RFA: correctly detect overlapping overlays

RFA: correctly detect overlapping overlays

[Jim Blandy]

Same here.

2001-08-12  Jim Blandy  <jimb@redhat.com>

	* symfile.c (sections_overlap): New function.
	(map_overlay_command): Call sections_overlap, instead of using
	incorrect logic to recognize overlapping sections.

Index: gdb/symfile.c
===================================================================
RCS file: /cvs/cvsfiles/devo/gdb/symfile.c,v
retrieving revision 1.237
diff -c -r1.237 symfile.c
*** gdb/symfile.c	2001/06/14 21:03:18	1.237
--- gdb/symfile.c	2001/08/12 17:28:35
***************
*** 2486,2491 ****
--- 2486,2492 ----
     section_is_overlay(sect):      true if section's VMA != LMA
     pc_in_mapped_range(pc,sec):    true if pc belongs to section's VMA
     pc_in_unmapped_range(...):     true if pc belongs to section's LMA
+    sections_overlap(sec1, sec2):  true if mapped sec1 and sec2 ranges overlap
     overlay_mapped_address(...):   map an address from section's LMA to VMA
     overlay_unmapped_address(...): map an address from section's VMA to LMA
     symbol_overlayed_address(...): Return a "current" address for symbol:
***************
*** 2625,2630 ****
--- 2626,2645 ----
    return 0;
  }
  
+ 
+ /* Return true if the mapped ranges of sections A and B overlap, false
+    otherwise.  */
+ int
+ sections_overlap (asection *a, asection *b)
+ {
+   CORE_ADDR a_start = a->vma;
+   CORE_ADDR a_end   = a->vma + bfd_get_section_size_before_reloc (a);
+   CORE_ADDR b_start = b->vma;
+   CORE_ADDR b_end   = b->vma + bfd_get_section_size_before_reloc (b);
+ 
+   return (a_start < b_end && b_start < a_end);
+ }
+ 
  /* Function: overlay_unmapped_address (PC, SECTION)
     Returns the address corresponding to PC in the unmapped (load) range.
     May be the same as PC.  */
***************
*** 2803,2813 ****
        /* Next, make a pass and unmap any sections that are
           overlapped by this new section: */
        ALL_OBJSECTIONS (objfile2, sec2)
! 	if (sec2->ovly_mapped &&
! 	    sec != sec2 &&
! 	    sec->the_bfd_section != sec2->the_bfd_section &&
! 	    (pc_in_mapped_range (sec2->addr, sec->the_bfd_section) ||
! 	     pc_in_mapped_range (sec2->endaddr, sec->the_bfd_section)))
  	{
  	  if (info_verbose)
  	    printf_filtered ("Note: section %s unmapped by overlap\n",
--- 2818,2828 ----
        /* Next, make a pass and unmap any sections that are
           overlapped by this new section: */
        ALL_OBJSECTIONS (objfile2, sec2)
! 	if (sec2->ovly_mapped
!             && sec != sec2
!             && sec->the_bfd_section != sec2->the_bfd_section
!             && sections_overlap (sec->the_bfd_section,
!                                  sec2->the_bfd_section))
  	{
  	  if (info_verbose)
  	    printf_filtered ("Note: section %s unmapped by overlap\n",

Re: RFA: correctly detect overlapping overlays

[Andrew Cagney]

> +   CORE_ADDR a_start = a->vma;
> +   CORE_ADDR a_end   = a->vma + bfd_get_section_size_before_reloc (a);
> +   CORE_ADDR b_start = b->vma;
> +   CORE_ADDR b_end   = b->vma + bfd_get_section_size_before_reloc (b);


Jim, just a reminder on coding style.
There isn't any reason to (and probably good reason to not) lay out the 
code so carefully.  It just makes the task of someone running indent 
over the code all the more confusing.

	Andrew

Re: RFA: correctly detect overlapping overlays

[Jim Blandy]

Andrew Cagney <ac131313@cygnus.com> writes:
> > +   CORE_ADDR a_start = a->vma;
> > +   CORE_ADDR a_end   = a->vma + bfd_get_section_size_before_reloc (a);
> > +   CORE_ADDR b_start = b->vma;
> > +   CORE_ADDR b_end   = b->vma + bfd_get_section_size_before_reloc (b);
> 
> 
> Jim, just a reminder on coding style.
> There isn't any reason to (and probably good reason to not) lay out the 
> code so carefully.  It just makes the task of someone running indent 
> over the code all the more confusing.

I hadn't thought of people running indent.  I'll fix that.

But I do think it's a shame.  When layout can expose symmetry in what
you're doing, it makes it easier to read the code.

