This is the mail archive of the
mailing list for the GDB project.
Re: [RFC] identifying if a variable/msymbol has a symbol...
- From: Jan Kratochvil <jan dot kratochvil at redhat dot com>
- To: Matt Rice <ratmice at gmail dot com>
- Cc: gdb-patches at sourceware dot org
- Date: Wed, 12 Oct 2011 23:45:55 +0200
- Subject: Re: [RFC] identifying if a variable/msymbol has a symbol...
- References: <CACTLOFp1cgYybBLZgcYC=Pfq6o3yOnmWbUCGpwCDxSOuCBsS3w@mail.gmail.com>
On Wed, 28 Sep 2011 03:39:11 +0200, Matt Rice wrote:
> i wrote it instead of the addrmap, because things like overlays end up
> hitting slow paths because
> addrmap doesn't support overlapping intervals,
I think this is the exact case like addrmap now covers code, it should just
cover also variables. For overlaps currently addrmap is already disabled
falling to the slow path (see find_pc_sect_psymtab) and I think it should
apply also to the variables. Therefore the new rbtree data structure is not
> I now see there is a per-objfile
> addrmap, not sure if that is .gdb-index specific though.
addrmap existed already before .gdb_index. .gdb_index just also supports
filling in addrmap and looking it up by dw2_find_pc_sect_symtab. There would
need to be some new quick_symbol_functions method for looking up variables (it
may be enough to overload the current find_pc_sect_symtab method also for
variables, not sure).