[PATCH] Hotplug CPU 3/4: locking

Rusty Russell (rusty@rustcorp.com.au)
Fri, 20 Sep 2002 16:56:04 +1000


Name: set_cpus_allowed can fail
Author: Rusty Russell
Status: Tested on 2.5.30 SMP
Depends: Hotcpu/cpumask.patch.gz

D: Simply moves the "is this mask valid?" check inside set_cpus_allowed,
D: since it might be racy otherwise in the context of adding/removing
D: CPUs. This also introduces the CPU brlock.

diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .3670-linux-2.5.36/arch/i386/kernel/smpboot.c .3670-linux-2.5.36.updated/arch/i386/kernel/smpboot.c
--- .3670-linux-2.5.36/arch/i386/kernel/smpboot.c 2002-09-18 14:44:00.000000000 +1000
+++ .3670-linux-2.5.36.updated/arch/i386/kernel/smpboot.c 2002-09-18 14:44:45.000000000 +1000
@@ -45,6 +45,7 @@

#include <linux/delay.h>
#include <linux/mc146818rtc.h>
+#include <linux/brlock.h>
#include <asm/pgalloc.h>
#include <asm/tlbflush.h>
#include <asm/smpboot.h>
@@ -456,8 +457,9 @@ int __init start_secondary(void *unused)
* the local TLBs too.
*/
local_flush_tlb();
- set_bit(smp_processor_id(), cpu_online_map);
- wmb();
+ br_write_lock_irq(BR_CPU_LOCK);
+ set_bit(smp_processor_id(), cpu_online_map);
+ br_write_unlock_irq(BR_CPU_LOCK);
return cpu_idle();
}

diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .3670-linux-2.5.36/arch/ppc/kernel/smp.c .3670-linux-2.5.36.updated/arch/ppc/kernel/smp.c
--- .3670-linux-2.5.36/arch/ppc/kernel/smp.c 2002-09-18 14:44:00.000000000 +1000
+++ .3670-linux-2.5.36.updated/arch/ppc/kernel/smp.c 2002-09-18 14:44:22.000000000 +1000
@@ -432,7 +432,9 @@ int __cpu_up(unsigned int cpu)
printk("Processor %d found.\n", cpu);

smp_ops->give_timebase();
- set_bit(cpu, cpu_online_map);
+ br_write_lock_irq(BR_CPU_LOCK);
+ set_bit(cpu, &cpu_online_map);
+ br_write_unlock_irq(BR_CPU_LOCK);
return 0;
}

diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .3670-linux-2.5.36/include/linux/brlock.h .3670-linux-2.5.36.updated/include/linux/brlock.h
--- .3670-linux-2.5.36/include/linux/brlock.h 2002-08-28 09:29:52.000000000 +1000
+++ .3670-linux-2.5.36.updated/include/linux/brlock.h 2002-09-18 14:44:22.000000000 +1000
@@ -34,6 +34,7 @@
/* Register bigreader lock indices here. */
enum brlock_indices {
BR_NETPROTO_LOCK,
+ BR_CPU_LOCK,
__BR_END
};

diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .3670-linux-2.5.36/include/linux/sched.h .3670-linux-2.5.36.updated/include/linux/sched.h
--- .3670-linux-2.5.36/include/linux/sched.h 2002-09-18 14:44:00.000000000 +1000
+++ .3670-linux-2.5.36.updated/include/linux/sched.h 2002-09-18 14:44:22.000000000 +1000
@@ -444,9 +444,14 @@ do { if (atomic_dec_and_test(&(tsk)->usa
#define _STK_LIM (8*1024*1024)

