Richard Hirst
|
fe63b9899e
|
Rework to process commands as they are issued, rather than on close()
|
2012-11-23 22:31:42 +00:00 |
|
Robin Mallinson
|
e59ad99d57
|
Change dev_read() to output data suitable for direct write back through the dev file.
|
2012-11-21 22:46:23 +00:00 |
|
Robin Mallinson
|
e712bd86ba
|
All works as intended now. Commands can be written partially, and a single
write to the device file may contain multiple commands.
To achieve both of these the driver now only parses the commands on
dev_close().
|
2012-11-21 03:15:12 +00:00 |
|
Robin Mallinson
|
0439809e27
|
Use max_idx rather than hardcoded data.
|
2012-11-21 01:48:14 +00:00 |
|
Robin Mallinson
|
1b8dbeed21
|
Increase space available for cmd_str ready for taking multiple commands in one go.
|
2012-11-21 01:07:17 +00:00 |
|
Robin Mallinson
|
ce367d489a
|
Factor out process_command_string()
|
2012-11-21 01:00:36 +00:00 |
|
Robin Mallinson
|
983e36f7c1
|
Following test and bug fix code now works as intended on partial writes.
|
2012-11-21 00:48:27 +00:00 |
|
Robin Mallinson
|
b1bb608d4c
|
Allow dev_write to work per-process.
|
2012-11-18 23:35:32 +00:00 |
|
Robin Mallinson
|
ef87ef88a7
|
Add code to allow dev_write to cope with partial writes (as will sometimes
occur when using languages such as python to write to the dev file).
Currently static data so same for all processes.
|
2012-11-18 23:15:35 +00:00 |
|
Robin Mallinson
|
b14044eba1
|
Refactor dev_write prior to allowing it to take partial strings.
|
2012-11-18 21:52:28 +00:00 |
|
Robin Mallinson
|
e993a088eb
|
Change dev_read to use per-process data.
|
2012-11-18 21:12:36 +00:00 |
|
Robin Mallinson
|
38101f8a62
|
fix dumb typo
|
2012-11-18 14:57:55 +00:00 |
|
Robin Mallinson
|
90dfe77fd5
|
Sketch out temporary data
|
2012-11-18 14:55:27 +00:00 |
|
Robin Mallinson
|
d6e1b2efe5
|
Add skeleton of allocation code for kmalloc and kfree. Currently allocates a dummy storage area.
|
2012-11-18 14:36:05 +00:00 |
|
Robin Mallinson
|
4b87d814df
|
Add crude 'written_data' array so dev_read simply echos verbatim what was written to dev_write without probing the actual DMA control block data. This gives no clue as to whether the DMA control bloack has been programmed but does mean that Richard's control code - which has a special case for delay=0 - does not require changing. Next steps are to have an array per-process.
|
2012-11-18 14:09:45 +00:00 |
|
Robin Mallinson
|
94c2aab61b
|
Revert to original code re: setting delay, so we again have a delay of 1
for a pulse of zero. Will frig the display issues by using an array that
stores written cnt rather than deriving it from actual register data.
|
2012-11-18 13:47:30 +00:00 |
|
Robin Mallinson
|
62f326001c
|
Refactor to amke the only difference between the code paths a little more obvious.
|
2012-11-13 00:02:11 +00:00 |
|
Robin Mallinson
|
e3230cb8d2
|
Make servoblaster write pulse high length even when cnt=0. Also change default tick high from 1 to zero.
|
2012-11-12 23:49:39 +00:00 |
|
Robin Mallinson
|
8002904e4e
|
Enable dev_read code, plus fix a couple of syntax errors (the code hadn't been compiled).
It appears to work at a cursory level:
pi@raspberrypi ~/src/PiBits/ServoBlaster $ sudo insmod ./servoblaster.ko
pi@raspberrypi ~/src/PiBits/ServoBlaster $ ls -l /dev/servoblaster
crw-rw-rw- 1 root root 251, 0 Nov 11 19:22 /dev/servoblaster
pi@raspberrypi ~/src/PiBits/ServoBlaster $ cat /dev/servoblaster
0 1
1 1
2 1
3 1
4 1
5 1
6 1
7 1
pi@raspberrypi ~/src/PiBits/ServoBlaster $
|
2012-11-11 19:23:41 +00:00 |
|
Robin Mallinson
|
fd2d771d1e
|
Add dev_read() function to driver to allow it to report current settings
for each of the 8 PWMs via /dev/servoblaster.
|
2012-11-05 22:23:45 +00:00 |
|
Richard Hirst
|
2b9f76dd88
|
Updates to allow tweaking timing via module parameters
|
2012-08-30 21:58:53 +01:00 |
|
Richard Hirst
|
d8df9b7625
|
Added ServoBlaster
|
2012-08-19 17:11:33 +01:00 |
|