From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Tue, 16 Sep 2014 12:03:16 -0500 From: Felipe Balbi To: Felipe Balbi CC: Jonathan Cameron , , Subject: Re: [RESEND PATCH] iio: light: add support for TI's opt3001 light sensor Message-ID: <20140916170316.GD19010@saruman.home> Reply-To: References: <1409671066-25747-1-git-send-email-balbi@ti.com> <54147632.9090204@kernel.org> <20140915052137.GA11866@saruman.home> MIME-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="TybLhxa8M7aNoW+V" In-Reply-To: <20140915052137.GA11866@saruman.home> List-ID: --TybLhxa8M7aNoW+V Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable ping On Mon, Sep 15, 2014 at 12:21:37AM -0500, Felipe Balbi wrote: > Hi, >=20 > On Sat, Sep 13, 2014 at 05:52:02PM +0100, Jonathan Cameron wrote: > > On 02/09/14 16:17, Felipe Balbi wrote: > > > TI's opt3001 light sensor is a simple and yet powerful > > > little device. The device provides 99% IR rejection, > > > Automatic full-scale, very low power consumption and > > > measurements from 0.01 to 83k lux. > > >=20 > > > This patch adds support for that device using the IIO > > > framework. > > >=20 > > > Signed-off-by: Felipe Balbi > > > --- > > >=20 > > > Resending as I saw no changes on the thread. > > Hi Felipe, > >=20 > > Sorry for the delay on this, entirely my fault - been busy and forgot > > I still had questions about what was going on in here (yup its the > > hysteresis bit again!) >=20 > right, this is starting to become way too much headache for such a > simple device. Sorry will not help me getting this driver upstream. When > I first sent this (August 6), we didn't even have v3.17-rc1, now we're > about to tag -rc5 and I'm worried this driver will not hit v3.18 merge > window. >=20 > > Anyhow, I'm afraid I am still a little confused about the meaning you > > have assigned to Hysteresis in this driver. > >=20 > > Let me conjecture on what might be going on here (I may be entirely > > wrong). > >=20 > > Normally a hysteresis value in IIO is defined as the 'distance' back > > from a threshold that a signal must go before it may retrip the > > threshold. > > This threshold value is separately controlled. Thus if we have a > > rising threshold of 10 and an hysteresis of 2 - to get two events the > > signal must first rise past 10, then drop back below 8 and rise again > > past 10. > > If it drops below 10 but not 8 and rises again past 10 then we will > > not get an event. > >=20 > > So having the same register for both the hysteresis and the threshold > > doesn't with this description make much sense. It would mean that you > > could only have a threshold of say 10 and a hysteresis of 10, thus in > > effect meaning the signal would always have to cross 0 before the next > > event whatever the combined threshold / hysteresis value? > >=20 > > Perhaps instead the device is automatically adjusting the threshold > > when we cross it and the 'hysteresis' here is with respect to a the > > previous threshold? > >=20 > > Thus if we start with a value of 0 and hysteresis is set to 2 it will > > trigger an event at: > >=20 > > 2, 4, 6, 8, 10 as we rise? > >=20 > > This sort of auto adjustment of levels isn't uncommon in light sensors > > (where the point of the interrupt is to notify the operating system > > that a 'significant' change has occurred and things like screen > > brightness may need adjusting. > >=20 > > If so then the current hysteresis interface does not apply, nor does > > the Rate of Change (ROC) interface as this is dependent on amount of > > change, not how fast it changed. Hence we needs something new to > > handle this cleanly. I would suggest a new event type. Perhaps > > something with sysfs attr naming along the lines of > > What: /sys/.../iio:deviceX/events/in_light_change_rising_en > > What: /sys/.../iio:deviceX/events/in_light_change_rising_value > >=20 > > etc? >=20 > will you just tell me what you want ? I really cannot give a crap > anymore. This has already taken me over a month of my time for such a > simple little device, not to mention your confusing and contradicting > change requests. >=20 > (could you also trim your responses ? it's very annoying to scroll so > much) >=20 > > > +#define OPT3001_RESULT 0x00 > > > +#define OPT3001_CONFIGURATION 0x01 > > > +#define OPT3001_LOW_LIMIT 0x02 > > > +#define OPT3001_HIGH_LIMIT 0x03 > > > +#define OPT3001_MANUFACTURER_ID 0x7e > > > +#define OPT3001_DEVICE_ID 0x7f > > > + > > > +#define OPT3001_CONFIGURATION_RN_MASK (0xf << 12) > > > +#define OPT3001_CONFIGURATION_RN_AUTO (0xc << 12) > > > + > > > +#define OPT3001_CONFIGURATION_CT BIT(11) > > > + > > > +#define OPT3001_CONFIGURATION_M_MASK (3 << 9) > > > +#define OPT3001_CONFIGURATION_M_SHUTDOWN (0 << 9) > > > +#define OPT3001_CONFIGURATION_M_SINGLE (1 << 9) > > > +#define OPT3001_CONFIGURATION_M_CONTINUOUS (2 << 9) /* also 3 << 9 */ > > > + > >=20 > > I guess this naming is straight off the datasheet, but it is rather > > more cryptic than perhaps it needs to be! That's kind of an issue > > with the datasheet choices rather than what you have here however! >=20 > man, are you nit-picky!! These are named as such to make grepping on > documentation easier. It's better to have something that matches > documentation, don't you think ? otherwise, future users/developers of > these driver will need either a shit ton of comments explaining that A > maps to B in docs, or will need a fscking crystal ball to read my mind. > Assuming I'll still remember what I meant. >=20 > > > +static int opt3001_remove(struct i2c_client *client) > > > +{ > > > + struct iio_dev *iio =3D i2c_get_clientdata(client); > > > + struct opt3001 *opt =3D iio_priv(iio); > > > + int ret; > > > + u16 reg; > > > + > > > + free_irq(client->irq, iio); > > > + iio_device_unregister(iio); > > > + > > > + ret =3D i2c_smbus_read_word_swapped(opt->client, OPT3001_CONFIGURAT= ION); > > > + if (ret < 0) { > > > + dev_err(opt->dev, "failed to read register %02x\n", > > > + OPT3001_CONFIGURATION); > > > + return ret; > > > + } > > > + > > > + reg =3D ret; > > > + opt3001_set_mode(opt, ®, OPT3001_CONFIGURATION_M_SHUTDOWN); > > > + > > > + ret =3D i2c_smbus_write_word_swapped(opt->client, OPT3001_CONFIGURA= TION, > > > + reg); > > > + if (ret < 0) { > > > + dev_err(opt->dev, "failed to write register %02x\n", > > > + OPT3001_CONFIGURATION); > > > + return ret; > > > + } > > > + > > > + iio_device_free(iio); > > > > Use the devm_iio_device_alloc and you can drop the need to free it. > > I don't really mind, but I'll almost guarantee that someone will post > > a follow up patch doing this if you don't. As it will be ever so > > slightly cleaner, I'll probably take that patch. >=20 > here's the original driver: >=20 > http://www.spinics.net/lists/linux-iio/msg14331.html >=20 > notice that it *WAS* *USING* devm_iio_device_alloc(), until: >=20 > http://www.spinics.net/lists/linux-iio/msg14421.html >=20 > you *SPECIFICALLY* asked for *NON* *DEVM* versions!! >=20 > So figure out what you really want, let me know and I'll code it all up > quickly and hopefully still get this into v3.18 merge window. I sent > this driver *very* early to be doubly sure it would hit v3.18 and there > was a long hiatus from yourself which is now jeopardizing what I was > expecting. That, coupled with your contradicting requests, has just put > me in a bad mood, even before Monday, hurray. >=20 > cheers >=20 > --=20 > balbi --=20 balbi --TybLhxa8M7aNoW+V Content-Type: application/pgp-signature; name="signature.asc" Content-Description: Digital signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1 iQIcBAEBAgAGBQJUGG1UAAoJEIaOsuA1yqREHaoP/2QwovBmVpEMKRtBA8+6PwSB /Y157FlnE0iTvqhN1wnOt3HM40N8/EkxDP+VkEBvdODGvj+4BrIXtRdEAwKSdK9d fLbYAnEuC5XEyuBaR+fiqF07hltu6RJlcFq88ccbk37kVB278AIMSIpZ9euuZcUp dgOFgl5GnXzKsMEV73S2BymR8NyB4i8omT75SDcTdeuOEq3G9upLioJuS6nJDRDk 3cUsAtiYpTzk+hF8kCMrGxDqVoZDXMgtHNkKa9DpecrDyY48w2gpA9F5NCMAhhIr aKnmwkmjhXbFAbVMx/AoYkTaeScWwXkeqlCAAIWSIVeWx8G78O0MG2W8LCbIAc9j MmvuKoneOwI3vO6yPLMMhECQ4PdL80YAZsb9qQdqL+vJdl3+15pfHstw798f81Jf KKE2jZY+GyNT5aXwQge79WR3FRqDuQkbW+BhxDRKjV5hb1N954Rd62CZUdG6kqH5 8qjwsrbw/dlC+Wbp2sKJB27u9THc6C3CYMid3ALGeUPC4Pc453tSRFaJcgEBueB2 q4kmjssJsteQqOYWFtjho6+1SZ5zLrOL1upyIbgKDGE0dhUB0bFix5VnLNaBtnlA GpeQgpcsW4TVQE8U4s1YZYq7C7yVTfQRqml3Yo6Gq9G68Jwl/51KQIJXuPkZAt/z N23r5HCXppU+TG8nScXj =m8G8 -----END PGP SIGNATURE----- --TybLhxa8M7aNoW+V--