It turns out that typedefing a struct makes the #includes really hairy
(asm/smp.h needs the definition which is in linux/sched.h), so I just
made it a per-arch typedef. So, please discard the and_region patch.
I also fixed the migration code (it needs to combine the online map
with the allowed map to choose another CPU).
You can revert ChangeSet 1.490.1.66 (adds cpu_online_map for UP), as
noone should be using it directly now (outside per-arch, of course).
Name: CPU mask patch
Author: Rusty Russell
Status: Trivial
D: This patch adds a cpu_mask_t for easy future expansion of cpu masks,
D: and an easy "migrate_to_cpu()" call for simple in-kernel affinity. It
D: also changes any_online_cpu to return NR_CPUS on failure, like
D: find_first_bit.
diff -urN -I \$.*\$ --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5.23/Documentation/sched-coding.txt working-2.5.23-sched/Documentation/sched-coding.txt
--- linux-2.5.23/Documentation/sched-coding.txt Thu May 30 10:00:45 2002
+++ working-2.5.23-sched/Documentation/sched-coding.txt Fri Jun 21 00:48:15 2002
@@ -103,7 +103,10 @@
Sets the "nice" value of task p to the given value.
int setscheduler(pid_t pid, int policy, struct sched_param *param)
Sets the scheduling policy and parameters for the given pid.
-void set_cpus_allowed(task_t *p, unsigned long new_mask)
+void migrate_to_cpu(unsigned int cpu)
+ Migrates the current task to the cpu and nails it there. This
+ is a more convenient form of:
+void set_cpus_allowed(task_t *p, cpu_mask_t new_mask)
Sets a given task's CPU affinity and migrates it to a proper cpu.
Callers must have a valid reference to the task and assure the
task not exit prematurely. No locks can be held during the call.
diff -urN -I \$.*\$ --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5.23/arch/ppc64/kernel/rtasd.c working-2.5.23-sched/arch/ppc64/kernel/rtasd.c
--- linux-2.5.23/arch/ppc64/kernel/rtasd.c Mon Jun 10 16:03:47 2002
+++ working-2.5.23-sched/arch/ppc64/kernel/rtasd.c Fri Jun 21 00:48:15 2002
@@ -221,7 +221,7 @@
#endif
cpu = 0;
- set_cpus_allowed(current, 1UL << cpu_logical_map(cpu));
+ migrate_to_cpu(cpu);
while(1) {
do {
@@ -255,8 +255,7 @@
cpu = 0;
}
- set_cpus_allowed(current, 1UL << cpu_logical_map(cpu));
-
+ migrate_to_cpu(cpu);
/* Check all cpus for pending events before sleeping*/
if (!first_pass) {
diff -urN -I \$.*\$ --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5.23/include/asm-i386/smp.h working-2.5.23-sched/include/asm-i386/smp.h
--- linux-2.5.23/include/asm-i386/smp.h Thu Jun 20 01:28:51 2002
+++ working-2.5.23-sched/include/asm-i386/smp.h Fri Jun 21 01:52:45 2002
@@ -99,14 +99,6 @@
return hweight32(cpu_online_map);
}
-extern inline int any_online_cpu(unsigned int mask)
-{
- if (mask & cpu_online_map)
- return __ffs(mask & cpu_online_map);
-
- return -1;
-}
-
static __inline int hard_smp_processor_id(void)
{
/* we don't want to mark this access volatile - bad code generation */
@@ -117,6 +109,23 @@
{
/* we don't want to mark this access volatile - bad code generation */
return GET_APIC_LOGICAL_ID(*(unsigned long *)(APIC_BASE+APIC_LDR));
+}
+
+/* This has to be bitop capable, so for most archs, an unsigned long */
+typedef unsigned long cpu_mask_t;
