**** BEGIN LOGGING AT Fri Aug 05 02:59:58 2016
Aug 05 17:17:09 <ZeekHuge> ^^
Aug 05 18:36:08 <nerdboy> what
Aug 05 21:03:28 <ds2> hmmm
Aug 05 22:46:32 <chanakya_vc> m_w, There?
Aug 05 22:53:30 <m_w> chanakya_vc: hey, feeling better?
Aug 05 22:55:39 <chanakya_vc> Yeah a lot better. Problem is that the medicines that the docs have prescribed make me very sleepy. m_w
Aug 05 22:56:11 <m_w> making any progress on your project?
Aug 05 22:56:22 <chanakya_vc> Yup
Aug 05 22:56:43 <chanakya_vc> m_w, Wanted to know if you could  guide me to a driver for i2c
Aug 05 22:56:52 <chanakya_vc> I mean if there is any example
Aug 05 22:57:00 <chanakya_vc> like omap-spi
Aug 05 22:57:42 <chanakya_vc> As far I have read on the net, a driver for i2c should be a bit easier than spi to understand.
Aug 05 22:59:09 <chanakya_vc> Like I have written the basic functions in the firmware, but I would need to know exactly what I need to do in the driver to complete it and then pick data as I did for SPI
Aug 05 22:59:22 <m_w> okay
Aug 05 22:59:44 <m_w> the i2c drivers are in drivers/i2c
Aug 05 23:01:16 <m_w> you are adding a new i2c bus so drivers/i2c/busses
Aug 05 23:02:27 <m_w> the simplest driver to use as an example would probably be i2c-gpio.c
Aug 05 23:02:40 <m_w> http://lxr.free-electrons.com/source/drivers/i2c/busses/i2c-gpio.c
Aug 05 23:02:56 <chanakya_vc> Okay m_w . I was reading this though:http://www.linuxjournal.com/article/7136
Aug 05 23:03:51 <m_w> this is a nice walk through, it is a little dated though
Aug 05 23:04:45 <m_w> as with other subsystems there is documentation within the kernel
Aug 05 23:05:13 <m_w> http://lxr.free-electrons.com/source/Documentation/i2c/
Aug 05 23:06:21 <chanakya_vc> Okay thanks a lot m_w . I think if I work hard enough, I will be able to finish this in the weekend itself. Looks much simpler to me than SPI
Aug 05 23:08:01 <chanakya_vc> m_w, http://lxr.free-electrons.com/source/drivers/i2c/busses/i2c-gpio.c#L281  Why this need to have a init function? I mean they could have registered in the probe function itself right?
Aug 05 23:09:36 <chanakya_vc> Or by calling the macro module_platform_driver() m_w
Aug 05 23:09:43 <chanakya_vc> Can be done this way right m_w ?
Aug 05 23:09:49 <m_w> sure
Aug 05 23:10:45 <m_w> the import pieces are defining the adapter and calling i2c_bit_add_numbered_bus
Aug 05 23:11:53 <m_w> the gpio driver is doing a bunch of things specific to bit banging the GPIO which you do not need to be concerned with
Aug 05 23:16:37 <chanakya_vc> Okay m_w I got that. Ah the probe function is mostly about setting up the GPIO pins and stuff I guess.
Aug 05 23:17:17 <chanakya_vc> So i2c_bit_add_numbered bus would basically be registering the bus with i2c core right?
Aug 05 23:17:20 <chanakya_vc> m_w,
Aug 05 23:17:25 <m_w> yes
Aug 05 23:17:59 <m_w> you will need to define a pass a static const struct i2c_algorithm
Aug 05 23:18:30 <m_w> this will point to callback functions the will communicate with the pru
Aug 05 23:20:12 <m_w> .master_xfer is the function that does the magic
Aug 05 23:21:26 <chanakya_vc> Okay i2c core basically calls upon this .master_xfer function.And I tell the i2c core that such a function exists via the struct i2c algorithm right m_w ?
Aug 05 23:21:42 <m_w> yes
Aug 05 23:24:32 <m_w> https://www.kernel.org/doc/htmldocs/device-drivers/API-struct-i2c-algorithm.html
Aug 05 23:26:36 <m_w> i2c-gpio.c is a bit style adapter which is more host processor intensive
Aug 05 23:27:38 <chanakya_vc> Okay m_w . I was looking for this structure there :P
Aug 05 23:27:53 <m_w> yeah it is not in there
Aug 05 23:28:04 <m_w> sorry
Aug 05 23:28:19 <chanakya_vc> Arre np m_w . I think I have got it
Aug 05 23:29:20 <m_w> http://lxr.free-electrons.com/source/drivers/i2c/busses/i2c-robotfuzz-osif.c
Aug 05 23:29:28 <m_w> this is a nice short one
Aug 05 23:29:43 <m_w> that uses master_xfer
Aug 05 23:30:09 <m_w> great have at and let me know if you need help along the way
Aug 05 23:33:48 <chanakya_vc> m_w, This osif driver has read and write functions. I would not need those right?
