Check for null result from gdb_demangle
Commit Message
* Simon Marchi <simark@simark.ca> [2020-02-18 14:54:54 -0500]:
> On 2020-02-18 5:39 a.m., Andrew Burgess wrote:
> > * Keith Moyer <kmoy@google.com> [2020-02-13 11:20:46 -0800]:
> >
> >> On Thu, Feb 13, 2020 at 2:52 AM Andrew Burgess <andrew.burgess@embecosm.com>
> >> wrote:
> >>
> >>> If you are able to find an example of a symbol that triggers the crash
> >>>
> >>
> >> The symbol where I ran into the problem was "<anon>". I see references to
> >> this in GCC sources, but I'm not entirely clear on when this is emitted
> >> instead of "<anonymous struct>", "<anonymous union>", "<anonymous>", or
> >> similar.
> >>
> >> then it should be pretty easy to add a test, see for example the last
> >>> few lines of gdb.cp/demangle.exp for C++ demangling tests. Having a
> >>> test would help something like this happening again.
> >>>
> >>
> >> Given that the bug here is that this code doesn't check for the possibility
> >> that gdb_demangle can fail (not that it failed), what kind of test would
> >> you recommend (and where)?
> >
> > You're absolutely right. Sorry for the confusion. I withdraw my
> > request.
>
> I might be missing something, but shouldn't it be possible to write a test
> in gdb.dwarf2 that creates a struct with linkage name <anon> (or any
> undemanglable identifier) and exercise this?
Yes. Sorry, I think I must be having a bad week.
Ali,
Please find below a test for this bug. If Pedro's points have been
addressed then please feel free to commit this test and the fix as a
single commit to GDB.
Thanks,
Andrew
----
commit e45daded868575ab25178c45cb075d2f77f244a6
Author: Andrew Burgess <andrew.burgess@embecosm.com>
Date: Wed Feb 19 14:06:45 2020 +0000
gdb/testsuite: New test for demangle issue
gdb/testsuite/ChangeLog:
* gdb.dwarf2/cpp-linkage-name.c: New file.
* gdb.dwarf2/cpp-linkage-name.exp: New file.
Comments
>>>>> "Andrew" == Andrew Burgess <andrew.burgess@embecosm.com> writes:
Andrew> Please find below a test for this bug. If Pedro's points have been
Andrew> addressed then please feel free to commit this test and the fix as a
Andrew> single commit to GDB.
FWIW I agree with this. Please go ahead.
Tom
new file mode 100644
@@ -0,0 +1,29 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+ Copyright 2020 Free Software Foundation, Inc.
+
+ This program is free software; you can redistribute it and/or modify
+ it under the terms of the GNU General Public License as published by
+ the Free Software Foundation; either version 3 of the License, or
+ (at your option) any later version.
+
+ This program is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU General Public License for more details.
+
+ You should have received a copy of the GNU General Public License
+ along with this program. If not, see <http://www.gnu.org/licenses/>. */
+
+volatile struct
+{
+ int x;
+ int y;
+} global_var;
+
+int
+main (void)
+{
+ asm ("main_label: .globl main_label");
+ return global_var.x + global_var.y;
+}
new file mode 100644
@@ -0,0 +1,96 @@
+# Copyright 2020 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program. If not, see <http://www.gnu.org/licenses/>.
+
+# Some compilers give anonymous structures a linkage name, and that
+# linkage name doesn't demangle (within GDB calling gdb_demangle
+# return NULL). At one point this caused GDB to crash due to
+# dereferencing a NULL pointer.
+
+load_lib dwarf.exp
+
+# This test can only be run on targets which support DWARF-2 and use gas.
+if {![dwarf2_support]} {
+ return 0
+}
+
+standard_testfile cpp-linkage-name.c cpp-linkage-name-debug.S
+
+# Set up the DWARF for the test.
+set asm_file [standard_output_file $srcfile2]
+Dwarf::assemble $asm_file {
+ global srcdir subdir srcfile
+
+ lassign [function_range main ${srcdir}/${subdir}/${srcfile}] \
+ main_start main_length
+
+ cu {} {
+ DW_TAG_compile_unit {
+ {DW_AT_language @DW_LANG_C_plus_plus}
+ {DW_AT_name ada-linkage-name.c}
+ {DW_AT_comp_dir /tmp}
+
+ } {
+ declare_labels a_l b_l
+
+ a_l: DW_TAG_base_type {
+ {DW_AT_byte_size 4 DW_FORM_sdata}
+ {DW_AT_encoding @DW_ATE_signed}
+ {DW_AT_name int}
+ }
+
+ # To expose the bug that existed at one point this
+ # structure must have a linkage name, but no name, and the
+ # linkage name is something that doesn't demangle.
+ b_l: DW_TAG_structure_type {
+ {DW_AT_byte_size 8 DW_FORM_sdata}
+ {DW_AT_encoding @DW_ATE_signed}
+ {DW_AT_linkage_name <anon>}
+ } {
+ member {
+ {name x}
+ {type :$a_l}
+ {data_member_location 0 data1}
+ }
+ member {
+ {name y}
+ {type :$a_l}
+ {data_member_location 0 data1}
+ }
+ }
+ DW_TAG_subprogram {
+ {name "main"}
+ {low_pc $main_start addr}
+ {high_pc "$main_start + $main_length" addr}
+ {type :$a_l}
+ }
+ DW_TAG_variable {
+ {type :$b_l}
+ {external 1 flag}
+ {DW_AT_name global_var}
+ {DW_AT_location {
+ DW_OP_addr [gdb_target_symbol global_var]
+ } SPECIAL_expr}
+ }
+ }
+ }
+}
+
+if { [prepare_for_testing "failed to prepare" ${testfile} \
+ [list $srcfile $asm_file] {nodebug}] } {
+ return -1
+}
+
+gdb_assert [runto_main] "run to main"
+gdb_test "p global_var" " = {x = 0, y = 0}"