Right now, symbol domains generally follow C. That is, there is a domain for tags, some for non-C things (Fortran stuff), a domain for labels, and then VAR_DOMAIN for everything else -- types, functions, variables. This leads to bad results, like bug #30158, where an attempt to look up a function instead finds a namespace. I think it would be better to drastically overhaul this code. There should be many more domains, as many as we think we'll need. Types, variables, and functions should all be separate. Probably namespaces should also be their own thing. Then, the symbol lookup functions should accept an enum flag type of all the domains that should be searched. This way, the C parser can implement its own semantics by searching the relevant C domains -- but other language parsers can do as they like.
These values are baked into the .gdb_index format. That isn't fatal but it does mean the index would be bit less efficient. Moving more to .debug_names or changing the format of .gdb_index would both be options here.
I've been slowly working on this. Lately I've been thinking that perhaps STRUCT_DOMAIN could be removed. It's only needed for C and is the source of a hack in symbol_matches_domain. Instead, any C-specific type-lookup code could just search TYPE_DOMAIN and look to see if the type is "tagged".
(In reply to Tom Tromey from comment #2) > I've been slowly working on this. > > Lately I've been thinking that perhaps STRUCT_DOMAIN could > be removed. It's only needed for C and is the source of a > hack in symbol_matches_domain. On further reflection, I don't think this will work properly, because in C it is fine to have a 'struct name' and a typedef for 'name' that are different -- they really are separate namespaces.
The rot here goes really deep :( symbol_matches_domain has a C++-specific hack. (That later was extended to other languages) However, the stabs reader, and some other readers, take care to handle the C++ typedef case in a more principled way: by creating a typedef symbol. One wonders why this wasn't done for DWARF...
https://sourceware.org/pipermail/gdb-patches/2023-November/204295.html
The master branch has been updated by Tom Tromey <tromey@sourceware.org>: https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=974b36c2ae2b351d022cc62579656f722da6e17a commit 974b36c2ae2b351d022cc62579656f722da6e17a Author: Tom Tromey <tom@tromey.com> Date: Thu Mar 2 07:44:11 2023 -0700 Use the new symbol domains This patch changes the DWARF reader to use the new symbol domains. It also adjusts many bits of associated code to adapt to this change. The non-DWARF readers are updated on a best-effort basis. This is somewhat simpler since most of them only support C and C++. I have no way to test a few of these. I went back and forth a few times on how to handle the "tag" situation. The basic problem is that C has a special namespace for tags, which is separate from the type namespace. Other languages don't do this. So, the question is, should a DW_TAG_structure_type end up in the tag domain, or the type domain, or should it be language-dependent? I settled on making it language-dependent using a thought experiment. Suppose there was a Rust compiler that only emitted nameless DW_TAG_structure_type objects, and specified all structure type names using DW_TAG_typedef. This DWARF would be correct, in that it faithfully represents the source language -- but would not work with a purely struct-domain implementation in gdb. Therefore gdb would be wrong. Now, this approach is a little tricky for C++, which uses tags but also enters a typedef for them. I notice that some other readers -- like stabsread -- actually emit a typedef symbol as well. And, I think this is a reasonable approach. It uses more memory, but it makes the internals simpler. However, DWARF never did this for whatever reason, and so in the interest of keeping the series slightly shorter, I've left some C++-specific hacks in place here. Note that this patch includes language_minimal as a language that uses tags. I did this to avoid regressing gdb.dwarf2/debug-names-tu.exp, which doesn't specify the language for a type unit. Arguably this test case is wrong. Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30164
Fixed.