Created attachment 15758 [details]
patch adding test-case gdb.dwarf2/dwz-2.exp

I rebased this patch series ( https://sourceware.org/pipermail/gdb-patches/2023-December/205054.html ) on top of trunk, and was left with two patches, variants of:
- [gdb/symtab] Recurse into c++ DW_TAG_subprogram DIEs for cooked index
- [gdb/symtab] Keep track of all parents for cooked index

Since the first one is in question (see PR30728), I tried to write a test-case for the second one that does not involve DW_TAG_inlined_subroutine.

I managed to write a test-case, see attached patch.

With gdb trunk, we get either:
...
$ gdb -q -batch outputs/gdb.dwarf2/dwz-2/dwz-2 \
    -ex "set lang c++" \
    -ex "print ns::foo" \
    -ex "print ns::bar" 
No symbol "foo" in namespace "ns".
$1 = {int (void)} 0x4101b4 <ns::bar()>
...
or:
...
$ gdb -q -batch outputs/gdb.dwarf2/dwz-2/dwz-2 \
    -ex "set lang c++" \
    -ex "print ns::foo" \
    -ex "print ns::bar" 
$1 = {int (void)} 0x4101ac <ns::foo()>
No symbol "bar" in namespace "ns".
...

With readnow:
...
$ gdb -readnow -q -batch outputs/gdb.dwarf2/dwz-2/dwz-2 \
    -ex "set lang c++" \
    -ex "print ns::foo" \
    -ex "print ns::bar" 
$1 = {int (void)} 0x4101ac <ns::foo()>
$2 = {int (void)} 0x4101b4 <ns::bar()>
...

Also, without the "set lang c++":
...
$ gdb -q -batch dwz-2 -ex "print ns::foo" -ex "print ns::bar"
$1 = {int (void)} 0x4101ac <ns::foo()>
$2 = {int (void)} 0x4101b4 <ns::bar()>
...
which is necessary because the dwarf assembly does not mention main and it's language.

The original series does not seem to fix this either.