fix(draw/opengles): cache compare image src by content

[borgeuz]

Summary

Fixes #9903.

The OpenGL ES texture cache used lv_memcmp on the entire lv_draw_image_dsc_t, which compared the bytes of the src pointer instead of the file path it points at. Because
lv_image_set_src() lv_strdups the path per widget, two widgets pointing at the same file ended up with different pointers and produced a cache miss on every screen recreation,
even though the underlying texture was identical.

Changes (src/draw/opengles/lv_draw_opengles.c)

• Added task_type to cache_data_t; the compare callback now partitions the RB tree by it before dispatching, so image_dsc_compare only ever runs on two image descriptors and
  the ordering stays symmetric.
• New image_dsc_compare does a field-by-field comparison for LV_DRAW_TASK_TYPE_IMAGE. src is resolved via lv_image_src_get_type: lv_strcmp for FILE/SYMBOL, pointer
  compare for VARIABLE. All other task types keep the existing lv_memcmp path.
• The cache now owns the file path: lv_strdup on insert (draw_to_texture), lv_free on eviction (opengles_texture_cache_free_cb). Without this, lv_strcmp would
  dereference a path the widget had already freed — swapping the cache miss for a use-after-free.
• Moved colorkey from content compare to pointer compare. The cache doesn't own the colorkey struct, so dereferencing it would create the same UAF class as src. Same policy
  already used for bitmap_mask_src. Comments aligned.
• sup is intentionally skipped (decoder-derived, would always miss); bitmap_mask_src keeps pointer compare. Behaviour unchanged.

[cubic-dev-ai]

1 issue found across 1 file

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="src/draw/opengles/lv_draw_opengles.c">

<violation number="1" location="src/draw/opengles/lv_draw_opengles.c:283">
P2: Cache comparator uses relational ordering on unrelated pointers, making key ordering non-portable/undefined in C.</violation>
</file>

_{Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review.}

[github-actions]

Hi 👋, thank you for your PR!

We've run benchmarks in an emulated environment. Here are the results:

ARM Emulated 32b - lv_conf_perf32b

Scene Name	Avg CPU (%)	Avg FPS	Avg Time (ms)	Render Time (ms)	Flush Time (ms)
All scenes avg.	27 (-1)	37	7	7	0

Detailed Results Per Scene

Scene Name	Avg CPU (%)	Avg FPS	Avg Time (ms)	Render Time (ms)	Flush Time (ms)
Empty screen	11	33	0	0	0
Moving wallpaper	2	33	1	1	0
Single rectangle	0	50	0	0	0
Multiple rectangles	0	33 (-1)	0	0	0
Multiple RGB images	0	39	0	0	0
Multiple ARGB images	11 (-5)	40 (+2)	3 (-1)	3 (-1)	0
Rotated ARGB images	55 (-4)	44	15	15	0
Multiple labels	6 (+3)	35 (+2)	0	0	0
Screen sized text	81	45	17	17	0
Multiple arcs	40 (+1)	33	7	7	0
Containers	4 (+1)	37 (-1)	0	0	0
Containers with overlay	86 (-4)	21	44	44	0
Containers with opa	17 (+3)	37	1	1	0
Containers with opa_layer	18	34	5	5	0
Containers with scrolling	44	45	12	12	0
Widgets demo	71	39 (-1)	17	17	0
All scenes avg.	27 (-1)	37	7	7	0

ARM Emulated 64b - lv_conf_perf64b

Scene Name	Avg CPU (%)	Avg FPS	Avg Time (ms)	Render Time (ms)	Flush Time (ms)
All scenes avg.	25	37	6	6	0

Detailed Results Per Scene

Scene Name	Avg CPU (%)	Avg FPS	Avg Time (ms)	Render Time (ms)	Flush Time (ms)
Empty screen	11	33	0	0	0
Moving wallpaper	1	33	0	0	0
Single rectangle	0	50	0	0	0
Multiple rectangles	0	35	0	0	0
Multiple RGB images	0	39	0	0	0
Multiple ARGB images	11	42	0	0	0
Rotated ARGB images	29	33	9	9	0
Multiple labels	2	35	0	0	0
Screen sized text	85	46	18	18	0
Multiple arcs	33	33	6	6	0
Containers	4	37 (-1)	0	0	0
Containers with overlay	98 (+1)	22	41	41	0
Containers with opa	16 (+1)	37 (-1)	0	0	0
Containers with opa_layer	8 (+1)	35 (-1)	2 (+1)	2 (+1)	0
Containers with scrolling	47 (-1)	48 (-1)	12	12	0
Widgets demo	67	40	15	15	0
All scenes avg.	25	37	6	6	0

Disclaimer: These benchmarks were run in an emulated environment using QEMU with instruction counting mode.
The timing values represent relative performance metrics within this specific virtualized setup and should
not be interpreted as absolute real-world performance measurements. Values are deterministic and useful for
comparing different LVGL features and configurations, but may not correlate directly with performance on
physical hardware. The measurements are intended for comparative analysis only.

---

🤖 This comment was automatically generated by a bot.

[Copilot]

Pull request overview

Fixes OpenGL ES texture cache misses for file/symbol-backed images by comparing image sources by content (path/symbol string) rather than by pointer value, addressing #9903.

Changes:

• Added task_type to cache keys and used it to partition comparisons so different task descriptors are never compared with mismatched logic.
• Implemented an image-specific descriptor comparator that compares src by string content for FILE/SYMBOL and by pointer for VARIABLE, while keeping other task types on the existing lv_memcmp path.
• Made the texture cache own duplicated FILE/SYMBOL src strings (strdup on insert, free on eviction) to avoid use-after-free.

[borgeuz]

addressed in commit 00509d

[AndreCostaaa]

Hi. Thank you for the PR

Please address Copilot's remarks