Aug 05 23:35:12 <m_w> those are hardware specific
Aug 05 23:35:24 <m_w> be right back rebooting
Aug 06 00:29:42 <chanakya_vc> m_w, In this: http://lxr.free-electrons.com/source/drivers/i2c/busses/i2c-robotfuzz-osif.c#L134 ,there is no  struct i2c_adapter()
Aug 06 00:29:52 <chanakya_vc> i2c adapter
Aug 06 00:33:02 <m_w> lemme see
Aug 06 00:34:11 <m_w> it is in the private data struct
Aug 06 00:34:17 <m_w> http://lxr.free-electrons.com/source/drivers/i2c/busses/i2c-robotfuzz-osif.c#L35
Aug 06 00:36:35 <chanakya_vc> Okay m_w got that. Ah then there is no call to i2c_bit_add_numbered_bus() here but there is a call to i2c_add_adapter() ?
Aug 06 00:37:18 <m_w> they are using i2c_add_adapter instead
Aug 06 00:37:45 <chanakya_vc> Okay so I can use any of the two
Aug 06 00:39:08 <m_w> if you use i2c_add_adapter it will dynamically assign a bus number
Aug 06 00:39:10 <m_w> http://lxr.free-electrons.com/source/drivers/i2c/i2c-core.c#L1651
Aug 06 00:40:38 <chanakya_vc> Okay m_w , a couple of more questions, here we are not calling the platform_set_drv_data()?
Aug 06 00:42:41 <chanakya_vc> Like I want to do it the same way I did SPI, declare the pointers in pru_i2c struct and then set the pointers in the probe function.
Aug 06 00:43:31 <chanakya_vc> But then I would need to call these pointers in the master_xfer function m_w .
Aug 06 00:45:06 <m_w> this this specific driver they are pointing back the the private data with algo_data
Aug 06 00:45:17 <m_w> http://lxr.free-electrons.com/source/drivers/i2c/busses/i2c-robotfuzz-osif.c#L153
Aug 06 00:46:20 <m_w> here is one of many other i2c drivers that use platform_set_drvdata:
Aug 06 00:46:23 <m_w> http://lxr.free-electrons.com/source/drivers/i2c/busses/i2c-mxs.c#L854
Aug 06 00:47:09 <chanakya_vc> Okay so store the private data in .algo adapter and then call it again in xfer_function
Aug 06 00:47:39 <m_w> you could do it that way if you want
Aug 06 00:48:30 <m_w> i2c_get_adapdata(adap)
Aug 06 00:51:27 <m_w> after i2c_set_adapdata(adap, yourstruct) in probe
Aug 06 00:53:36 <m_w> platform_get_drvdata can only be used if pdev is passed to the function
Aug 06 00:54:08 <m_w> you can use it in your remove function
Aug 06 01:05:24 <chanakya_vc> Okay m_w . The .algo_adpater way looks easier to me.
Aug 06 01:06:14 <chanakya_vc> Last question, what are the buffers that the xfer_function provides me that I would have to write in memory? Like in spi I had rx_buf and tx_buf
Aug 06 01:12:35 <m_w> msg->buf
Aug 06 01:13:43 <chanakya_vc> struct i2c_msg right? m_w ?
Aug 06 01:13:55 <chanakya_vc> http://lxr.free-electrons.com/source/include/uapi/linux/i2c.h#L68
Aug 06 01:14:20 <chanakya_vc> So for writing the address somewhere I basically write to memory msg->addr?
Aug 06 01:15:07 <m_w> i2c sends the address as part of the transaction
Aug 06 01:15:38 <m_w> it determine the device it is talking to
Aug 06 01:16:13 <chanakya_vc> Yes I know that. But My firmware would need to pick up the address separately from the data in the memory
Aug 06 01:16:38 <chanakya_vc> like when it starts a transaction, it will first bit bang the address
Aug 06 01:16:56 <m_w> yes it would have to be passed to the pru in another memory location
Aug 06 01:17:20 <chanakya_vc> so msg->addr is the buffer with the address I presume
Aug 06 01:17:34 <chanakya_vc> m_w,
Aug 06 01:17:48 <m_w> it is the slave address
Aug 06 01:18:12 <chanakya_vc> Also in 7 bit addressing I read that the last bit represents whether the master wants to read or write to the slave?
Aug 06 01:18:28 <chanakya_vc> But here it is defined in the flags I guess
Aug 06 01:19:07 <m_w> it is part of the i2c_msg struct
Aug 06 01:19:36 <m_w> the flags are stored in another u16
Aug 06 01:20:19 <m_w> the defines are used to tell what bit corresponds to what flag
Aug 06 01:20:53 <chanakya_vc> Yeah so m_w , there I only check if flags==1? If yes the master wants to read data, if false then master wants to write data?
Aug 06 01:21:14 <m_w> yup
Aug 06 01:21:46 <chanakya_vc> Okay and this goes right after I have bitbanged the address right?
Aug 06 01:22:53 <m_w> yeah
Aug 06 01:23:30 <chanakya_vc> One more thing where do I write the data that I get from the slave? When master says it wants to read?
Aug 06 01:23:44 <chanakya_vc> Like I see no receiving buffer
Aug 06 01:23:58 <chanakya_vc> in struct i2c_msg
Aug 06 01:24:23 <m_w> it is half duplex so it uses the same buffer I believe
Aug 06 01:24:56 <chanakya_vc> Okay.
Aug 06 01:26:02 <chanakya_vc> Thanks for your help m_w . Will you be online during the weekend?
Aug 06 01:26:13 <m_w> yeah
Aug 06 01:26:53 <chanakya_vc> Okay m_w . I better get busy writing code now.
Aug 06 01:27:11 <m_w> good luck
**** ENDING LOGGING AT Sat Aug 06 02:59:58 2016