Kconfig and MicroPython config is not updated yet.
* start to refactor lv_conf.h
* further mem refactoring
* create lv_mem_builtin.c/h
* update lv_conf_internal.h
* add lv_strlen and lv_strncpy
* rename LV_DRAW_COMPLEX to LV_USE_DRAW_MASK
* update lv_conf_template
* minor fix
* feat(pxp/vglite) Attach NXP GPUs to the new draw context.
Create a single NXP draw layer on top of PXP and VGLITE.
Extra changes:
1. Add VGLITE image blits acceleration.
2. Reenable blit split workaround for quality issue in RT500.
3. Increase threshold from 32 to 5000 px to fill/blit with both PXP and VGLITE.
4. Allow to enable both PXP and VGLITE. Add a fallback mechanism:
- by default the PXP will try to accelerate. if that is not supported (or fails
due to threshold limit condition from 3.) then it will fallback to VGLITE.
- if VGLITE does not support that feature (or fails due to threshold limit
condition from 3.) the it will fallback to CPU.
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* feat(vglite) Add VGLITE support to draw the backgroud of rectangles.
optim: draw only a circle when radius has value LV_RADIUS_CIRCLE.
optim: to draw rounded corners, use cubic bezier curves
Signed-off-by: Seb Fagard <sebastien.fagard@nxp.com>
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* feat(vglite) Add VGLITE support to draw arcs.
Use up to 4 bezier curves to optimize the drawing of an arc.
The arc curve has to be constant when growing the angle:
for this we compute sub-arc based on a best approximation of quarter-arc.
use dichotomy to find the sub-arc 't' param, instead of tangent approximation.
Signed-off-by: Seb Fagard <sebastien.fagard@nxp.com>
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* feat(pxp) Add ARGB88888 format support for PXP backend.
Supports per pixel and global alpha blending, and combination of alpha blending with recolor feature.
Signed-off-by: Jerome Evillard <jerome.evillard@nxp.com>
Signed-off-by: Seb Fagard <sebastien.fagard@nxp.com>
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* feat(vglite) Add the support of ARGB 32bits color format.
Signed-off-by: Seb Fagard <sebastien.fagard@nxp.com>
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* feat(vglite) Add VGLITE acceleration support for rotation and zoom.
Signed-off-by: Seb Fagard <sebastien.fagard@nxp.com>
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* feat(pxp) Add PXP acceleratin to rotate in blit.
Applies the rotation on the pxp output.
Signed-off-by: Seb Fagard <sebastien.fagard@nxp.com>
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* fix(pxp/vglite) Avoid calling the blend callback from image decoded callback.
Add lv_gpu_nxp_pxp_blit_transform() and lv_gpu_nxp_vglite_blit_transform().
This will simplify a lot the fallback mechanism and the way of adding new
image decoded features.
(MGG-884)
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* feat(pxp) Add image rotate 90x.
Simplify the two steps process by adding the blit_cover and blit_opa functions.
In order to rotate or recolor with opacity, two steps must be fallowed:
1. Run the operation without opa.
2. Blend the result by applying the opa.
(MGG-469)
Obs:
Recolor and rotate is currently not supported with opa or alpha channel.(MGG-883)
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* fix(vglite) Fix incorrect slider widget indicator when vglite acceleration is enabled
(MGG-863)
Signed-off-by: Stefan Babatie <stefan.babatie@nxp.com>
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* fix(pxp) Fix pxp fill (simple rect draw) on 32 bit color depth.
(MGG-890)
Signed-off-by: Stefan Babatie <stefan.babatie@nxp.com>
* fix(pxp) Separate blit simple by the blit with transformation.
While we run into blit with transformation we have to
decide if the operation needs to be done in one or two steps.
Blit with color format (opa or alpha or chroma key) - require one step: blit_cf().
Blit with rotate or recolor but no color format - require one step: blit_cover().
Blit with rotate or recolor + opa or alpha - require two steps:
blit_opa() = blit_cover() + blit_cf().
Blit with rotate or recolor + chroma key - not supported yet.
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* feat(pxp) Add support for recolor with chroma-keying.
(MGG-434)
Signed-off-by: Stefan Babatie <stefan.babatie@nxp.com>
* fix(pxp) Fix temporary buffer allocation limit.
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* fix(pxp) Add PXP limitation while rotating images not aligned to 16x16 blocks.
