linux-stable-rt/arch/x86/kernel/irq_64.c

143 lines
3.2 KiB
C
Raw Normal View History

/*
* Copyright (C) 1992, 1998 Linus Torvalds, Ingo Molnar
*
* This file contains the lowest level x86_64-specific interrupt
* entry and irq statistics code. All the remaining irq logic is
* done by the generic kernel/irq/ code and in the
* x86_64-specific irq controller code. (e.g. i8259.c and
* io_apic.c.)
*/
#include <linux/kernel_stat.h>
#include <linux/interrupt.h>
#include <linux/seq_file.h>
#include <linux/module.h>
#include <linux/delay.h>
#include <linux/ftrace.h>
#include <linux/uaccess.h>
#include <linux/smp.h>
#include <asm/io_apic.h>
#include <asm/idle.h>
#include <asm/apic.h>
DEFINE_PER_CPU_SHARED_ALIGNED(irq_cpustat_t, irq_stat);
EXPORT_PER_CPU_SYMBOL(irq_stat);
DEFINE_PER_CPU(struct pt_regs *, irq_regs);
EXPORT_PER_CPU_SYMBOL(irq_regs);
/*
* Probabilistic stack overflow check:
*
* Only check the stack in process context, because everything else
* runs on the big interrupt stacks. Checking reliably is too expensive,
* so we just check from interrupts.
*/
static inline void stack_overflow_check(struct pt_regs *regs)
{
#ifdef CONFIG_DEBUG_STACKOVERFLOW
u64 curbase = (u64)task_stack_page(current);
WARN_ONCE(regs->sp >= curbase &&
regs->sp <= curbase + THREAD_SIZE &&
regs->sp < curbase + sizeof(struct thread_info) +
sizeof(struct pt_regs) + 128,
"do_IRQ: %s near stack overflow (cur:%Lx,sp:%lx)\n",
current->comm, curbase, regs->sp);
#endif
}
bool handle_irq(unsigned irq, struct pt_regs *regs)
{
struct irq_desc *desc;
stack_overflow_check(regs);
desc = irq_to_desc(irq);
if (unlikely(!desc))
return false;
generic_handle_irq_desc(irq, desc);
return true;
}
#ifdef CONFIG_HOTPLUG_CPU
/* A cpu has been removed from cpu_online_mask. Reset irq affinities. */
void fixup_irqs(void)
{
unsigned int irq;
static int warned;
struct irq_desc *desc;
for_each_irq_desc(irq, desc) {
x86_64 irq: use mask/unmask and proper locking in fixup_irqs() Force irq migration path during cpu offline, is not using proper locks and irq_chip mask/unmask routines. This will result in some races(especially the device generating the interrupt can see some inconsistent state, resulting in issues like stuck irq,..). Appended patch fixes the issue by taking proper lock and encapsulating irq_chip set_affinity() with a mask() before and an unmask() after. This fixes a MSI irq stuck issue reported by Darrick Wong. There are several more general bugs in this area(irq migration in the process context). For example, 1. Possibility of missing edge triggered irq. 2. Reliable method of migrating level triggered irq in the process context. We plan to look and close these in the near future. Eric says: In addition even with the fix from Suresh there is still at least one nasty hardware race in fixup_irqs(). However we exercise that code path rarely enough that we are unlikely to hit it in the real world, and that race seems to have existed since the code was merged. And a fix for that is not coming soon as it is an open investigation area if we can fix irq migration to work outside of irq context or if we have to rework the requirements imposed by the generic cpu hotplug and layer on fixup_irqs(). So this may come up again. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Reported-and-tested-by: Darrick Wong <djwong@us.ibm.com> Cc: Andi Kleen <ak@suse.de> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-06-26 06:52:35 +08:00
int break_affinity = 0;
int set_affinity = 1;
const struct cpumask *affinity;
x86_64 irq: use mask/unmask and proper locking in fixup_irqs() Force irq migration path during cpu offline, is not using proper locks and irq_chip mask/unmask routines. This will result in some races(especially the device generating the interrupt can see some inconsistent state, resulting in issues like stuck irq,..). Appended patch fixes the issue by taking proper lock and encapsulating irq_chip set_affinity() with a mask() before and an unmask() after. This fixes a MSI irq stuck issue reported by Darrick Wong. There are several more general bugs in this area(irq migration in the process context). For example, 1. Possibility of missing edge triggered irq. 2. Reliable method of migrating level triggered irq in the process context. We plan to look and close these in the near future. Eric says: In addition even with the fix from Suresh there is still at least one nasty hardware race in fixup_irqs(). However we exercise that code path rarely enough that we are unlikely to hit it in the real world, and that race seems to have existed since the code was merged. And a fix for that is not coming soon as it is an open investigation area if we can fix irq migration to work outside of irq context or if we have to rework the requirements imposed by the generic cpu hotplug and layer on fixup_irqs(). So this may come up again. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Reported-and-tested-by: Darrick Wong <djwong@us.ibm.com> Cc: Andi Kleen <ak@suse.de> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-06-26 06:52:35 +08:00
if (!desc)
continue;
if (irq == 2)
continue;
