[cairo] 回复： cairo memory leak

[Maarten Bosmans]

Please, keep conversation on the list. (use reply-all)

Maarten

2011/10/14 唐博 <hitiger.2008 at yahoo.com.cn>:
> Thank you  very much. I have checked it , it stilled leak 10 bytes. I have no idea with it.
>
> ----- 原邮件 -----
> 发件人： Maarten Bosmans <mkbosmans at gmail.com>
> 收件人： 唐博 <hitiger.2008 at yahoo.com.cn>
> 抄送： "cairo at cairographics.org" <cairo at cairographics.org>
> 发送日期： 2011年10月14日, 星期五, 下午 5:51
> 主题: Re: [cairo] cairo memory leak
>
> 2011/10/14 唐博 <hitiger.2008 at yahoo.com.cn>:
>> I use cairo as this:
>>
>> cairo_move_to(cr, pPolyline->pPoints[0].x, pPolyline->pPoints[0].y);
>> for (s = 1; s < pPolyline->count; ++s)
>> {
>>   cairo_line_to(cr, pPolyline->pPoints[s].x, pPolyline->pPoints[s].y);
>> }
>> cairo_stroke(cr);
>> it will Memory leak. HookLogger said:
>>>>>>_tr_malloc <C:\systemapi\src\system\memory\malloc.cpp #7> + 8 - Navidog2X.exe! 4de78a6b()
>> __pixman_image_allocate <C:\workspaces\map9.1.5\src\XPixman_20110814\src\pixman-image.c #59> + 10 - Navidog2X.exe! 4df505bc()
>> _pixman_image_create_solid_fill <C:\workspaces\map9.1.5\src\XPixman_20110814\src\pixman-solid-fill.c #77> + 5 - Navidog2X.exe! 4df56c94()
>> __cairo_format_bits_per_pixel <C:\workspaces\map9.1.5\src\XCairo_20110815\src\cairo-image-surface.c #1076> + 9 - Navidog2X.exe! 4df3a088()
>> __cairo_format_bits_per_pixel <C:\workspaces\map9.1.5\src\XCairo_20110815\src\cairo-image-surface.c #1670> + 8 - Navidog2X.exe! 4df3ba28()
>> __cairo_format_bits_per_pixel <C:\workspaces\map9.1.5\src\XCairo_20110815\src\cairo-image-surface.c #3515> + 27 - Navidog2X.exe! 4df3f1e0()
>> __cairo_format_bits_per_pixel <C:\workspaces\map9.1.5\src\XCairo_20110815\src\cairo-image-surface.c #2327> + 44 - Navidog2X.exe! 4df3cd61()
>> __cairo_format_bits_per_pixel <C:\workspaces\map9.1.5\src\XCairo_20110815\src\cairo-image-surface.c #3579> + 32 - Navidog2X.exe! 4df3f396()
>> __cairo_format_bits_per_pixel <C:\workspaces\map9.1.5\src\XCairo_20110815\src\cairo-image-surface.c #3680> + 36 - Navidog2X.exe! 4df3f659()
>> __cairo_surface_stroke <C:\workspaces\map9.1.5\src\XCairo_20110815\src\cairo-surface.c #2184> + 41 - Navidog2X.exe! 4df471b4()
>> __cairo_gstate_stroke <C:\workspaces\map9.1.5\src\XCairo_20110815\src\cairo-gstate.c #1145> + 96 - Navidog2X.exe! 4df75ffc()
>> _cairo_stroke_preserve <C:\workspaces\map9.1.5\src\XCairo_20110815\src\cairo.c #2416> + 20 - Navidog2X.exe! 4df4181d()
>> _cairo_stroke <C:\workspaces\map9.1.5\src\XCairo_20110815\src\cairo.c #2389> + 8
>
> You need to compile with -O0 for this backtrace to be useful.
> cairo_format_bits_per_pixel hardly a function that spans several
> thousands lines of code and pixman_image_create_solid_fill is not
> called from within cairo-image-surface.c, so the information you show
> above is incorrect.
>
>> I didn't find some wrong in pixman-image.c and cairo-image-surface.c . Who can tell me the reason.
>
> My first guess would be a solid image cache.
> Are you sure this is a real leak (if you draw and stroke the polygon
> inside a loop, does the memory leak get larger?)
>
> Maarten
>
>> --
>> cairo mailing list
>> cairo at cairographics.org
>> http://lists.cairographics.org/mailman/listinfo/cairo
>
>
>