Commit 2f75d9e1 authored by Thomas Gleixner's avatar Thomas Gleixner
Browse files

genirq: Implement bitmap matrix allocator 



Implement the infrastructure for a simple bitmap based allocator, which
will replace the x86 vector allocator. It's in the core code as other
architectures might be able to reuse/extend it. For now it only implements
allocations for single CPUs, but it's simple to add multi CPU allocation
support if required.

The concept is rather simple:

 Global information:
 	system_vector bitmap
	global accounting

 PerCPU information:
 	allocation bitmap
	managed allocation bitmap
	local accounting

The system vector bitmap is used to exclude vectors system wide from the
allocation space.

The allocation bitmap is used to keep track of per cpu used vectors.

The managed allocation bitmap is used to reserve vectors for managed
interrupts.

When a regular (non managed) interrupt allocation happens then the
following rule applies:

      tmpmap = system_map | alloc_map | managed_map
      find_zero_bit(tmpmap)

Oring the bitmaps together gives the real available space. The same rule
applies for reserving a managed interrupt vector. But contrary to the
regular interrupts the reservation only marks the bit in the managed map
and therefor excludes it from the regular allocations. The managed map is
only cleaned out when the a managed interrupt is completely released and it
stays alive accross CPU offline/online operations.

For managed interrupt allocations the rule is:

      tmpmap = managed_map & ~alloc_map
      find_first_bit(tmpmap)

This returns the first bit which is in the managed map, but not yet
allocated in the allocation map. The allocation marks it in the allocation
map and hands it back to the caller for use.

The rest of the code are helper functions to handle the various
requirements and the accounting which are necessary to replace the x86
vector allocation code. The result is a single patch as the evolution of
this infrastructure cannot be represented in bits and pieces.

Signed-off-by: default avatarThomas Gleixner <tglx@linutronix.de>
Tested-by: default avatarJuergen Gross <jgross@suse.com>
Tested-by: default avatarYu Chen <yu.c.chen@intel.com>
Acked-by: default avatarJuergen Gross <jgross@suse.com>
Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Alok Kataria <akataria@vmware.com>
Cc: Joerg Roedel <joro@8bytes.org>
Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Rui Zhang <rui.zhang@intel.com>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Chris Metcalf <cmetcalf@mellanox.com>
Cc: Len Brown <lenb@kernel.org>
Link: https://lkml.kernel.org/r/20170913213153.185437174@linutronix.de
parent 22d0b12f

include/linux/irq.h

+22 −0
Original line number Diff line number Diff line
@@ -1113,6 +1113,28 @@ static inline u32 irq_reg_readl(struct irq_chip_generic *gc, 
		return readl(gc->reg_base + reg_offset);

}



struct irq_matrix;

struct irq_matrix *irq_alloc_matrix(unsigned int matrix_bits,

				    unsigned int alloc_start,

				    unsigned int alloc_end);

void irq_matrix_online(struct irq_matrix *m);

void irq_matrix_offline(struct irq_matrix *m);

void irq_matrix_assign_system(struct irq_matrix *m, unsigned int bit, bool replace);

int irq_matrix_reserve_managed(struct irq_matrix *m, const struct cpumask *msk);

void irq_matrix_remove_managed(struct irq_matrix *m, const struct cpumask *msk);

int irq_matrix_alloc_managed(struct irq_matrix *m, unsigned int cpu);

void irq_matrix_reserve(struct irq_matrix *m);

void irq_matrix_remove_reserved(struct irq_matrix *m);

int irq_matrix_alloc(struct irq_matrix *m, const struct cpumask *msk,

		     bool reserved, unsigned int *mapped_cpu);

void irq_matrix_free(struct irq_matrix *m, unsigned int cpu,

		     unsigned int bit, bool managed);

void irq_matrix_assign(struct irq_matrix *m, unsigned int bit);

unsigned int irq_matrix_available(struct irq_matrix *m, bool cpudown);

unsigned int irq_matrix_allocated(struct irq_matrix *m);

unsigned int irq_matrix_reserved(struct irq_matrix *m);

void irq_matrix_debug_show(struct seq_file *sf, struct irq_matrix *m, int ind);



