https://devzone.nordicsemi.com/f/nordic-q-a/19725/app_fifo-wrapping-problemThe app_fifo support e.g. in SDK 12.1.0, appears to have a serious problem when the read_pos/write_pos values wrap. i.e. when the fifo has been used over a period of time, regardless of the configured fifo size, since the pos vars are continually incrementeden-USTelligent Community 13Wed, 22 Feb 2017 11:08:25 GMThttps://devzone.nordicsemi.com/thread/76751?ContentTypeID=1Wed, 22 Feb 2017 11:08:25 GMT137ad170-7792-4731-bb38-c0d22fbe4515:567a64c6-a748-4d71-a00d-f21165c35437Will<p>Confirmed</p> <p>Many Thanks</p><div style="clear:both;"></div>https://devzone.nordicsemi.com/thread/76750?ContentTypeID=1Thu, 16 Feb 2017 10:40:36 GMT137ad170-7792-4731-bb38-c0d22fbe4515:01eef35b-63d6-4301-9f2a-e1bfaac9de22Ole Bauck<p>write_pos and read_pos are uint32_t (unsigned 32 bit integers) so</p> <pre><code>write_pos - read_pos </code></pre> <p>is essentially the same as</p> <pre><code>write_pos - read_pos + UINT32_MAX + 1 </code></pre> <p>Where UINT32_MAX is 0xFFFFFFFF. Therefore the check</p> <pre><code>FIFO_LENGTH() &lt;= p_fifo-&gt;buf_size_mask </code></pre> <p>Will still be valid.</p> <p>For example see this <a href="http://stackoverflow.com/questions/2760502/question-about-c-behaviour-for-unsigned-integer-underflow">post</a>.</p><div style="clear:both;"></div>