Bug 55812

Ugh, having to copy the Cr buffer! Let's hope userptr doesn't miss this kernel cycle as well...

commit ae6ae91baa0574a7a23ab76afac5e2d827c49c20
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Tue Oct 9 18:36:59 2012 +0100

    sna: Check the map exists before trying to release it
    
    Reported-by: Clemens Eisserer <linuxhippy@gmail.com>
    Bugzilla: ttps://bugs.freedesktop.org/show_bug.cgi?id=55812
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

Your bugfix latency is incredible ;)

Another one, however only happens when debug!=full so probably timing related:

Program received signal SIGSEGV, Segmentation fault.
sna_render_composite_redirect_done (sna=sna@entry=0x7fb926193010, op=op@entry=0x7fff74c71e90) at sna_render.c:1916
1916				sna_damage_combine(t->real_damage, t->damage,
(gdb) bt
#0  sna_render_composite_redirect_done (sna=sna@entry=0x7fb926193010, op=op@entry=0x7fff74c71e90) at sna_render.c:1916
#1  0x00007fb9264ecd71 in gen6_render_fill_boxes (sna=0x7fb926193010, op=<optimized out>, format=<optimized out>, color=0x7fb924c65210, 
    dst=<optimized out>, dst_bo=0x23aaa60, box=<optimized out>, n=0) at gen6_render.c:3777
#2  0x00007fb92648bc6c in sna_composite_rectangles (op=1 '\001', dst=0x2791c00, color=0x7fb924c65210, num_rects=1, rects=0x7fb924c65218)
    at sna_composite.c:874
#3  0x00000000004fb165 in ?? ()
#4  0x000000000043447a in ?? ()
#5  0x0000000000423485 in ?? ()
#6  0x0000003ff1421735 in __libc_start_main (main=0x423110, argc=8, ubp_av=0x7fff74c72328, init=<optimized out>, fini=<optimized out>, 
    rtld_fini=<optimized out>, stack_end=0x7fff74c72318) at libc-start.c:226
#7  0x000000000042375d in _start ()

Furthermore I noticed google-chrome's shows a lot artifacts when its windows ~7500px, however I have to admit, I don't care about this ;)

composite_redirect on SNB, what sites do you frequent?!!!

google-chrome itself tends to break down with large windows, but it seems like you are having a lot of fun along code paths I haven't run since userptr :)

> composite_redirect on SNB, what sites do you frequent?!!!
Actually, just the chrome start "page" ;)

> google-chrome itself tends to break down with large windows
Good to know...

Whilst you have it around, can you do a 'bt full'? which I hope will be enough to see how we ended up here.

As you wish:

