
}
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
static int gic_shared_irq_domain_map(struct irq_domain *d, unsigned int virq,
irq_hw_number_t hw, unsigned int cpu)
irqchip: mips-gic: Support local interrupts The MIPS GIC supports 7 local interrupts, 2 of which are the GIC local watchdog and count/compare timer. The remainder are CPU interrupts which may optionally be re-routed through the GIC. GIC hardware IRQs 0-6 are now used for local interrupts while hardware IRQs 7+ are used for external (shared) interrupts. Note that the 5 CPU interrupts may not be re-routable through the GIC. In that case mapping will fail and the vectors reported in C0_IntCtl should be used instead. gic_get_c0_compare_int() and gic_get_c0_perfcount_int() will return the correct IRQ number to use for the C0 timer and perfcounter interrupts based on the routability of those interrupts through the GIC. A separate irq_chip, with callbacks that mask/unmask the local interrupt on all CPUs, is used for the C0 timer and performance counter interrupts since all other platforms do not use the percpu IRQ API for those interrupts. Malta, SEAD-3, and the GIC clockevent driver have been updated to use local interrupts and the R4K clockevent driver has been updated to poll for C0 timer interrupts through the GIC when the GIC is present. Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Acked-by: Jason Cooper <jason@lakedaemon.net> Reviewed-by: Qais Yousef <qais.yousef@imgtec.com> Tested-by: Qais Yousef <qais.yousef@imgtec.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jeffrey Deans <jeffrey.deans@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Jonas Gorski <jogo@openwrt.org> Cc: John Crispin <blogic@openwrt.org> Cc: David Daney <ddaney.cavm@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/7819/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2014-09-18 21:47:27 +00:00
{
int intr = GIC_HWIRQ_TO_SHARED(hw);
struct irq_data *data;
unsigned long flags;
data = irq_get_irq_data(virq);
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
irq_data_update_effective_affinity(data, cpumask_of(cpu));
raw_spin_lock_irqsave(&gic_lock, flags);
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
/* Route the interrupt to its VP(E) */
if (gic_irq_lock_cluster(data)) {
write_gic_redir_map_pin(intr,
GIC_MAP_PIN_MAP_TO_PIN | gic_cpu_pin);
write_gic_redir_map_vp(intr, BIT(mips_cm_vp_id(cpu)));
mips_cm_unlock_other();
} else {
write_gic_map_pin(intr, GIC_MAP_PIN_MAP_TO_PIN | gic_cpu_pin);
write_gic_map_vp(intr, BIT(mips_cm_vp_id(cpu)));
}
raw_spin_unlock_irqrestore(&gic_lock, flags);
return 0;
}
static int gic_irq_domain_xlate(struct irq_domain *d, struct device_node *ctrlr,
const u32 *intspec, unsigned int intsize,
irq_hw_number_t *out_hwirq,
unsigned int *out_type)
{
if (intsize != 3)
return -EINVAL;
if (intspec[0] == GIC_SHARED)
*out_hwirq = GIC_SHARED_TO_HWIRQ(intspec[1]);
else if (intspec[0] == GIC_LOCAL)
*out_hwirq = GIC_LOCAL_TO_HWIRQ(intspec[1]);
else
return -EINVAL;
*out_type = intspec[2] & IRQ_TYPE_SENSE_MASK;
return 0;
}
static int gic_irq_domain_map(struct irq_domain *d, unsigned int virq,
irq_hw_number_t hwirq)
{
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
struct gic_all_vpes_chip_data *cd;
irqchip: mips-gic: Inline gic_local_irq_domain_map() The gic_local_irq_domain_map() function has only one callsite in gic_irq_domain_map(), and the split between the two functions makes it unclear that they duplicate calculations & checks. Inline gic_local_irq_domain_map() into gic_irq_domain_map() in order to clean this up. Doing this makes the following small issues obvious, and the patch tidies them up: - Both functions used GIC_HWIRQ_TO_LOCAL() to convert a hwirq number to a local IRQ number. We now only do this once. Although the compiler ought to have optimised this away before anyway, the change leaves us with less duplicate code. - gic_local_irq_domain_map() had a check for invalid local interrupt numbers (intr > GIC_LOCAL_INT_FDC). This condition can never occur because any hwirq higher than those used for local interrupts is a shared interrupt, which gic_irq_domain_map() already handles separately. We therefore remove this check. - The decision of whether to map the interrupt to gic_cpu_pin or timer_cpu_pin can be handled within the existing switch statement in gic_irq_domain_map(), shortening the code a little. The change additionally prepares us nicely for the following patch of the series which would otherwise need to duplicate the check for whether a local interrupt should be percpu_devid or just percpu (ie. the switch statement from gic_irq_domain_map()) in gic_local_irq_domain_map(). Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:44 +00:00
