Re: [PATCH v2 2/2] gdb/c++: Improve error messages in overload resolution

[Bruno Larsen <blarsen@redhat.com>]

On 10/11/2022 16:19, Simon Marchi wrote:
>
> On 11/4/22 11:47, Bruno Larsen via Gdb-patches wrote:
>> When resolving overloaded functions, GDB relies on knowing relationships
>> between types, i.e. if a type inherits from another. However, some
>> compilers may not add complete information for given types as a way to
>> reduce unnecessary debug information. In these cases, GDB would just say
>> that it couldn't resolve the method or function, with no extra
>> information.
>>
>> The problem is that sometimes the user may not know that the type
>> information is incomplete, and may just assume that there is a bug in
>> GDB. To improve the user experience, we attempt to detect if the
>> overload match failed because of an incomplete type, and warn the user
>> of this.
>>
>> This commit also adds a testcase confirming that the message is only
>> triggered in the correct scenario. This test was not developed as an
>> expansion of gdb.cp/overload.cc because it needed the dwarf assembler,
>> and porting all of overload.cc seemed unnecessary.
> I get some new failures starting from this patch (tested both on Ubuntu
> 22.04 and a fairly recent Arch Linux):
>
> FAIL: gdb.cp/converts.exp: p foo3_1 (0, 1)
> FAIL: gdb.cp/converts.exp: p foo1_7(ta)
> FAIL: gdb.cp/converts.exp: strict type checking on: p foo1_type_check (123)
> FAIL: gdb.cp/converts.exp: strict type checking on: p foo2_type_check (0, 1)
> FAIL: gdb.cp/converts.exp: strict type checking on: p foo2_type_check (1, 0)
> FAIL: gdb.cp/converts.exp: strict type checking on: p foo2_type_check (1, 1)
> FAIL: gdb.cp/converts.exp: strict type checking on: p foo3_type_check (0, 0, 1)
> FAIL: gdb.cp/converts.exp: strict type checking on: p foo3_type_check (0, 1, 0)
> FAIL: gdb.cp/converts.exp: strict type checking on: p foo3_type_check (1, 0, 0)
> FAIL: gdb.cp/converts.exp: strict type checking on: p foo3_type_check (0, 1, 1)
> FAIL: gdb.cp/converts.exp: strict type checking on: p foo3_type_check (1, 1, 0)
> FAIL: gdb.cp/converts.exp: strict type checking on: p foo3_type_check (1, 1, 1)

/me facepalms

I'm very sorry, got tunnel visioned in fixing the issue, forgot to run 
the rest of the tests. Will fix all of these.

> FAIL: gdb.cp/enum-class.exp: print overload2(77)
> FAIL: gdb.cp/enum-class.exp: print overload3(E1::THERE)
> FAIL: gdb.cp/koenig.exp: p foo(ix)
> FAIL: gdb.cp/koenig.exp: p foo (p_union)
> FAIL: gdb.cp/local-static.exp: c++: print S::method()
> FAIL: gdb.cp/local-static.exp: c++: print S::inline_method()
> FAIL: gdb.cp/local-static.exp: c++: print S2<int>::method()
> FAIL: gdb.cp/local-static.exp: c++: print S2<int>::inline_method()
> FAIL: gdb.cp/operator.exp: namespace alias
> FAIL: gdb.cp/operator.exp: imported declaration
> FAIL: gdb.cp/pr12028.exp: p D::foo(b)
> FAIL: gdb.cp/rvalue-ref-overload.exp: passing lvalue arg to rvalue parameter
However, I'm not getting these fails... is gcc only emitting delaration 
DIEs for these tests, or is it something else?

-- 
Cheers,
Bruno

> Simon