linux/arch/ia64/kernel/process.c

742 lines
19 KiB
C
Raw Normal View History

/*
* Architecture-specific setup.
*
* Copyright (C) 1998-2003 Hewlett-Packard Co
* David Mosberger-Tang <davidm@hpl.hp.com>
* 04/11/17 Ashok Raj <ashok.raj@intel.com> Added CPU Hotplug Support
[IA64] Extend notify_die() hooks for IA64 notify_die() added for MCA_{MONARCH,SLAVE,RENDEZVOUS}_{ENTER,PROCESS,LEAVE} and INIT_{MONARCH,SLAVE}_{ENTER,PROCESS,LEAVE}. We need multiple notification points for these events because they can take many seconds to run which has nasty effects on the behaviour of the rest of the system. DIE_SS replaced by a generic DIE_FAULT which checks the vector number, to allow interception of faults other than SS. DIE_MACHINE_{HALT,RESTART} added to allow last minute close down processing, especially when the halt/restart routines are called from error handlers. DIE_OOPS added. The check for kprobe's break numbers has been moved from traps.c to kprobes.c, allowing DIE_BREAK to be used for any additional break numbers, i.e. it is no longer kprobes specific. Hooks for kernel debuggers and kernel dumpers added, ENTER and LEAVE. Both of these disable the system for long periods which impact on watchdogs and heartbeat systems in general. More patches to come that use these events to reset watchdogs and heartbeats. unregister_die_notifier() added and both routines exported. Requested by Dean Nelson. Lock removed from {un,}register_die_notifier. notifier_chain_register() already takes a lock. Also the generic notifier chain locking is being reworked to distinguish between callbacks that can block and those that cannot, the lock in {un,}register_die_notifier would interfere with that change. http://marc.theaimsgroup.com/?l=linux-kernel&m=113018709002036&w=2 Leading white space removed from arch/ia64/kernel/kprobes.c. Typo in mca.c in original version of this patch found & fixed by Dean Nelson. Signed-off-by: Keith Owens <kaos@sgi.com> Acked-by: Dean Nelson <dcn@sgi.com> Acked-by: Anil Keshavamurthy <anil.s.keshavamurthy@intel.com> Signed-off-by: Tony Luck <tony.luck@intel.com>
2005-11-08 03:27:13 +08:00
*
* 2005-10-07 Keith Owens <kaos@sgi.com>
* Add notify_die() hooks.
*/
#include <linux/cpu.h>
#include <linux/pm.h>
#include <linux/elf.h>
#include <linux/errno.h>
#include <linux/kallsyms.h>
#include <linux/kernel.h>
#include <linux/mm.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 16:04:11 +08:00
#include <linux/slab.h>
#include <linux/module.h>
#include <linux/notifier.h>
#include <linux/personality.h>
#include <linux/sched.h>
#include <linux/stddef.h>
#include <linux/thread_info.h>
#include <linux/unistd.h>
#include <linux/efi.h>
#include <linux/interrupt.h>
#include <linux/delay.h>
#include <linux/kdebug.h>
#include <linux/utsname.h>
#include <linux/tracehook.h>
#include <linux/rcupdate.h>
#include <asm/cpu.h>
#include <asm/delay.h>
#include <asm/elf.h>
#include <asm/irq.h>
#include <asm/kexec.h>
#include <asm/pgalloc.h>
#include <asm/processor.h>
#include <asm/sal.h>
#include <asm/switch_to.h>
#include <asm/tlbflush.h>
#include <asm/uaccess.h>
#include <asm/unwind.h>
#include <asm/user.h>
#include "entry.h"
#ifdef CONFIG_PERFMON
# include <asm/perfmon.h>
#endif
#include "sigframe.h"
void (*ia64_mark_idle)(int);
unsigned long boot_option_idle_override = IDLE_NO_OVERRIDE;
EXPORT_SYMBOL(boot_option_idle_override);
void (*pm_idle) (void);
EXPORT_SYMBOL(pm_idle);
void (*pm_power_off) (void);
EXPORT_SYMBOL(pm_power_off);
void
ia64_do_show_stack (struct unw_frame_info *info, void *arg)
{
unsigned long ip, sp, bsp;
char buf[128]; /* don't make it so big that it overflows the stack! */
printk("\nCall Trace:\n");
do {
unw_get_ip(info, &ip);
if (ip == 0)
break;
unw_get_sp(info, &sp);
unw_get_bsp(info, &bsp);
snprintf(buf, sizeof(buf),
" [<%016lx>] %%s\n"
" sp=%016lx bsp=%016lx\n",
ip, sp, bsp);
print_symbol(buf, ip);
} while (unw_unwind(info) >= 0);
}
void
show_stack (struct task_struct *task, unsigned long *sp)
{
if (!task)
unw_init_running(ia64_do_show_stack, NULL);
else {
struct unw_frame_info info;
unw_init_from_blocked_task(&info, task);
ia64_do_show_stack(&info, NULL);
}
}
void
dump_stack (void)
{
show_stack(NULL, NULL);
}
EXPORT_SYMBOL(dump_stack);
void
show_regs (struct pt_regs *regs)
{
unsigned long ip = regs->cr_iip + ia64_psr(regs)->ri;
print_modules();
printk("\nPid: %d, CPU %d, comm: %20s\n", task_pid_nr(current),
smp_processor_id(), current->comm);
printk("psr : %016lx ifs : %016lx ip : [<%016lx>] %s (%s)\n",
regs->cr_ipsr, regs->cr_ifs, ip, print_tainted(),
init_utsname()->release);
print_symbol("ip is at %s\n", ip);
printk("unat: %016lx pfs : %016lx rsc : %016lx\n",
regs->ar_unat, regs->ar_pfs, regs->ar_rsc);
printk("rnat: %016lx bsps: %016lx pr : %016lx\n",
regs->ar_rnat, regs->ar_bspstore, regs->pr);
printk("ldrs: %016lx ccv : %016lx fpsr: %016lx\n",
regs->loadrs, regs->ar_ccv, regs->ar_fpsr);
printk("csd : %016lx ssd : %016lx\n", regs->ar_csd, regs->ar_ssd);
printk("b0 : %016lx b6 : %016lx b7 : %016lx\n", regs->b0, regs->b6, regs->b7);
printk("f6 : %05lx%016lx f7 : %05lx%016lx\n",
regs->f6.u.bits[1], regs->f6.u.bits[0],
regs->f7.u.bits[1], regs->f7.u.bits[0]);
printk("f8 : %05lx%016lx f9 : %05lx%016lx\n",
regs->f8.u.bits[1], regs->f8.u.bits[0],
regs->f9.u.bits[1], regs->f9.u.bits[0]);
printk("f10 : %05lx%016lx f11 : %05lx%016lx\n",
regs->f10.u.bits[1], regs->f10.u.bits[0],
regs->f11.u.bits[1], regs->f11.u.bits[0]);
printk("r1 : %016lx r2 : %016lx r3 : %016lx\n", regs->r1, regs->r2, regs->r3);
printk("r8 : %016lx r9 : %016lx r10 : %016lx\n", regs->r8, regs->r9, regs->r10);
printk("r11 : %016lx r12 : %016lx r13 : %016lx\n", regs->r11, regs->r12, regs->r13);
printk("r14 : %016lx r15 : %016lx r16 : %016lx\n", regs->r14, regs->r15, regs->r16);
printk("r17 : %016lx r18 : %016lx r19 : %016lx\n", regs->r17, regs->r18, regs->r19);
printk("r20 : %016lx r21 : %016lx r22 : %016lx\n", regs->r20, regs->r21, regs->r22);
printk("r23 : %016lx r24 : %016lx r25 : %016lx\n", regs->r23, regs->r24, regs->r25);
printk("r26 : %016lx r27 : %016lx r28 : %016lx\n", regs->r26, regs->r27, regs->r28);
printk("r29 : %016lx r30 : %016lx r31 : %016lx\n", regs->r29, regs->r30, regs->r31);
if (user_mode(regs)) {
/* print the stacked registers */
unsigned long val, *bsp, ndirty;
int i, sof, is_nat = 0;
sof = regs->cr_ifs & 0x7f; /* size of frame */
ndirty = (regs->loadrs >> 19);
bsp = ia64_rse_skip_regs((unsigned long *) regs->ar_bspstore, ndirty);
for (i = 0; i < sof; ++i) {
get_user(val, (unsigned long __user *) ia64_rse_skip_regs(bsp, i));
printk("r%-3u:%c%016lx%s", 32 + i, is_nat ? '*' : ' ', val,
((i == sof - 1) || (i % 3) == 2) ? "\n" : " ");
}
} else
show_stack(NULL, NULL);
}
/* local support for deprecated console_print */
void
console_print(const char *s)
{
printk(KERN_EMERG "%s", s);
}
void
do_notify_resume_user(sigset_t *unused, struct sigscratch *scr, long in_syscall)
{
if (fsys_mode(current, &scr->pt)) {
/*
* defer signal-handling etc. until we return to
* privilege-level 0.
