DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org E5A4221B64
Date: Sun, 3 Sep 2017 17:24:37 +0100
From: Jonathan Cameron <jic23@kernel.org>
To: harinath Nampally <harinath922@gmail.com>
Cc: Martin Kepplinger <martink@posteo.de>, knaack.h@gmx.de,
        lars@metafoo.de, Peter Meerwald-Stadler <pmeerw@pmeerw.net>,
        Greg KH <gregkh@linuxfoundation.org>, linux-iio@vger.kernel.org,
        linux-kernel@vger.kernel.org, Alison Schofield <amsfield22@gmail.com>
Subject: Re: [PATCH v5] iio: accel: mma8452: improvements to handle multiple
 events
Message-ID: <20170903172437.0aaec34d@archlinux>
In-Reply-To: <CAAGUq_rMONY=upVapudHC+fpM2PQC-zUWnm9yJ4YCt4FQ_HjxQ@mail.gmail.com>
References: <1503879782-11945-1-git-send-email-harinath922@gmail.com>
        <1699ec2db96b498478b95a3cbcfe73ca@posteo.de>
        <CAAGUq_ryozztz1B_-zuQ4SgR5Yh7Lznk=mrk7pd4YtDuScLyJw@mail.gmail.com>
        <CAAGUq_rMONY=upVapudHC+fpM2PQC-zUWnm9yJ4YCt4FQ_HjxQ@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8BIT
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 3029
Lines: 73

On Tue, 29 Aug 2017 23:01:16 -0400
harinath Nampally <harinath922@gmail.com> wrote:

> > We should never say "transient is for rising
> > direction" or "ff_mt is for falling direction". any combination is fine.  
> 
> Ok I agree that there is no hard and fast rule that "transient is for rising
> direction" or "ff_mt is for falling direction".
> But in our case, datasheet for these chips define these events based on
> acceleration magnitude rising or falling below a set threshold value.
> 
> For quick reference, below excerpts are from fxls8471 datasheet:
> Motion Event: "When the acceleration exceeds a set threshold for a set
> amount of time,
> the motion interrupt is asserted."
> 
> Freefall event: "The detection of “Freefall” involves the monitoring
> of the X, Y, and Z axes
> for the condition where the acceleration magnitude is below a
> user-specified threshold
> for a user-definable amount of time"
> 
> Transient event: "When the high-pass filter is bypassed, the
> functionality becomes
> similar to the motion-detection function; in this mode, acceleration
> greater than
> a programmable threshold is detected (along an axis)."
> 
> Therefore I think in this driver freefall event is defined as
> 'falling' event type and
> motion event is defined as 'rising' event type and Transient is also defined as
> 'rising' event type.
>  As you might already know that mma8562 and mma8563 doesn't have
> transient event support
> but they do have freefall and motion event support which are defined
> as 'fall' and 'rise'
> event types respectively. Please note in this driver, motion event is
> enabled/configured only
> for mma8652 and mma8653.
> Therefore if I read/write sysfs node for 'rise' it should use the
> FF_MT registers for mma8652 and mma853, but for all others like
> mma8451, mma8452 and
> mma8453 which has transient event support it picks the Transient
> registers if enabled. Also please
> note transient event is enabled(but not motion event) for mma8451,
> mma8452 and mma8453.
> The problem seems like we have two different events(motion and
> transient) that are defined
> as same event type 'rising' but in fact both motion and transient are
> pretty much similar as they
> both raise interrupt flag when the acceleration magnitude rises above
> the threshold.
> Only difference is transient event has its own event config registers
> with High pass filter.
> If HPF bypassed using config register transient event acts like motion
> detection event.

> 
> That was my understanding but please correct me if I am wrong.

I agree with your understanding.  It's a rising threshold, just that the input
will only reflect high frequency changes in the signal.

> 
> > Only freefall mode needs one fix: remembering to which set of registers to fall back when
> > disabling it.  
> 
> I don't quite understand what you mean by 'to fall back when disabling
> it'. Please elaborate. I would
> appreciate if you could suggest your logic in the form of pseudo-code.
> Thanks for your time
> 
...