[Bug 70461] [sna suse] ddx assertion hits

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=70461

--- Comment #138 from Chris Wilson <chris at chris-wilson.co.uk> ---
To try and capture my thoughts:

It appears to be associated with resizing the window.

[ 40132.873] sna_set_window_pixmap: window=23068682, old pixmap=2883 new
pixmap=2896
[ 40132.873] sna_dri2_decouple_window: window=23068682
...
[ 40132.877] sna_dri2_create_buffer pixmap=2896, (attachment=1, format=32,
drawable=532x495), window?=1
[ 40132.877] sna_dri2_create_buffer: creating back buffer 532x495, suitable for
scanout? 0
[ 40132.877] kgem_create_2d(532x495, bpp=32, tiling=1, exact=1, inactive=0,
cpu-mapping=0, gtt-mapping=0, scanout?=0, prime?=0, temp?=0)
[ 40132.877] kgem_surface_size: tile_width=512, tile_height=8 => aligned
pitch=2560, height=496
[ 40132.878] kgem_set_tiling: handle=42, tiling=1 [1], pitch=2560 [2560]: 1
[ 40132.878] kgem_bo_remove_from_inactive: removing handle=42 from inactive
[ 40132.878]   from inactive: pitch=2560, tiling=1: handle=42, id=1730
[ 40132.878] kgem_bo_flink: flinked handle=42 to name=20, marking non-reusable
[ 40132.878] sna_dri2_create_buffer pixmap=2896, (attachment=0, format=32,
drawable=532x495), window?=1
[ 40132.878] sna_pixmap_set_dri: attaching DRI client to pixmap=2896
[ 40132.878] sna_pixmap_move_to_gpu(pixmap=2896, usage=0), flags=c2
[ 40132.878] sna_pixmap_move_to_gpu: already all-damaged
[ 40132.878] sna_pixmap_mark_active: pixmap=2896, handle=50
[ 40132.878] sna_dri2_create_buffer: attaching to front buffer 532x495
[0x14e1030:1], scanout? 0
[ 40132.878] kgem_bo_flink: flinked handle=50 to name=22, marking non-reusable
[ 40132.878] sna_dri2_create_buffer: adding flush hint to handle=50
[ 40132.878] sna_accel_watch_flush: enable=1
[ 40132.878] _sna_dri2_destroy_buffer: 0x11cac10 [handle=37] -- refcnt=1,
pixmap=0, proxy?=0
[ 40132.878] sna_dri2_cache_bo(handle=37, name=17)
[ 40132.878] sna_dri2_cache_bo: wrong size [532x42], releasing handle
[ 40132.878] _kgem_bo_destroy: handle=37, proxy? 0
[ 40132.878] __kgem_bo_destroy: handle=37, size=122880
[ 40132.878] __kgem_bo_destroy: handle=37, not reusable
[ 40132.878] kgem_bo_free: handle=37, size=122880
[ 40132.878] kgem_bo_free: releasing 0x0:0x0 vma for handle=37, count=0
[ 40132.878] _sna_dri2_destroy_buffer: 0x11af9e0 [handle=35] -- refcnt=1,
pixmap=2883, proxy?=0
[ 40132.878] _sna_dri2_destroy_buffer: removing active_scanout=1 from pixmap
handle=35
[ 40132.878] _sna_dri2_destroy_buffer: releasing last DRI pixmap=2883,
scanout?=0
[ 40132.878] _sna_dri2_destroy_buffer: dropping flush hint from handle=35
...
[ 40132.886] sna_wakeup_handler
[ 40132.886] _sna_dri2_destroy_buffer: 0x11b1000 [handle=19] -- refcnt=1,
pixmap=0, proxy?=0
[ 40132.886] sna_dri2_cache_bo(handle=19, name=3)
[ 40132.886] sna_dri2_cache_bo: not a window, releasing handle=19
[ 40132.886] _kgem_bo_destroy: handle=19, proxy? 0
[ 40132.886] __kgem_bo_destroy: handle=19, size=122880
[ 40132.886] __kgem_bo_destroy: handle=19, not reusable
[ 40132.886] kgem_bo_free: handle=19, size=122880
[ 40132.886] kgem_bo_free: releasing 0x0:0x0 vma for handle=19, count=0
[ 40132.886] _sna_dri2_destroy_buffer: 0x11af9e0 [handle=35] -- refcnt=0,
pixmap=2883, proxy?=0
ASSERT!

The oddities that I need to explain:

Where did the last unref come from, the only clue in stacktrace is
FreeResource(), suggesting that it came from the DRI2 core.

Why didn't sna_dri2_decouple_window() mark the pending events as dead?

Do I have dangling private pointers from the window?

-- 
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/20160315/cc4454b5/attachment.html>