+#define CPU_MASK_ALL (~0UL)
+#define CPU_MASK_NONE (0UL)
+
+static inline void cpu_online_mask(cpu_mask_t *res, cpu_mask_t mask)
+{
+ *res = (mask & cpu_online_map);
+}
+
+static inline unsigned int any_online_cpu(cpu_mask_t mask)
+{
+ if (mask & cpu_online_map)
+ return __ffs(mask & cpu_online_map);
+ return NR_CPUS;
}
#endif /* !__ASSEMBLY__ */
diff -urN -I \$.*\$ --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5.23/include/asm-ia64/smp.h working-2.5.23-sched/include/asm-ia64/smp.h
--- linux-2.5.23/include/asm-ia64/smp.h Thu Jun 20 01:28:51 2002
+++ working-2.5.23-sched/include/asm-ia64/smp.h Fri Jun 21 01:53:00 2002
@@ -51,12 +51,22 @@
return hweight64(cpu_online_map);
}
-extern inline int any_online_cpu(unsigned int mask)
+/* This has to be bitop capable, so for most archs, an unsigned long */
+typedef unsigned long cpu_mask_t;
+#define CPU_MASK_ALL (~0UL)
+#define CPU_MASK_NONE (0UL)
+
+static inline void cpu_online_mask(cpu_mask_t *res, cpu_mask_t mask)
+{
+ *res = (mask & cpu_online_map);
+}
+
+static inline unsigned int any_online_cpu(cpu_mask_t mask)
{
if (mask & cpu_online_map)
return __ffs(mask & cpu_online_map);
- return -1;
+ return NR_CPUS;
}
/*
diff -urN -I \$.*\$ --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5.23/include/asm-ppc/smp.h working-2.5.23-sched/include/asm-ppc/smp.h
--- linux-2.5.23/include/asm-ppc/smp.h Thu Jun 20 01:28:51 2002
+++ working-2.5.23-sched/include/asm-ppc/smp.h Fri Jun 21 01:53:12 2002
@@ -54,12 +54,21 @@
return hweight32(cpu_online_map);
}
-extern inline int any_online_cpu(unsigned int mask)
+typedef unsigned long cpu_mask_t;
+#define CPU_MASK_ALL (~0UL)
+#define CPU_MASK_NONE (0UL)
+
+static inline void cpu_online_mask(cpu_mask_t *res, cpu_mask_t mask)
+{
+ *res = (mask & cpu_online_map);
+}
+
+static inline unsigned int any_online_cpu(cpu_mask_t mask)
{
if (mask & cpu_online_map)
return __ffs(mask & cpu_online_map);
- return -1;
+ return NR_CPUS;
}
extern int smp_hw_index[];
diff -urN -I \$.*\$ --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5.23/include/linux/init_task.h working-2.5.23-sched/include/linux/init_task.h
--- linux-2.5.23/include/linux/init_task.h Sun May 19 12:07:34 2002
+++ working-2.5.23-sched/include/linux/init_task.h Fri Jun 21 01:50:32 2002
@@ -48,7 +48,7 @@
prio: MAX_PRIO-20, \
static_prio: MAX_PRIO-20, \
policy: SCHED_OTHER, \
- cpus_allowed: -1, \
+ cpus_allowed: CPU_MASK_ALL, \
mm: NULL, \
active_mm: &init_mm, \
run_list: LIST_HEAD_INIT(tsk.run_list), \
diff -urN -I \$.*\$ --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5.23/include/linux/sched.h working-2.5.23-sched/include/linux/sched.h
--- linux-2.5.23/include/linux/sched.h Thu Jun 20 01:28:52 2002
+++ working-2.5.23-sched/include/linux/sched.h Fri Jun 21 01:36:32 2002
@@ -21,6 +21,7 @@
#include <asm/page.h>
#include <asm/ptrace.h>
#include <asm/mmu.h>
+#include <asm/current.h>
#include <linux/smp.h>
#include <linux/sem.h>
@@ -264,7 +265,7 @@
unsigned long sleep_timestamp;
unsigned long policy;
- unsigned long cpus_allowed;
+ cpu_mask_t cpus_allowed;
unsigned int time_slice;
struct list_head tasks;
@@ -408,9 +409,17 @@
#define _STK_LIM (8*1024*1024)
#if CONFIG_SMP
-extern void set_cpus_allowed(task_t *p, unsigned long new_mask);
