Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH 05/15] mm: allow read-ahead with IOCB_NOWAIT set
To:     Dave Chinner <david@fromorbit.com>
Cc:     Matthew Wilcox <willy@infradead.org>, io-uring@vger.kernel.org,
        linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
        linux-mm@kvack.org, akpm@linux-foundation.org,
        Johannes Weiner <hannes@cmpxchg.org>
References: <20200618144355.17324-1-axboe@kernel.dk>
 <20200618144355.17324-6-axboe@kernel.dk>
 <20200624010253.GB5369@dread.disaster.area>
 <20200624014645.GJ21350@casper.infradead.org>
 <bad52be9-ae44-171b-8dbf-0d98eedcadc0@kernel.dk>
 <70b0427c-7303-8f45-48bd-caa0562a2951@kernel.dk>
 <20200624164127.GP21350@casper.infradead.org>
 <8835b6f2-b3c5-c9a0-2119-1fb161cf87dd@kernel.dk>
 <20200810225601.GE2079@dread.disaster.area>
From:   Jens Axboe <axboe@kernel.dk>
Message-ID: <cf2384d3-8707-3a83-a667-8a0024867cdb@kernel.dk>
Date:   Mon, 10 Aug 2020 17:03:27 -0600
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.10.0
MIME-Version: 1.0
In-Reply-To: <20200810225601.GE2079@dread.disaster.area>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 8/10/20 4:56 PM, Dave Chinner wrote:
> On Wed, Jun 24, 2020 at 10:44:21AM -0600, Jens Axboe wrote:
>> On 6/24/20 10:41 AM, Matthew Wilcox wrote:
>>> On Wed, Jun 24, 2020 at 09:35:19AM -0600, Jens Axboe wrote:
>>>> On 6/24/20 9:00 AM, Jens Axboe wrote:
>>>>> On 6/23/20 7:46 PM, Matthew Wilcox wrote:
>>>>>> I'd be quite happy to add a gfp_t to struct readahead_control.
>>>>>> The other thing I've been looking into for other reasons is adding
>>>>>> a memalloc_nowait_{save,restore}, which would avoid passing down
>>>>>> the gfp_t.
>>>>>
>>>>> That was my first thought, having the memalloc_foo_save/restore for
>>>>> this. I don't think adding a gfp_t to readahead_control is going
>>>>> to be super useful, seems like the kind of thing that should be
>>>>> non-blocking by default.
>>>>
>>>> We're already doing memalloc_nofs_save/restore in
>>>> page_cache_readahead_unbounded(), so I think all we need is to just do a
>>>> noio dance in generic_file_buffered_read() and that should be enough.
>>>
>>> I think we can still sleep though, right?  I was thinking more
>>> like this:
>>>
>>> http://git.infradead.org/users/willy/linux.git/shortlog/refs/heads/memalloc
>>
>> Yeah, that's probably better. How do we want to handle this? I've already
>> got the other bits queued up. I can either add them to the series, or
>> pull a branch that'll go into Linus as well.
> 
> Jens, Willy,
> 
> Now that this patch has been merged and IOCB_NOWAIT semantics ifor
> buffered reads are broken in Linus' tree, what's the plan to get
> this regression fixed before 5.9 releases?

Not sure where Willy's work went on this topic, but it is on my radar. But
I think we can do something truly simple now that we have IOCB_NOIO:


diff --git a/include/linux/fs.h b/include/linux/fs.h
index bd7ec3eaeed0..f1cca4bfdd7b 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -3293,7 +3293,7 @@ static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
 	if (flags & RWF_NOWAIT) {
 		if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
 			return -EOPNOTSUPP;
-		kiocb_flags |= IOCB_NOWAIT;
+		kiocb_flags |= IOCB_NOWAIT | IOCB_NOIO;
 	}
 	if (flags & RWF_HIPRI)
 		kiocb_flags |= IOCB_HIPRI;

-- 
Jens Axboe