Home Original page

[Patch 2/4] Rewrite of the alpha-vms bfd back-end (bfd)

Nick Clifton nickc@redhat.com
Tue Apr 13 11:13:00 GMT 2010 
Hi Tristan,

> bfd/
> 2010-04-09  Tristan Gingold<gingold@adacore.com>
>
> 	* Makefile.am (BFD32_BACKENDS): Remove vms-gsd.lo, vms-hdr.lo,
> 	vms-tir.lo, vms.lo and add vms-alpha.lo
> 	(BFD32_BACKENDS_CFILES): Remove vms-gsd.c, vms-hdr.c,
> 	vms-tir.c, vms.c and add vms-alpha.c
> 	* Makefile.in: Regenerate.
> 	* configure.in (TDEFINES): Adjust file list for vms_alpha_vec.
> 	Remove vms_vax_vec.
> 	* configure: Regenerate.
> 	* targets.c (vms_vax_vec): Remove the declaration.
> 	(_bfd_target_vector): Remove vms_vax_vec.
> 	* vms-alpha.c: New file.
> 	* vms-gsd.c: Removed, rewritten in vms-alpha.c
> 	* vms-hdr.c: Ditto.
> 	* vms-tir.c: Ditto.
> 	* vms.c: Ditto.
> 	* vms-misc.c: Fix indentation and comments.  Replace most of
> 	#if VMS_DEBUG/vms_debug with vms_debug2.
> 	(_bfd_vms_hash_newfunc): Moved to vms-alpha.c
> 	(hash_string): Ditto.
> 	(_bfd_vms_length_hash_symbol): Ditto.
> 	(maybe_adjust_record_pointer_for_object): Ditto.
> 	(_bfd_vms_get_object_record): Ditto.
> 	(vms_get_remaining_object_record): Ditto.
> 	(_bfd_vms_push): Ditto.
> 	(_bfd_vms_pop): Ditto.
> 	(_bfd_vms_get_header_values): Removed.
> 	(_bfd_vms_get_first_record): Removed.
> 	(vms_get_remaining_image_record): Removed.
> 	(new_symbol): Removed.
> 	(_bfd_vms_enter_symbol): Removed.
> 	(_bfd_vms_save_sized_string): Use memcpy instead of strncpy.
> 	(_bfd_vms_output_begin): Remove rechead parameter.  Replace bfd
> 	parameter with struct vms_rec_wr.
> 	(_bfd_vms_output_push): Removed and replaced by ...
> 	(_bfd_vms_output_begin_subrec): ... new function.
> 	(_bfd_vms_output_alignment): Replace bfd parameter with
> 	struct vms_rec_wr, and adjust.
> 	(_bfd_vms_output_check): Ditto
> 	(_bfd_vms_output_byte): Ditto.
> 	(_bfd_vms_output_short): Ditto.
> 	(_bfd_vms_output_long): Ditto.
> 	(_bfd_vms_output_quad): Ditto.
> 	(_bfd_vms_output_counted): Ditto.
> 	(_bfd_vms_output_dump): Ditto.
> 	(_bfd_vms_output_fill): Ditto.
> 	(_bfd_vms_output_pop): Removed and replaced by ...
> 	(_bfd_vms_output_end_subrec): ... new function.
> 	(_bfd_vms_output_flush): Removed.
> 	(_bfd_vms_output_align): New function.
> 	(_bfd_vms_output_end): Add recwr parameter.  Adjust for this new
> 	parameter.
> 	(vms_convert_to_var): New function imported from vms.c
> 	(vms_convert_to_var_1): Ditto.
> 	(vms_convert_to_var_unix_filename): Ditto.
> 	(vms_get_module_name): Ditto.
> 	(get_vms_time_string): Ditto.
> 	(vms_time_to_time_t): Ditto.
> 	(vms_rawtime_to_time_t): Ditto.
> 	* vms.h: All macros for the VMS file format are now in include/vms.
> 	Prototypes for vms.c, vms-gsd.c, vms-misc.c, vms-hdr.c, vms-tir.c
> 	have been removed.
> 	(struct vms_symbol_struct, struct stack_struct): Moved to vms-alpha.c
> 	(struct fileinfo, struct srecinfo, struct lineinfo): Ditto.
> 	(struct funcinfo, struct vms_private_data_struct): Ditto.
> 	(struct vms_section_data_struct): Ditto.
> 	(struct vms_rec_rd, stryct vms_rec_wr): New declarations.
> 	(vms_get_module_name, get_vms_time_string): New declarations.
> 	(vms_time_to_time_t, vms_rawtime_to_time_t): Ditto.
> 	(_bfd_vms_output_begin_subrec, _bfd_vms_output_end_subrec): Ditto.
> 	(_bfd_vms_save_sized_string, _bfd_vms_save_counted_string): Adjusted.
> 	(_bfd_vms_output_begin, _bfd_vms_output_alignment): Ditto.
> 	(_bfd_vms_output_end,_bfd_vms_output_check): Ditto.
> 	(_bfd_vms_output_byte, _bfd_vms_output_short): Ditto.
> 	(_bfd_vms_output_long, _bfd_vms_output_quad): Ditto.
> 	(_bfd_vms_output_counted, _bfd_vms_output_dump): Ditto.
> 	(_bfd_vms_output_fill): Ditto.
> 	(bfd_vms_set_section_flags): Ditto.

Approved - please apply.

Some comments:

+/* Read & process EIHD record.
+   Return 0 on success, -1 on error  */
+
+static int
+_bfd_vms_slurp_eihd (bfd *abfd, unsigned int *eisd_offset,
+                     unsigned int *eihs_offset)

Why not use bfd_boolean as the return type here ?  (And for other, 
similar functions ...)


+	  sprintf (name, "$%s_%03d$", pfx, section_count++);

Is it (theoretically) possible to have more than 999 EISD records ?  If 
so then maybe an assertion or range check would be appropriate here.


Cheers
   Nick
More information about the Binutils mailing list