+extern void set_cpus_allowed(task_t *p, cpu_mask_t new_mask);
+static inline void migrate_to_cpu(unsigned int cpu)
+{
+ cpu_mask_t mask = CPU_MASK_NONE;
+ BUG_ON(!cpu_online(cpu));
+ __set_bit(cpu, (unsigned long *)&mask);
+ set_cpus_allowed(current, mask);
+}
#else
-# define set_cpus_allowed(p, new_mask) do { } while (0)
+#define set_cpus_allowed(p, new_mask) do { } while (0)
+#define migrate_to_cpu(cpu) do { BUG_ON((cpu) != 0); } while(0)
#endif
extern void set_user_nice(task_t *p, long nice);
@@ -477,8 +486,6 @@
/* per-UID process charging. */
extern struct user_struct * alloc_uid(uid_t);
extern void free_uid(struct user_struct *);
-
-#include <asm/current.h>
extern unsigned long itimer_ticks;
extern unsigned long itimer_next;
diff -urN -I \$.*\$ --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5.23/include/linux/smp.h working-2.5.23-sched/include/linux/smp.h
--- linux-2.5.23/include/linux/smp.h Thu Jun 20 01:28:52 2002
+++ working-2.5.23-sched/include/linux/smp.h Fri Jun 21 01:24:07 2002
@@ -91,6 +91,7 @@
#define __per_cpu_data
#define per_cpu(var, cpu) var
#define this_cpu(var) var
+#define cpu_online_mask(res,mask) do { *(res) = ((mask) & 1); } while(0)
#endif /* !SMP */
diff -urN -I \$.*\$ --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5.23/kernel/sched.c working-2.5.23-sched/kernel/sched.c
--- linux-2.5.23/kernel/sched.c Thu Jun 20 01:28:52 2002
+++ working-2.5.23-sched/kernel/sched.c Fri Jun 21 01:49:47 2002
@@ -367,7 +367,8 @@
*/
if (unlikely(sync && (rq->curr != p) &&
(p->thread_info->cpu != smp_processor_id()) &&
- (p->cpus_allowed & (1UL << smp_processor_id())))) {
+ test_bit(smp_processor_id(),
+ (unsigned long *)&p->cpus_allowed))) {
p->thread_info->cpu = smp_processor_id();
task_rq_unlock(rq, &flags);
@@ -647,7 +648,7 @@
#define CAN_MIGRATE_TASK(p,rq,this_cpu) \
((jiffies - (p)->sleep_timestamp > cache_decay_ticks) && \
((p) != (rq)->curr) && \
- ((p)->cpus_allowed & (1UL << (this_cpu))))
+ (test_bit((this_cpu), (unsigned long *)&(p)->cpus_allowed)))
if (!CAN_MIGRATE_TASK(tmp, busiest, this_cpu)) {
curr = curr->next;
@@ -1319,7 +1320,7 @@
asmlinkage int sys_sched_setaffinity(pid_t pid, unsigned int len,
unsigned long *user_mask_ptr)
{
- unsigned long new_mask;
+ cpu_mask_t online_mask, new_mask;
int retval;
task_t *p;
@@ -1329,8 +1330,8 @@
if (copy_from_user(&new_mask, user_mask_ptr, sizeof(new_mask)))
return -EFAULT;
- new_mask &= cpu_online_map;
- if (!new_mask)
+ cpu_online_mask(&online_mask, new_mask);
+ if (find_first_bit((unsigned long *)online_mask, NR_CPUS) == NR_CPUS)
return -EINVAL;
read_lock(&tasklist_lock);
@@ -1355,7 +1356,7 @@
goto out_unlock;
retval = 0;
- set_cpus_allowed(p, new_mask);
+ set_cpus_allowed(p, online_mask);
out_unlock:
put_task_struct(p);
@@ -1372,7 +1373,7 @@
unsigned long *user_mask_ptr)
{
unsigned int real_len;
- unsigned long mask;
+ cpu_mask_t mask;
int retval;
task_t *p;
@@ -1388,7 +1389,7 @@
goto out_unlock;
retval = 0;
- mask = p->cpus_allowed & cpu_online_map;
+ cpu_online_mask(&mask, p->cpus_allowed);
out_unlock:
read_unlock(&tasklist_lock);
@@ -1734,16 +1735,12 @@
* task must not exit() & deallocate itself prematurely. The
* call is not atomic; no spinlocks may be held.
