[RFA 1/2] Make line tables independent of progspace

Simon Marchi simark@simark.ca
Sat Mar 24 03:40:00 GMT 2018


Hi Tom,

I haven't looked at in details, but I just have one comment for now:

On 2018-03-21 01:18 PM, Tom Tromey wrote:
> diff --git a/gdb/symtab.h b/gdb/symtab.h
> index f9d52e7697..2b3d8018a7 100644
> --- a/gdb/symtab.h
> +++ b/gdb/symtab.h
> @@ -1228,9 +1228,23 @@ struct rust_vtable_symbol : public symbol
>  struct linetable_entry
>  {
>    int line;
> -  CORE_ADDR pc;
> +
> +  /* Note that the PC as stored is unrelocated.  The appropriate
> +     offset must be applied before it can be used.  */
> +  CORE_ADDR m_pc;
>  };
>  
> +#define SET_LINETABLE_ENTRY_ADDRESS(ENTRY, PC)		\
> +  ((ENTRY).m_pc = (PC))
> +#define LINETABLE_ENTRY_RAW_ADDRESS(ENTRY)		\
> +  ((ENTRY).m_pc + 0)
> +#define LINETABLE_ENTRY_ADDRESS(SYMTAB, ENTRY)				\
> +  gdbarch_addr_bits_remove						\
> +    (get_objfile_arch (SYMTAB_OBJFILE (SYMTAB)),			\
> +     ((ENTRY).m_pc							\
> +      + ANOFFSET (SYMTAB_OBJFILE (SYMTAB)->section_offsets,		\
> +		  SYMTAB_COMPUNIT (SYMTAB)->block_line_section)))

Is it possible to make these functions instead of macros?  They could even be
made into methods of struct linetable_entry without making the structure
non-pod.  I think it would also make the code more readable:

  LINETABLE_ENTRY_RAW_ADDRESS (e)
  SET_LINETABLE_ENTRY_ADDRESS (e, pc);
  LINETABLE_ENTRY_ADDRESS (s, e)

vs

  e.raw_address ()
  e.set_address (pc)
  e.address (s)

I think it makes a good difference in terms of readability in those big ifs:

      if (LINETABLE_ENTRY_ADDRESS (iter_s, *item) > pc
	  && (!alt || (LINETABLE_ENTRY_ADDRESS (iter_s, *item)
		       < LINETABLE_ENTRY_ADDRESS (alt_symtab, *alt))))
Simon



More information about the Gdb-patches mailing list