On 03/21/2019 06:00 PM, Peter Zijlstra wrote:
> On Thu, Mar 21, 2019 at 05:45:12PM -0400, Waiman Long wrote:
>> If the freeing queue has many objects, freeing all of them consecutively
>> may cause soft lockup especially on a debug kernel. So kmem_free_up_q()
>> is modified to call cond_resched() if running in the process context.
>>
>> Signed-off-by: Waiman Long <longman@xxxxxxxxxx>
>> ---
>> mm/slab_common.c | 11 ++++++++++-
>> 1 file changed, 10 insertions(+), 1 deletion(-)
>>
>> diff --git a/mm/slab_common.c b/mm/slab_common.c
>> index dba20b4208f1..633a1d0f6d20 100644
>> --- a/mm/slab_common.c
>> +++ b/mm/slab_common.c
>> @@ -1622,11 +1622,14 @@ EXPORT_SYMBOL_GPL(kmem_free_q_add);
>> * kmem_free_up_q - free all the objects in the freeing queue
>> * @head: freeing queue head
>> *
>> - * Free all the objects in the freeing queue.
>> + * Free all the objects in the freeing queue. The caller cannot hold any
>> + * non-sleeping locks.
>> */
>> void kmem_free_up_q(struct kmem_free_q_head *head)
>> {
>> struct kmem_free_q_node *node, *next;
>> + bool do_resched = !in_irq();
>> + int cnt = 0;
>>
>> for (node = head->first; node; node = next) {
>> next = node->next;
>> @@ -1634,6 +1637,12 @@ void kmem_free_up_q(struct kmem_free_q_head *head)
>> kmem_cache_free(node->cachep, node);
>> else
>> kfree(node);
>> + /*
>> + * Call cond_resched() every 256 objects freed when in
>> + * process context.
>> + */
>> + if (do_resched && !(++cnt & 0xff))
>> + cond_resched();
> Why not just: cond_resched() ?
cond_resched() calls ___might_sleep(). So it is prudent to check for
process context first to avoid erroneous message. Yes, I can call
cond_resched() after every free. I added the count just to not call it
too frequently.
Cheers,
Longman