unsigned int intr;
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
int err, cpu;
irqchip: mips-gic: Inline gic_local_irq_domain_map() The gic_local_irq_domain_map() function has only one callsite in gic_irq_domain_map(), and the split between the two functions makes it unclear that they duplicate calculations & checks. Inline gic_local_irq_domain_map() into gic_irq_domain_map() in order to clean this up. Doing this makes the following small issues obvious, and the patch tidies them up: - Both functions used GIC_HWIRQ_TO_LOCAL() to convert a hwirq number to a local IRQ number. We now only do this once. Although the compiler ought to have optimised this away before anyway, the change leaves us with less duplicate code. - gic_local_irq_domain_map() had a check for invalid local interrupt numbers (intr > GIC_LOCAL_INT_FDC). This condition can never occur because any hwirq higher than those used for local interrupts is a shared interrupt, which gic_irq_domain_map() already handles separately. We therefore remove this check. - The decision of whether to map the interrupt to gic_cpu_pin or timer_cpu_pin can be handled within the existing switch statement in gic_irq_domain_map(), shortening the code a little. The change additionally prepares us nicely for the following patch of the series which would otherwise need to duplicate the check for whether a local interrupt should be percpu_devid or just percpu (ie. the switch statement from gic_irq_domain_map()) in gic_local_irq_domain_map(). Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:44 +00:00
u32 map;
if (hwirq >= GIC_SHARED_HWIRQ_BASE) {
#ifdef CONFIG_GENERIC_IRQ_IPI
/* verify that shared irqs don't conflict with an IPI irq */
if (test_bit(GIC_HWIRQ_TO_SHARED(hwirq), ipi_resrv))
return -EBUSY;
#endif /* CONFIG_GENERIC_IRQ_IPI */
err = irq_domain_set_hwirq_and_chip(d, virq, hwirq,
&gic_level_irq_controller,
NULL);
if (err)
return err;
irqd_set_single_target(irq_desc_get_irq_data(irq_to_desc(virq)));
return gic_shared_irq_domain_map(d, virq, hwirq, 0);
}
irqchip: mips-gic: Inline gic_local_irq_domain_map() The gic_local_irq_domain_map() function has only one callsite in gic_irq_domain_map(), and the split between the two functions makes it unclear that they duplicate calculations & checks. Inline gic_local_irq_domain_map() into gic_irq_domain_map() in order to clean this up. Doing this makes the following small issues obvious, and the patch tidies them up: - Both functions used GIC_HWIRQ_TO_LOCAL() to convert a hwirq number to a local IRQ number. We now only do this once. Although the compiler ought to have optimised this away before anyway, the change leaves us with less duplicate code. - gic_local_irq_domain_map() had a check for invalid local interrupt numbers (intr > GIC_LOCAL_INT_FDC). This condition can never occur because any hwirq higher than those used for local interrupts is a shared interrupt, which gic_irq_domain_map() already handles separately. We therefore remove this check. - The decision of whether to map the interrupt to gic_cpu_pin or timer_cpu_pin can be handled within the existing switch statement in gic_irq_domain_map(), shortening the code a little. The change additionally prepares us nicely for the following patch of the series which would otherwise need to duplicate the check for whether a local interrupt should be percpu_devid or just percpu (ie. the switch statement from gic_irq_domain_map()) in gic_local_irq_domain_map(). Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:44 +00:00
intr = GIC_HWIRQ_TO_LOCAL(hwirq);
map = GIC_MAP_PIN_MAP_TO_PIN | gic_cpu_pin;
/*
* If adding support for more per-cpu interrupts, keep the
* array in gic_all_vpes_irq_cpu_online() in sync.