*/
if (!ia64_psr(&scr->pt)->lp)
ia64_psr(&scr->pt)->lp = 1;
return;
}
#ifdef CONFIG_PERFMON
if (current->thread.pfm_needs_checking)
/*
* Note: pfm_handle_work() allow us to call it with interrupts
* disabled, and may enable interrupts within the function.
*/
pfm_handle_work();
#endif
/* deal with pending signal delivery */
if (test_thread_flag(TIF_SIGPENDING)) {
local_irq_enable(); /* force interrupt enable */
ia64_do_signal(scr, in_syscall);
}
if (test_and_clear_thread_flag(TIF_NOTIFY_RESUME)) {
local_irq_enable(); /* force interrupt enable */
tracehook_notify_resume(&scr->pt);
}
/* copy user rbs to kernel rbs */
if (unlikely(test_thread_flag(TIF_RESTORE_RSE))) {
local_irq_enable(); /* force interrupt enable */
ia64_sync_krbs();
}
local_irq_disable(); /* force interrupt disable */
}
[IA64] perfmon & PAL_HALT again The pmu_active test is based on the values of PSR.up. THIS IS THE PROBLEM as it does not take into account the lazy restore logic which is as follow (simplified): context switch out: save PMDs clear psr.up release ownership context switch in: if (ctx->last_cpu == smp_processor_id() && ctx->cpu_activation == cpu_activation) { set psr.up return } restore PMD restore PMC ctx->last_cpu = smp_processor_id(); ctx->activation = ++cpu_activation; set psr.up The key here is that on context switch out, we clear psr.up and on context switch in we check if nobody else used the PMU on that processor since last time we came. In that case, we assume the PMD/PMC are ours and we simply reactivate. The Caliper problem is that between the moment we context switch out and the moment we come back, nobody effectively used the PMU BUT the processor went idle. Normally this would have no incidence but PAL_HALT does alter the PMU registers. In default_idle(), the test on psr.up is not strong enough to cover this case and we go into PAL which trashed the PMU resgisters. When we come back we falsely assume that this is our state yet it is corrupted. Very nasty indeed. To avoid the problem it is necessary to forbid going to PAL_HALT as soon as perfmon installs some valid state in the PMU registers. This happens with an application attaches a context to a thread or CPU. It is not enough to check the psr/dcr bits. Hence I propose the attached patch. It adds a callback in process.c to modify the condition to enter PAL on idle. Basically, now it is conditional to pal_halt=1 AND perfmon saying it is okay. Signed-off-by: Tony Luck <tony.luck@intel.com>
2005-04-12 04:45:00 +08:00
static int pal_halt = 1;
static int can_do_pal_halt = 1;
static int __init nohalt_setup(char * str)
{
pal_halt = can_do_pal_halt = 0;
return 1;
}
__setup("nohalt", nohalt_setup);
void
[IA64] perfmon & PAL_HALT again The pmu_active test is based on the values of PSR.up. THIS IS THE PROBLEM as it does not take into account the lazy restore logic which is as follow (simplified): context switch out: save PMDs clear psr.up release ownership context switch in: if (ctx->last_cpu == smp_processor_id() && ctx->cpu_activation == cpu_activation) { set psr.up return } restore PMD restore PMC ctx->last_cpu = smp_processor_id(); ctx->activation = ++cpu_activation; set psr.up The key here is that on context switch out, we clear psr.up and on context switch in we check if nobody else used the PMU on that processor since last time we came. In that case, we assume the PMD/PMC are ours and we simply reactivate. The Caliper problem is that between the moment we context switch out and the moment we come back, nobody effectively used the PMU BUT the processor went idle. Normally this would have no incidence but PAL_HALT does alter the PMU registers. In default_idle(), the test on psr.up is not strong enough to cover this case and we go into PAL which trashed the PMU resgisters. When we come back we falsely assume that this is our state yet it is corrupted. Very nasty indeed. To avoid the problem it is necessary to forbid going to PAL_HALT as soon as perfmon installs some valid state in the PMU registers. This happens with an application attaches a context to a thread or CPU. It is not enough to check the psr/dcr bits. Hence I propose the attached patch. It adds a callback in process.c to modify the condition to enter PAL on idle. Basically, now it is conditional to pal_halt=1 AND perfmon saying it is okay. Signed-off-by: Tony Luck <tony.luck@intel.com>
2005-04-12 04:45:00 +08:00
update_pal_halt_status(int status)
{
can_do_pal_halt = pal_halt && status;
}
/*
* We use this if we don't have any better idle routine..
*/
void
default_idle (void)
{
local_irq_enable();
[PATCH] sched: resched and cpu_idle rework Make some changes to the NEED_RESCHED and POLLING_NRFLAG to reduce confusion, and make their semantics rigid. Improves efficiency of resched_task and some cpu_idle routines. * In resched_task: - TIF_NEED_RESCHED is only cleared with the task's runqueue lock held, and as we hold it during resched_task, then there is no need for an atomic test and set there. The only other time this should be set is when the task's quantum expires, in the timer interrupt - this is protected against because the rq lock is irq-safe. - If TIF_NEED_RESCHED is set, then we don't need to do anything. It won't get unset until the task get's schedule()d off. - If we are running on the same CPU as the task we resched, then set TIF_NEED_RESCHED and no further action is required. - If we are running on another CPU, and TIF_POLLING_NRFLAG is *not* set after TIF_NEED_RESCHED has been set, then we need to send an IPI. Using these rules, we are able to remove the test and set operation in resched_task, and make clear the previously vague semantics of POLLING_NRFLAG. * In idle routines: - Enter cpu_idle with preempt disabled. When the need_resched() condition becomes true, explicitly call schedule(). This makes things a bit clearer (IMO), but haven't updated all architectures yet. - Many do a test and clear of TIF_NEED_RESCHED for some reason. According to the resched_task rules, this isn't needed (and actually breaks the assumption that TIF_NEED_RESCHED is only cleared with the runqueue lock held). So remove that. Generally one less locked memory op when switching to the idle thread. - Many idle routines clear TIF_POLLING_NRFLAG, and only set it in the inner most polling idle loops. The above resched_task semantics allow it to be set until before the last time need_resched() is checked before going into a halt requiring interrupt wakeup. Many idle routines simply never enter such a halt, and so POLLING_NRFLAG can be always left set, completely eliminating resched IPIs when rescheduling the idle task. POLLING_NRFLAG width can be increased, to reduce the chance of resched IPIs. Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Ingo Molnar <mingo@elte.hu> Cc: Con Kolivas <kernel@kolivas.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-11-09 13:39:04 +08:00
while (!need_resched()) {
if (can_do_pal_halt) {
local_irq_disable();
if (!need_resched()) {
safe_halt();
}
local_irq_enable();
} else
cpu_relax();
[PATCH] sched: resched and cpu_idle rework Make some changes to the NEED_RESCHED and POLLING_NRFLAG to reduce confusion, and make their semantics rigid. Improves efficiency of resched_task and some cpu_idle routines. * In resched_task: - TIF_NEED_RESCHED is only cleared with the task's runqueue lock held, and as we hold it during resched_task, then there is no need for an atomic test and set there. The only other time this should be set is when the task's quantum expires, in the timer interrupt - this is protected against because the rq lock is irq-safe. - If TIF_NEED_RESCHED is set, then we don't need to do anything. It won't get unset until the task get's schedule()d off. - If we are running on the same CPU as the task we resched, then set TIF_NEED_RESCHED and no further action is required. - If we are running on another CPU, and TIF_POLLING_NRFLAG is *not* set after TIF_NEED_RESCHED has been set, then we need to send an IPI. Using these rules, we are able to remove the test and set operation in resched_task, and make clear the previously vague semantics of POLLING_NRFLAG. * In idle routines: - Enter cpu_idle with preempt disabled. When the need_resched() condition becomes true, explicitly call schedule(). This makes things a bit clearer (IMO), but haven't updated all architectures yet. - Many do a test and clear of TIF_NEED_RESCHED for some reason. According to the resched_task rules, this isn't needed (and actually breaks the assumption that TIF_NEED_RESCHED is only cleared with the runqueue lock held). So remove that. Generally one less locked memory op when switching to the idle thread. - Many idle routines clear TIF_POLLING_NRFLAG, and only set it in the inner most polling idle loops. The above resched_task semantics allow it to be set until before the last time need_resched() is checked before going into a halt requiring interrupt wakeup. Many idle routines simply never enter such a halt, and so POLLING_NRFLAG can be always left set, completely eliminating resched IPIs when rescheduling the idle task. POLLING_NRFLAG width can be increased, to reduce the chance of resched IPIs. Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Ingo Molnar <mingo@elte.hu> Cc: Con Kolivas <kernel@kolivas.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-11-09 13:39:04 +08:00
}
}
#ifdef CONFIG_HOTPLUG_CPU
/* We don't actually take CPU down, just spin without interrupts. */
static inline void play_dead(void)
{
unsigned int this_cpu = smp_processor_id();
/* Ack it */
__get_cpu_var(cpu_state) = CPU_DEAD;
max_xtp();
local_irq_disable();
idle_task_exit();
ia64_jump_to_sal(&sal_boot_rendez_state[this_cpu]);
/*
* The above is a point of no-return, the processor is
* expected to be in SAL loop now.
