[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <492B77F3.3020303@kernel.org>
Date: Mon, 24 Nov 2008 19:58:43 -0800
From: Yinghai Lu <yinghai@...nel.org>
To: Ingo Molnar <mingo@...e.hu>, Thomas Gleixner <tglx@...utronix.de>,
"H. Peter Anvin" <hpa@...or.com>,
Andrew Morton <akpm@...ux-foundation.org>
CC: "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: [PATCH 2/2] irq: move irq_desc according to smp_affinity v2
impact: new feature move irq_desc with sparseirq
if CONFIG_MOVE_IRQ_DESC is set
make irq_desc to go with affinity aka irq_desc moving etc
call move_irq_desc in irq_complete_move()
legacy irq_desc is not moved, because they are allocated via static array
for logical apic mode, need to add move_desc_in_progress_in_same_domain. otherwise it will not get moved. ==> also could need two phase to get irq_desc moved.
for example: 0xff is old affinity, and need to set 0xf, and then set to 0xf0.
[ or we need to change domain definition to cpus on the same node ? ]
LBSuse:~ # cat /proc/irq/22/smp_affinity
00000000,00000000,00000000,000000ff
LBSuse:~ # echo f > /proc/irq/22/smp_affinity
LBSuse:~ # cat /proc/irq/22/smp_affinity
00000000,00000000,00000000,0000000f
LBSuse:~ # tail /var/log/messages
...
Oct 27 12:35:34 LBSuse kernel: klogd 1.4.1, log source = /proc/kmsg started.
Oct 27 12:35:34 LBSuse kernel: eth0: no IPv6 routers present
LBSuse:~ # echo f0 > /proc/irq/22/smp_affinity
LBSuse:~ # tail /var/log/messages
Oct 27 12:35:34 LBSuse kernel: klogd 1.4.1, log source = /proc/kmsg started.
Oct 27 12:35:34 LBSuse kernel: eth0: no IPv6 routers present
Oct 27 12:36:46 LBSuse kernel: move irq_desc for 22 aka 0x16 to cpu 7 node 1
Oct 27 12:36:46 LBSuse kernel: alloc kstat_irqs on cpu 7 node 1
Oct 27 12:36:46 LBSuse kernel: alloc irq_cfg on cpu 7 node 1
Oct 27 12:36:46 LBSuse kernel: alloc irq_2_pin on cpu 7 node 1
so assume the user space program should update /proc/irq/XX/smp_affinity to 03 or 0f at first on boot
or we change irq_default_affinity ?
for physical apic is much simple
on 4 sockets 16 cores system
irq_desc is moving..
when
# echo 10 > /proc/irq/134483967/smp_affinity
# echo 100 > /proc/irq/134483967/smp_affinity
# echo 1000 > /proc/irq/134483967/smp_affinity
got
Nov 9 21:39:51 LBSuse kernel: move irq_desc for 134483967 aka 0x8040fff to cpu 4 node 1
Nov 9 21:39:51 LBSuse kernel: alloc kstat_irqs on cpu 4 node 1
Nov 9 21:39:51 LBSuse kernel: alloc irq_cfg on cpu 4 node 1
Nov 9 21:40:05 LBSuse kernel: move irq_desc for 134483967 aka 0x8040fff to cpu 8 node 2
Nov 9 21:40:05 LBSuse kernel: alloc kstat_irqs on cpu 8 node 2
Nov 9 21:40:05 LBSuse kernel: alloc irq_cfg on cpu 8 node 2
Nov 9 21:40:18 LBSuse kernel: move irq_desc for 134483967 aka 0x8040fff to cpu 12 node 3
Nov 9 21:40:18 LBSuse kernel: alloc kstat_irqs on cpu 12 node 3
Nov 9 21:40:18 LBSuse kernel: alloc irq_cfg on cpu 12 node 3
Signed-off-by: Yinghai Lu <yinghai@...nel.org>
---
arch/x86/Kconfig | 9 ++
arch/x86/kernel/io_apic.c | 143 +++++++++++++++++++++++++++++++++++++++++++++-
kernel/irq/handle.c | 111 +++++++++++++++++++++++++++++++++++
3 files changed, 262 insertions(+), 1 deletion(-)
Index: linux-2.6/arch/x86/Kconfig
===================================================================
--- linux-2.6.orig/arch/x86/Kconfig
+++ linux-2.6/arch/x86/Kconfig
@@ -251,6 +251,15 @@ config SPARSE_IRQ
If you don't know what to do here, say Y.
+config MOVE_IRQ_DESC
+ bool "Move irq desc when changing irq smp_affinity"
+ depends on SPARSE_IRQ && SMP
+ default y
+ help
+ This enables moving irq_desc to cpu/node that irq will use handled.
+
+ If you don't know what to do here, say Y.
