A separate and intentionally suboptimal patch to simplify the review of this and the next changes.
And the new helpers, stop_work_alloc(cpumask) and stop_work_free(cpumask), should be called if the caller is going to use cpu_stopper->stop_work's. Note that 2 callers can never deadlock even if their cpumask's overlap, they always "lock" cpu_stopper->stop_owner's in the same order as if we had another per-cpu mutex. This is obviously greatly inefficient, this will be fixed later. Signed-off-by: Oleg Nesterov <[email protected]> --- kernel/stop_machine.c | 50 +++++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 50 insertions(+), 0 deletions(-) diff --git a/kernel/stop_machine.c b/kernel/stop_machine.c index 6212208..3d5d810 100644 --- a/kernel/stop_machine.c +++ b/kernel/stop_machine.c @@ -42,11 +42,61 @@ struct cpu_stopper { struct list_head works; /* list of pending works */ struct cpu_stop_work stop_work; /* for stop_cpus */ + struct task_struct *stop_owner; }; static DEFINE_PER_CPU(struct cpu_stopper, cpu_stopper); static bool stop_machine_initialized = false; +static DECLARE_WAIT_QUEUE_HEAD(stop_work_wq); + +static void stop_work_free_one(int cpu) +{ + struct cpu_stopper *stopper = &per_cpu(cpu_stopper, cpu); + /* Can be NULL if stop_work_alloc(wait => false) fails */ + if (likely(stopper->stop_owner == current)) + stopper->stop_owner = NULL; +} + +static void stop_work_free(const struct cpumask *cpumask) +{ + int cpu; + + for_each_cpu(cpu, cpumask) + stop_work_free_one(cpu); + wake_up_all(&stop_work_wq); +} + +static struct cpu_stop_work *stop_work_alloc_one(int cpu, bool wait) +{ + struct cpu_stopper *stopper = &per_cpu(cpu_stopper, cpu); + + if (cmpxchg(&stopper->stop_owner, NULL, current) == NULL) + goto done; + + if (!wait) + return NULL; + + __wait_event(stop_work_wq, + cmpxchg(&stopper->stop_owner, NULL, current) == NULL); +done: + return &stopper->stop_work; +} + +static bool stop_work_alloc(const struct cpumask *cpumask, bool wait) +{ + int cpu; + + for_each_cpu(cpu, cpumask) { + if (stop_work_alloc_one(cpu, wait)) + continue; + stop_work_free(cpumask); + return false; + } + + return true; +} + /* * Avoids a race between stop_two_cpus and global stop_cpus, where * the stoppers could get queued up in reverse order, leading to -- 1.5.5.1 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