#if CONFIG_SMP
-extern void set_cpus_allowed(task_t *p, const unsigned long new_mask[]);
+extern int set_cpus_allowed(task_t *p, const unsigned long new_mask[]);
#else
-# define set_cpus_allowed(p, new_mask) do { } while (0)
+static inline int set_cpus_allowed(struct task_struct *p,
+ const unsigned long *new_mask)
+{
+ if (!(*new_mask & 1)) return -EINVAL;
+ return 0;
+}
#endif
#define CPU_MASK_NONE { 0 }
#define CPU_MASK_ALL { [0 ... BITS_TO_LONG(NR_CPUS)-1 ] = ~0UL }
@@ -454,9 +459,9 @@ extern void set_cpus_allowed(task_t *p,
static inline void migrate_to_cpu(unsigned int cpu)
{
DECLARE_BITMAP(mask, NR_CPUS) = CPU_MASK_NONE;
- BUG_ON(!cpu_online(cpu));
__set_bit(cpu, mask);
- set_cpus_allowed(current, mask);
+ if (set_cpus_allowed(current, mask) != 0)
+ BUG();
}

extern void set_user_nice(task_t *p, long nice);
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .3670-linux-2.5.36/kernel/sched.c .3670-linux-2.5.36.updated/kernel/sched.c
--- .3670-linux-2.5.36/kernel/sched.c 2002-09-18 14:44:00.000000000 +1000
+++ .3670-linux-2.5.36.updated/kernel/sched.c 2002-09-18 14:44:22.000000000 +1000
@@ -30,6 +30,7 @@
#include <linux/notifier.h>
#include <linux/delay.h>
#include <linux/bitops.h>
+#include <linux/brlock.h>

/*
* Convert user-nice values [ -20 ... 0 ... 19 ]
@@ -1559,9 +1560,6 @@ asmlinkage int sys_sched_setaffinity(pid
if (copy_from_user(&new_mask, user_mask_ptr, sizeof(new_mask)))
return -EFAULT;

- if (any_online_cpu(new_mask) == NR_CPUS)
- return -EINVAL;
-
read_lock(&tasklist_lock);

p = find_process_by_pid(pid);
@@ -1583,8 +1581,7 @@ asmlinkage int sys_sched_setaffinity(pid
!capable(CAP_SYS_NICE))
goto out_unlock;

- retval = 0;
- set_cpus_allowed(p, new_mask);
+ retval = set_cpus_allowed(p, new_mask);

out_unlock:
put_task_struct(p);
@@ -1921,29 +1918,30 @@ typedef struct {
* task must not exit() & deallocate itself prematurely. The
* call is not atomic; no spinlocks may be held.
*/
-void set_cpus_allowed(task_t *p, const unsigned long new_mask[])
+int set_cpus_allowed(task_t *p, const unsigned long new_mask[])
{
unsigned long flags;
migration_req_t req;
runqueue_t *rq;
-
-#if 0 /* FIXME: Grab cpu_lock, return error on this case. --RR */
- new_mask &= cpu_online_map;
- if (!new_mask)
- BUG();
-#endif
+ int ret = 0;

preempt_disable();
rq = task_rq_lock(p, &flags);
+ /* Let's not deal with cpus leaving us */
+ br_read_lock(BR_CPU_LOCK);
+ if (any_online_cpu(new_mask) == NR_CPUS) {
+ ret = -EINVAL;
+ goto out_unlock;
+ }
+
memcpy(p->cpus_allowed, new_mask, sizeof(p->cpus_allowed));
/*
* Can the task run on the task's current CPU? If not then
* migrate the thread off to a proper CPU.
*/
- if (test_bit(task_cpu(p), new_mask)) {
- task_rq_unlock(rq, &flags);
- goto out;
- }
+ if (test_bit(task_cpu(p), new_mask))
+ goto out_unlock;
+
/*
* If the task is not on a runqueue (and not running), then
* it is sufficient to simply update the task's cpu field.
@@ -1951,17 +1949,24 @@ void set_cpus_allowed(task_t *p, const u
if (!p->array && !task_running(rq, p)) {
set_task_cpu(p, any_online_cpu(p->cpus_allowed));
task_rq_unlock(rq, &flags);
- goto out;
+ goto out_unlock;
}
init_completion(&req.done);
req.task = p;
list_add(&req.list, &rq->migration_queue);
+ br_read_unlock(BR_CPU_LOCK);
task_rq_unlock(rq, &flags);
wake_up_process(rq->migration_thread);

wait_for_completion(&req.done);
-out:
preempt_enable();
+ return 0;
+
+ out_unlock:
+ br_read_unlock(BR_CPU_LOCK);
+ task_rq_unlock(rq, &flags);
+ preempt_enable();
+ return ret;
}

/*

--
  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/