From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ppsw-50.csi.cam.ac.uk ([131.111.8.150]:41490 "EHLO ppsw-50.csi.cam.ac.uk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751960Ab0JEL5D (ORCPT ); Tue, 5 Oct 2010 07:57:03 -0400 Message-ID: <4CAB13DC.9080600@cam.ac.uk> Date: Tue, 05 Oct 2010 13:02:36 +0100 From: Jonathan Cameron MIME-Version: 1.0 To: "Hennerich, Michael" CC: "linux-iio@vger.kernel.org" , Drivers Subject: Re: IIO_CONST_ATTR_SCAN_EL_TYPE add result shift? References: <544AC56F16B56944AEC3BD4E3D5917712F0DB8ACA8@LIMKCMBX1.ad.analog.com> <4CAB0EF0.3010606@cam.ac.uk> <544AC56F16B56944AEC3BD4E3D5917712F0DB8ACC8@LIMKCMBX1.ad.analog.com> In-Reply-To: <544AC56F16B56944AEC3BD4E3D5917712F0DB8ACC8@LIMKCMBX1.ad.analog.com> Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-iio-owner@vger.kernel.org List-Id: linux-iio@vger.kernel.org On 10/05/10 12:52, Hennerich, Michael wrote: > Jonathan Cameron wrote on 2010-10-05: >> On 10/05/10 12:24, Hennerich, Michael wrote: >>> /** >>> * IIO_CONST_ATTR_SCAN_EL_TYPE - attr to specify the data format of a >>> scan el * @name: the scan el name (may be more general and cover a set >>> of scan elements * @_sign: either s or u for signed or unsigned * >>> @_bits: number of actual bits occuplied by the value * @_storagebits: >>> number of bits _bits is padded to when read out of >>> buffer **/ #define IIO_CONST_ATTR_SCAN_EL_TYPE(_name, _sign, _bits, >>> _storagebits) \ >>> IIO_CONST_ATTR(_name##_type, #_sign#_bits"/"#_storagebits); >>> There are cases where bits < storagebits and the result is not alligned >>> to the LSB. Lets say we have an 10-bit ADC. The output is stored in a >>> 16-bit word, with four leading zeros, then the 10-bit datum, followed >>> by 2 don't care bits. >>> >>> So it would be good to provide an additional shift value? >>> Or how shall we handle these cases? >> This was discussed in the original debate and we simply proposed >> leaving it until we had a user. So the question is how to describe >> the shift. >> >> So taking your example, how about: >> >> u10/16>>2 >> >> Where the shift is optional. >> >> Other suggestions welcome given I made that up with 2 seconds thought. > > I think this is sufficient. I can only think about right shifts being useful. > This notation would consistently allow both, that's why I like it. It would be decidedly odd for right shift to make sense. > > A bit off topic - some common defines for signed/unsigned types might also be useful. > > #define IIO_SCAN_EL_TYPE_SIGNED 's' > #define IIO_SCAN_EL_TYPE_UNSIGNED 'u' Fine with me. > > Greetings, > Michael > > Analog Devices GmbH Wilhelm-Wagenfeld-Str. 6 80807 Muenchen > Sitz der Gesellschaft Muenchen, Registergericht Muenchen HRB 4036 Geschaeftsfuehrer Thomas Wessel, William A. Martin, Margaret Seif > > >