[Bug 110796] [REGRESSION] [BISECTED] [GLESCTS] race between destruction of types and shader compilation (?)

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=110796

--- Comment #20 from Lionel Landwerlin <lionel.g.landwerlin at linux.intel.com> ---
Can get some hints in valgrind now :

==26855== Invalid read of size 8
==26855==    at 0x6192980: prototype_string(glsl_type const*, char const*,
exec_list*) (ast_function.cpp:92)
==26855==    by 0x619440C: print_function_prototypes(_mesa_glsl_parse_state*,
YYLTYPE*, ir_function*) (ast_function.cpp:777)
==26855==    by 0x61945CE: no_matching_function_error(char const*, YYLTYPE*,
exec_list*, _mesa_glsl_parse_state*) (ast_function.cpp:812)
==26855==    by 0x6198F77: ast_function_expression::hir(exec_list*,
_mesa_glsl_parse_state*) (ast_function.cpp:2386)
==26855==    by 0x619C056: ast_expression::do_hir(exec_list*,
_mesa_glsl_parse_state*, bool) (ast_to_hir.cpp:1408)
==26855==    by 0x619BE9A: ast_expression::hir_no_rvalue(exec_list*,
_mesa_glsl_parse_state*) (ast_to_hir.cpp:1310)
==26855==    by 0x619F18F: ast_expression_statement::hir(exec_list*,
_mesa_glsl_parse_state*) (ast_to_hir.cpp:2236)
==26855==    by 0x619F202: ast_compound_statement::hir(exec_list*,
_mesa_glsl_parse_state*) (ast_to_hir.cpp:2252)
==26855==    by 0x61A6FFD: ast_function_definition::hir(exec_list*,
_mesa_glsl_parse_state*) (ast_to_hir.cpp:6223)
==26855==    by 0x6199A07: _mesa_ast_to_hir(exec_list*,
_mesa_glsl_parse_state*) (ast_to_hir.cpp:159)
==26855==    by 0x60E0BE9: _mesa_glsl_compile_shader
(glsl_parser_extras.cpp:2157)
==26855==    by 0x5DBFCE0: _mesa_compile_shader (shaderapi.c:1200)
==26855==  Address 0x13a78550 is 16 bytes inside a block of size 224 free'd
==26855==    at 0x48369AB: free (vg_replace_malloc.c:540)
==26855==    by 0x5D00BC8: unsafe_free (ralloc.c:299)
==26855==    by 0x5D00B8D: unsafe_free (ralloc.c:292)
==26855==    by 0x5D00B8D: unsafe_free (ralloc.c:292)
==26855==    by 0x5D00AB2: ralloc_free (ralloc.c:262)
==26855==    by 0x60E163B: glsl_symbol_table::operator delete(void*)
(glsl_symbol_table.h:43)
==26855==    by 0x60E0DAF: _mesa_glsl_compile_shader
(glsl_parser_extras.cpp:2191)
==26855==    by 0x5DBFCE0: _mesa_compile_shader (shaderapi.c:1200)
==26855==    by 0x5DC0D55: _mesa_CompileShader (shaderapi.c:1569)
==26855==    by 0x12DDA16: glcts::TestCaseBase::buildProgramVA(unsigned int,
bool*, unsigned int, ...) (esextcTestCaseBase.cpp:424)
==26855==    by 0x12DE1C9: glcts::TestCaseBase::buildProgram(unsigned int,
unsigned int, unsigned int, char const* const*, unsigned int, unsigned int,
char const* const*, unsigned int, unsigned int, char const* const*, bool*)
(esextcTestCaseBase.cpp:581)
==26855==    by 0x1393DAE:
glcts::GeometryShaderFlatInterpolationTest::initProgram()
(esextcGeometryShaderQualifiers.cpp:181)
==26855==  Block was alloc'd at
==26855==    at 0x483577F: malloc (vg_replace_malloc.c:309)
==26855==    by 0x5D00665: ralloc_size (ralloc.c:119)
==26855==    by 0x5D00721: rzalloc_size (ralloc.c:151)
==26855==    by 0x5FBF655: exec_node::operator new(unsigned long, void*)
(list.h:58)
==26855==    by 0x60C806C: (anonymous
namespace)::builtin_variable_generator::add_const(char const*, int, int)
(builtin_variables.cpp:655)
==26855==    by 0x60C7968: (anonymous
namespace)::builtin_variable_generator::add_const(char const*, int)
(builtin_variables.cpp:454)
==26855==    by 0x60C8813: (anonymous
namespace)::builtin_variable_generator::generate_constants()
(builtin_variables.cpp:817)
==26855==    by 0x60CB436: _mesa_glsl_initialize_variables(exec_list*,
_mesa_glsl_parse_state*) (builtin_variables.cpp:1543)
==26855==    by 0x619993E: _mesa_ast_to_hir(exec_list*,
_mesa_glsl_parse_state*) (ast_to_hir.cpp:131)
==26855==    by 0x60E0BE9: _mesa_glsl_compile_shader
(glsl_parser_extras.cpp:2157)
==26855==    by 0x5DBFCE0: _mesa_compile_shader (shaderapi.c:1200)
==26855==    by 0x5DC0D55: _mesa_CompileShader (shaderapi.c:1569)
==26855==

-- 
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-3d-bugs/attachments/20190730/1aee29a0/attachment.html>