x86_64 irq: use mask/unmask and proper locking in fixup_irqs() Force irq migration path during cpu offline, is not using proper locks and irq_chip mask/unmask routines. This will result in some races(especially the device generating the interrupt can see some inconsistent state, resulting in issues like stuck irq,..). Appended patch fixes the issue by taking proper lock and encapsulating irq_chip set_affinity() with a mask() before and an unmask() after. This fixes a MSI irq stuck issue reported by Darrick Wong. There are several more general bugs in this area(irq migration in the process context). For example, 1. Possibility of missing edge triggered irq. 2. Reliable method of migrating level triggered irq in the process context. We plan to look and close these in the near future. Eric says: In addition even with the fix from Suresh there is still at least one nasty hardware race in fixup_irqs(). However we exercise that code path rarely enough that we are unlikely to hit it in the real world, and that race seems to have existed since the code was merged. And a fix for that is not coming soon as it is an open investigation area if we can fix irq migration to work outside of irq context or if we have to rework the requirements imposed by the generic cpu hotplug and layer on fixup_irqs(). So this may come up again. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Reported-and-tested-by: Darrick Wong <djwong@us.ibm.com> Cc: Andi Kleen <ak@suse.de> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-06-26 06:52:35 +08:00
/* interrupt's are disabled at this point */
spin_lock(&desc->lock);
x86_64 irq: use mask/unmask and proper locking in fixup_irqs() Force irq migration path during cpu offline, is not using proper locks and irq_chip mask/unmask routines. This will result in some races(especially the device generating the interrupt can see some inconsistent state, resulting in issues like stuck irq,..). Appended patch fixes the issue by taking proper lock and encapsulating irq_chip set_affinity() with a mask() before and an unmask() after. This fixes a MSI irq stuck issue reported by Darrick Wong. There are several more general bugs in this area(irq migration in the process context). For example, 1. Possibility of missing edge triggered irq. 2. Reliable method of migrating level triggered irq in the process context. We plan to look and close these in the near future. Eric says: In addition even with the fix from Suresh there is still at least one nasty hardware race in fixup_irqs(). However we exercise that code path rarely enough that we are unlikely to hit it in the real world, and that race seems to have existed since the code was merged. And a fix for that is not coming soon as it is an open investigation area if we can fix irq migration to work outside of irq context or if we have to rework the requirements imposed by the generic cpu hotplug and layer on fixup_irqs(). So this may come up again. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Reported-and-tested-by: Darrick Wong <djwong@us.ibm.com> Cc: Andi Kleen <ak@suse.de> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-06-26 06:52:35 +08:00
affinity = desc->affinity;
x86_64 irq: use mask/unmask and proper locking in fixup_irqs() Force irq migration path during cpu offline, is not using proper locks and irq_chip mask/unmask routines. This will result in some races(especially the device generating the interrupt can see some inconsistent state, resulting in issues like stuck irq,..). Appended patch fixes the issue by taking proper lock and encapsulating irq_chip set_affinity() with a mask() before and an unmask() after. This fixes a MSI irq stuck issue reported by Darrick Wong. There are several more general bugs in this area(irq migration in the process context). For example, 1. Possibility of missing edge triggered irq. 2. Reliable method of migrating level triggered irq in the process context. We plan to look and close these in the near future. Eric says: In addition even with the fix from Suresh there is still at least one nasty hardware race in fixup_irqs(). However we exercise that code path rarely enough that we are unlikely to hit it in the real world, and that race seems to have existed since the code was merged. And a fix for that is not coming soon as it is an open investigation area if we can fix irq migration to work outside of irq context or if we have to rework the requirements imposed by the generic cpu hotplug and layer on fixup_irqs(). So this may come up again. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Reported-and-tested-by: Darrick Wong <djwong@us.ibm.com> Cc: Andi Kleen <ak@suse.de> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-06-26 06:52:35 +08:00
if (!irq_has_action(irq) ||
cpumask_equal(affinity, cpu_online_mask)) {
spin_unlock(&desc->lock);
x86_64 irq: use mask/unmask and proper locking in fixup_irqs() Force irq migration path during cpu offline, is not using proper locks and irq_chip mask/unmask routines. This will result in some races(especially the device generating the interrupt can see some inconsistent state, resulting in issues like stuck irq,..). Appended patch fixes the issue by taking proper lock and encapsulating irq_chip set_affinity() with a mask() before and an unmask() after. This fixes a MSI irq stuck issue reported by Darrick Wong. There are several more general bugs in this area(irq migration in the process context). For example, 1. Possibility of missing edge triggered irq. 2. Reliable method of migrating level triggered irq in the process context. We plan to look and close these in the near future. Eric says: In addition even with the fix from Suresh there is still at least one nasty hardware race in fixup_irqs(). However we exercise that code path rarely enough that we are unlikely to hit it in the real world, and that race seems to have existed since the code was merged. And a fix for that is not coming soon as it is an open investigation area if we can fix irq migration to work outside of irq context or if we have to rework the requirements imposed by the generic cpu hotplug and layer on fixup_irqs(). So this may come up again. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Reported-and-tested-by: Darrick Wong <djwong@us.ibm.com> Cc: Andi Kleen <ak@suse.de> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-06-26 06:52:35 +08:00
