|Deletions are marked like this.||Additions are marked like this.|
|Line 79:||Line 79:|
|Task 03:||Task 04:|
|Line 86:||Line 86:|
|Task 04:||Task 05:|
|Line 97:||Line 97:|
|Task 05:||Task 06:|
|Line 104:||Line 104:|
|Task 06:||Task 07:|
Welcome to IIO tasks page
This page is dedicated to students applying for GSoC/Outreachy programs.
It also aims to help newcomers getting started with the Industrial IO subsystem, offering pointers to documentation and small tasks.
For GSoC/Outreachy students:
Please use a single email thread for communication. During the application always reply to this thread if you have a question or sending the solution for a task. The email should have the following format:
subject: GSoC IIO project: Your Name
body: <short description of content: e.g Task 01 / Question about the schedule / etc> <the actual email body>
Experimenting with IIO subsystem
For this we will use two kernel modules found in drivers/iio/dummy:
iio_dummy.ko - example IIO driver to demonstrate existing functionality
core implementation can be found in iio_simple_dummy.c
buffer functionality is implemented in iio_simple_dummy_buffer.c
events functionality is implemented in iio_simple_dummy_events.c
iio_dummy_evgen.ko - generates fake events interrupts to be used by the iio_dummy example driver
implementation for this module is in iio_dummy_evgen.c
Dummy modules compilation
You need to select the following config options:
CONFIG_IIO_CONFIGFS - for creating the dummy device under configfs
CONFIG_IIO_SIMPLE_DUMMY - for building iio_dummy kernel module
CONFIG_IIO_SIMPLE_DUMMY_EVENTS, CONFIG_IIO_SIMPLE_DUMMY_BUFFER should be selected for events and buffer functionality.
CONFIG_IIO_DUMMY_EVGEN - for building iio_dummy_evgen kernel module
- Mount the configfs filesystem: read Documentation/iio/iio_configfs.txt
Use the following commands for modules compilation:
$ make drivers/iio/dummy/iio_dummy_evgen.ko
$ make drivers/iio/dummy/iio_dummy.ko
Use the following commands for module loading:
$ insmod iio_dummy_evgen.ko
$ insmod iio_dummy.ko
Use the following command to create your dummy device under the configfs filesystem:
$ mkdir /config/iio/devices/dummy/my_dummy_device
- Show that config options for IIO dummy modules were correctly selected
- send the content of the following command grep IIO .config
- Show that the modules were successfully loaded.
- lsmod | grep dummy
- ls -l /config/iio/devices/dummy/
- ls -l /sys/bus/iio/devices/iio:device0/
- ls -l /sys/bus/iio/devices/iio_evgen/
- Add channels for a 3-axis compass to iio_simple_dummy module.
- Show that channels were successfully added
- ls -l /sys/bus/iio/devices/iio:device0/
- ls -l /sys/bus/iio/devices/iio:device0/scan_elements
- create a patch with your changes
- channel type for a compass is IIO_MAGN
- users should be able to read raw readings from each axis
- users should be able to read a shared scale
- users should be able to access data via a buffer
- data is unsigned, resolution is 16 bits, storage is 16 bits
- compass doesn't support events
IIO event monitor
IIO event monitor is an user space example application which reads events from IIO layer and pretty prints the results. Implementation can be found in iio_event_monitor.c under tools/iio/.
compile iio_event_monitor.c to obtain an executable called iio_event_monitor.
- send us the command(s) used to successfully compile iio_event_monitor.c
Read events using iio_event_monitor
- run iio_event_monitor without arguments to figure out how it should be used
- read events from iio_dummy module
- Events are generated by iio_dummy_evgen via sysfs (look inside /sys/bus/iio/devices/iio_evgen/)
- send us the commands used to read/generate the events
Create triggers using configfs interface.
- read Documentation/iio/iio_configfs.txt in order to create an hrtimer software trigger named t1.
- where in the sysfs hierarchy does the trigger resides?
- sends us the commands used to create the trigger
Read samples from buffer generated by the iio_dummy module.
- compile iio_generic_buffer.c from tools/iio. This program will be used to read data from buffer.
have a look at ./iio_generic_buffer -h options. You will use the trigger t1 created with the previous tasks and iio_dummy_part_no for IIO device.
- send us the full command history
Note for all IIO Coding Tasks:
- Please use the task page to claim the driver you are going to patch.
Please 'CC firstname.lastname@example.org on all questions related to this so everyone can benefit.
Coding Task 1:
Clean-up idea: IIO Headers: This was previously posted to the email@example.com
See description and follow-ups here: https://groups.google.com/forum/#!topic/outreachy-kernel/LtI90_SwjHE
I believe this task is completed or close to being completed. If you find a driver which you think applies here, be sure to search for it in the outreachy group mailing list to see if it has already been done or disregarded.
Coding Task 2:
Coding Task 2: Remove driver usages of mlock
The locking scheme in the IIO subsystem includes a lock in the iio_dev structure called mlock. The usage of iio_dev->mlock is being redefined as protecting operating mode changes - as in changes between BUFFER* and DIRECT modes.
Notice how the struct iio_dev fields are labelled [INTERN] or [DRIVER]. The mlock will revert to an [INTERN] field once all the non-conforming usages are removed from drivers. The IIO core functions in (drivers/iio/industrialio-*.c) will be the only users of mlock. Drivers will use helper functions to control operating mode changes.
That's a bunch of background that you don't need to totally grasp to do this task, but it's nice to see the bigger picture.
The piece of the migration this task focuses on is removing the usages of mlock that don't meet the new model. I suspect we'll find that these drivers were using mlock as it was previously defined: "to protect simultaneous device *state* changes." Typically this means they are changing some configuration bits in the hardware. Those changes are important to protect, just do it with a driver private lock, not mlock.
Review this recently submitted patch: staging: iio: ad9832: replace mlock with driver private lock http://marc.info/?l=linux-iio&m=148943215125820&w=2
This is probably the simplest case patch for this task. I want to emphasize simple, because I did pass over a few others on my way to creating an example patch. I saved the more interesting drivers for you
Jonathan pleads: "Make sure that they fully review the locking whilst at it though as chances are some of it may be less than ideal!"
You may uncover issues with the usage or placement of locks in general. Take the time to understand the usage of the locks. This is not intended as a search&replace exercise.
Your reviewer may notice other things that need tidying up. If they're not immediately required for this patch, save it for a follow up patch.
Your mission is to remove the usage of mlock from the IIO drivers in staging. A simple 'git grep mlock' finds the drivers and I've posted that list on the Outreachy tasks page.
Please sign up for one driver at a time, take it all the way through ACK, before selecting another one. Be sure to follow along as these are posted and learn from others patches.
Sign up is on the OutreachyTaskPage
PATCHES need to be sent to all of:
the IIO Maintainer, IIO Reviewers, and IIO Mailing List,
Jonathan Cameron <firstname.lastname@example.org> (maintainer:IIO SUBSYSTEM AND DRIVERS) Hartmut Knaack <email@example.com> (reviewer:IIO SUBSYSTEM AND DRIVERS) Lars-Peter Clausen <firstname.lastname@example.org> (reviewer:IIO SUBSYSTEM AND DRIVERS) Peter Meerwald <email@example.com> (reviewer:IIO SUBSYSTEM AND DRIVERS) firstname.lastname@example.org (open list:IIO SUBSYSTEM AND DRIVERS)
and, for good measure, Jonathan likes to make sure the original Author(s) are directly copied on all patches. So, grab the MODULE_AUTHOR found at the tail end of the driver source.
Post questions to email@example.com or the #linux-iio IRC channel (server irc.oftc.net).