Re: ad7793 iio-ring test read issue - Reg

[Jonathan Cameron <jic23@cam.ac.uk>]

On 08/23/11 14:48, malas wrote:
> On Tue, 23 Aug 2011 14:41:57 +0100, Jonathan Cameron wrote
>> On 08/23/11 14:24, malas wrote:
>>> On Tue, 23 Aug 2011 13:58:34 +0100, Jonathan Cameron wrote
>>>> On 08/23/11 13:43, malas wrote:
>>>>> YES : Gregs updated kernel taken from link below:
>>>>>
>>>>> http://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging-2.6.git
>>>> Right now that's pretty much mainline.
>>>>
>>>> See comments below and please reply inline as it makes this sort
>>>> of discussion much easier to follow.
>>>>>
>>>>>
>>>>> On Tue, 23 Aug 2011 13:39:12 +0100, Jonathan Cameron wrote
>>>>>> On 08/23/11 13:17, malas wrote:
>>>>>>>
>>>>>>>
>>>>>>> Sorry, I did not explicitly mention that. I did use the new staging kernel
> referred by
>>>>>>> Michael Hennerich and integrated with 2010RC5 uclinux dist .
>>>>>>
>>>>>> Hmm. As in current contents of Greg KH's staging/staging-next?
>>>>>>
>>>>>> I'll assume so... See below.
>>>>>>
>>>>>>>
>>>>>>> On Tue, 23 Aug 2011 12:39:53 +0100, Hennerich, Michael wrote
>>>>>>>> malas wrote on 2011-08-23:
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Dear Jonathan Cameron,
>>>>>>>>>
>>>>>>>>> The kernel version I am using is :
>>>>>>>>
>>>>>>>> Didn't you say that you use the staging kernel?
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>
>>>
> http://blackfin.uclinux.org/gf/project/uclinux-dist/forum/?_forum_action=MessageReply&message_id=102839&action=ForumBrowse
>>>>>>>>
>>>>>>>> Or copied drivers/staging/iio over to your Blackfin kernel tree?
>>>>>>>>
>>>>>>>>>
>>>>>>>>> ---------------------------------------------------------------------
>>>>>>>>> midas@ubuntu:~/Mala/Telemedicine/uclinux_source/uc2010_source/uclinux/2
>>>>>>>>> 010R1$ svn info Path: . URL:
>>>>>>>>> svn://blackfin.uclinux.org/uclinux-dist/branches/2010R1 Repository Root:
>>>>>>>>> svn://blackfin.uclinux.org/uclinux-dist Repository UUID:
>>>>>>>>> 3e2a5524-39e8-4cb9-84a9-c49da69cba3a Revision: 10161 Node Kind:
>>>>>>>>> directory Schedule: normal Last Changed Author: vapier Last Changed Rev:
>>>>>>>>> 10136 Last Changed Date: 2011-04-02 22:36:15 +0530 (Sat, 02 Apr 2011)
>>>>>>>>> ---------------------------------------------------------------------
>>>>>>>>>
>>>>>>>>> Regards
>>>>>>>>> Mala
>>>>>>>>>
>>>>>>>>> On Tue, 23 Aug 2011 12:21:25 +0100, Jonathan Cameron wrote
>>>>>>>>>> On 08/23/11 09:18, malas wrote:
>>>>>>>>>>>
>>>>>>>>>>> Dear Michael Hennerich,
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> I shall post iio quesris in this mail list as per your suggestions.
>>>>>>>>>>>
>>>>>>>>>>> AD7793 driver is built into kernel. The chip gets configured and am
>>>>>>>>>>> able to read data on continuous mode via spi interface using ad7793
>>>>>>>>>>> driver at kernel level.
>>>>>>>>>>>
>>>>>>>>>>> While testing userspace test iio-ring to read data from ring buffer,
>>>>>>>>>>> generic_buffer code was executed, it  proceeds till opening
>>>>>>>>>>> /sys/bus/iio/devices/device0:buffer0 and Read call fails.
>>>>>>>>>> Question 1: What's the kernel and where did you get it from?
>>>>>>>>>> (Interfaces are moving fast so it's plausible you've fallen in a hole
>>>>>>>>>> where userspace wasn't matching kernelspace).
>>>>>>>>>>>
>>>>>>>>>>> In fs I can list the iio device ad7793 with following properties
>>>>>>>>>>> get/set.
>>>>>>>>>>>
>>>>>>>>>>> root:/> ls /sys/bus/iio/devices/device0\:buffer0/
>>>>>>>>>>> bytes_per_datum  enable           power            subsystem
>>>>>>>>>>> dev              length           scan_elements    uevent
>>>>>>>>>>> root:/>
>>>>>>>>>>>
>>>>>>>>>>> I find access0 and event0 missing in this list. Is that the reason
>>>>>>>>> for read() fails.
>>>>>>>>>> Yes.  Question is why and without kernel version I don't know where to
>>>>>>>>>> look I'm afraid.
>>>>>> Ah, this does look like a missmatch in interfaces between userspace and kernel.
>>>>>>
>>>>>> At this point in time we have scrapped the event0 chrdev (done via polling on
>>>>>> the access one) and the access one has be flattened into device0\:buffer0/
>>>>>> (read dev).
>>>>>>
>>>>>> Looks like the generic_buffer.c example does line up though. Where did the
>>>>>> 'test iio-ring' come from? Is that one yours Michael? 
>>>
>>>
>>> iio-ring.c under uclinux2010/user/blkfin-test/iio-ring-test. The file is in the
>>> attached. I did not use that since i didn't find access0 or event0 files under
>>> /sys/bus/iio/devices/device0:buffer0.
>>>
>>> I have tested with the generic_buffer.c example (under iio/documentaion folder) to get
>>> Ad7793 received samples to user space from ring buffer.
>>> The app opens: /dev/device0:buffer0 for poll and reading samples.
>>> Is the device name for the open() correct.
>> Yes - that should be right.
> *************************
> Open fails for "/dev/device0:buffer0" device name
>  I tried changing to "/sys/bus/iio/devices/device0:buffer0"- open() succeeded then but
> then read fails.
Read must be the one under dev.  The other is a directory with description of the buffer.

What is the error on the open of /dev/device0:buffer0?