*/
irqchip: mips-gic: Inline gic_local_irq_domain_map() The gic_local_irq_domain_map() function has only one callsite in gic_irq_domain_map(), and the split between the two functions makes it unclear that they duplicate calculations & checks. Inline gic_local_irq_domain_map() into gic_irq_domain_map() in order to clean this up. Doing this makes the following small issues obvious, and the patch tidies them up: - Both functions used GIC_HWIRQ_TO_LOCAL() to convert a hwirq number to a local IRQ number. We now only do this once. Although the compiler ought to have optimised this away before anyway, the change leaves us with less duplicate code. - gic_local_irq_domain_map() had a check for invalid local interrupt numbers (intr > GIC_LOCAL_INT_FDC). This condition can never occur because any hwirq higher than those used for local interrupts is a shared interrupt, which gic_irq_domain_map() already handles separately. We therefore remove this check. - The decision of whether to map the interrupt to gic_cpu_pin or timer_cpu_pin can be handled within the existing switch statement in gic_irq_domain_map(), shortening the code a little. The change additionally prepares us nicely for the following patch of the series which would otherwise need to duplicate the check for whether a local interrupt should be percpu_devid or just percpu (ie. the switch statement from gic_irq_domain_map()) in gic_local_irq_domain_map(). Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:44 +00:00
switch (intr) {
case GIC_LOCAL_INT_TIMER:
case GIC_LOCAL_INT_PERFCTR:
case GIC_LOCAL_INT_FDC:
/*
* HACK: These are all really percpu interrupts, but
* the rest of the MIPS kernel code does not use the
* percpu IRQ API for them.
*/
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
cd = &gic_all_vpes_chip_data[intr];
cd->map = map;
err = irq_domain_set_hwirq_and_chip(d, virq, hwirq,
&gic_all_vpes_local_irq_controller,
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
cd);
if (err)
return err;
irq_set_handler(virq, handle_percpu_irq);
break;
default:
err = irq_domain_set_hwirq_and_chip(d, virq, hwirq,
&gic_local_irq_controller,
NULL);
if (err)
return err;
irq_set_handler(virq, handle_percpu_devid_irq);
irq_set_percpu_devid(virq);
break;
}
irqchip: mips-gic: Inline gic_local_irq_domain_map() The gic_local_irq_domain_map() function has only one callsite in gic_irq_domain_map(), and the split between the two functions makes it unclear that they duplicate calculations & checks. Inline gic_local_irq_domain_map() into gic_irq_domain_map() in order to clean this up. Doing this makes the following small issues obvious, and the patch tidies them up: - Both functions used GIC_HWIRQ_TO_LOCAL() to convert a hwirq number to a local IRQ number. We now only do this once. Although the compiler ought to have optimised this away before anyway, the change leaves us with less duplicate code. - gic_local_irq_domain_map() had a check for invalid local interrupt numbers (intr > GIC_LOCAL_INT_FDC). This condition can never occur because any hwirq higher than those used for local interrupts is a shared interrupt, which gic_irq_domain_map() already handles separately. We therefore remove this check. - The decision of whether to map the interrupt to gic_cpu_pin or timer_cpu_pin can be handled within the existing switch statement in gic_irq_domain_map(), shortening the code a little. The change additionally prepares us nicely for the following patch of the series which would otherwise need to duplicate the check for whether a local interrupt should be percpu_devid or just percpu (ie. the switch statement from gic_irq_domain_map()) in gic_local_irq_domain_map(). Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:44 +00:00
if (!gic_local_irq_is_routable(intr))
return -EPERM;
if (mips_cps_multicluster_cpus()) {
for_each_online_cpu_gic(cpu, &gic_lock)
write_gic_vo_map(mips_gic_vx_map_reg(intr), map);
}
