On 9/4/24 11:53 AM, Hou Tao wrote:
>
>
> On 9/3/2024 5:34 PM, Jingbo Xu wrote:
>>
>> On 8/31/24 5:37 PM, Hou Tao wrote:
>>> From: Hou Tao <houtao1@xxxxxxxxxx>
>>>
>>> When invoking virtio_fs_enqueue_req() through kworker, both the
>>> allocation of the sg array and the bounce buffer still use GFP_ATOMIC.
>>> Considering the size of the sg array may be greater than PAGE_SIZE, use
>>> GFP_NOFS instead of GFP_ATOMIC to lower the possibility of memory
>>> allocation failure and to avoid unnecessarily depleting the atomic
>>> reserves. GFP_NOFS is not passed to virtio_fs_enqueue_req() directly,
>>> GFP_KERNEL and memalloc_nofs_{save|restore} helpers are used instead.
>>>
>>> It may seem OK to pass GFP_NOFS to virtio_fs_enqueue_req() as well when
>>> queuing the request for the first time, but this is not the case. The
>>> reason is that fuse_request_queue_background() may call
>>> ->queue_request_and_unlock() while holding fc->bg_lock, which is a
>>> spin-lock. Therefore, still use GFP_ATOMIC for it.
>> Actually, .wake_pending_and_unlock() is called under fiq->lock and
>> GFP_ATOMIC is requisite.
>
> Er, but virtio_fs_wake_pending_and_unlock() unlocks fiq->lock before
> queuing the request.
Alright, I missed that :(
--
Thanks,
Jingbo