+
config X86_FIND_SMP_CONFIG
def_bool y
depends on X86_MPPARSE || X86_VOYAGER
Index: linux-2.6/arch/x86/kernel/io_apic.c
===================================================================
--- linux-2.6.orig/arch/x86/kernel/io_apic.c
+++ linux-2.6/arch/x86/kernel/io_apic.c
@@ -141,6 +141,9 @@ struct irq_cfg {
unsigned move_cleanup_count;
u8 vector;
u8 move_in_progress : 1;
+#ifdef CONFIG_MOVE_IRQ_DESC
+ u8 move_desc_in_progress_in_same_domain : 1;
+#endif
};
/* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */
@@ -230,6 +233,122 @@ void arch_init_chip_data(struct irq_desc
}
}
+#ifdef CONFIG_MOVE_IRQ_DESC
+
+static void init_copy_irq_2_pin(struct irq_cfg *old_cfg, struct irq_cfg *cfg,
+ int cpu)
+{
+ struct irq_pin_list *old_entry, *head, *tail, *entry;
+
+ cfg->irq_2_pin = NULL;
+ old_entry = old_cfg->irq_2_pin;
+ if (!old_entry)
+ return;
+
+ entry = get_one_free_irq_2_pin(cpu);
+ if (!entry)
+ return;
+
+ entry->apic = old_entry->apic;
+ entry->pin = old_entry->pin;
+ head = entry;
+ tail = entry;
+ old_entry = old_entry->next;
+
+ while (old_entry) {
+ entry = get_one_free_irq_2_pin(cpu);
+ if (!entry) {
+ entry = head;
+ while (entry) {
+ head = entry->next;
+ kfree(entry);
+ entry = head;
+ }
+ /* still use the old one */
+ return;
+ }
+ entry->apic = old_entry->apic;
+ entry->pin = old_entry->pin;
+ tail->next = entry;
+ tail = entry;
+ old_entry = old_entry->next;
+ }
+
+ tail->next = NULL;
+ cfg->irq_2_pin = head;
+}
+
+static void free_irq_2_pin(struct irq_cfg *old_cfg, struct irq_cfg *cfg)
+{
+ struct irq_pin_list *entry, *next;
+
+ if (old_cfg->irq_2_pin == cfg->irq_2_pin)
+ return;
+
+ entry = old_cfg->irq_2_pin;
+
+ while (entry) {
+ next = entry->next;
+ kfree(entry);
+ entry = next;
+ }
+ old_cfg->irq_2_pin = NULL;
+}
+
+void arch_init_copy_chip_data(struct irq_desc *old_desc,
+ struct irq_desc *desc, int cpu)
+{
+ struct irq_cfg *cfg;
+ struct irq_cfg *old_cfg;
+
+ cfg = get_one_free_irq_cfg(cpu);
+
+ if (!cfg)
+ return;
+
+ desc->chip_data = cfg;
+
+ old_cfg = old_desc->chip_data;
+
+ memcpy(cfg, old_cfg, sizeof(struct irq_cfg));
+
+ init_copy_irq_2_pin(old_cfg, cfg, cpu);
+}
+
+static void free_irq_cfg(struct irq_cfg *old_cfg)
+{
+ kfree(old_cfg);
+}
+
+void arch_free_chip_data(struct irq_desc *old_desc, struct irq_desc *desc)
+{
+ struct irq_cfg *old_cfg, *cfg;
+
+ old_cfg = old_desc->chip_data;
+ cfg = desc->chip_data;
+
+ if (old_cfg == cfg)
+ return;
+
+ if (old_cfg) {
+ free_irq_2_pin(old_cfg, cfg);
+ free_irq_cfg(old_cfg);
+ old_desc->chip_data = NULL;
+ }
+}
+
+static void set_extra_move_desc(struct irq_desc *desc, cpumask_t mask)
+{
+ struct irq_cfg *cfg = desc->chip_data;
+
+ if (!cfg->move_in_progress) {
+ /* it means that domain is not changed */
+ if (!cpus_intersects(desc->affinity, mask))
+ cfg->move_desc_in_progress_in_same_domain = 1;
+ }
+}
+#endif
+
#else
static struct irq_cfg *irq_cfg(unsigned int irq)
{
@@ -238,9 +357,11 @@ static struct irq_cfg *irq_cfg(unsigned
#endif
+#ifndef CONFIG_MOVE_IRQ_DESC
static inline void set_extra_move_desc(struct irq_desc *desc, cpumask_t mask)
{
}
+#endif
struct io_apic {
unsigned int index;
@@ -2358,14 +2479,34 @@ static void irq_complete_move(struct irq
struct irq_cfg *cfg = desc->chip_data;
unsigned vector, me;
- if (likely(!cfg->move_in_progress))
+ if (likely(!cfg->move_in_progress)) {
+#ifdef CONFIG_MOVE_IRQ_DESC
+ if (likely(!cfg->move_desc_in_progress_in_same_domain))
+ return;
+
+ /* domain is not change, but affinity is changed */
+ me = smp_processor_id();
+ if (cpu_isset(me, desc->affinity)) {
+ *descp = desc = move_irq_desc(desc, me);
+ /* get the new one */
+ cfg = desc->chip_data;
+ cfg->move_desc_in_progress_in_same_domain = 0;
+ }
+#endif
return;
+ }
vector = ~get_irq_regs()->orig_ax;
me = smp_processor_id();
