Crash in libGLESv2 when running texture-from-pixmap-glesv2 in fallback mode

[Alexandros Frantzis]

When running the texture-from-pixmap-glesv2 example:
http://people.canonical.com/~afrantzis/texture-from-pixmap-glesv2/texture-from-pixmap-glesv2.tar.gz
using the fallback texture upload method (using XImage instead of EGLImage) there is a crash in the libGLESv2 library. This happens as soon as the program starts (before any user input) in a call to glTexSubImage2D().

$ gdb --args ./texture_from_pixmap_glesv2 --no-egl-image
GNU gdb (GDB) 7.2-ubuntu
Copyright (C) 2010 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law. Type "show copying"
and "show warranty" for details.
This GDB was configured as "arm-linux-gnueabi".
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>...
Reading symbols from /home/alf/work/texture-from-pixmap-glesv2/texture_from_pixmap_glesv2...done.
(gdb) r
Starting program: /home/alf/work/texture-from-pixmap-glesv2/texture_from_pixmap_glesv2 --no-egl-image
[Thread debugging using libthread_db enabled]
EGL extensions: eglCreatePbufferFromClientBuffer EGL_KHR_image EGL_KHR_image_base EGL_KHR_image_pixmap EGL_AMD_create_image
GL extensions: GL_AMD_compressed_3DC_texture GL_AMD_compressed_ATC_texture GL_AMD_performance_monitor GL_AMD_program_binary_Z400 GL_AMD_tiled_rendering GL_EXT_texture_filter_anisotropic GL_EXT_texture_type_2_10_10_10_REV GL_EXT_bgra GL_OES_compressed_ETC1_RGB8_texture GL_OES_compressed_paletted_texture GL_OES_depth_texture GL_OES_depth24 GL_OES_EGL_image GL_OES_EGL_image_external GL_OES_element_index_uint GL_OES_fbo_render_mipmap GL_OES_fragment_precision_high GL_OES_get_program_binary GL_OES_packed_depth_stencil GL_OES_rgb8_rgba8 GL_OES_standard_derivatives GL_OES_texture_3D GL_OES_texture_float GL_OES_texture_half_float GL_OES_texture_half_float_linear GL_OES_texture_npot GL_OES_vertex_half_float GL_OES_vertex_type_10_10_10_2 GL_NV_fence
vertex shader info:
fragment shader info:
info: Link was successful.

==> Draw something on the left with the mouse! <==

Program received signal SIGSEGV, Segmentation fault.
0x2abc14d6 in _rb_tile_with_range_check () from /usr/lib/libGLESv2.so.2.0
(gdb) bt
#0 0x2abc14d6 in _rb_tile_with_range_check () from /usr/lib/libGLESv2.so.2.0
#1 0x2abc14ba in _rb_tile_with_range_check () from /usr/lib/libGLESv2.so.2.0
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb)

[Jammy Zhou (jammy-zhou)]

For the March release of maverick installer, this problem also doesn't happen on my system. But there is some corruption on the cube when run the test application with --no-egl-image

[Matt Sealey (mwsealey)]

We've had some bugs in the same place running the EFL Expedite benchmark with the GL backend, but they magically went away somehow. Nice to know it's still in there.. but it is sort of random :/

[Matt Sealey (mwsealey)]

This actually looks like part of the GL stack or Xorg is walking over the texture data and headers.. which is worrying.

[Matt Sealey (mwsealey)]

All I'm seeing here on Smarttop is "Error: Couldn't get X visual".

I'm running in 32-bit rather than the 16-bit default mode on the Smartbook, could this be a problem?

[Matt Sealey (mwsealey)]

Note: from discussion with bug reporter, this seems fixed except for that graphical corruption, but I still cannot run the benchmark to find out if that's true on the latest code.