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* fix(nxp) Add makefiles.
Tested with:
working-directory: tests/makefile
run: make test_file
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* feat(nxp) Update NXP github documentation.
(MGG-864)
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* fix(nxp) Move the API comments of global functions only in the H files to make maintenance simpler.
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* fix(nxp/vglite) Fixed some warnings. Remove unused variables.
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* fix(nxp) lv_draw_nxp_ctx_deinit() shall simply fallback to sofware call.
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* fix(nxp) Fallback to software callbacks if need for argb8565 support.
During rendering, LVGL might initializes new draw_ctxs and start drawing into
a separate buffer (called layer). If the content to be rendered has "holes",
e.g. rounded corner, LVGL temporarily sets the disp_drv.screen_transp flag.
It means the renderers should draw into an ARGB buffer.
With 32 bit color depth it's not a big problem but with 16 bit color depth
the target pixel format is ARGB8565 which is not supported by the GPU.
In this case, the NXP callbacks should fallback to SW rendering.
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* fix(pxp) Fix wrong demo widget transparency.
1. Fix alpha inverted calculation for fill with opacity.
2. Fix also the ratio of recoloring for chroma key.
Signed-off-by: Stefan Babatie <stefan.babatie@nxp.com>
* fix(vglite) Remove software pre-multiplication when hardware pre-multiplication is available.
(MGG-886)
Signed-off-by: Stefan Babatie <stefan.babatie@nxp.com>
* doc(vglite) Add vglite initialization info.
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* fix(pxp/vglite) Fix unused variable warnings when PXP/VGLite are not enabled simultaneously.
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
* fix(pxp) Fixed the color key recoloring on 16 bits color depth.
1. Fixed the arguments of lv_color_mix(), previously was using an inversed logic. But it works anyway.
2. Use LV_COLOR_SET_X for adjusting the channels on both 16 and 32 bits. Fixed the max values.
Signed-off-by: Nicușor Cîțu <nicusor.citu@nxp.com>
Co-authored-by: Stefan Babatie <stefan.babatie@nxp.com>
* Update lv_gpu_stm32_dma2d.c
fix stm32 DMA2D blend。
* add SWM341 DMA2D support for lvgl
* add SWM341 DMA2D support for lvgl
* add SWM341 DMA2D support for lvgl
* add SWM341 DMA2D support for lvgl
* add SWM341 DMA2D support for lvgl
When the event is bubbling, use lv_event_get_current_target
to get the obj that needs to be processed
Signed-off-by: wangxuedong <wangxuedong@xiaomi.com>
Co-authored-by: wangxuedong <wangxuedong@xiaomi.com>
* perf(draw) reimplement circle drawing algorithms
Imporve the speed of circle drawing
Add circle draw caching
Various other speed improvements
* docs describe how to use masks
* fix(draw) add missing GC root usage
While it's possible to try initialize LVGL after it is already initialized, it warns 'lv_init: already inited' in such case. Therefore it is useful to be able to check on runtime whether LVGL was initialized or not before calling lv_init().
Add lv_<widget>_draw_part_type_t to widgets to precisly describe the hooked drawings.
Also add class_p element to lv_obj_draw_part_dsc_t to show what widgets lv_<widget>_draw_part_type_t needs to be used.
Related to: https://forum.lvgl.io/t/how-to-add-minor-division-lines-to-a-chart/5366/
includes updates from NXP MCU SDK 2.10 with adaptation for lvgl v8:
Updated cache handling
- range limited cache flushing changed to complete cache flush, which is
faster and shold be safe
- flushing done via callback system, so OS specific code is removed
from LVGL
Signed-off-by: Jozef Bastek <jozef.bastek@nxp.com>
gpu: pxp: Buffer sync with BLITs
Added buffer synchronisation with BLITs instead of memcpy for PXP.
In LVGL v8, buffer synchronisation acceleration is moved to littlevgl_support.c.
Signed-off-by: Jozef Bastek <jozef.bastek@nxp.com>
gpu: pxp: updates for v8
cache callback type changed in display driver struct
chroma key name changed
Signed-off-by: Seb Fagard <sebastien.fagard@nxp.com>
gpu: pxp: new log header file for v8
Signed-off-by: Seb Fagard <sebastien.fagard@nxp.com>
core: init PXP accelerator
Signed-off-by: Seb Fagard <sebastien.fagard@nxp.com>