*/
BUG();
}
#else
static inline void play_dead(void)
{
BUG();
}
#endif /* CONFIG_HOTPLUG_CPU */
void __attribute__((noreturn))
cpu_idle (void)
{
void (*mark_idle)(int) = ia64_mark_idle;
[PATCH] sched: resched and cpu_idle rework Make some changes to the NEED_RESCHED and POLLING_NRFLAG to reduce confusion, and make their semantics rigid. Improves efficiency of resched_task and some cpu_idle routines. * In resched_task: - TIF_NEED_RESCHED is only cleared with the task's runqueue lock held, and as we hold it during resched_task, then there is no need for an atomic test and set there. The only other time this should be set is when the task's quantum expires, in the timer interrupt - this is protected against because the rq lock is irq-safe. - If TIF_NEED_RESCHED is set, then we don't need to do anything. It won't get unset until the task get's schedule()d off. - If we are running on the same CPU as the task we resched, then set TIF_NEED_RESCHED and no further action is required. - If we are running on another CPU, and TIF_POLLING_NRFLAG is *not* set after TIF_NEED_RESCHED has been set, then we need to send an IPI. Using these rules, we are able to remove the test and set operation in resched_task, and make clear the previously vague semantics of POLLING_NRFLAG. * In idle routines: - Enter cpu_idle with preempt disabled. When the need_resched() condition becomes true, explicitly call schedule(). This makes things a bit clearer (IMO), but haven't updated all architectures yet. - Many do a test and clear of TIF_NEED_RESCHED for some reason. According to the resched_task rules, this isn't needed (and actually breaks the assumption that TIF_NEED_RESCHED is only cleared with the runqueue lock held). So remove that. Generally one less locked memory op when switching to the idle thread. - Many idle routines clear TIF_POLLING_NRFLAG, and only set it in the inner most polling idle loops. The above resched_task semantics allow it to be set until before the last time need_resched() is checked before going into a halt requiring interrupt wakeup. Many idle routines simply never enter such a halt, and so POLLING_NRFLAG can be always left set, completely eliminating resched IPIs when rescheduling the idle task. POLLING_NRFLAG width can be increased, to reduce the chance of resched IPIs. Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Ingo Molnar <mingo@elte.hu> Cc: Con Kolivas <kernel@kolivas.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-11-09 13:39:04 +08:00
int cpu = smp_processor_id();
/* endless idle loop with no priority at all */
while (1) {
rcu_idle_enter();
[PATCH] sched: fix bad missed wakeups in the i386, x86_64, ia64, ACPI and APM idle code Fernando Lopez-Lezcano reported frequent scheduling latencies and audio xruns starting at the 2.6.18-rt kernel, and those problems persisted all until current -rt kernels. The latencies were serious and unjustified by system load, often in the milliseconds range. After a patient and heroic multi-month effort of Fernando, where he tested dozens of kernels, tried various configs, boot options, test-patches of mine and provided latency traces of those incidents, the following 'smoking gun' trace was captured by him: _------=> CPU# / _-----=> irqs-off | / _----=> need-resched || / _---=> hardirq/softirq ||| / _--=> preempt-depth |||| / ||||| delay cmd pid ||||| time | caller \ / ||||| \ | / IRQ_19-1479 1D..1 0us : __trace_start_sched_wakeup (try_to_wake_up) IRQ_19-1479 1D..1 0us : __trace_start_sched_wakeup <<...>-5856> (37 0) IRQ_19-1479 1D..1 0us : __trace_start_sched_wakeup (c01262ba 0 0) IRQ_19-1479 1D..1 0us : resched_task (try_to_wake_up) IRQ_19-1479 1D..1 0us : __spin_unlock_irqrestore (try_to_wake_up) ... <idle>-0 1...1 11us!: default_idle (cpu_idle) ... <idle>-0 0Dn.1 602us : smp_apic_timer_interrupt (c0103baf 1 0) ... <...>-5856 0D..2 618us : __switch_to (__schedule) <...>-5856 0D..2 618us : __schedule <<idle>-0> (20 162) <...>-5856 0D..2 619us : __spin_unlock_irq (__schedule) <...>-5856 0...1 619us : trace_stop_sched_switched (__schedule) <...>-5856 0D..1 619us : trace_stop_sched_switched <<...>-5856> (37 0) what is visible in this trace is that CPU#1 ran try_to_wake_up() for PID:5856, it placed PID:5856 on CPU#0's runqueue and ran resched_task() for CPU#0. But it decided to not send an IPI that no CPU - due to TS_POLLING. But CPU#0 never woke up after its NEED_RESCHED bit was set, and only rescheduled to PID:5856 upon the next lapic timer IRQ. The result was a 600+ usecs latency and a missed wakeup! the bug turned out to be an idle-wakeup bug introduced into the mainline kernel this summer via an optimization in the x86_64 tree: commit 495ab9c045e1b0e5c82951b762257fe1c9d81564 Author: Andi Kleen <ak@suse.de> Date: Mon Jun 26 13:59:11 2006 +0200 [PATCH] i386/x86-64/ia64: Move polling flag into thread_info_status During some profiling I noticed that default_idle causes a lot of memory traffic. I think that is caused by the atomic operations to clear/set the polling flag in thread_info. There is actually no reason to make this atomic - only the idle thread does it to itself, other CPUs only read it. So I moved it into ti->status. the problem is this type of change: if (!hlt_counter && boot_cpu_data.hlt_works_ok) { - clear_thread_flag(TIF_POLLING_NRFLAG); + current_thread_info()->status &= ~TS_POLLING; smp_mb__after_clear_bit(); while (!need_resched()) { local_irq_disable(); this changes clear_thread_flag() to an explicit clearing of TS_POLLING. clear_thread_flag() is defined as: clear_bit(flag, &ti->flags); and clear_bit() is a LOCK-ed atomic instruction on all x86 platforms: static inline void clear_bit(int nr, volatile unsigned long * addr) { __asm__ __volatile__( LOCK_PREFIX "btrl %1,%0" hence smp_mb__after_clear_bit() is defined as a simple compile barrier: #define smp_mb__after_clear_bit() barrier() but the explicit TS_POLLING clearing introduced by the patch: + current_thread_info()->status &= ~TS_POLLING; is not an atomic op! So the clearing of the TS_POLLING bit is freely reorderable with the reading of the NEED_RESCHED bit - and both now reside in different memory addresses. CPU idle wakeup very much depends on ordered memory ops, the clearing of the TS_POLLING flag must always be done before we test need_resched() and hit the idle instruction(s). [Symmetrically, the wakeup code needs to set NEED_RESCHED before it tests the TS_POLLING flag, so memory ordering is paramount.] Fernando's dual-core Athlon64 system has a sufficiently advanced memory ordering model so that it triggered this scenario very often. ( And it also turned out that the reason why these latencies never triggered on my testsystems is that i routinely use idle=poll, which was the only idle variant not affected by this bug. ) The fix is to change the smp_mb__after_clear_bit() to an smp_mb(), to act as an absolute barrier between the TS_POLLING write and the NEED_RESCHED read. This affects almost all idling methods (default, ACPI, APM), on all 3 x86 architectures: i386, x86_64, ia64. Signed-off-by: Ingo Molnar <mingo@elte.hu> Tested-by: Fernando Lopez-Lezcano <nando@ccrma.Stanford.EDU> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-22 17:11:56 +08:00
if (can_do_pal_halt) {
current_thread_info()->status &= ~TS_POLLING;
