[Bug translator/27933] New: Use of unitialized functioncall synthetic field in
mark at klomp dot org
sourceware-bugzilla@sourceware.org
Mon May 31 12:21:28 GMT 2021
https://sourceware.org/bugzilla/show_bug.cgi?id=27933
Bug ID: 27933
Summary: Use of unitialized functioncall synthetic field in
Product: systemtap
Version: unspecified
Status: NEW
Severity: normal
Priority: P2
Component: translator
Assignee: systemtap at sourceware dot org
Reporter: mark at klomp dot org
Target Milestone: ---
Running the helloworld.stp example under valgrind gives:
==113473== Conditional jump or move depends on uninitialised value(s)
==113473== at 0x1EE356:
functioncall_security_check::visit_embeddedcode(embeddedcode*)
(elaborate.cxx:3140)
==113473== by 0x1EAC50: UnknownInlinedFun (elaborate.cxx:3110)
==113473== by 0x1EAC50: symresolution_info::find_functions(functioncall*,
std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >
const&, unsigned int, token const*) (elaborate.cxx:3268)
==113473== by 0x1ED93B:
symresolution_info::visit_functioncall(functioncall*) (elaborate.cxx:2946)
==113473== by 0x1E93BB: symresolution_info::visit_block(block*)
(elaborate.cxx:2610)
==113473== by 0x1E93BB: symresolution_info::visit_block(block*)
(elaborate.cxx:2610)
==113473== by 0x1E83E4: semantic_pass_symbols(systemtap_session&) [clone
.lto_priv.0] (elaborate.cxx:1977)
==113473== by 0x19BCCE: UnknownInlinedFun (elaborate.cxx:2540)
==113473== by 0x19BCCE: passes_0_4(systemtap_session&) (main.cxx:1049)
==113473== by 0x18FB73: main (main.cxx:1534)
==113473== Uninitialised value was created by a heap allocation
==113473== at 0x4840FF5: operator new(unsigned long)
(vg_replace_malloc.c:417)
==113473== by 0x1C4E22: UnknownInlinedFun (parse.cxx:4044)
==113473== by 0x1C4E22: parser::parse_value() (parse.cxx:3795)
==113473== by 0x1C59E9: parser::parse_dwarf_value() (parse.cxx:3732)
==113473== by 0x1C6C0D: UnknownInlinedFun (parse.cxx:3690)
==113473== by 0x1C6C0D: parser::parse_unary() (parse.cxx:3664)
==113473== by 0x1C6ECF: parser::parse_multiplicative() (parse.cxx:3625)
==113473== by 0x1C7023: parser::parse_additive() (parse.cxx:3602)
==113473== by 0x1BB25F: parser::parse_concatenation() (parse.cxx:3578)
==113473== by 0x1BB39F: parser::parse_shift() (parse.cxx:3555)
==113473== by 0x1BBCDB: UnknownInlinedFun (parse.cxx:3513)
==113473== by 0x1BBCDB: parser::parse_array_in() (parse.cxx:3465)
==113473== by 0x1BC2CF: parser::parse_boolean_and() (parse.cxx:3421)
==113473== by 0x1BC40F: parser::parse_boolean_xor() (parse.cxx:3399)
==113473== by 0x1BC54F: parser::parse_boolean_or() (parse.cxx:3377)
valgrind seems right, functioncall_security_check::visit_embeddedcode does:
// Don't allow /* guru */ functions unless caller is privileged.
if (!call->synthetic && !call->tok->location.file->privileged &&
s->tagged_p ("/* guru */"))
throw SEMANTIC_ERROR (_("function may not be used unless -g is specified"),
call->tok);
Checking the functioncall synthetic field.
But that seems never have been initialized at parse.cxx:4044 we see:
struct functioncall* f = new functioncall;
f->tok = t;
f->function = name;
There are some code paths that set functioncall synthetic to true. Maybe the
default initialization should be false for this field?
--
You are receiving this mail because:
You are the assignee for the bug.
More information about the Systemtap
mailing list