irqchip: mips-gic: Inline gic_local_irq_domain_map() The gic_local_irq_domain_map() function has only one callsite in gic_irq_domain_map(), and the split between the two functions makes it unclear that they duplicate calculations & checks. Inline gic_local_irq_domain_map() into gic_irq_domain_map() in order to clean this up. Doing this makes the following small issues obvious, and the patch tidies them up: - Both functions used GIC_HWIRQ_TO_LOCAL() to convert a hwirq number to a local IRQ number. We now only do this once. Although the compiler ought to have optimised this away before anyway, the change leaves us with less duplicate code. - gic_local_irq_domain_map() had a check for invalid local interrupt numbers (intr > GIC_LOCAL_INT_FDC). This condition can never occur because any hwirq higher than those used for local interrupts is a shared interrupt, which gic_irq_domain_map() already handles separately. We therefore remove this check. - The decision of whether to map the interrupt to gic_cpu_pin or timer_cpu_pin can be handled within the existing switch statement in gic_irq_domain_map(), shortening the code a little. The change additionally prepares us nicely for the following patch of the series which would otherwise need to duplicate the check for whether a local interrupt should be percpu_devid or just percpu (ie. the switch statement from gic_irq_domain_map()) in gic_local_irq_domain_map(). Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:44 +00:00
return 0;
}
static int gic_irq_domain_alloc(struct irq_domain *d, unsigned int virq,
unsigned int nr_irqs, void *arg)
{
struct irq_fwspec *fwspec = arg;
irq_hw_number_t hwirq;
if (fwspec->param[0] == GIC_SHARED)
hwirq = GIC_SHARED_TO_HWIRQ(fwspec->param[1]);
else
hwirq = GIC_LOCAL_TO_HWIRQ(fwspec->param[1]);
return gic_irq_domain_map(d, virq, hwirq);
}
static void gic_irq_domain_free(struct irq_domain *d, unsigned int virq,
unsigned int nr_irqs)
{
}
static const struct irq_domain_ops gic_irq_domain_ops = {
.xlate = gic_irq_domain_xlate,
.alloc = gic_irq_domain_alloc,
.free = gic_irq_domain_free,
.map = gic_irq_domain_map,
};
#ifdef CONFIG_GENERIC_IRQ_IPI
static int gic_ipi_domain_xlate(struct irq_domain *d, struct device_node *ctrlr,
const u32 *intspec, unsigned int intsize,
irq_hw_number_t *out_hwirq,
unsigned int *out_type)
{
/*
* There's nothing to translate here. hwirq is dynamically allocated and
* the irq type is always edge triggered.
* */
*out_hwirq = 0;
*out_type = IRQ_TYPE_EDGE_RISING;
return 0;
}
static int gic_ipi_domain_alloc(struct irq_domain *d, unsigned int virq,
unsigned int nr_irqs, void *arg)
{
struct cpumask *ipimask = arg;
irq_hw_number_t hwirq, base_hwirq;
int cpu, ret, i;
base_hwirq = find_first_bit(ipi_available, gic_shared_intrs);
if (base_hwirq == gic_shared_intrs)
return -ENOMEM;
/* check that we have enough space */
for (i = base_hwirq; i < nr_irqs; i++) {
if (!test_bit(i, ipi_available))
return -EBUSY;
}
bitmap_clear(ipi_available, base_hwirq, nr_irqs);
/* map the hwirq for each cpu consecutively */
i = 0;
for_each_cpu(cpu, ipimask) {
hwirq = GIC_SHARED_TO_HWIRQ(base_hwirq + i);
ret = irq_domain_set_hwirq_and_chip(d, virq + i, hwirq,
&gic_edge_irq_controller,
NULL);
if (ret)
goto error;
ret = irq_domain_set_hwirq_and_chip(d->parent, virq + i, hwirq,
&gic_edge_irq_controller,
NULL);
if (ret)
goto error;