/* Contrary to Linux irqs, for hardware irqs the irq number 0 is valid */

#define INVALID_HWIRQ	(~0UL)

irq_hw_number_t ipi_get_hwirq(unsigned int irq, unsigned int cpu);

kernel/irq/Kconfig

+3 −0
Original line number Diff line number Diff line
@@ -97,6 +97,9 @@ config HANDLE_DOMAIN_IRQ 
config IRQ_TIMINGS

	bool



config GENERIC_IRQ_MATRIX_ALLOCATOR

	bool



config IRQ_DOMAIN_DEBUG

	bool "Expose hardware/virtual IRQ mapping via debugfs"

	depends on IRQ_DOMAIN && DEBUG_FS

kernel/irq/Makefile

+1 −0
Original line number Diff line number Diff line
@@ -13,3 +13,4 @@ obj-$(CONFIG_GENERIC_MSI_IRQ) += msi.o 
obj-$(CONFIG_GENERIC_IRQ_IPI) += ipi.o

obj-$(CONFIG_SMP) += affinity.o

obj-$(CONFIG_GENERIC_IRQ_DEBUGFS) += debugfs.o

obj-$(CONFIG_GENERIC_IRQ_MATRIX_ALLOCATOR) += matrix.o

kernel/irq/matrix.c

0 → 100644
+428 −0
Original line number Diff line number Diff line 
/*

 * Copyright (C) 2017 Thomas Gleixner <tglx@linutronix.de>

 *

 * SPDX-License-Identifier: GPL-2.0

 */

#include <linux/spinlock.h>

#include <linux/seq_file.h>

#include <linux/bitmap.h>

#include <linux/percpu.h>

#include <linux/cpu.h>

#include <linux/irq.h>



#define IRQ_MATRIX_SIZE	(BITS_TO_LONGS(IRQ_MATRIX_BITS) * sizeof(unsigned long))



struct cpumap {

	unsigned int		available;

	unsigned int		allocated;

	unsigned int		managed;

	bool			online;

	unsigned long		alloc_map[IRQ_MATRIX_SIZE];

	unsigned long		managed_map[IRQ_MATRIX_SIZE];

};



struct irq_matrix {

	unsigned int		matrix_bits;

	unsigned int		alloc_start;

	unsigned int		alloc_end;

	unsigned int		alloc_size;

	unsigned int		global_available;

	unsigned int		global_reserved;

	unsigned int		systembits_inalloc;

	unsigned int		total_allocated;

	unsigned int		online_maps;

	struct cpumap __percpu	*maps;

	unsigned long		scratch_map[IRQ_MATRIX_SIZE];

	unsigned long		system_map[IRQ_MATRIX_SIZE];

};



/**

 * irq_alloc_matrix - Allocate a irq_matrix structure and initialize it

 * @matrix_bits:	Number of matrix bits must be <= IRQ_MATRIX_BITS

 * @alloc_start:	From which bit the allocation search starts

 * @alloc_end:		At which bit the allocation search ends, i.e first

 *			invalid bit

 */

__init struct irq_matrix *irq_alloc_matrix(unsigned int matrix_bits,

					   unsigned int alloc_start,

					   unsigned int alloc_end)

{

	struct irq_matrix *m;



	if (matrix_bits > IRQ_MATRIX_BITS)

		return NULL;



	m = kzalloc(sizeof(*m), GFP_KERNEL);

	if (!m)

		return NULL;



	m->matrix_bits = matrix_bits;

	m->alloc_start = alloc_start;

	m->alloc_end = alloc_end;

	m->alloc_size = alloc_end - alloc_start;

	m->maps = alloc_percpu(*m->maps);

	if (!m->maps) {

		kfree(m);

		return NULL;

	}

	return m;