[PATCH] sched: fix bad missed wakeups in the i386, x86_64, ia64, ACPI and APM idle code Fernando Lopez-Lezcano reported frequent scheduling latencies and audio xruns starting at the 2.6.18-rt kernel, and those problems persisted all until current -rt kernels. The latencies were serious and unjustified by system load, often in the milliseconds range. After a patient and heroic multi-month effort of Fernando, where he tested dozens of kernels, tried various configs, boot options, test-patches of mine and provided latency traces of those incidents, the following 'smoking gun' trace was captured by him: _------=> CPU# / _-----=> irqs-off | / _----=> need-resched || / _---=> hardirq/softirq ||| / _--=> preempt-depth |||| / ||||| delay cmd pid ||||| time | caller \ / ||||| \ | / IRQ_19-1479 1D..1 0us : __trace_start_sched_wakeup (try_to_wake_up) IRQ_19-1479 1D..1 0us : __trace_start_sched_wakeup <<...>-5856> (37 0) IRQ_19-1479 1D..1 0us : __trace_start_sched_wakeup (c01262ba 0 0) IRQ_19-1479 1D..1 0us : resched_task (try_to_wake_up) IRQ_19-1479 1D..1 0us : __spin_unlock_irqrestore (try_to_wake_up) ... <idle>-0 1...1 11us!: default_idle (cpu_idle) ... <idle>-0 0Dn.1 602us : smp_apic_timer_interrupt (c0103baf 1 0) ... <...>-5856 0D..2 618us : __switch_to (__schedule) <...>-5856 0D..2 618us : __schedule <<idle>-0> (20 162) <...>-5856 0D..2 619us : __spin_unlock_irq (__schedule) <...>-5856 0...1 619us : trace_stop_sched_switched (__schedule) <...>-5856 0D..1 619us : trace_stop_sched_switched <<...>-5856> (37 0) what is visible in this trace is that CPU#1 ran try_to_wake_up() for PID:5856, it placed PID:5856 on CPU#0's runqueue and ran resched_task() for CPU#0. But it decided to not send an IPI that no CPU - due to TS_POLLING. But CPU#0 never woke up after its NEED_RESCHED bit was set, and only rescheduled to PID:5856 upon the next lapic timer IRQ. The result was a 600+ usecs latency and a missed wakeup! the bug turned out to be an idle-wakeup bug introduced into the mainline kernel this summer via an optimization in the x86_64 tree: commit 495ab9c045e1b0e5c82951b762257fe1c9d81564 Author: Andi Kleen <ak@suse.de> Date: Mon Jun 26 13:59:11 2006 +0200 [PATCH] i386/x86-64/ia64: Move polling flag into thread_info_status During some profiling I noticed that default_idle causes a lot of memory traffic. I think that is caused by the atomic operations to clear/set the polling flag in thread_info. There is actually no reason to make this atomic - only the idle thread does it to itself, other CPUs only read it. So I moved it into ti->status. the problem is this type of change: if (!hlt_counter && boot_cpu_data.hlt_works_ok) { - clear_thread_flag(TIF_POLLING_NRFLAG); + current_thread_info()->status &= ~TS_POLLING; smp_mb__after_clear_bit(); while (!need_resched()) { local_irq_disable(); this changes clear_thread_flag() to an explicit clearing of TS_POLLING. clear_thread_flag() is defined as: clear_bit(flag, &ti->flags); and clear_bit() is a LOCK-ed atomic instruction on all x86 platforms: static inline void clear_bit(int nr, volatile unsigned long * addr) { __asm__ __volatile__( LOCK_PREFIX "btrl %1,%0" hence smp_mb__after_clear_bit() is defined as a simple compile barrier: #define smp_mb__after_clear_bit() barrier() but the explicit TS_POLLING clearing introduced by the patch: + current_thread_info()->status &= ~TS_POLLING; is not an atomic op! So the clearing of the TS_POLLING bit is freely reorderable with the reading of the NEED_RESCHED bit - and both now reside in different memory addresses. CPU idle wakeup very much depends on ordered memory ops, the clearing of the TS_POLLING flag must always be done before we test need_resched() and hit the idle instruction(s). [Symmetrically, the wakeup code needs to set NEED_RESCHED before it tests the TS_POLLING flag, so memory ordering is paramount.] Fernando's dual-core Athlon64 system has a sufficiently advanced memory ordering model so that it triggered this scenario very often. ( And it also turned out that the reason why these latencies never triggered on my testsystems is that i routinely use idle=poll, which was the only idle variant not affected by this bug. ) The fix is to change the smp_mb__after_clear_bit() to an smp_mb(), to act as an absolute barrier between the TS_POLLING write and the NEED_RESCHED read. This affects almost all idling methods (default, ACPI, APM), on all 3 x86 architectures: i386, x86_64, ia64. Signed-off-by: Ingo Molnar <mingo@elte.hu> Tested-by: Fernando Lopez-Lezcano <nando@ccrma.Stanford.EDU> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-22 17:11:56 +08:00
/*
* TS_POLLING-cleared state must be visible before we
* test NEED_RESCHED:
*/
smp_mb();
} else {
current_thread_info()->status |= TS_POLLING;
[PATCH] sched: fix bad missed wakeups in the i386, x86_64, ia64, ACPI and APM idle code Fernando Lopez-Lezcano reported frequent scheduling latencies and audio xruns starting at the 2.6.18-rt kernel, and those problems persisted all until current -rt kernels. The latencies were serious and unjustified by system load, often in the milliseconds range. After a patient and heroic multi-month effort of Fernando, where he tested dozens of kernels, tried various configs, boot options, test-patches of mine and provided latency traces of those incidents, the following 'smoking gun' trace was captured by him: _------=> CPU# / _-----=> irqs-off | / _----=> need-resched || / _---=> hardirq/softirq ||| / _--=> preempt-depth |||| / ||||| delay cmd pid ||||| time | caller \ / ||||| \ | / IRQ_19-1479 1D..1 0us : __trace_start_sched_wakeup (try_to_wake_up) IRQ_19-1479 1D..1 0us : __trace_start_sched_wakeup <<...>-5856> (37 0) IRQ_19-1479 1D..1 0us : __trace_start_sched_wakeup (c01262ba 0 0) IRQ_19-1479 1D..1 0us : resched_task (try_to_wake_up) IRQ_19-1479 1D..1 0us : __spin_unlock_irqrestore (try_to_wake_up) ... <idle>-0 1...1 11us!: default_idle (cpu_idle) ... <idle>-0 0Dn.1 602us : smp_apic_timer_interrupt (c0103baf 1 0) ... <...>-5856 0D..2 618us : __switch_to (__schedule) <...>-5856 0D..2 618us : __schedule <<idle>-0> (20 162) <...>-5856 0D..2 619us : __spin_unlock_irq (__schedule) <...>-5856 0...1 619us : trace_stop_sched_switched (__schedule) <...>-5856 0D..1 619us : trace_stop_sched_switched <<...>-5856> (37 0) what is visible in this trace is that CPU#1 ran try_to_wake_up() for PID:5856, it placed PID:5856 on CPU#0's runqueue and ran resched_task() for CPU#0. But it decided to not send an IPI that no CPU - due to TS_POLLING. But CPU#0 never woke up after its NEED_RESCHED bit was set, and only rescheduled to PID:5856 upon the next lapic timer IRQ. The result was a 600+ usecs latency and a missed wakeup! the bug turned out to be an idle-wakeup bug introduced into the mainline kernel this summer via an optimization in the x86_64 tree: commit 495ab9c045e1b0e5c82951b762257fe1c9d81564 Author: Andi Kleen <ak@suse.de> Date: Mon Jun 26 13:59:11 2006 +0200 [PATCH] i386/x86-64/ia64: Move polling flag into thread_info_status During some profiling I noticed that default_idle causes a lot of memory traffic. I think that is caused by the atomic operations to clear/set the polling flag in thread_info. There is actually no reason to make this atomic - only the idle thread does it to itself, other CPUs only read it. So I moved it into ti->status. the problem is this type of change: if (!hlt_counter && boot_cpu_data.hlt_works_ok) { - clear_thread_flag(TIF_POLLING_NRFLAG); + current_thread_info()->status &= ~TS_POLLING; smp_mb__after_clear_bit(); while (!need_resched()) { local_irq_disable(); this changes clear_thread_flag() to an explicit clearing of TS_POLLING. clear_thread_flag() is defined as: clear_bit(flag, &ti->flags); and clear_bit() is a LOCK-ed atomic instruction on all x86 platforms: static inline void clear_bit(int nr, volatile unsigned long * addr) { __asm__ __volatile__( LOCK_PREFIX "btrl %1,%0" hence smp_mb__after_clear_bit() is defined as a simple compile barrier: #define smp_mb__after_clear_bit() barrier() but the explicit TS_POLLING clearing introduced by the patch: + current_thread_info()->status &= ~TS_POLLING; is not an atomic op! So the clearing of the TS_POLLING bit is freely reorderable with the reading of the NEED_RESCHED bit - and both now reside in different memory addresses. CPU idle wakeup very much depends on ordered memory ops, the clearing of the TS_POLLING flag must always be done before we test need_resched() and hit the idle instruction(s). [Symmetrically, the wakeup code needs to set NEED_RESCHED before it tests the TS_POLLING flag, so memory ordering is paramount.] Fernando's dual-core Athlon64 system has a sufficiently advanced memory ordering model so that it triggered this scenario very often. ( And it also turned out that the reason why these latencies never triggered on my testsystems is that i routinely use idle=poll, which was the only idle variant not affected by this bug. ) The fix is to change the smp_mb__after_clear_bit() to an smp_mb(), to act as an absolute barrier between the TS_POLLING write and the NEED_RESCHED read. This affects almost all idling methods (default, ACPI, APM), on all 3 x86 architectures: i386, x86_64, ia64. Signed-off-by: Ingo Molnar <mingo@elte.hu> Tested-by: Fernando Lopez-Lezcano <nando@ccrma.Stanford.EDU> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-22 17:11:56 +08:00