irqchip/mips-gic: Multi-cluster support The MIPS I6500 CPU & CM (Coherence Manager) 3.5 introduce the concept of multiple clusters to the system. In these systems, each cluster contains its own GIC, so the GIC isn't truly global any longer. Access to registers in the GICs of remote clusters is possible using a redirect register block much like the redirect register blocks provided by the CM & CPC, and configured through the same GCR_REDIRECT register that mips_cm_lock_other() abstraction builds upon. It is expected that external interrupts are connected identically on all clusters. That is, if there is a device providing an interrupt connected to GIC interrupt pin 0 then it should be connected to pin 0 of every GIC in the system. For the most part, the GIC can be treated as though it is still truly global, so long as interrupts in the cluster are configured properly. Introduce support for such multi-cluster systems in the MIPS GIC irqchip driver. A newly introduced gic_irq_lock_cluster() function allows: 1) Configure access to a GIC in a remote cluster via the redirect register block, using mips_cm_lock_other(). Or: 2) Detect that the interrupt in question is affine to the local cluster and plain old GIC register access to the GIC in the local cluster should be used. It is possible to access the local cluster's GIC registers via the redirect block, but keeping the special case for them is both good for performance (because we avoid the locking & indirection overhead of using the redirect block) and necessary to maintain compatibility with systems using CM revisions prior to 3.5 which don't support the redirect block. The gic_irq_lock_cluster() function relies upon an IRQs effective affinity in order to discover which cluster the IRQ is affine to. In order to track this & allow it to be updated at an appropriate point during gic_set_affinity() select the generic support for effective affinity using CONFIG_GENERIC_IRQ_EFFECTIVE_AFF_MASK. gic_set_affinity() is the one function which gains much complexity. It now deconfigures routing to any VP(E), ie. CPU, on the old cluster when moving affinity to a new cluster. gic_shared_irq_domain_map() moves its update of the IRQs effective affinity to before its use of gic_irq_lock_cluster(), to ensure that operation is on the cluster the IRQ is affine to. The remaining changes are straightforward use of the gic_irq_lock_cluster() function to select between local cluster & remote cluster code-paths when configuring interrupts. Signed-off-by: Paul Burton <paulburton@kernel.org> Signed-off-by: Chao-ying Fu <cfu@wavecomp.com> Signed-off-by: Dragan Mladjenovic <dragan.mladjenovic@syrmia.com> Signed-off-by: Aleksandar Rikalo <arikalo@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Serge Semin <fancer.lancer@gmail.com> Tested-by: Gregory CLEMENT <gregory.clement@bootlin.com> Link: https://lore.kernel.org/all/20241028175935.51250-5-arikalo@gmail.com
2024-10-28 17:59:26 +00:00
/* Set affinity to cpu. */
irq_data_update_effective_affinity(irq_get_irq_data(virq + i),
cpumask_of(cpu));
ret = irq_set_irq_type(virq + i, IRQ_TYPE_EDGE_RISING);
if (ret)
goto error;
ret = gic_shared_irq_domain_map(d, virq + i, hwirq, cpu);
if (ret)
goto error;
i++;
}
return 0;
error:
bitmap_set(ipi_available, base_hwirq, nr_irqs);
return ret;
}
static void gic_ipi_domain_free(struct irq_domain *d, unsigned int virq,
unsigned int nr_irqs)
{
irq_hw_number_t base_hwirq;
struct irq_data *data;
data = irq_get_irq_data(virq);
if (!data)
return;
base_hwirq = GIC_HWIRQ_TO_SHARED(irqd_to_hwirq(data));
bitmap_set(ipi_available, base_hwirq, nr_irqs);
}
static int gic_ipi_domain_match(struct irq_domain *d, struct device_node *node,
enum irq_domain_bus_token bus_token)
{
bool is_ipi;
switch (bus_token) {
case DOMAIN_BUS_IPI:
is_ipi = d->bus_token == bus_token;
irqchip/mips-gic: Match IPI IRQ domain by bus token only Commit fbde2d7d8290 ("MIPS: Add generic SMP IPI support") introduced code which calls irq_find_matching_host with a NULL node parameter in order to discover IPI IRQ domains which are not associated with the DT root node's interrupt parent. This suggests that implementations of IPI IRQ domains should effectively ignore the node parameter if it is NULL and search purely based upon the bus token. Commit 2af70a962070 ("irqchip/mips-gic: Add a IPI hierarchy domain") did not do this when implementing the GIC IPI IRQ domain, and on MIPS Boston boards this leads to no IPI domain being discovered and a NULL pointer