}



/**

 * irq_matrix_online - Bring the local CPU matrix online

 * @m:		Matrix pointer

 */

void irq_matrix_online(struct irq_matrix *m)

{

	struct cpumap *cm = this_cpu_ptr(m->maps);



	BUG_ON(cm->online);



	bitmap_zero(cm->alloc_map, m->matrix_bits);

	cm->available = m->alloc_size - (cm->managed + m->systembits_inalloc);

	cm->allocated = 0;

	m->global_available += cm->available;

	cm->online = true;

	m->online_maps++;

}



/**

 * irq_matrix_offline - Bring the local CPU matrix offline

 * @m:		Matrix pointer

 */

void irq_matrix_offline(struct irq_matrix *m)

{

	struct cpumap *cm = this_cpu_ptr(m->maps);



	/* Update the global available size */

	m->global_available -= cm->available;

	cm->online = false;

	m->online_maps--;

}



static unsigned int matrix_alloc_area(struct irq_matrix *m, struct cpumap *cm,

				      unsigned int num, bool managed)

{

	unsigned int area, start = m->alloc_start;

	unsigned int end = m->alloc_end;



	bitmap_or(m->scratch_map, cm->managed_map, m->system_map, end);

	bitmap_or(m->scratch_map, m->scratch_map, cm->alloc_map, end);

	area = bitmap_find_next_zero_area(m->scratch_map, end, start, num, 0);

	if (area >= end)

		return area;

	if (managed)

		bitmap_set(cm->managed_map, area, num);

	else

		bitmap_set(cm->alloc_map, area, num);

	return area;

}



/**

 * irq_matrix_assign_system - Assign system wide entry in the matrix

 * @m:		Matrix pointer

 * @bit:	Which bit to reserve

 * @replace:	Replace an already allocated vector with a system

 *		vector at the same bit position.

 *

 * The BUG_ON()s below are on purpose. If this goes wrong in the

 * early boot process, then the chance to survive is about zero.

 * If this happens when the system is life, it's not much better.

 */

void irq_matrix_assign_system(struct irq_matrix *m, unsigned int bit,

			      bool replace)

{

	struct cpumap *cm = this_cpu_ptr(m->maps);



	BUG_ON(bit > m->matrix_bits);

	BUG_ON(m->online_maps > 1 || (m->online_maps && !replace));



	set_bit(bit, m->system_map);

	if (replace) {

		BUG_ON(!test_and_clear_bit(bit, cm->alloc_map));

		cm->allocated--;

		m->total_allocated--;

	}

	if (bit >= m->alloc_start && bit < m->alloc_end)

		m->systembits_inalloc++;

}



/**

 * irq_matrix_reserve_managed - Reserve a managed interrupt in a CPU map

 * @m:		Matrix pointer

 * @msk:	On which CPUs the bits should be reserved.

 *

 * Can be called for offline CPUs. Note, this will only reserve one bit

 * on all CPUs in @msk, but it's not guaranteed that the bits are at the

 * same offset on all CPUs

 */

int irq_matrix_reserve_managed(struct irq_matrix *m, const struct cpumask *msk)

{

	unsigned int cpu, failed_cpu;



	for_each_cpu(cpu, msk) {

		struct cpumap *cm = per_cpu_ptr(m->maps, cpu);

		unsigned int bit;



		bit = matrix_alloc_area(m, cm, 1, true);

		if (bit >= m->alloc_end)

			goto cleanup;

		cm->managed++;

		if (cm->online) {

			cm->available--;

			m->global_available--;

		}

	}

	return 0;

cleanup:

	failed_cpu = cpu;

	for_each_cpu(cpu, msk) {

		if (cpu == failed_cpu)

			break;

		irq_matrix_remove_managed(m, cpumask_of(cpu));

	}

	return -ENOSPC;

}



/**

 * irq_matrix_remove_managed - Remove managed interrupts in a CPU map

 * @m:		Matrix pointer

 * @msk:	On which CPUs the bits should be removed

 *

 * Can be called for offline CPUs

 *

 * This removes not allocated managed interrupts from the map. It does

 * not matter which one because the managed interrupts free their

 * allocation when they shut down. If not, the accounting is screwed,

 * but all what can be done at this point is warn about it.