}
[PATCH] sched: resched and cpu_idle rework Make some changes to the NEED_RESCHED and POLLING_NRFLAG to reduce confusion, and make their semantics rigid. Improves efficiency of resched_task and some cpu_idle routines. * In resched_task: - TIF_NEED_RESCHED is only cleared with the task's runqueue lock held, and as we hold it during resched_task, then there is no need for an atomic test and set there. The only other time this should be set is when the task's quantum expires, in the timer interrupt - this is protected against because the rq lock is irq-safe. - If TIF_NEED_RESCHED is set, then we don't need to do anything. It won't get unset until the task get's schedule()d off. - If we are running on the same CPU as the task we resched, then set TIF_NEED_RESCHED and no further action is required. - If we are running on another CPU, and TIF_POLLING_NRFLAG is *not* set after TIF_NEED_RESCHED has been set, then we need to send an IPI. Using these rules, we are able to remove the test and set operation in resched_task, and make clear the previously vague semantics of POLLING_NRFLAG. * In idle routines: - Enter cpu_idle with preempt disabled. When the need_resched() condition becomes true, explicitly call schedule(). This makes things a bit clearer (IMO), but haven't updated all architectures yet. - Many do a test and clear of TIF_NEED_RESCHED for some reason. According to the resched_task rules, this isn't needed (and actually breaks the assumption that TIF_NEED_RESCHED is only cleared with the runqueue lock held). So remove that. Generally one less locked memory op when switching to the idle thread. - Many idle routines clear TIF_POLLING_NRFLAG, and only set it in the inner most polling idle loops. The above resched_task semantics allow it to be set until before the last time need_resched() is checked before going into a halt requiring interrupt wakeup. Many idle routines simply never enter such a halt, and so POLLING_NRFLAG can be always left set, completely eliminating resched IPIs when rescheduling the idle task. POLLING_NRFLAG width can be increased, to reduce the chance of resched IPIs. Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Ingo Molnar <mingo@elte.hu> Cc: Con Kolivas <kernel@kolivas.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-11-09 13:39:04 +08:00
if (!need_resched()) {
void (*idle)(void);
#ifdef CONFIG_SMP
min_xtp();
#endif
rmb();
if (mark_idle)
(*mark_idle)(1);
idle = pm_idle;
if (!idle)
idle = default_idle;
(*idle)();
[PATCH] sched: resched and cpu_idle rework Make some changes to the NEED_RESCHED and POLLING_NRFLAG to reduce confusion, and make their semantics rigid. Improves efficiency of resched_task and some cpu_idle routines. * In resched_task: - TIF_NEED_RESCHED is only cleared with the task's runqueue lock held, and as we hold it during resched_task, then there is no need for an atomic test and set there. The only other time this should be set is when the task's quantum expires, in the timer interrupt - this is protected against because the rq lock is irq-safe. - If TIF_NEED_RESCHED is set, then we don't need to do anything. It won't get unset until the task get's schedule()d off. - If we are running on the same CPU as the task we resched, then set TIF_NEED_RESCHED and no further action is required. - If we are running on another CPU, and TIF_POLLING_NRFLAG is *not* set after TIF_NEED_RESCHED has been set, then we need to send an IPI. Using these rules, we are able to remove the test and set operation in resched_task, and make clear the previously vague semantics of POLLING_NRFLAG. * In idle routines: - Enter cpu_idle with preempt disabled. When the need_resched() condition becomes true, explicitly call schedule(). This makes things a bit clearer (IMO), but haven't updated all architectures yet. - Many do a test and clear of TIF_NEED_RESCHED for some reason. According to the resched_task rules, this isn't needed (and actually breaks the assumption that TIF_NEED_RESCHED is only cleared with the runqueue lock held). So remove that. Generally one less locked memory op when switching to the idle thread. - Many idle routines clear TIF_POLLING_NRFLAG, and only set it in the inner most polling idle loops. The above resched_task semantics allow it to be set until before the last time need_resched() is checked before going into a halt requiring interrupt wakeup. Many idle routines simply never enter such a halt, and so POLLING_NRFLAG can be always left set, completely eliminating resched IPIs when rescheduling the idle task. POLLING_NRFLAG width can be increased, to reduce the chance of resched IPIs. Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Ingo Molnar <mingo@elte.hu> Cc: Con Kolivas <kernel@kolivas.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-11-09 13:39:04 +08:00
if (mark_idle)
(*mark_idle)(0);
#ifdef CONFIG_SMP
[PATCH] sched: resched and cpu_idle rework Make some changes to the NEED_RESCHED and POLLING_NRFLAG to reduce confusion, and make their semantics rigid. Improves efficiency of resched_task and some cpu_idle routines. * In resched_task: - TIF_NEED_RESCHED is only cleared with the task's runqueue lock held, and as we hold it during resched_task, then there is no need for an atomic test and set there. The only other time this should be set is when the task's quantum expires, in the timer interrupt - this is protected against because the rq lock is irq-safe. - If TIF_NEED_RESCHED is set, then we don't need to do anything. It won't get unset until the task get's schedule()d off. - If we are running on the same CPU as the task we resched, then set TIF_NEED_RESCHED and no further action is required. - If we are running on another CPU, and TIF_POLLING_NRFLAG is *not* set after TIF_NEED_RESCHED has been set, then we need to send an IPI. Using these rules, we are able to remove the test and set operation in resched_task, and make clear the previously vague semantics of POLLING_NRFLAG. * In idle routines: - Enter cpu_idle with preempt disabled. When the need_resched() condition becomes true, explicitly call schedule(). This makes things a bit clearer (IMO), but haven't updated all architectures yet. - Many do a test and clear of TIF_NEED_RESCHED for some reason. According to the resched_task rules, this isn't needed (and actually breaks the assumption that TIF_NEED_RESCHED is only cleared with the runqueue lock held). So remove that. Generally one less locked memory op when switching to the idle thread. - Many idle routines clear TIF_POLLING_NRFLAG, and only set it in the inner most polling idle loops. The above resched_task semantics allow it to be set until before the last time need_resched() is checked before going into a halt requiring interrupt wakeup. Many idle routines simply never enter such a halt, and so POLLING_NRFLAG can be always left set, completely eliminating resched IPIs when rescheduling the idle task. POLLING_NRFLAG width can be increased, to reduce the chance of resched IPIs. Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Ingo Molnar <mingo@elte.hu> Cc: Con Kolivas <kernel@kolivas.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-11-09 13:39:04 +08:00