Re: RFA: correctly detect overlapping overlays

[Andrew Cagney]

> But I do think it's a shame.  When layout can expose symmetry in what
> you're doing, it makes it easier to read the code.


Order it as:

	start_a
	start_b
	end_a
	end_b

enjoy,
	Andrew

Re: RFA: correctly detect overlapping overlays

[Michael Snyder]

Jim Blandy wrote:
> 
> Andrew Cagney <ac131313@cygnus.com> writes:
> > > +   CORE_ADDR a_start = a->vma;
> > > +   CORE_ADDR a_end   = a->vma + bfd_get_section_size_before_reloc (a);
> > > +   CORE_ADDR b_start = b->vma;
> > > +   CORE_ADDR b_end   = b->vma + bfd_get_section_size_before_reloc (b);
> >
> >
> > Jim, just a reminder on coding style.
> > There isn't any reason to (and probably good reason to not) lay out the
> > code so carefully.  It just makes the task of someone running indent
> > over the code all the more confusing.
> 
> I hadn't thought of people running indent.  I'll fix that.
> 
> But I do think it's a shame.  When layout can expose symmetry in what
> you're doing, it makes it easier to read the code.

I agree, I like to do the kind of formatting that jim is doing.

Re: RFA: correctly detect overlapping overlays

[Michael Snyder]

Jim Blandy wrote:
> 
> Same here.

Approved.


> 
> 2001-08-12  Jim Blandy  <jimb@redhat.com>
> 
>         * symfile.c (sections_overlap): New function.
>         (map_overlay_command): Call sections_overlap, instead of using
>         incorrect logic to recognize overlapping sections.
> 
> Index: gdb/symfile.c
> ===================================================================
> RCS file: /cvs/cvsfiles/devo/gdb/symfile.c,v
> retrieving revision 1.237
> diff -c -r1.237 symfile.c
> *** gdb/symfile.c       2001/06/14 21:03:18     1.237
> --- gdb/symfile.c       2001/08/12 17:28:35
> ***************
> *** 2486,2491 ****
> --- 2486,2492 ----
>      section_is_overlay(sect):      true if section's VMA != LMA
>      pc_in_mapped_range(pc,sec):    true if pc belongs to section's VMA
>      pc_in_unmapped_range(...):     true if pc belongs to section's LMA
> +    sections_overlap(sec1, sec2):  true if mapped sec1 and sec2 ranges overlap
>      overlay_mapped_address(...):   map an address from section's LMA to VMA
>      overlay_unmapped_address(...): map an address from section's VMA to LMA
>      symbol_overlayed_address(...): Return a "current" address for symbol:
> ***************
> *** 2625,2630 ****
> --- 2626,2645 ----
>     return 0;
>   }
> 
> +
> + /* Return true if the mapped ranges of sections A and B overlap, false
> +    otherwise.  */
> + int
> + sections_overlap (asection *a, asection *b)
> + {
> +   CORE_ADDR a_start = a->vma;
> +   CORE_ADDR a_end   = a->vma + bfd_get_section_size_before_reloc (a);
> +   CORE_ADDR b_start = b->vma;
> +   CORE_ADDR b_end   = b->vma + bfd_get_section_size_before_reloc (b);
> +
> +   return (a_start < b_end && b_start < a_end);
> + }
> +
>   /* Function: overlay_unmapped_address (PC, SECTION)
>      Returns the address corresponding to PC in the unmapped (load) range.
>      May be the same as PC.  */
> ***************
> *** 2803,2813 ****
>         /* Next, make a pass and unmap any sections that are
>            overlapped by this new section: */
>         ALL_OBJSECTIONS (objfile2, sec2)
> !       if (sec2->ovly_mapped &&
> !           sec != sec2 &&
> !           sec->the_bfd_section != sec2->the_bfd_section &&
> !           (pc_in_mapped_range (sec2->addr, sec->the_bfd_section) ||
> !            pc_in_mapped_range (sec2->endaddr, sec->the_bfd_section)))
>         {
>           if (info_verbose)
>             printf_filtered ("Note: section %s unmapped by overlap\n",
> --- 2818,2828 ----
>         /* Next, make a pass and unmap any sections that are
>            overlapped by this new section: */
>         ALL_OBJSECTIONS (objfile2, sec2)
> !       if (sec2->ovly_mapped
> !             && sec != sec2
> !             && sec->the_bfd_section != sec2->the_bfd_section
> !             && sections_overlap (sec->the_bfd_section,
> !                                  sec2->the_bfd_section))
>         {
>           if (info_verbose)
>             printf_filtered ("Note: section %s unmapped by overlap\n",