 */

void irq_matrix_remove_managed(struct irq_matrix *m, const struct cpumask *msk)

{

	unsigned int cpu;



	for_each_cpu(cpu, msk) {

		struct cpumap *cm = per_cpu_ptr(m->maps, cpu);

		unsigned int bit, end = m->alloc_end;



		if (WARN_ON_ONCE(!cm->managed))

			continue;



		/* Get managed bit which are not allocated */

		bitmap_andnot(m->scratch_map, cm->managed_map, cm->alloc_map, end);



		bit = find_first_bit(m->scratch_map, end);

		if (WARN_ON_ONCE(bit >= end))

			continue;



		clear_bit(bit, cm->managed_map);



		cm->managed--;

		if (cm->online) {

			cm->available++;

			m->global_available++;

		}

	}

}



/**

 * irq_matrix_alloc_managed - Allocate a managed interrupt in a CPU map

 * @m:		Matrix pointer

 * @cpu:	On which CPU the interrupt should be allocated

 */

int irq_matrix_alloc_managed(struct irq_matrix *m, unsigned int cpu)

{

	struct cpumap *cm = per_cpu_ptr(m->maps, cpu);

	unsigned int bit, end = m->alloc_end;



	/* Get managed bit which are not allocated */

	bitmap_andnot(m->scratch_map, cm->managed_map, cm->alloc_map, end);

	bit = find_first_bit(m->scratch_map, end);

	if (bit >= end)

		return -ENOSPC;

	set_bit(bit, cm->alloc_map);

	cm->allocated++;

	m->total_allocated++;

	return bit;

}



/**

 * irq_matrix_assign - Assign a preallocated interrupt in the local CPU map

 * @m:		Matrix pointer

 * @bit:	Which bit to mark

 *

 * This should only be used to mark preallocated vectors

 */

void irq_matrix_assign(struct irq_matrix *m, unsigned int bit)

{

	struct cpumap *cm = this_cpu_ptr(m->maps);



	if (WARN_ON_ONCE(bit < m->alloc_start || bit >= m->alloc_end))

		return;

	if (WARN_ON_ONCE(test_and_set_bit(bit, cm->alloc_map)))

		return;

	cm->allocated++;

	m->total_allocated++;

	cm->available--;

	m->global_available--;

}



/**

 * irq_matrix_reserve - Reserve interrupts

 * @m:		Matrix pointer

 *

 * This is merily a book keeping call. It increments the number of globally

 * reserved interrupt bits w/o actually allocating them. This allows to

 * setup interrupt descriptors w/o assigning low level resources to it.

 * The actual allocation happens when the interrupt gets activated.

 */

void irq_matrix_reserve(struct irq_matrix *m)

{

	if (m->global_reserved <= m->global_available &&

	    m->global_reserved + 1 > m->global_available)

		pr_warn("Interrupt reservation exceeds available resources\n");



	m->global_reserved++;

}



/**

 * irq_matrix_remove_reserved - Remove interrupt reservation

 * @m:		Matrix pointer

 *

 * This is merily a book keeping call. It decrements the number of globally

 * reserved interrupt bits. This is used to undo irq_matrix_reserve() when the

 * interrupt was never in use and a real vector allocated, which undid the

 * reservation.