normal_xtp();
#endif
[PATCH] sched: resched and cpu_idle rework Make some changes to the NEED_RESCHED and POLLING_NRFLAG to reduce confusion, and make their semantics rigid. Improves efficiency of resched_task and some cpu_idle routines. * In resched_task: - TIF_NEED_RESCHED is only cleared with the task's runqueue lock held, and as we hold it during resched_task, then there is no need for an atomic test and set there. The only other time this should be set is when the task's quantum expires, in the timer interrupt - this is protected against because the rq lock is irq-safe. - If TIF_NEED_RESCHED is set, then we don't need to do anything. It won't get unset until the task get's schedule()d off. - If we are running on the same CPU as the task we resched, then set TIF_NEED_RESCHED and no further action is required. - If we are running on another CPU, and TIF_POLLING_NRFLAG is *not* set after TIF_NEED_RESCHED has been set, then we need to send an IPI. Using these rules, we are able to remove the test and set operation in resched_task, and make clear the previously vague semantics of POLLING_NRFLAG. * In idle routines: - Enter cpu_idle with preempt disabled. When the need_resched() condition becomes true, explicitly call schedule(). This makes things a bit clearer (IMO), but haven't updated all architectures yet. - Many do a test and clear of TIF_NEED_RESCHED for some reason. According to the resched_task rules, this isn't needed (and actually breaks the assumption that TIF_NEED_RESCHED is only cleared with the runqueue lock held). So remove that. Generally one less locked memory op when switching to the idle thread. - Many idle routines clear TIF_POLLING_NRFLAG, and only set it in the inner most polling idle loops. The above resched_task semantics allow it to be set until before the last time need_resched() is checked before going into a halt requiring interrupt wakeup. Many idle routines simply never enter such a halt, and so POLLING_NRFLAG can be always left set, completely eliminating resched IPIs when rescheduling the idle task. POLLING_NRFLAG width can be increased, to reduce the chance of resched IPIs. Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Ingo Molnar <mingo@elte.hu> Cc: Con Kolivas <kernel@kolivas.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-11-09 13:39:04 +08:00
}
rcu_idle_exit();
schedule_preempt_disabled();
check_pgt_cache();
[PATCH] sched: resched and cpu_idle rework Make some changes to the NEED_RESCHED and POLLING_NRFLAG to reduce confusion, and make their semantics rigid. Improves efficiency of resched_task and some cpu_idle routines. * In resched_task: - TIF_NEED_RESCHED is only cleared with the task's runqueue lock held, and as we hold it during resched_task, then there is no need for an atomic test and set there. The only other time this should be set is when the task's quantum expires, in the timer interrupt - this is protected against because the rq lock is irq-safe. - If TIF_NEED_RESCHED is set, then we don't need to do anything. It won't get unset until the task get's schedule()d off. - If we are running on the same CPU as the task we resched, then set TIF_NEED_RESCHED and no further action is required. - If we are running on another CPU, and TIF_POLLING_NRFLAG is *not* set after TIF_NEED_RESCHED has been set, then we need to send an IPI. Using these rules, we are able to remove the test and set operation in resched_task, and make clear the previously vague semantics of POLLING_NRFLAG. * In idle routines: - Enter cpu_idle with preempt disabled. When the need_resched() condition becomes true, explicitly call schedule(). This makes things a bit clearer (IMO), but haven't updated all architectures yet. - Many do a test and clear of TIF_NEED_RESCHED for some reason. According to the resched_task rules, this isn't needed (and actually breaks the assumption that TIF_NEED_RESCHED is only cleared with the runqueue lock held). So remove that. Generally one less locked memory op when switching to the idle thread. - Many idle routines clear TIF_POLLING_NRFLAG, and only set it in the inner most polling idle loops. The above resched_task semantics allow it to be set until before the last time need_resched() is checked before going into a halt requiring interrupt wakeup. Many idle routines simply never enter such a halt, and so POLLING_NRFLAG can be always left set, completely eliminating resched IPIs when rescheduling the idle task. POLLING_NRFLAG width can be increased, to reduce the chance of resched IPIs. Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Ingo Molnar <mingo@elte.hu> Cc: Con Kolivas <kernel@kolivas.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-11-09 13:39:04 +08:00
if (cpu_is_offline(cpu))
play_dead();
}
}
void
ia64_save_extra (struct task_struct *task)
{
#ifdef CONFIG_PERFMON
unsigned long info;
#endif
if ((task->thread.flags & IA64_THREAD_DBG_VALID) != 0)
ia64_save_debug_regs(&task->thread.dbr[0]);
#ifdef CONFIG_PERFMON
if ((task->thread.flags & IA64_THREAD_PM_VALID) != 0)
pfm_save_regs(task);
info = __get_cpu_var(pfm_syst_info);
if (info & PFM_CPUINFO_SYST_WIDE)
pfm_syst_wide_update_task(task, info, 0);
#endif
}
void
ia64_load_extra (struct task_struct *task)
{
#ifdef CONFIG_PERFMON
unsigned long info;
#endif
if ((task->thread.flags & IA64_THREAD_DBG_VALID) != 0)
ia64_load_debug_regs(&task->thread.dbr[0]);
#ifdef CONFIG_PERFMON
if ((task->thread.flags & IA64_THREAD_PM_VALID) != 0)
pfm_load_regs(task);
info = __get_cpu_var(pfm_syst_info);
if (info & PFM_CPUINFO_SYST_WIDE)
pfm_syst_wide_update_task(task, info, 1);
#endif
}
/*
* Copy the state of an ia-64 thread.
*
* We get here through the following call chain:
*
* from user-level: from kernel:
*
* <clone syscall> <some kernel call frames>
* sys_clone :
* do_fork do_fork
* copy_thread copy_thread
*
* This means that the stack layout is as follows:
*
* +---------------------+ (highest addr)
* | struct pt_regs |
* +---------------------+
* | struct switch_stack |
* +---------------------+
* | |
* | memory stack |
* | | <-- sp (lowest addr)
* +---------------------+
*
* Observe that we copy the unat values that are in pt_regs and switch_stack. Spilling an
* integer to address X causes bit N in ar.unat to be set to the NaT bit of the register,
* with N=(X & 0x1ff)/8. Thus, copying the unat value preserves the NaT bits ONLY if the
* pt_regs structure in the parent is congruent to that of the child, modulo 512. Since
* the stack is page aligned and the page size is at least 4KB, this is always the case,
* so there is nothing to worry about.
*/
int
copy_thread(unsigned long clone_flags,
unsigned long user_stack_base, unsigned long user_stack_size,
struct task_struct *p, struct pt_regs *regs)
{
extern char ia64_ret_from_clone;
struct switch_stack *child_stack, *stack;
unsigned long rbs, child_rbs, rbs_size;
struct pt_regs *child_ptregs;
int retval = 0;
child_ptregs = (struct pt_regs *) ((unsigned long) p + IA64_STK_OFFSET) - 1;
child_stack = (struct switch_stack *) child_ptregs - 1;
rbs = (unsigned long) current + IA64_RBS_OFFSET;
child_rbs = (unsigned long) p + IA64_RBS_OFFSET;
/* copy parts of thread_struct: */
p->thread.ksp = (unsigned long) child_stack - 16;
/*
* NOTE: The calling convention considers all floating point
* registers in the high partition (fph) to be scratch. Since
* the only way to get to this point is through a system call,
* we know that the values in fph are all dead. Hence, there
* is no need to inherit the fph state from the parent to the
* child and all we have to do is to make sure that
* IA64_THREAD_FPH_VALID is cleared in the child.