dereference when attempting to send an IPI: CPU 0 Unable to handle kernel paging request at virtual address 0000000000000040, epc == ffffffff8016e70c, ra == ffffffff8010ff5c Oops[#1]: CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.7.0-rc6-00223-gad0d1b6 #945 task: a8000000ff066fc0 ti: a8000000ff068000 task.ti: a8000000ff068000 $ 0 : 0000000000000000 0000000000000001 ffffffff80730000 0000000000000003 $ 4 : 0000000000000000 ffffffff8057e5b0 a800000001e3ee00 0000000000000000 $ 8 : 0000000000000000 0000000000000023 0000000000000001 0000000000000001 $12 : 0000000000000000 ffffffff803323d0 0000000000000000 0000000000000000 $16 : 0000000000000000 0000000000000000 0000000000000001 ffffffff801108fc $20 : 0000000000000000 ffffffff8057e5b0 0000000000000001 0000000000000000 $24 : 0000000000000000 ffffffff8012de28 $28 : a8000000ff068000 a8000000ff06fbc0 0000000000000000 ffffffff8010ff5c Hi : ffffffff8014c174 Lo : a800000001e1e140 epc : ffffffff8016e70c __ipi_send_mask+0x24/0x11c ra : ffffffff8010ff5c mips_smp_send_ipi_mask+0x68/0x178 Status: 140084e2 KX SX UX KERNEL EXL Cause : 00800008 (ExcCode 02) BadVA : 0000000000000040 PrId : 0001a920 (MIPS I6400) Process swapper/0 (pid: 1, threadinfo=a8000000ff068000, task=a8000000ff066fc0, tls=0000000000000000) Stack : 0000000000000000 0000000000000000 0000000000000001 ffffffff801108fc 0000000000000000 ffffffff8057e5b0 0000000000000001 ffffffff8010ff5c 0000000000000001 0000000000000020 0000000000000000 0000000000000000 0000000000000000 ffffffff801108fc 0000000000000000 0000000000000001 0000000000000001 0000000000000000 0000000000000000 ffffffff801865e8 a8000000ff0c7500 a8000000ff06fc90 0000000000000001 0000000000000002 ffffffff801108fc ffffffff801868b8 0000000000000000 ffffffff801108fc 0000000000000000 0000000000000003 ffffffff8068c700 0000000000000001 ffffffff80730000 0000000000000001 a8000000ff00a290 ffffffff80110c50 0000000000000003 a800000001e48308 0000000000000003 0000000000000008 ... Call Trace: [<ffffffff8016e70c>] __ipi_send_mask+0x24/0x11c [<ffffffff8010ff5c>] mips_smp_send_ipi_mask+0x68/0x178 [<ffffffff801865e8>] generic_exec_single+0x150/0x170 [<ffffffff801868b8>] smp_call_function_single+0x108/0x160 [<ffffffff80110c50>] cps_boot_secondary+0x328/0x394 [<ffffffff80110534>] __cpu_up+0x38/0x90 [<ffffffff8012de4c>] bringup_cpu+0x24/0xac [<ffffffff8012df40>] cpuhp_up_callbacks+0x58/0xdc [<ffffffff8012e648>] cpu_up+0x118/0x18c [<ffffffff806dc158>] smp_init+0xbc/0xe8 [<ffffffff806d4c18>] kernel_init_freeable+0xa0/0x228 [<ffffffff8056c908>] kernel_init+0x10/0xf0 [<ffffffff80105098>] ret_from_kernel_thread+0x14/0x1c Fix this by allowing the GIC IPI IRQ domain to match purely based upon the bus token if the node provided is NULL. Fixes: 2af70a962070 ("irqchip/mips-gic: Add a IPI hierarchy domain") Signed-off-by: Paul Burton <paul.burton@imgtec.com> Cc: linux-mips@linux-mips.org Cc: Jason Cooper <jason@lakedaemon.net> Cc: Qais Yousef <qsyousef@gmail.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: stable@vger.kernel.org Link: http://lkml.kernel.org/r/20160705132600.27730-2-paul.burton@imgtec.com Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2016-07-05 13:26:00 +00:00
return (!node || to_of_node(d->fwnode) == node) && is_ipi;
break;
default:
return 0;
}
}
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,
.match = gic_ipi_domain_match,
};
static int gic_register_ipi_domain(struct device_node *node)
{
struct irq_domain *gic_ipi_domain;
unsigned int v[2], num_ipis;
gic_ipi_domain = irq_domain_add_hierarchy(gic_irq_domain,
IRQ_DOMAIN_FLAG_IPI_PER_CPU,
GIC_NUM_LOCAL_INTRS + gic_shared_intrs,
node, &gic_ipi_domain_ops, NULL);
if (!gic_ipi_domain) {
pr_err("Failed to add IPI domain");
return -ENXIO;
}
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)) {
bitmap_set(ipi_resrv, v[0], v[1]);
} else {
/*
* Reserve 2 interrupts per possible CPU/VP for use as IPIs,
* meeting the requirements of arch/mips SMP.
