Re: [RFC][gdb/symtab] Lazy expansion of full symbol table

[Tom Tromey <tom@tromey.com>]

>>>>> "Tom" == Tom de Vries <tdevries@suse.de> writes:

Tom> I managed to fix that using this patch:
Tom> ...
Tom> diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c
Tom> index 265ebee1de8..f98acfa33ed 100644
Tom> --- a/gdb/dwarf2/read.c
Tom> +++ b/gdb/dwarf2/read.c
Tom> @@ -7204,7 +7204,8 @@ skip_one_die (const struct die_reader_specs
Tom> *reader, const gdb_b
Tom> yte *info_ptr,
Tom>        /* We only handle DW_FORM_ref4 here.  */
Tom>        const gdb_byte *sibling_data = info_ptr + abbrev->sibling_offset;
Tom>        unsigned int offset = read_4_bytes (abfd, sibling_data);
Tom> -      const gdb_byte *sibling_ptr = buffer + offset;
Tom> +      const gdb_byte *sibling_ptr
Tom> +       = buffer + to_underlying (cu->header.sect_off) + offset;
Tom>        if (sibling_ptr >= info_ptr && sibling_ptr < reader->buffer_end)
Tom>         return sibling_ptr;
Tom>        /* Fall through to the slow way.  */
Tom> ...

Tom> The DW_FORM_ref4 offset is relative to the start of the CU, but buffer
Tom> points to the start of .debug_info.

Thanks.  I think I copied this without realizing or remembering that
read_attribute_value adds in the section offset automatically.

Tom