*/
-void set_cpus_allowed(task_t *p, unsigned long new_mask)
+void set_cpus_allowed(task_t *p, cpu_mask_t new_mask)
{
unsigned long flags;
migration_req_t req;
runqueue_t *rq;
- new_mask &= cpu_online_map;
- if (!new_mask)
- BUG();
-
preempt_disable();
rq = task_rq_lock(p, &flags);
p->cpus_allowed = new_mask;
@@ -1751,7 +1748,7 @@
* Can the task run on the task's current CPU? If not then
* migrate the process off to a proper CPU.
*/
- if (new_mask & (1UL << p->thread_info->cpu)) {
+ if (test_bit(p->thread_info->cpu, (unsigned long *)&new_mask)) {
task_rq_unlock(rq, &flags);
goto out;
}
@@ -1760,7 +1757,8 @@
* it is sufficient to simply update the task's cpu field.
*/
if (!p->array && (p != rq->curr)) {
- p->thread_info->cpu = __ffs(p->cpus_allowed);
+ p->thread_info->cpu = any_online_cpu(p->cpus_allowed);
+ BUG_ON(p->thread_info->cpu >= NR_CPUS);
task_rq_unlock(rq, &flags);
goto out;
}
@@ -1795,7 +1793,7 @@
if (cpu != 0) {
while (!cpu_rq(0)->migration_thread)
yield();
- set_cpus_allowed(current, 1UL << cpu);
+ migrate_to_cpu(cpu);
}
printk("migration_task %d on cpu=%d\n", cpu, smp_processor_id());
ret = setscheduler(0, SCHED_FIFO, ¶m);
@@ -1826,7 +1825,7 @@
spin_unlock_irqrestore(&rq->lock, flags);
p = req->task;
- cpu_dest = __ffs(p->cpus_allowed);
+ cpu_dest = any_online_cpu(p->cpus_allowed);
rq_dest = cpu_rq(cpu_dest);
repeat:
cpu_src = p->thread_info->cpu;
@@ -1857,7 +1856,12 @@
{
int cpu;
- current->cpus_allowed = 1UL << 0;
+ /* Must not be preempted with cpus_allowed all zero. */
+ cpu = get_cpu();
+ current->cpus_allowed = CPU_MASK_NONE;
+ __set_bit(cpu, (unsigned long *)¤t->cpus_allowed);
+ put_cpu();
+
for (cpu = 0; cpu < NR_CPUS; cpu++) {
if (!cpu_online(cpu))
continue;
@@ -1865,7 +1869,7 @@
CLONE_FS | CLONE_FILES | CLONE_SIGNAL) < 0)
BUG();
}
- current->cpus_allowed = -1L;
+ current->cpus_allowed = CPU_MASK_ALL;
for (cpu = 0; cpu < NR_CPUS; cpu++) {
if (!cpu_online(cpu))
diff -urN -I \$.*\$ --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal linux-2.5.23/kernel/softirq.c working-2.5.23-sched/kernel/softirq.c
--- linux-2.5.23/kernel/softirq.c Thu Jun 20 01:28:52 2002
+++ working-2.5.23-sched/kernel/softirq.c Fri Jun 21 00:48:15 2002
@@ -370,8 +370,7 @@
current->flags |= PF_IOTHREAD;
sigfillset(¤t->blocked);
- /* Migrate to the right CPU */
- set_cpus_allowed(current, 1UL << cpu);
+ migrate_to_cpu(cpu);
if (smp_processor_id() != cpu)
BUG();
-- Anyone who quotes me in their sig is an idiot. -- Rusty Russell. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sun Jun 23 2002 - 22:00:22 EST