**** BEGIN LOGGING AT Thu Jun 02 02:59:58 2016 Jun 02 16:25:01 Logs down or silent here today? Jun 02 16:27:00 Hey bradfa,Regarding the code,as you suggested I am going to be giving a switch case.Currently switch cases based on a user defined variable.Later will migrate to ioctls. Jun 02 16:29:21 bradfa, Also had a doubt regarding the parameters + data.I am thinking of defining a flag which when set as zero would treat the information it receives as parameters and then set the same var as 1 so that all the information sent then is data? Jun 02 16:29:27 what do you think? Jun 02 16:29:37 Log working :) Jun 02 16:37:48 all is quiet on the beagle front Jun 02 17:05:31 m_w: hi, now I'm setting McSPI register Jun 02 17:09:21 m_w: Where should be located basic user setting e.g. cs polarity etc when driver is not find this setting in dts. Jun 02 17:13:39 pmezydlo1: default settings? Jun 02 17:16:17 I mean user setting which if not defined in DTS, shoud be defalut value. https://github.com/pmezydlo/SPI_slave_driver_implementation/blob/master/driver/spi-mcspi-slave.h#L11 Jun 02 17:17:29 now This setting is located in platform device but i don't know whether it should be there. Jun 02 17:19:26 sorry not in platform device but platform data Jun 02 17:20:19 pmezydlo1: if the user does not set the num_cs in device tree just error out Jun 02 17:20:55 pmezydlo1: what is num_cs? are you going to support multiple slave devices on the same bus? Jun 02 17:21:42 no, will be only one channel Jun 02 17:22:18 slave mode does not allow more channels Jun 02 17:22:25 then this parameter needs to be ignored Jun 02 17:23:26 or only one of them must be selected Jun 02 17:23:41 does the slave require a native chip select? Jun 02 17:23:42 it's an optional property anyways Jun 02 17:24:20 Yes I know, When I wrote this code part, I thought that would be needed Jun 02 17:24:26 cs polarity? Jun 02 17:25:20 You can be freely set Jun 02 17:26:40 cs polarity should default active low Jun 02 17:27:37 ok, but I will give the ability to freely set Jun 02 17:27:42 in dts Jun 02 17:28:04 that should be fine Jun 02 17:28:59 bits_per_word is part of the spi_transfer struct Jun 02 17:30:32 or is that different for slave mode? Jun 02 17:31:45 I don't use structures with master spi. Jun 02 17:32:58 https://www.kernel.org/doc/htmldocs/device-drivers/API-struct-spi-transfer.html Jun 02 17:33:45 in master mode you can have devices with different word sizes on the same bus, so you want to set it for each transfer, not just when the driver loads Jun 02 17:33:50 I don't know yet how it will look like part of transfer. Jun 02 17:34:37 pmezydlo1: I'm just looking at this doc, which describes all the required and optional properties for master and slave mode spi: https://www.kernel.org/doc/Documentation/devicetree/bindings/spi/spi-bus.txt Jun 02 17:35:31 * alexhiam knows very little about the spi slave mode implementation in linux though Jun 02 17:38:05 I think that SPI slave isn't supported on linux Jun 02 17:38:58 "SPI slave nodes must be children of the SPI master node and can contain the following properties..." Jun 02 17:39:54 I'm going the other way Jun 02 17:40:36 SPI slave was not supported in linux for a long time, but eventually they added it though not a whole lot of slaves last I looked Jun 02 17:41:06 mcspi omap2 master's code is not good looking Jun 02 17:41:07 Wormo: you know when it got added? Jun 02 17:41:24 this is on 4.6: http://lxr.free-electrons.com/source/include/linux/spi/spi.h#L32 Jun 02 17:41:38 No, I'd have to do git investigation Jun 02 17:41:55 that comment could be outdated Jun 02 17:42:31 yes outdated Jun 02 17:43:58 Wormo: what SPI slaves are supported? Jun 02 17:44:09 I'll look... Jun 02 17:45:05 recall seeing a TI driver Jun 02 17:48:10 https://github.com/torvalds/linux/blob/master/drivers/spi/Kconfig#L739 Jun 02 17:48:42 apparently not upstreamed, so n/m Jun 02 17:49:50 pmezydlo1: I would still recommend going by that doc Jun 02 17:51:30 ok Jun 02 17:54:28 other possibility is I saw it in freescale kernel, I was working on an imx6 recently Jun 02 17:55:49 but I don't know how and when my driver with few restrictions will be upstreamed Jun 02 17:56:14 I did just see some talk of slave support for an imx while googling Jun 02 17:57:05 pmezydlo1: shorter track to upstreaming if you go by that doc, I would think Jun 02 17:57:34 