continue;
}
if (cpumask_any_and(affinity, cpu_online_mask) >= nr_cpu_ids) {
x86_64 irq: use mask/unmask and proper locking in fixup_irqs() Force irq migration path during cpu offline, is not using proper locks and irq_chip mask/unmask routines. This will result in some races(especially the device generating the interrupt can see some inconsistent state, resulting in issues like stuck irq,..). Appended patch fixes the issue by taking proper lock and encapsulating irq_chip set_affinity() with a mask() before and an unmask() after. This fixes a MSI irq stuck issue reported by Darrick Wong. There are several more general bugs in this area(irq migration in the process context). For example, 1. Possibility of missing edge triggered irq. 2. Reliable method of migrating level triggered irq in the process context. We plan to look and close these in the near future. Eric says: In addition even with the fix from Suresh there is still at least one nasty hardware race in fixup_irqs(). However we exercise that code path rarely enough that we are unlikely to hit it in the real world, and that race seems to have existed since the code was merged. And a fix for that is not coming soon as it is an open investigation area if we can fix irq migration to work outside of irq context or if we have to rework the requirements imposed by the generic cpu hotplug and layer on fixup_irqs(). So this may come up again. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Reported-and-tested-by: Darrick Wong <djwong@us.ibm.com> Cc: Andi Kleen <ak@suse.de> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-06-26 06:52:35 +08:00
break_affinity = 1;
affinity = cpu_all_mask;
}
x86_64 irq: use mask/unmask and proper locking in fixup_irqs() Force irq migration path during cpu offline, is not using proper locks and irq_chip mask/unmask routines. This will result in some races(especially the device generating the interrupt can see some inconsistent state, resulting in issues like stuck irq,..). Appended patch fixes the issue by taking proper lock and encapsulating irq_chip set_affinity() with a mask() before and an unmask() after. This fixes a MSI irq stuck issue reported by Darrick Wong. There are several more general bugs in this area(irq migration in the process context). For example, 1. Possibility of missing edge triggered irq. 2. Reliable method of migrating level triggered irq in the process context. We plan to look and close these in the near future. Eric says: In addition even with the fix from Suresh there is still at least one nasty hardware race in fixup_irqs(). However we exercise that code path rarely enough that we are unlikely to hit it in the real world, and that race seems to have existed since the code was merged. And a fix for that is not coming soon as it is an open investigation area if we can fix irq migration to work outside of irq context or if we have to rework the requirements imposed by the generic cpu hotplug and layer on fixup_irqs(). So this may come up again. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Reported-and-tested-by: Darrick Wong <djwong@us.ibm.com> Cc: Andi Kleen <ak@suse.de> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-06-26 06:52:35 +08:00
if (desc->chip->mask)
desc->chip->mask(irq);
x86_64 irq: use mask/unmask and proper locking in fixup_irqs() Force irq migration path during cpu offline, is not using proper locks and irq_chip mask/unmask routines. This will result in some races(especially the device generating the interrupt can see some inconsistent state, resulting in issues like stuck irq,..). Appended patch fixes the issue by taking proper lock and encapsulating irq_chip set_affinity() with a mask() before and an unmask() after. This fixes a MSI irq stuck issue reported by Darrick Wong. There are several more general bugs in this area(irq migration in the process context). For example, 1. Possibility of missing edge triggered irq. 2. Reliable method of migrating level triggered irq in the process context. We plan to look and close these in the near future. Eric says: In addition even with the fix from Suresh there is still at least one nasty hardware race in fixup_irqs(). However we exercise that code path rarely enough that we are unlikely to hit it in the real world, and that race seems to have existed since the code was merged. And a fix for that is not coming soon as it is an open investigation area if we can fix irq migration to work outside of irq context or if we have to rework the requirements imposed by the generic cpu hotplug and layer on fixup_irqs(). So this may come up again. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Reported-and-tested-by: Darrick Wong <djwong@us.ibm.com> Cc: Andi Kleen <ak@suse.de> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-06-26 06:52:35 +08:00
