[Bug 105720] screen corruption using SNA and TearFree on Intel GeminiLake

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=105720

--- Comment #8 from Chris Wilson <chris at chris-wilson.co.uk> ---
Towards the end of the log, we see a sequence like

[    55.623] sna_poly_fill_rect: fallback - fbPolyFillRect
[    55.623] sna_gc_move_to_gpu(0x55e0f751c330)
[    55.623] sna_change_window_attributes
[    55.623] sna_change_window_attributes: flushing background pixmap
[    55.623] sna_validate_pixmap: target bpp=32, source bpp=32
[    55.623] sna_validate_gc(0x55e0f6f690c0) changes=7fffff, previous
serial=80000000, drawable=247
[    55.623] sna_validate_gc: recomputing clip
[    55.623] sna_validate_gc: composite clip=1x[(3090, 1211), (3106, 1227)]
[0x55e0f73a4f30]
[    55.623] sna_poly_fill_rect(n=1, PlaneMask: ffffffff (solid 1), solid fill:
0 [style=1, tileIsPixel=0], alu=3)
[    55.623] sna_poly_fill_rect_extents: [0] = (0, 0)x(16, 16)
[    55.623] sna_poly_fill_rect: extents(3090, 1211), (3106, 1227), flags=1
[    55.623] sna_poly_fill_rect: dropping last-cpu hint
[    55.623] sna_drawable_use_bo pixmap=8, box=((3090, 1211), (3106, 1227)),
flags=9...
[    55.623] sna_drawable_use_bo: pinned, never REPLACES
[    55.623] sna_drawable_use_bo: flush=0, shm=0, cpu=0 => flags=9
[    55.623] sna_drawable_use_bo: use GPU fast path (all-damaged)
[    55.623] sna_drawable_use_bo: applying move-to-gpu override
[    55.623] sna_pixmap_discard_shadow_damage: discarding region 1x[(3090,
1211), (3106, 1227)] from damage 1x[(2989, 1209], (3106, 1229)]
[    55.623] wait_for_shadow: enabled? 1 waiting? 0, flags=3, flips=2, pixmap=8
[front?=1], handle=10, shadow=39
[    55.626] wait_for_shadow: 2 flips still pending, shadow flip_active=2
[    55.626] wait_for_shadow: after waiting 2 flips outstanding, flip_active=2
[    55.626] kgem_create_2d(5760x2160, bpp=32, tiling=1, exact=1, inactive=0,
cpu-mapping=0, gtt-mapping=0, scanout?=1, prime?=0, temp?=0)
[    55.626] kgem_surface_size: tile_width=512, tile_height=8 => aligned
pitch=23040, height=2160
[    55.626]   1:from scanout: pitch=23040, tiling=1, handle=11, id=447
[    55.626] wait_for_shadow: replacing still-attached GPU bo handle=1, flips=2
[    55.626] wait_for_shadow: copying existing GPU damage: 4x(0, 0), (5760,
2160)
[    55.626] gen9_render_copy_boxes (0, 0)->(0, 0) x 4, alu=3, flags=0,
self-copy=0, overlaps? 0

The tricky part here is sna_pixmap_discard_shadow_damage. So the idea behind
the earlier test was to remove that discard and force the full repaint. It's
still the best idea I have as to where it goes wrong...

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20180326/7cc2b607/attachment.html>