seems like there's at least an accepted outline of what slave mode will look like Jun 02 17:58:04 well I take care of it Jun 02 17:59:18 that's it, it was freescale driver that I was thinking of Jun 02 17:59:20 https://lkml.org/lkml/2015/9/25/444 Jun 02 17:59:51 alexhiam: Are you heard about serious hardware limitation? Jun 02 17:59:54 it's in a vendor tree somewhere and has been posted upstream, not accepted yet apparently Jun 02 18:00:14 pmezydlo1: no, what limitation? Jun 02 18:03:03 am335x datasheet See 24.3.3.1 Jun 02 18:03:14 Its own transmitter register MCSPI_TX on top of the common shift register. Jun 02 18:03:14 If the transmitter register is empty, the status bit TXS of the register Jun 02 18:03:14 MCSPI_CH0STAT is set. When McSPI is selected by an external master (active Jun 02 18:03:14 signal on the SPIEN port assigned to channel 0), the transmitter register Jun 02 18:03:14 content of channel0 is always loaded in shift register whether it has been Jun 02 18:03:14 updated or not. The transmitter register should be loaded before McSPI Jun 02 18:03:14 is selected by a master.' Jun 02 18:03:46 ah Jun 02 18:04:37 and next receive massage must be loaded before transaction Jun 02 18:06:16 I doubt that the mainline will accept spi slave drivers without much convincing Jun 02 18:06:23 it is described in more detail https://github.com/pmezydlo/SPI_slave_driver_implementation/wiki Jun 02 18:07:23 so I guess that puts a limit on the max clock speed, and would maybe require the master to pause between TXing and RXing in certain circumstances (which is not unheard of - some sensors need time to sample, etc.) Jun 02 18:08:05 m_w It will probably help if TI and freescale vendor trees could carry slave drivers that are compatible with each other for a while Jun 02 18:09:31 not to say they won't ask for a bunch of API changes, but having slave drivers used in 2 different places could start the ball rolling Jun 02 18:12:26 chanakya_vc: I think it might be best to send messages between Linux and the PRU to set parameters but then have ring buffers set aside to act as data fifos. I know this was the "harder" of the two options you prestented before, but maybe it's something you should consider after you get a basic rpmsg-only thing running (your idea of defining messages and data is reasonable to get started with). Jun 02 18:12:50 chanakya_vc: sorry but I'm quite busy today but if you send me messages I'll read them, just might be a huge delay, sorry Jun 02 18:13:46 chanakya_vc: if you can send email I'll be more able to respond better later tonight Jun 02 18:15:46 bradfa, Okay.I am currently working on that only.First aim is to get a rpmsg only model up I guess. I will email once I am done with this. I might need another day. Jun 02 18:16:05 bradfa, Let me be first sure of the idea that I want to pitch you : ) Jun 02 18:17:00 bradfa, And the exact problems.Because I am discovering them as I write the code. Jun 02 18:17:16 kiran4399: how's it going? Jun 02 18:18:21 alexhiam: yeah.. good.. I saw your mail today morning. Jun 02 18:18:59 alexhiam: Will you be free tomorrow at this time?? Jun 02 18:19:08 shall we have a meeting again? Jun 02 18:19:52 kiran4399: yeah, I could do that. m_w, Abhishek_: you able to talk blue APIs tomorrow at this time? Jun 02 18:20:30 alexhiam: is karki my mentor? Jun 02 18:21:08 kiran4399: I'm not sure... haven't heard much from karki recently... Jun 02 18:21:21 kiran4399: have you tested the pressure sensor? Jun 02 18:22:04 btw, did you get the rev D cape? Because according to the schematics it should be a bmp280 and mpu9250: https://github.com/StrawsonDesign/Robotics_Cape_Documentation Jun 02 18:22:16 which would mean the same parts as on the blue Jun 02 18:24:33 alexhiam: I got the cape.. but I did not test it though.. Jun 02 18:24:41 I'll do it tomorrow morning.. Jun 02 18:26:09 ok... remember we are expecting somewhere on the order of content published 5 days/week, whether it be commits pushed to github or documentation on github wiki pages Jun 02 18:27:26 so if you're spending time wrangling kernel drivers, that should be documented on a wiki page Jun 02 18:28:54 and if you get stuck you ask for help here immediately Jun 02 18:59:14 jkridner: i thought i saw you post a tracking # for TI eval boards Jun 02 18:59:22 now i can't find it **** ENDING LOGGING AT Fri Jun 03 02:59:58 2016