*
* XXX We could push this optimization a bit further by
* clearing IA64_THREAD_FPH_VALID on ANY system call.
* However, it's not clear this is worth doing. Also, it
* would be a slight deviation from the normal Linux system
* call behavior where scratch registers are preserved across
* system calls (unless used by the system call itself).
*/
# define THREAD_FLAGS_TO_CLEAR (IA64_THREAD_FPH_VALID | IA64_THREAD_DBG_VALID \
| IA64_THREAD_PM_VALID)
# define THREAD_FLAGS_TO_SET 0
p->thread.flags = ((current->thread.flags & ~THREAD_FLAGS_TO_CLEAR)
| THREAD_FLAGS_TO_SET);
ia64_drop_fpu(p); /* don't pick up stale state from a CPU's fph */
if (unlikely(p->flags & PF_KTHREAD)) {
if (unlikely(!user_stack_base)) {
/* fork_idle() called us */
return 0;
}
memset(child_stack, 0, sizeof(*child_ptregs) + sizeof(*child_stack));
child_stack->r4 = user_stack_base; /* payload */
child_stack->r5 = user_stack_size; /* argument */
/*
* Preserve PSR bits, except for bits 32-34 and 37-45,
* which we can't read.
*/
child_ptregs->cr_ipsr = ia64_getreg(_IA64_REG_PSR) | IA64_PSR_BN;
/* mark as valid, empty frame */
child_ptregs->cr_ifs = 1UL << 63;
child_stack->ar_fpsr = child_ptregs->ar_fpsr
= ia64_getreg(_IA64_REG_AR_FPSR);
child_stack->pr = (1 << PRED_KERNEL_STACK);
child_stack->ar_bspstore = child_rbs;
child_stack->b0 = (unsigned long) &ia64_ret_from_clone;
/* stop some PSR bits from being inherited.
* the psr.up/psr.pp bits must be cleared on fork but inherited on execve()
* therefore we must specify them explicitly here and not include them in
* IA64_PSR_BITS_TO_CLEAR.
*/
child_ptregs->cr_ipsr = ((child_ptregs->cr_ipsr | IA64_PSR_BITS_TO_SET)
& ~(IA64_PSR_BITS_TO_CLEAR | IA64_PSR_PP | IA64_PSR_UP));
return 0;
}
stack = ((struct switch_stack *) regs) - 1;
/* copy parent's switch_stack & pt_regs to child: */
memcpy(child_stack, stack, sizeof(*child_ptregs) + sizeof(*child_stack));
/* copy the parent's register backing store to the child: */
rbs_size = stack->ar_bspstore - rbs;
memcpy((void *) child_rbs, (void *) rbs, rbs_size);
if (clone_flags & CLONE_SETTLS)
child_ptregs->r13 = regs->r16; /* see sys_clone2() in entry.S */
if (user_stack_base) {
child_ptregs->r12 = user_stack_base + user_stack_size - 16;
child_ptregs->ar_bspstore = user_stack_base;
child_ptregs->ar_rnat = 0;
child_ptregs->loadrs = 0;
}
child_stack->ar_bspstore = child_rbs + rbs_size;
child_stack->b0 = (unsigned long) &ia64_ret_from_clone;
/* stop some PSR bits from being inherited.
* the psr.up/psr.pp bits must be cleared on fork but inherited on execve()
* therefore we must specify them explicitly here and not include them in
* IA64_PSR_BITS_TO_CLEAR.
*/
child_ptregs->cr_ipsr = ((child_ptregs->cr_ipsr | IA64_PSR_BITS_TO_SET)
& ~(IA64_PSR_BITS_TO_CLEAR | IA64_PSR_PP | IA64_PSR_UP));
#ifdef CONFIG_PERFMON
if (current->thread.pfm_context)
pfm_inherit(p, child_ptregs);
#endif
return retval;
}
static void
do_copy_task_regs (struct task_struct *task, struct unw_frame_info *info, void *arg)
{
unsigned long mask, sp, nat_bits = 0, ar_rnat, urbs_end, cfm;
unsigned long uninitialized_var(ip); /* GCC be quiet */
elf_greg_t *dst = arg;
struct pt_regs *pt;
char nat;
int i;
memset(dst, 0, sizeof(elf_gregset_t)); /* don't leak any kernel bits to user-level */
if (unw_unwind_to_user(info) < 0)
return;
unw_get_sp(info, &sp);
pt = (struct pt_regs *) (sp + 16);
urbs_end = ia64_get_user_rbs_end(task, pt, &cfm);
if (ia64_sync_user_rbs(task, info->sw, pt->ar_bspstore, urbs_end) < 0)
return;
ia64_peek(task, info->sw, urbs_end, (long) ia64_rse_rnat_addr((long *) urbs_end),
&ar_rnat);
/*
* coredump format:
* r0-r31
* NaT bits (for r0-r31; bit N == 1 iff rN is a NaT)
* predicate registers (p0-p63)
* b0-b7
* ip cfm user-mask
* ar.rsc ar.bsp ar.bspstore ar.rnat
* ar.ccv ar.unat ar.fpsr ar.pfs ar.lc ar.ec
*/
/* r0 is zero */
for (i = 1, mask = (1UL << i); i < 32; ++i) {
unw_get_gr(info, i, &dst[i], &nat);
if (nat)
nat_bits |= mask;
mask <<= 1;
}
dst[32] = nat_bits;
unw_get_pr(info, &dst[33]);
for (i = 0; i < 8; ++i)
unw_get_br(info, i, &dst[34 + i]);
unw_get_rp(info, &ip);
dst[42] = ip + ia64_psr(pt)->ri;
dst[43] = cfm;
dst[44] = pt->cr_ipsr & IA64_PSR_UM;
unw_get_ar(info, UNW_AR_RSC, &dst[45]);
/*
* For bsp and bspstore, unw_get_ar() would return the kernel
* addresses, but we need the user-level addresses instead:
*/
dst[46] = urbs_end; /* note: by convention PT_AR_BSP points to the end of the urbs! */
dst[47] = pt->ar_bspstore;
dst[48] = ar_rnat;
unw_get_ar(info, UNW_AR_CCV, &dst[49]);
unw_get_ar(info, UNW_AR_UNAT, &dst[50]);
unw_get_ar(info, UNW_AR_FPSR, &dst[51]);
dst[52] = pt->ar_pfs; /* UNW_AR_PFS is == to pt->cr_ifs for interrupt frames */
unw_get_ar(info, UNW_AR_LC, &dst[53]);
unw_get_ar(info, UNW_AR_EC, &dst[54]);
unw_get_ar(info, UNW_AR_CSD, &dst[55]);
unw_get_ar(info, UNW_AR_SSD, &dst[56]);
}
void
do_dump_task_fpu (struct task_struct *task, struct unw_frame_info *info, void *arg)
{
elf_fpreg_t *dst = arg;
int i;
memset(dst, 0, sizeof(elf_fpregset_t)); /* don't leak any "random" bits */
if (unw_unwind_to_user(info) < 0)
return;
/* f0 is 0.0, f1 is 1.0 */
for (i = 2; i < 32; ++i)
unw_get_fr(info, i, dst + i);
ia64_flush_fph(task);
if ((task->thread.flags & IA64_THREAD_FPH_VALID) != 0)
memcpy(dst + 32, task->thread.fph, 96*16);
}
void
do_copy_regs (struct unw_frame_info *info, void *arg)
{
do_copy_task_regs(current, info, arg);
}
void
do_dump_fpu (struct unw_frame_info *info, void *arg)
{
do_dump_task_fpu(current, info, arg);
}
void
ia64_elf_core_copy_regs (struct pt_regs *pt, elf_gregset_t dst)
{
unw_init_running(do_copy_regs, dst);
}
int
dump_fpu (struct pt_regs *pt, elf_fpregset_t dst)
{
unw_init_running(do_dump_fpu, dst);
return 1; /* f0-f31 are always valid so we always return 1 */
}
/*
* Flush thread state. This is called when a thread does an execve().
