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: [BUG] About "io_uring: add more uring info to fdinfo for debug"
To:     Hao Xu <haoxu@linux.alibaba.com>,
        Eric Dumazet <edumazet@google.com>,
        Eric Dumazet <eric.dumazet@gmail.com>
Cc:     LKML <linux-kernel@vger.kernel.org>
References: <9ac22d4d-c841-651a-fdd5-9fb3a65c57ab@gmail.com>
 <e72cdd44-c027-e51c-8b57-24f39bf21d87@kernel.dk>
 <3b7dbb90-7fd5-d30f-cfe0-dd83f2cd4482@kernel.dk>
 <CANn89iJv6cRbK95Gwj=rrz2+X+Q4KQ7EtEcpz0Md0Vs1_iABPw@mail.gmail.com>
 <30b7018d-f60d-0c57-b3d6-92c1cf0762b9@kernel.dk>
 <03e9b953-a9c2-c3bc-c265-b3fff76d7739@linux.alibaba.com>
From:   Jens Axboe <axboe@kernel.dk>
Message-ID: <7c6ec470-0985-2ee8-4cbd-1f9649c7b69c@kernel.dk>
Date:   Fri, 29 Oct 2021 11:56:53 -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: <03e9b953-a9c2-c3bc-c265-b3fff76d7739@linux.alibaba.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 10/29/21 11:54 AM, Hao Xu wrote:
> Hi Jens,
> I was thinking that print all the entries including the unused ones
> would make the list very long(real product use hundreds or thousands
> while there may be just several useful sqes). And that may be confused,
> a example is: sq_array[0-999] = 0, we have only one sqe sqes[0] in use.
> then we'll print sqe[0] 1000 times, though we can get to know the real
> useful one by sq_head and tail info after calculation.
> And if only print the useful ones, exposing the original counters will
> make the loop condition a bit simpler than exposing the ring index(since
> the original counters depend on natural overflow) though we cannot do
> that now since realizing issues like what Eric reports.

We can revisit in the merge window, I wrote up the pull requests for
this branch (among others) already. The most important part was fixing
the looping to appropriately cap it.

-- 
Jens Axboe