*/
num_ipis = 2 * num_possible_cpus();
bitmap_set(ipi_resrv, gic_shared_intrs - num_ipis, num_ipis);
}
bitmap_copy(ipi_available, ipi_resrv, GIC_MAX_INTRS);
return 0;
}
#else /* !CONFIG_GENERIC_IRQ_IPI */
static inline int gic_register_ipi_domain(struct device_node *node)
{
return 0;
}
#endif /* !CONFIG_GENERIC_IRQ_IPI */
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
static int gic_cpu_startup(unsigned int cpu)
{
/* Enable or disable EIC */
change_gic_vl_ctl(GIC_VX_CTL_EIC,
cpu_has_veic ? GIC_VX_CTL_EIC : 0);
irqchip: mips-gic: Mask local interrupts when CPUs come online We currently walk through the range 0..gic_vpes-1, expecting these values all to be valid Linux CPU numbers to provide to mips_cm_vp_id(), and masking all routable local interrupts during boot. This approach has a few drawbacks: - In multi-cluster systems we won't have access to all CPU's GIC local registers when the driver is probed, since clusters (and their GICs) may be powered down at this point & only brought online later. - In multi-cluster systems we may power down clusters at runtime, for example if we offline all CPUs within it via hotplug, and the cluster's GIC may lose state. We therefore need to reinitialise it when powering back up, which this approach does not take into account. - The range 0..gic_vpes-1 may not all be valid Linux CPU numbers, for example if we run a kernel configured to support fewer CPUs than the system it is running on actually has. In this case we'll get garbage values from mips_cm_vp_id() as we read past the end of the cpu_data array. Fix this and simplify the code somewhat by writing an all-bits-set value to the VP-local reset mask register when a CPU is brought online, before any local interrupts are configured for it. This removes the need for us to access all CPUs during driver probe, removing all of the problems described above. In the name of simplicity we drop the checks for routability of interrupts and simply clear the mask bits for all interrupts. Bits for non-routable local interrupts will have no effect so there's no point performing extra work to avoid modifying them. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:46 +00:00
/* Clear all local IRQ masks (ie. disable all local interrupts) */
write_gic_vl_rmask(~0);
/* Enable desired interrupts */
gic_all_vpes_irq_cpu_online();
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
return 0;
}
static int __init gic_of_init(struct device_node *node,
struct device_node *parent)
{
unsigned int cpu_vec, i, gicconfig, cl, nclusters;
unsigned long reserved;
phys_addr_t gic_base;
struct resource res;
size_t gic_len;
int ret;
/* Find the first available CPU vector. */
i = 0;
reserved = (C_SW0 | C_SW1) >> __ffs(C_SW0);
while (!of_property_read_u32_index(node, "mti,reserved-cpu-vectors",
i++, &cpu_vec))
reserved |= BIT(cpu_vec);
cpu_vec = find_first_zero_bit(&reserved, hweight_long(ST0_IM));
if (cpu_vec == hweight_long(ST0_IM)) {
pr_err("No CPU vectors available\n");
return -ENODEV;
}
if (of_address_to_resource(node, 0, &res)) {
/*
* Probe the CM for the GIC base address if not specified
* in the device-tree.
