Re: [PATCH] gdb/dwarf: dump cooked index contents in cooked_index_functions::dump

[Tom Tromey <tom@tromey.com>]

>>>>> "Simon" == Simon Marchi via Gdb-patches <gdb-patches@sourceware.org> writes:

Simon> The cooked_index_functions::dump function is called as part of the
Simon> "maintenance print objfiles" command.  I tried to make the output
Simon> well structured and indented to help readability, as this prints a lot
Simon> of text.

Thanks for doing this.

Simon> diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c
Simon> index cd937f24ee74..73aafbed80b4 100644
Simon> --- a/gdb/dwarf2/read.c
Simon> +++ b/gdb/dwarf2/read.c
Simon> @@ -93,6 +93,7 @@
Simon>  #include "split-name.h"
Simon>  #include "gdbsupport/parallel-for.h"
Simon>  #include "gdbsupport/thread-pool.h"
Simon> +#include "inferior.h"
 
Simon>  /* When == 1, print basic high level tracing messages.
Simon>     When > 1, be more verbose.
Simon> @@ -18588,7 +18589,78 @@ struct cooked_index_functions : public dwarf2_base_index_functions
 
Simon>    void dump (struct objfile *objfile) override
Simon>    {
Simon> -    gdb_printf ("Cooked index in use\n");
Simon> +    gdb_printf ("Cooked index in use:\n");
Simon> +    gdb_printf ("\n");

I think I'd rather have the bulk of the dumping in the cooked-index code
instead of here.

Simon> +
Simon> +    dwarf2_per_objfile *per_objfile = get_dwarf2_per_objfile (objfile);
Simon> +
Simon> +    cooked_index_vector *table
Simon> +      = (gdb::checked_static_cast<cooked_index_vector *>
Simon> +         (per_objfile->per_bfd->index_table.get ()));

Like right here could be a method call.

The cooked index dumper should probably start with a call to 'wait' to
ensure that no background work is being done.

Simon> +	    const char *start_addr_str
Simon> +	      = paddress (current_inferior ()->gdbarch, start_addr);

I was curious why the inferior.h include was needed.
I suspect the objfile's arch is preferred here.

Tom