if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) {
cpumask_t cleanup_mask;
+#ifdef CONFIG_MOVE_IRQ_DESC
+ *descp = desc = move_irq_desc(desc, me);
+ /* get the new one */
+ cfg = desc->chip_data;
+#endif
+
cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map);
cfg->move_cleanup_count = cpus_weight(cleanup_mask);
send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR);
Index: linux-2.6/kernel/irq/handle.c
===================================================================
--- linux-2.6.orig/kernel/irq/handle.c
+++ linux-2.6/kernel/irq/handle.c
@@ -90,6 +90,32 @@ static void init_kstat_irqs(struct irq_d
desc->kstat_irqs = (unsigned int *)ptr;
}
+#ifdef CONFIG_MOVE_IRQ_DESC
+static void init_copy_kstat_irqs(struct irq_desc *old_desc, struct irq_desc *desc,
+ int cpu, int nr)
+{
+ unsigned long bytes;
+
+ init_kstat_irqs(desc, cpu, nr);
+
+ if (desc->kstat_irqs != old_desc->kstat_irqs) {
+ /* Compute how many bytes we need per irq and allocate them */
+ bytes = nr * sizeof(unsigned int);
+
+ memcpy(desc->kstat_irqs, old_desc->kstat_irqs, bytes);
+ }
+}
+
+static void free_kstat_irqs(struct irq_desc *old_desc, struct irq_desc *desc)
+{
+ if (old_desc->kstat_irqs == desc->kstat_irqs)
+ return;
+
+ kfree(old_desc->kstat_irqs);
+ old_desc->kstat_irqs = NULL;
+}
+#endif
+
void __attribute__((weak)) arch_init_chip_data(struct irq_desc *desc, int cpu)
{
}
@@ -110,6 +136,23 @@ static void init_one_irq_desc(int irq, s
arch_init_chip_data(desc, cpu);
}
+#ifdef CONFIG_MOVE_IRQ_DESC
+static void init_copy_one_irq_desc(int irq, struct irq_desc *old_desc,
+ struct irq_desc *desc, int cpu)
+{
+ memcpy(desc, old_desc, sizeof(struct irq_desc));
+ desc->cpu = cpu;
+ lockdep_set_class(&desc->lock, &irq_desc_lock_class);
+ init_copy_kstat_irqs(old_desc, desc, cpu, nr_cpu_ids);
+ arch_init_copy_chip_data(old_desc, desc, cpu);
+}
+
+static void free_one_irq_desc(struct irq_desc *old_desc, struct irq_desc *desc)
+{
+ free_kstat_irqs(old_desc, desc);
+ arch_free_chip_data(old_desc, desc);
+}
+#endif
/*
* Protect the sparse_irqs:
*/
@@ -212,6 +255,74 @@ out_unlock:
return desc;
}
+#ifdef CONFIG_MOVE_IRQ_DESC
+static struct irq_desc *__real_move_irq_desc(struct irq_desc *old_desc,
+ int cpu)
+{
+ struct irq_desc *desc;
+ unsigned int irq;
+ unsigned long flags;
+ int node;
+
+ irq = old_desc->irq;
+
+ spin_lock_irqsave(&sparse_irq_lock, flags);
+
+ /* We have to check it to avoid races with another CPU */
+ desc = irq_desc_ptrs[irq];
+
+ if (desc && old_desc != desc)
+ goto out_unlock;
+
+ node = cpu_to_node(cpu);
+ desc = kzalloc_node(sizeof(*desc), GFP_ATOMIC, node);
+ printk(KERN_DEBUG " move irq_desc for %d to cpu %d node %d\n",
+ irq, cpu, node);
+ if (!desc) {
+ printk(KERN_ERR "can not get new irq_desc for moving\n");
+ /* still use old one */
+ desc = old_desc;
+ goto out_unlock;
+ }
+ init_copy_one_irq_desc(irq, old_desc, desc, cpu);
+
+ irq_desc_ptrs[irq] = desc;
+ list_replace_rcu(&old_desc->list, &desc->list);
+
+ /* free the old one */
+ free_one_irq_desc(old_desc, desc);
+ kfree(old_desc);
+
+out_unlock:
+ spin_unlock_irqrestore(&sparse_irq_lock, flags);
+
+ return desc;
+}
+
+struct irq_desc *move_irq_desc(struct irq_desc *desc, int cpu)
+{
+ int old_cpu;
+ int node, old_node;
+
+ /* those all static, do move them */
+ if (desc->irq < NR_IRQS_LEGACY)
+ return desc;
+
+ old_cpu = desc->cpu;
+ printk(KERN_DEBUG "try to move irq_desc from cpu %d to %d\n", old_cpu, cpu);
+ if (old_cpu != cpu) {
+ node = cpu_to_node(cpu);
+ old_node = cpu_to_node(old_cpu);
+ if (old_node != node)
+ desc = __real_move_irq_desc(desc, cpu);
+ else
+ desc->cpu = cpu;
+ }
+
+ return desc;
+}
+#endif
+
#else
struct irq_desc irq_desc[NR_IRQS] __cacheline_aligned_in_smp = {
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists