[Devel] [PATCH 1/2] fs/fuse kio: add FUSE_S_FAIL_IMMEDIATELY check in pcs_fuse_submit()

[Pavel Butsykin]

On 01.02.2019 18:13, Pavel Butsykin wrote:
> 
> 
> On 01.02.2019 17:24, Kirill Tkhai wrote:
>> On 24.01.2019 16:12, Pavel Butsykin wrote:
>>> Fuse file with FUSE_S_FAIL_IMMEDIATELY state should not allow to 
>>> execute new
>>> requests. But in case of kio requests it doesn't work because the 
>>> status check
>>> is located behind kio.op->req_send(). To fix this let's add the 
>>> status check
>>> in pcs_fuse_submit().
>>>
>>> Signed-off-by: Pavel Butsykin <pbutsykin at virtuozzo.com>
>>> ---
>>> Note:
>>> applies on top of "[PATCH v2] fs/fuse kio: invalidate files for kio"
>>>
>>>   fs/fuse/kio/pcs/pcs_fuse_kdirect.c | 5 +++++
>>>   1 file changed, 5 insertions(+)
>>>
>>> diff --git a/fs/fuse/kio/pcs/pcs_fuse_kdirect.c 
>>> b/fs/fuse/kio/pcs/pcs_fuse_kdirect.c
>>> index da4b5fba03fb..3ca1ce2d6bd5 100644
>>> --- a/fs/fuse/kio/pcs/pcs_fuse_kdirect.c
>>> +++ b/fs/fuse/kio/pcs/pcs_fuse_kdirect.c
>>> @@ -986,6 +986,11 @@ error:
>>>   submit:
>>>       spin_lock(&di->kq_lock);
>>> +    if (req->ff && test_bit(FUSE_S_FAIL_IMMEDIATELY, 
>>> &req->ff->ff_state)) {
>>
>> FUSE_S_FAIL_IMMEDIATELY is set under fc->lock, while it's checked 
>> under kq_lock.
> 
> Actually this lock is needed to protect listing fi->rw_files, not for
> set_bit.
> 
>> How do you guarantee visibility?
> 
> It's atomic operation, so I should just guarantee
> synchronization with fuse_invalidate_files() which is carried out by
> set_bit(FUSE_S_FAIL_IMMEDIATELY, ) + kpcs_kill_requests() and the locked
> check.

But perhaps for architectures "not x86" need barriers.

>>
>>> +        spin_unlock(&di->kq_lock);
>>> +        req->out.h.error = -EIO;
>>> +        goto error;
>>> +    }
>>>       list_add_tail(&req->list, &di->kq);
>>>       spin_unlock(&di->kq_lock);
>>>