Age | Commit message (Collapse) | Author |
|
After clearing the fix and var screeninfos fix.id should be reset to
the proper value. Move the relevant memset() and strncpy() calls into
a helper function which does the right thing.
Signed-off-by: Ville Syrjälä <ville.syrjala@nokia.com>
Conflicts:
drivers/video/omap2/omapfb/omapfb-main.c
|
|
When applying the settings to all the overlays ilace is initialzed to 0
in the beginning and then changed to 1 when the first overlay directed
to an interlaced display is found. If there are more overlays after that
one which are not directed to an interlaced display the code would still
pass ilace=1 to the the dispc code which would halve the output height of
the overlay. Fix the problem by correctly setting ilace for each overlay.
Signed-off-by: Ville Syrjälä <ville.syrjala@nokia.com>
Conflicts:
drivers/video/omap2/dss/manager.c
|
|
When downscaling the bottom field offset may be several source lines.
Unfortunately the registers for the initial accumulator value will only
hold the fractional part of of that offset so the integer part must be
added to the bottom field base address. The TV-out quality is
significantly improved as a result. Additionally this fixes the bottom
field offset with DMA rotation and mirroring enabled as it was
previously miscalculated.
Signed-off-by: Ville Syrjälä <ville.syrjala@nokia.com>
|
|
When SDI is disabled there is still ~1.6mA drain through the VSDI_CSI
regulator. This seems to be internal to the SDI module, since even
disconnecting the LCD panel doesn't get rid of the drain. Also
disabling/resetting the SDI or the DSS module doesn't help. The only
solution is to configure the SDI pads to safe mode with pull up
resistors enabled on them when SDI gets disabled.
Bootloader / platform code is still responsible for setting up the
muxing, so in the driver check if all required pads are in SDI mode
or all of them are in non-configured state.
Signed-off-by: Imre Deak <imre.deak@nokia.com>
Conflicts:
arch/arm/plat-omap/include/mach/display.h
|
|
This is needed by an upcoming patch, in which sdi->init_display can
fail.
Signed-off-by: Imre Deak <imre.deak@nokia.com>
Conflicts:
drivers/video/omap2/dss/display.c
drivers/video/omap2/dss/dpi.c
drivers/video/omap2/dss/dsi.c
drivers/video/omap2/dss/dss.h
drivers/video/omap2/dss/rfbi.c
drivers/video/omap2/dss/sdi.c
drivers/video/omap2/dss/venc.c
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Store manager config in similar as overlay config. Config
is applied with mgr->apply(). Also change transparency color key
variable names from color_key to trans_key.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
DSI needs vdds_dsi regulator
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Venc needs vdda_dac regulator
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Changes to generic, samsung and sharp panel drivers
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Use better model for handling DSS devices and drivers.
Remove omap_display, omap_ctrl and omap_panel structs, and add
omap_dss_device and omap_dss_driver. DSS devices are added to DSS bus.
The commit changes almost every part of DSS2, even if actual
functionality should be changed only in few parts.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
DSI bus lock protects the DSI bus. Normally dsi driver acquires the
lock, but if display drivers have their own workqueues, or similar,
they need to acquire the bus lock before sending data.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Proper transfer shutdown is still missing.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Do not reset the the WSS data every time the display is enabled. Only
reset it when the TV standard changes since PAL and NTSC use different
WSS data layouts.
Signed-off-by: Ville Syrjälä <ville.syrjala@nokia.com>
|
|
Spotted-by: Ville Syrjala <ville.syrjala@nokia.com>
Signed-off-by: Imre Deak <imre.deak@nokia.com>
|
|
Adds an IOCTL through which an application can iterate through the
color modes supported by an overlay. Instead of a simple mode ID
this will give the parameters needed to setup the FB, as this is what
the application will do anyway.
Signed-off-by: Mayuresh Janorkar <mayur@ti.com>
|
|
Replication logic will make colors a bit better on active matrix LCD
when the fb is in 16bpp mode and the LCD interface is 18/24bpp.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Actually make the venc register dump file visible in debugfs. Somehow
this hunk got dropped when the venc debugfs file was originally added.
Signed-off-by: Ville Syrjälä <ville.syrjala@nokia.com>
|
|
Fix the PAL/NTSC video timings. Only some of these values are actually
specified for digital video, the rest were calculated based on the the
analog timings. The analog timings differ from the digital timings
anyway but these should be close to the truth. These values aren't
actually needed by the VENC but it doesn't hurt to make them a bit more
accurate.
Signed-off-by: Ville Syrjälä <ville.syrjala@nokia.com>
|
|
Add support for setting the widescreen signalling (WSS) data via sysfs.
Signed-off-by: Ville Syrjälä <ville.syrjala@nokia.com>
|
|
Allow overlays attached to the same framebuffer to have different
rotation settings. A new sysfs file overlays_rotate can be used to
configure the rotation settings for each overlay. The total rotation
for a single overlay is now '(var.rotate + overlays_rotate) % 4'.
Signed-off-by: Ville Syrjälä <ville.syrjala@nokia.com>
|
|
Do not to change the actual rotation setting in check_fb_var() and
also remove the xres/yres swap from the same place. Applications must
now specify var.xres and var.yres in framebuffer coordinates. Eg. if
you have a 480x800 framebuffer and want it rotated cw/ccw to a 800x480
display xres must be 480 and yres must be 800. So the application
doesn't have to know if it's dealing with a 480x800 unrotated display
or a 800x480 rotated display.
Signed-off-by: Ville Syrjälä <ville.syrjala@nokia.com>
|
|
Outgoing LP clock is actually DSI fclk / 2 / clk_divisor.
Also don't use hardcoded LP clock frequency, but get it from
the board file.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
I hope this can be removed when the DSI displays work properly.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
This fixes the problem that when requesting a new irq
we could get the isr called too early in case irqstatus
already had that irq flag on.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
The field 0 and field 1 offset values were still incorrect when VRFB
rotation is used. Swap them to put the fields into proper order.
Signed-off-by: Ville Syrjälä <ville.syrjala@nokia.com>
|
|
Releasing an inactive context is valid, it can happen when
the application first blanks the screen then frees the
framebuffer.
Signed-off-by: Imre Deak <imre.deak@nokia.com>
|
|
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@nokia.com>
|
|
At the moment the VRFB context is restored at each core power domain
OFF->ON transition. This is not optimal since the VRFB might be unused
temporarily for example when the screen is blanked. Add a suspend /
resume function to mark these unused periods during which we'll avoid
thea the context restore.
Use atomic bitops for ctx_map for consistency.
Signed-off-by: Imre Deak <imre.deak@nokia.com>
|
|
Signed-off-by: Imre Deak <imre.deak@nokia.com>
|
|
The function returns int not unsigned since it can fail. Handle the
failing case as if the context had been lost. So now:
1. No get_last_off_on_transaction_id func in platform data->
never restore the context
2. Return val < 0 -> force the restore
3. Return val >= 0 do the restore only if the counter has changed.
Signed-off-by: Imre Deak <imre.deak@nokia.com>
|
|
|
|
The VRFB is part of the SMS and supplied by the core power domain; do
the context saving while VRFB is configured and restore it along with
the rest of the domain context.
This patch only implements the restore functionality, but not the
actual call to the restore. That belongs to PM functionality.
|
|
When calculating the required fclk rate for five tap filtering if the
display width and output width are equal div by zero could occur. The
TRM doesn't actually specify how this case should be handled but for
now just skip the calculation which would trigger the div by zero.
Signed-off-by: Ville Syrjälä <ville.syrjala@nokia.com>
|