[v3] SoC Camera: add driver for OMAP1 camera interface

  This is a V4L2 driver for TI OMAP1 SoC camera interface.

Both videobuf-dma versions are supported, contig and sg, selectable with a 
module option. The former uses less processing power, but often fails to 
allocate contignuous buffer memory. The latter is free of this problem, but 
generates tens of DMA interrupts per frame. If contig memory allocation ever 
fails, the driver falls back to sg automatically on next open, but still can 
be switched back to contig manually. Both paths work stable for me, even 
under heavy load, on my OMAP1510 based Amstrad Delta videophone, that is the 
oldest, least powerfull OMAP1 implementation.

The interface generally works in pass-through mode. Since input data byte 
endianess can be swapped, it provides up to two v4l2 pixel formats per each of 
several soc_mbus formats that have their swapped endian counterparts.

Boards using this driver can provide it with the following platform data:
- if and what freqency clock is expected by an on-board camera sensor,
- what is the maximum pixel clock that should be accepted from the sensor,
- what is the polarity of the sensor provided pixel clock,
- if the interface GPIO line is connected to a sensor reset/powerdown input 
  and what is the input polarity.

Created and tested against linux-2.6.36-rc5 on Amstrad Delta.

Signed-off-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
---
Guennadi,
Since 2.6.36-rc6 is out, I've decided to submit the driver, no longer trying 
to make it still better or waiting for more comments, in hope it's still not 
too late for it to be merged during the upcomming window. The redefined 
set_fmt and set_crop operations work for me as expected with the last version 
(v3) of my ov6650 sensor driver.

There are still two SG mode specific corner cases to be corrected, previously 
not detected because of poor sensor driver functionality: 1) frame size not 
exceeding one page, resulting in "unexpected end of frame" message and capture 
restart every frame, and 2) last sgbuf lenght less than bytes_per_line, 
resulting in unstable picture. I'm going to address those two with fixes.

Thanks,
Janusz

v2->v3 changes:

requested, suggested or inspired by Guennadi Liakhovetski (thanks again!):
- compare enum variable against one of enum values instead of relaying on the 
  fact that a possible value can be either 0 or not 0,
- put an argument inside parenthesis in a macro definition,
- CONTIG and SG are not good enough names to be defined in a header under 
  include/...,
- still better explain why suspend capture if pcdev->ready is NULL on buffer 
  completion,
- ephasize the fact and explain why not fetch new buffer from the queue 
  immediately, rather do it only when a next DMA interrupt occures,
- use more correct English wording in a few cases,
- explain why not yet call videobuf_done() right after the end of current 
  sglist is detected, but only on next DMA interrupt,
- don't split strings, even on print format specifier boudries,
- v2 updated subdev_call_with_sense() macro return value was no longer stored 
  in a variable which was still examined next,
- don't store 1 in a bool variable, use "true" instead,
- allocate register cache dynamically based on platform resource size instead 
  of sizing it staticaly with a predefined OMAP1_CAMERA_IOSIZE macro; move 
  this macro out of the <media/omap1_camera.h>,

suggested by Ralph Corderoy (thanks again!):
- a few hex print formats still not optimal,

other:
- correct a few typos still found,
- if a block consisting of a single command contains a comment as well, it 
  should be enclosed in braces,
- make the driver less noisy by lowering down levels of a fwe more messages,
- is_dma_aligned() and dma_align() results can depend on vb_mode istead of 
  always assuming CONTIG mode as a worth case,
- redefine set_crop and set_fmt algorithms to better follow the v4l2 model, 
  drop a few no longer used helper functions, create a new one, 
  set_mbus_format(), from the new code common to both operations.

v1->v2 changes:

requested by Guennadi Liakhovetski (thanks!):
- first try contig, and if it fails - fall back to sg; invalidates next two,
- invalidated: Kconfig: VIDEO_OMAP1_SG: not need "if", the "depends on" should 
  suffice,
- invalidated: include both <media/videobuf-dma-contig.h> and 
  <media/videobuf-dma-sg.h> headers,
- extensively document buffer manipulations, better yet clean it up,
- a copyright / licence header was missing form a header file, 
- need to #include <linux/bitops.h> if using BIT() macro,
- don't need macros representing frequencies - use numbers directly,
- add a few missing "static" qualifiers,
- use u32 type for register content handling,
- some cached registers were unnecessarily read from the hardware directly,
- use true/false constants instead of 0/1 for booleans,
- avoid assigning variables inside other constructs,
- don't need to test if RAZ_FIFO is cleared,
- no need to split "\n" to a new line, don't worry about > 80 characters,
- don't increment field_count in case of a VIDEOBUF_ERROR,
- adjust mbus format codes to the new names,
- make is_dma_aligned() return value a bool,
- no need to align frame line lenghts to integer number of DMA elements,
- drop a few superflous whitespace, braces, etc.,
- use correct multiline comment style,
- follow some good-style rules when defining a multi-line function-like macro,
- drop tests for impossible bool < 0,
- that's not an error when S_FMT sets a format different from what the user
  has requested, just use whatever you managed to configure,
- as a general rule, try_fmt shouldn't fail,
- use sensor_reset() instead of duplicating its code,
- first shut down the hardware, and only then unconfigure software,

suggested by Ralph Corderoy (thanks!):
- include vb->state's value in the debug messages instead of just "unknown",
- correct a few print formats,
- don't use goto if return is all that needs to be done,

other:
- correct a few issues reported with "checkpatch.pl --strict".
- refreshed against linux-2.6.36-rc3

 drivers/media/video/Kconfig        |    8
 drivers/media/video/Makefile       |    1
 drivers/media/video/omap1_camera.c | 1702 +++++++++++++++++++++++++++++++++++++
 include/media/omap1_camera.h       |   35
 4 files changed, 1746 insertions(+)

--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[v3] SoC Camera: add driver for OMAP1 camera interface

Commit Message

Comments

Patch