 */

void irq_matrix_remove_reserved(struct irq_matrix *m)

{

	m->global_reserved--;

}



/**

 * irq_matrix_alloc - Allocate a regular interrupt in a CPU map

 * @m:		Matrix pointer

 * @msk:	Which CPUs to search in

 * @reserved:	Allocate previously reserved interrupts

 * @mapped_cpu: Pointer to store the CPU for which the irq was allocated

 */

int irq_matrix_alloc(struct irq_matrix *m, const struct cpumask *msk,

		     bool reserved, unsigned int *mapped_cpu)

{

	unsigned int cpu;



	for_each_cpu(cpu, msk) {

		struct cpumap *cm = per_cpu_ptr(m->maps, cpu);

		unsigned int bit;



		if (!cm->online)

			continue;



		bit = matrix_alloc_area(m, cm, 1, false);

		if (bit < m->alloc_end) {

			cm->allocated++;

			cm->available--;

			m->total_allocated++;

			m->global_available--;

			if (reserved)

				m->global_reserved--;

			*mapped_cpu = cpu;

			return bit;

		}

	}

	return -ENOSPC;

}



/**

 * irq_matrix_free - Free allocated interrupt in the matrix

 * @m:		Matrix pointer

 * @cpu:	Which CPU map needs be updated

 * @bit:	The bit to remove

 * @managed:	If true, the interrupt is managed and not accounted

 *		as available.

 */

void irq_matrix_free(struct irq_matrix *m, unsigned int cpu,

		     unsigned int bit, bool managed)

{

	struct cpumap *cm = per_cpu_ptr(m->maps, cpu);



	if (WARN_ON_ONCE(bit < m->alloc_start || bit >= m->alloc_end))

		return;



	if (cm->online) {

		clear_bit(bit, cm->alloc_map);

		cm->allocated--;

		m->total_allocated--;

		if (!managed) {

			cm->available++;

			m->global_available++;

		}

	}

}



/**

 * irq_matrix_available - Get the number of globally available irqs

 * @m:		Pointer to the matrix to query

 * @cpudown:	If true, the local CPU is about to go down, adjust

 *		the number of available irqs accordingly

 */

unsigned int irq_matrix_available(struct irq_matrix *m, bool cpudown)

{

	struct cpumap *cm = this_cpu_ptr(m->maps);



	return m->global_available - cpudown ? cm->available : 0;

}



/**

 * irq_matrix_reserved - Get the number of globally reserved irqs

 * @m:		Pointer to the matrix to query

 */

unsigned int irq_matrix_reserved(struct irq_matrix *m)

{

	return m->global_reserved;

}



/**

 * irq_matrix_allocated - Get the number of allocated irqs on the local cpu

 * @m:		Pointer to the matrix to search

 *

 * This returns number of allocated irqs

 */

unsigned int irq_matrix_allocated(struct irq_matrix *m)

{

	struct cpumap *cm = this_cpu_ptr(m->maps);



	return cm->allocated;

}



#ifdef CONFIG_GENERIC_IRQ_DEBUGFS

/**

 * irq_matrix_debug_show - Show detailed allocation information

 * @sf:		Pointer to the seq_file to print to

 * @m:		Pointer to the matrix allocator

 * @ind:	Indentation for the print format

 *

 * Note, this is a lockless snapshot.

 */

void irq_matrix_debug_show(struct seq_file *sf, struct irq_matrix *m, int ind)

{

	unsigned int nsys = bitmap_weight(m->system_map, m->matrix_bits);

	int cpu;



	seq_printf(sf, "Online bitmaps:   %6u\n", m->online_maps);

	seq_printf(sf, "Global available: %6u\n", m->global_available);

	seq_printf(sf, "Global reserved:  %6u\n", m->global_reserved);

	seq_printf(sf, "Total allocated:  %6u\n", m->total_allocated);

	seq_printf(sf, "System: %u: %*pbl\n", nsys, m->matrix_bits,

		   m->system_map);

	seq_printf(sf, "%*s| CPU | avl | man | act | vectors\n", ind, " ");

	cpus_read_lock();

	for_each_online_cpu(cpu) {

		struct cpumap *cm = per_cpu_ptr(m->maps, cpu);



		seq_printf(sf, "%*s %4d  %4u  %4u  %4u  %*pbl\n", ind, " ",

			   cpu, cm->available, cm->managed, cm->allocated,

			   m->matrix_bits, cm->alloc_map);

	}

	cpus_read_unlock();

}

#endif

CRA Git | Maintained and supported by SUSTech CRA and CCSE