*/
if (mips_cm_present()) {
gic_base = read_gcr_gic_base() &
~CM_GCR_GIC_BASE_GICEN;
gic_len = 0x20000;
pr_warn("Using inherited base address %pa\n",
&gic_base);
} else {
pr_err("Failed to get memory range\n");
return -ENODEV;
}
} else {
gic_base = res.start;
gic_len = resource_size(&res);
}
if (mips_cm_present()) {
write_gcr_gic_base(gic_base | CM_GCR_GIC_BASE_GICEN);
/* Ensure GIC region is enabled before trying to access it */
__sync();
}
mips_gic_base = ioremap(gic_base, gic_len);
if (!mips_gic_base) {
pr_err("Failed to ioremap gic_base\n");
return -ENOMEM;
}
gicconfig = read_gic_config();
gic_shared_intrs = FIELD_GET(GIC_CONFIG_NUMINTERRUPTS, gicconfig);
gic_shared_intrs = (gic_shared_intrs + 1) * 8;
if (cpu_has_veic) {
/* Always use vector 1 in EIC mode */
gic_cpu_pin = 0;
set_vi_handler(gic_cpu_pin + GIC_PIN_TO_VEC_OFFSET,
__gic_irq_dispatch);
} else {
gic_cpu_pin = cpu_vec - GIC_CPU_PIN_OFFSET;
irq_set_chained_handler(MIPS_CPU_IRQ_BASE + cpu_vec,
gic_irq_dispatch);
}
gic_irq_domain = irq_domain_add_simple(node, GIC_NUM_LOCAL_INTRS +
gic_shared_intrs, 0,
&gic_irq_domain_ops, NULL);
if (!gic_irq_domain) {
pr_err("Failed to add IRQ domain");
return -ENXIO;
}
ret = gic_register_ipi_domain(node);
if (ret)
return ret;
board_bind_eic_interrupt = &gic_bind_eic_interrupt;
/*
* Initialise each cluster's GIC shared registers to sane default
* values.
* Otherwise, the IPI set up will be erased if we move code
* to gic_cpu_startup for each cpu.
*/
nclusters = mips_cps_numclusters();
for (cl = 0; cl < nclusters; cl++) {
if (cl == cpu_cluster(&current_cpu_data)) {
for (i = 0; i < gic_shared_intrs; i++) {
change_gic_pol(i, GIC_POL_ACTIVE_HIGH);
change_gic_trig(i, GIC_TRIG_LEVEL);
write_gic_rmask(i);
}
} else if (mips_cps_numcores(cl) != 0) {
mips_cm_lock_other(cl, 0, 0, CM_GCR_Cx_OTHER_BLOCK_GLOBAL);
for (i = 0; i < gic_shared_intrs; i++) {
change_gic_redir_pol(i, GIC_POL_ACTIVE_HIGH);
change_gic_redir_trig(i, GIC_TRIG_LEVEL);
write_gic_redir_rmask(i);
}
mips_cm_unlock_other();
} else {
pr_warn("No CPU cores on the cluster %d skip it\n", cl);
}
}
irqchip: mips-gic: Use irq_cpu_online to (un)mask all-VP(E) IRQs The gic_all_vpes_local_irq_controller chip currently attempts to operate on all CPUs/VPs in the system when masking or unmasking an interrupt. This has a few drawbacks: - In multi-cluster systems we may not always have access to all CPUs in the system. When all CPUs in a cluster are powered down that cluster's GIC may also power down, in which case we cannot configure its state. - Relatedly, if we power down a cluster after having configured interrupts for CPUs within it then the cluster's GIC may lose state & we need to reconfigure it. The current approach doesn't take this into account. - It's wasteful if we run Linux on fewer VPs than are present in the system. For example if we run a uniprocessor kernel on CPU0 of a system with 16 CPUs then there's no point in us configuring CPUs 1-15. - The implementation is also lacking in that it expects the range 0..gic_vpes-1 to represent valid Linux CPU numbers which may not always be the case - for example if we run on a system with more VPs than the kernel is configured to support. Fix all of these issues by only configuring the affected interrupts for CPUs which are online at the time, and recording the configuration in a new struct gic_all_vpes_chip_data for later use by CPUs being brought online. We register a CPU hotplug state (reusing CPUHP_AP_IRQ_GIC_STARTING which the ARM GIC driver uses, and which seems suitably generic for reuse with the MIPS GIC) and execute irq_cpu_online() in order to configure the interrupts on the newly onlined CPU. Signed-off-by: Paul Burton <paul.burton@mips.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-mips@linux-mips.org Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
2017-10-31 16:41:45 +00:00
return cpuhp_setup_state(CPUHP_AP_IRQ_MIPS_GIC_STARTING,
"irqchip/mips/gic:starting",
gic_cpu_startup, NULL);
}
IRQCHIP_DECLARE(mips_gic, "mti,gic", gic_of_init);