if (desc->chip->set_affinity)
desc->chip->set_affinity(irq, affinity);
x86_64 irq: use mask/unmask and proper locking in fixup_irqs() Force irq migration path during cpu offline, is not using proper locks and irq_chip mask/unmask routines. This will result in some races(especially the device generating the interrupt can see some inconsistent state, resulting in issues like stuck irq,..). Appended patch fixes the issue by taking proper lock and encapsulating irq_chip set_affinity() with a mask() before and an unmask() after. This fixes a MSI irq stuck issue reported by Darrick Wong. There are several more general bugs in this area(irq migration in the process context). For example, 1. Possibility of missing edge triggered irq. 2. Reliable method of migrating level triggered irq in the process context. We plan to look and close these in the near future. Eric says: In addition even with the fix from Suresh there is still at least one nasty hardware race in fixup_irqs(). However we exercise that code path rarely enough that we are unlikely to hit it in the real world, and that race seems to have existed since the code was merged. And a fix for that is not coming soon as it is an open investigation area if we can fix irq migration to work outside of irq context or if we have to rework the requirements imposed by the generic cpu hotplug and layer on fixup_irqs(). So this may come up again. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Reported-and-tested-by: Darrick Wong <djwong@us.ibm.com> Cc: Andi Kleen <ak@suse.de> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-06-26 06:52:35 +08:00
else if (!(warned++))
set_affinity = 0;
if (desc->chip->unmask)
desc->chip->unmask(irq);
x86_64 irq: use mask/unmask and proper locking in fixup_irqs() Force irq migration path during cpu offline, is not using proper locks and irq_chip mask/unmask routines. This will result in some races(especially the device generating the interrupt can see some inconsistent state, resulting in issues like stuck irq,..). Appended patch fixes the issue by taking proper lock and encapsulating irq_chip set_affinity() with a mask() before and an unmask() after. This fixes a MSI irq stuck issue reported by Darrick Wong. There are several more general bugs in this area(irq migration in the process context). For example, 1. Possibility of missing edge triggered irq. 2. Reliable method of migrating level triggered irq in the process context. We plan to look and close these in the near future. Eric says: In addition even with the fix from Suresh there is still at least one nasty hardware race in fixup_irqs(). However we exercise that code path rarely enough that we are unlikely to hit it in the real world, and that race seems to have existed since the code was merged. And a fix for that is not coming soon as it is an open investigation area if we can fix irq migration to work outside of irq context or if we have to rework the requirements imposed by the generic cpu hotplug and layer on fixup_irqs(). So this may come up again. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Reported-and-tested-by: Darrick Wong <djwong@us.ibm.com> Cc: Andi Kleen <ak@suse.de> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-06-26 06:52:35 +08:00
spin_unlock(&desc->lock);
x86_64 irq: use mask/unmask and proper locking in fixup_irqs() Force irq migration path during cpu offline, is not using proper locks and irq_chip mask/unmask routines. This will result in some races(especially the device generating the interrupt can see some inconsistent state, resulting in issues like stuck irq,..). Appended patch fixes the issue by taking proper lock and encapsulating irq_chip set_affinity() with a mask() before and an unmask() after. This fixes a MSI irq stuck issue reported by Darrick Wong. There are several more general bugs in this area(irq migration in the process context). For example, 1. Possibility of missing edge triggered irq. 2. Reliable method of migrating level triggered irq in the process context. We plan to look and close these in the near future. Eric says: In addition even with the fix from Suresh there is still at least one nasty hardware race in fixup_irqs(). However we exercise that code path rarely enough that we are unlikely to hit it in the real world, and that race seems to have existed since the code was merged. And a fix for that is not coming soon as it is an open investigation area if we can fix irq migration to work outside of irq context or if we have to rework the requirements imposed by the generic cpu hotplug and layer on fixup_irqs(). So this may come up again. Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Reported-and-tested-by: Darrick Wong <djwong@us.ibm.com> Cc: Andi Kleen <ak@suse.de> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-06-26 06:52:35 +08:00
if (break_affinity && set_affinity)
printk("Broke affinity for irq %i\n", irq);
else if (!set_affinity)
printk("Cannot set affinity for irq %i\n", irq);
}
/* That doesn't seem sufficient. Give it 1ms. */
local_irq_enable();
mdelay(1);
local_irq_disable();
}
#endif
extern void call_softirq(void);
asmlinkage void do_softirq(void)
{
__u32 pending;
unsigned long flags;
if (in_interrupt())
return;
local_irq_save(flags);
pending = local_softirq_pending();
/* Switch to interrupt stack */
if (pending) {
call_softirq();
WARN_ON_ONCE(softirq_count());
}
local_irq_restore(flags);
}