Subject: Re: [PATCH 4/4] perf core: Add backward attribute to perf event
To: Alexei Starovoitov <ast@kernel.org>,
        Arnaldo Carvalho de Melo <acme@redhat.com>,
        Peter Zijlstra <peterz@infradead.org>
References: <1459147292-239310-1-git-send-email-wangnan0@huawei.com>
 <1459147292-239310-5-git-send-email-wangnan0@huawei.com>
CC: <linux-kernel@vger.kernel.org>, Brendan Gregg <brendan.d.gregg@gmail.com>,
        He Kuang <hekuang@huawei.com>, Jiri Olsa <jolsa@kernel.org>,
        Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>,
        Namhyung Kim <namhyung@kernel.org>, <pi3orama@163.com>,
        Zefan Li <lizefan@huawei.com>
From: "Wangnan (F)" <wangnan0@huawei.com>
Message-ID: <56F9E1F4.2050807@huawei.com>
Date: Tue, 29 Mar 2016 10:01:24 +0800
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:38.0) Gecko/20100101
 Thunderbird/38.5.0
MIME-Version: 1.0
In-Reply-To: <1459147292-239310-5-git-send-email-wangnan0@huawei.com>
Content-Type: text/plain; charset="utf-8"; format=flowed
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1310
Lines: 51


On 2016/3/28 14:41, Wang Nan wrote:

[SNIP]

>
> To prevent this problem, we need to find a way to ensure the ring buffer
> is stable during reading. ioctl(PERF_EVENT_IOC_PAUSE_OUTPUT) is
> suggested because its overhead is lower than
> ioctl(PERF_EVENT_IOC_ENABLE).
>

Add comment:

By carefully verifying 'header' pointer, reader can avoid pausing the
ring-buffer. For example:

     /* A union of all possible events */
     union perf_event event;

     p = head = perf_mmap__read_head();
     while (true) {
         /* copy header of next event */
         fetch(&event.header, p, sizeof(event.header));

         /* read 'head' pointer */
         head = perf_mmap__read_head();

         /* check overwritten: is the header good? */
         if (!verify(sizeof(event.header), p, head))
             break;

         /* copy the whole event */
         fetch(&event, p, event.header.size);

         /* read 'head' pointer again */
         head = perf_mmap__read_head();

         /* is the whole event good? */
         if (!verify(event.header.size, p, head))
             break;
         p += event.header.size;
     }

However, the overhead is high because:

  a) In-place decoding is unsafe. Copy-verifying-decode is required.
  b) Fetching 'head' pointer requires additional synchronization.