Program received signal SIGSEGV, Segmentation fault.
sna_render_composite_redirect_done (sna=sna@entry=0x7f026e353010, 
    op=op@entry=0x7fff8bc58520) at sna_render.c:1916
1916				sna_damage_combine(t->real_damage, t->damage,
(gdb) bt full
#0  sna_render_composite_redirect_done (sna=sna@entry=0x7f026e353010, 
    op=op@entry=0x7fff8bc58520) at sna_render.c:1916
        t = 0x7fff8bc58650
        __PRETTY_FUNCTION__ = "sna_render_composite_redirect_done"
#1  0x00007f026e6acd71 in gen6_render_fill_boxes (sna=0x7f026e353010, 
    op=<optimized out>, format=<optimized out>, color=0x11042e0, 
    dst=<optimized out>, dst_bo=0x1042680, box=<optimized out>, n=0)
    at gen6_render.c:3777
        tmp = {blt = 0x7fff8bc585c0, 
          box = 0x7f026e643154 <sna_accel_wakeup_handler+68>, 
          boxes = 0x7fff8bc58570, done = 0x7df180, damage = 0x7fff8bc58660, 
          op = 6205, dst = {pixmap = 0xf86ac0, format = 537004168, 
            bo = 0xff2a50, x = -384, y = -96, width = 256, height = 16}, 
          src = {bo = 0xfd7e80, transform = 0x10000000e, width = 61696, 
            height = 125, pict_format = 0, card_format = 1852099344, 
            filter = 32514, repeat = 0, is_affine = 0, is_solid = 0, 
            is_linear = 0, is_opaque = 0, alpha_fixup = 0, rb_reversed = 0, 
            offset = {0, -9986}, scale = {4.55618183e-41, 2.61657624e-38}, 
            embedded_transform = {matrix = {{0, 3, 0}, {16280256, 0, 
                  1848979472}, {32514, 9252, 0}}}, u = {gen2 = {
                pixel = 1851972677, linear_dx = 4.55618183e-41, linear_dy = 0, 
                linear_offset = 0}, gen3 = {type = 1851972677, mode = 32514, 
                constants = 0}}}, mask = {bo = 0x0, transform = 0x0, 
---Type <return> to continue, or q <return> to quit---
            width = 65505, height = 79, pict_format = 0, card_format = 0, 
            filter = 0, repeat = 0, is_affine = 0, is_solid = 0, 
            is_linear = 0, is_opaque = 0, alpha_fixup = 0, rb_reversed = 0, 
            offset = {0, -31280}, scale = {0, 6.28430653e-39}, 
            embedded_transform = {matrix = {{0, 17529264, 0}, {5592138, 0, 0}, 
                {0, 17529264, 0}}}, u = {gen2 = {pixel = 17529496, 
                linear_dx = 0, linear_dy = -7.60845301e-32, 
                linear_offset = 4.59163468e-41}, gen3 = {type = 17529496, 
                mode = 0, constants = 2344978136}}}, is_affine = 0, 
          has_component_alpha = 0, need_magic_ca_pass = 0, rb_reversed = 0, 
          floats_per_vertex = 2, floats_per_rect = 6, prim_emit = 0x4fffe1, 
          redirect = {real_bo = 0x1042680, real_damage = 0x3, 
            damage = 0xfe1640, box = {x1 = 0, y1 = 0, x2 = 0, y2 = 0}}, u = {
            blt = {src_pixmap = 0x200980, sx = -13696, sy = 183, inplace = 0, 
              overwrites = 0, bpp = 0, cmd = 12052032, br13 = 0, pitch = {
                524288, 0}, pixel = 2344978280, bo = {0x101094ec0, 
                0x10b79b0}}, gen3 = {constants = {2.94214383e-39, 0, 
                1.68785391e-38, 0, 1.68884939e-38, 0, 7.34683969e-40, 0}, 
              num_constants = 2344978280}, gen4 = {wm_kernel = 2099584, 
              ve_id = 0}, gen5 = {wm_kernel = 2099584, ve_id = 0}, gen6 = {
              flags = 2099584}, gen7 = {flags = 2099584}}, priv = 0x56d990}
        pixel = 4278190335
#2  0x00007f026e64bc6c in sna_composite_rectangles (op=1 '\001', dst=0xf667b0, 
---Type <return> to continue, or q <return> to quit---
    color=0x11042e0, num_rects=1, rects=0x11042e8) at sna_composite.c:874
        sna = 0x7f026e353010
        pixmap = 0xf86ac0
        priv = 0x1cb
        bo = 0x1042680
        damage = <optimized out>
        region = {extents = {x1 = 465, y1 = 111, x2 = 622, y2 = 112}, 
          data = 0x0}
        boxes = 0x7fff8bc58770
        dst_x = <optimized out>
        dst_y = <optimized out>
        num_boxes = 1
        error = <optimized out>
        hint = <optimized out>
        __PRETTY_FUNCTION__ = "sna_composite_rectangles"
#3  0x00000000004fb165 in ?? ()
No symbol table info available.
#4  0x000000000043447a in ?? ()
No symbol table info available.
#5  0x0000000000423485 in ?? ()
No symbol table info available.
#6  0x0000003ff1421735 in __libc_start_main (main=0x423110, argc=8, 
    ubp_av=0x7fff8bc589b8, init=<optimized out>, fini=<optimized out>, 
---Type <return> to continue, or q <return> to quit---
    rtld_fini=<optimized out>, stack_end=0x7fff8bc589a8) at libc-start.c:226
        result = <optimized out>
        unwind_buf = {cancel_jmp_buf = {{jmp_buf = {0, -4728514540509556870, 
                4339508, 140735538366896, 0, 0, 4728751704611786618, 
                -4747053972283289734}, mask_was_saved = 0}}, priv = {pad = {
              0x0, 0x0, 0x3ff100eee0, 0x1b0839}, data = {prev = 0x0, 
              cleanup = 0x0, canceltype = -251597088}}}
        not_first_call = <optimized out>
#7  0x000000000042375d in _start ()
No symbol table info available.

commit d73f5b5bb1a81421f1fdc3ac3b460a8ad90be0ad
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Tue Oct 9 19:41:10 2012 +0100

    sna/gen6+: Initialize the damage for fill-boxes
    
    In case we need to redirect the rendering for a large render target, we
    need to initialize the damage pointer.
    
    Reported-by: Clemens Eisserer <linuxhippy@gmail.com>
    Bugzilla: ttps://bugs.freedesktop.org/show_bug.cgi?id=55812
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

Crashing fixed, thanks!