*/
void
flush_thread (void)
{
/* drop floating-point and debug-register state if it exists: */
current->thread.flags &= ~(IA64_THREAD_FPH_VALID | IA64_THREAD_DBG_VALID);
ia64_drop_fpu(current);
}
/*
* Clean up state associated with current thread. This is called when
* the thread calls exit().
*/
void
exit_thread (void)
{
[PATCH] Return probe redesign: ia64 specific implementation The following patch implements function return probes for ia64 using the revised design. With this new design we no longer need to do some of the odd hacks previous required on the last ia64 return probe port that I sent out for comments. Note that this new implementation still does not resolve the problem noted by Keith Owens where backtrace data is lost after a return probe is hit. Changes include: * Addition of kretprobe_trampoline to act as a dummy function for instrumented functions to return to, and for the return probe infrastructure to place a kprobe on on, gaining control so that the return probe handler can be called, and so that the instruction pointer can be moved back to the original return address. * Addition of arch_init(), allowing a kprobe to be registered on kretprobe_trampoline * Addition of trampoline_probe_handler() which is used as the pre_handler for the kprobe inserted on kretprobe_implementation. This is the function that handles the details for calling the return probe handler function and returning control back at the original return address * Addition of arch_prepare_kretprobe() which is setup as the pre_handler for a kprobe registered at the beginning of the target function by kernel/kprobes.c so that a return probe instance can be setup when a caller enters the target function. (A return probe instance contains all the needed information for trampoline_probe_handler to do it's job.) * Hooks added to the exit path of a task so that we can cleanup any left-over return probe instances (i.e. if a task dies while inside a targeted function then the return probe instance was reserved at the beginning of the function but the function never returns so we need to mark the instance as unused.) Signed-off-by: Rusty Lynch <rusty.lynch@intel.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 06:17:12 +08:00
ia64_drop_fpu(current);
#ifdef CONFIG_PERFMON
/* if needed, stop monitoring and flush state to perfmon context */
if (current->thread.pfm_context)
pfm_exit_thread(current);
/* free debug register resources */
if (current->thread.flags & IA64_THREAD_DBG_VALID)
pfm_release_debug_registers(current);
#endif
}
unsigned long
get_wchan (struct task_struct *p)
{
struct unw_frame_info info;
unsigned long ip;
int count = 0;
if (!p || p == current || p->state == TASK_RUNNING)
return 0;
/*
* Note: p may not be a blocked task (it could be current or
* another process running on some other CPU. Rather than
* trying to determine if p is really blocked, we just assume
* it's blocked and rely on the unwind routines to fail
* gracefully if the process wasn't really blocked after all.
* --davidm 99/12/15
*/
unw_init_from_blocked_task(&info, p);
do {
if (p->state == TASK_RUNNING)
return 0;
if (unw_unwind(&info) < 0)
return 0;
unw_get_ip(&info, &ip);
if (!in_sched_functions(ip))
return ip;
} while (count++ < 16);
return 0;
}
void
cpu_halt (void)
{
pal_power_mgmt_info_u_t power_info[8];
unsigned long min_power;
int i, min_power_state;
if (ia64_pal_halt_info(power_info) != 0)
return;
min_power_state = 0;
min_power = power_info[0].pal_power_mgmt_info_s.power_consumption;
for (i = 1; i < 8; ++i)
if (power_info[i].pal_power_mgmt_info_s.im
&& power_info[i].pal_power_mgmt_info_s.power_consumption < min_power) {
min_power = power_info[i].pal_power_mgmt_info_s.power_consumption;
min_power_state = i;
}
while (1)
ia64_pal_halt(min_power_state);
}
void machine_shutdown(void)
{
#ifdef CONFIG_HOTPLUG_CPU
int cpu;
for_each_online_cpu(cpu) {
if (cpu != smp_processor_id())
cpu_down(cpu);
}
#endif
#ifdef CONFIG_KEXEC
kexec_disable_iosapic();
#endif
}
void
machine_restart (char *restart_cmd)
{
[IA64] Extend notify_die() hooks for IA64 notify_die() added for MCA_{MONARCH,SLAVE,RENDEZVOUS}_{ENTER,PROCESS,LEAVE} and INIT_{MONARCH,SLAVE}_{ENTER,PROCESS,LEAVE}. We need multiple notification points for these events because they can take many seconds to run which has nasty effects on the behaviour of the rest of the system. DIE_SS replaced by a generic DIE_FAULT which checks the vector number, to allow interception of faults other than SS. DIE_MACHINE_{HALT,RESTART} added to allow last minute close down processing, especially when the halt/restart routines are called from error handlers. DIE_OOPS added. The check for kprobe's break numbers has been moved from traps.c to kprobes.c, allowing DIE_BREAK to be used for any additional break numbers, i.e. it is no longer kprobes specific. Hooks for kernel debuggers and kernel dumpers added, ENTER and LEAVE. Both of these disable the system for long periods which impact on watchdogs and heartbeat systems in general. More patches to come that use these events to reset watchdogs and heartbeats. unregister_die_notifier() added and both routines exported. Requested by Dean Nelson. Lock removed from {un,}register_die_notifier. notifier_chain_register() already takes a lock. Also the generic notifier chain locking is being reworked to distinguish between callbacks that can block and those that cannot, the lock in {un,}register_die_notifier would interfere with that change. http://marc.theaimsgroup.com/?l=linux-kernel&m=113018709002036&w=2 Leading white space removed from arch/ia64/kernel/kprobes.c. Typo in mca.c in original version of this patch found & fixed by Dean Nelson. Signed-off-by: Keith Owens <kaos@sgi.com> Acked-by: Dean Nelson <dcn@sgi.com> Acked-by: Anil Keshavamurthy <anil.s.keshavamurthy@intel.com> Signed-off-by: Tony Luck <tony.luck@intel.com>
2005-11-08 03:27:13 +08:00
(void) notify_die(DIE_MACHINE_RESTART, restart_cmd, NULL, 0, 0, 0);
(*efi.reset_system)(EFI_RESET_WARM, 0, 0, NULL);
}
void
machine_halt (void)
{
[IA64] Extend notify_die() hooks for IA64 notify_die() added for MCA_{MONARCH,SLAVE,RENDEZVOUS}_{ENTER,PROCESS,LEAVE} and INIT_{MONARCH,SLAVE}_{ENTER,PROCESS,LEAVE}. We need multiple notification points for these events because they can take many seconds to run which has nasty effects on the behaviour of the rest of the system. DIE_SS replaced by a generic DIE_FAULT which checks the vector number, to allow interception of faults other than SS. DIE_MACHINE_{HALT,RESTART} added to allow last minute close down processing, especially when the halt/restart routines are called from error handlers. DIE_OOPS added. The check for kprobe's break numbers has been moved from traps.c to kprobes.c, allowing DIE_BREAK to be used for any additional break numbers, i.e. it is no longer kprobes specific. Hooks for kernel debuggers and kernel dumpers added, ENTER and LEAVE. Both of these disable the system for long periods which impact on watchdogs and heartbeat systems in general. More patches to come that use these events to reset watchdogs and heartbeats. unregister_die_notifier() added and both routines exported. Requested by Dean Nelson. Lock removed from {un,}register_die_notifier. notifier_chain_register() already takes a lock. Also the generic notifier chain locking is being reworked to distinguish between callbacks that can block and those that cannot, the lock in {un,}register_die_notifier would interfere with that change. http://marc.theaimsgroup.com/?l=linux-kernel&m=113018709002036&w=2 Leading white space removed from arch/ia64/kernel/kprobes.c. Typo in mca.c in original version of this patch found & fixed by Dean Nelson. Signed-off-by: Keith Owens <kaos@sgi.com> Acked-by: Dean Nelson <dcn@sgi.com> Acked-by: Anil Keshavamurthy <anil.s.keshavamurthy@intel.com> Signed-off-by: Tony Luck <tony.luck@intel.com>
2005-11-08 03:27:13 +08:00
(void) notify_die(DIE_MACHINE_HALT, "", NULL, 0, 0, 0);
cpu_halt();
}
void
machine_power_off (void)
{
if (pm_power_off)
pm_power_off();
machine_halt();
}