Re: next release

[Jakub Jelinek <jakub at redhat dot com>]

On Wed, Jan 24, 2001 at 12:00:05AM +1100, Alan Modra wrote:
> On Sat, 20 Jan 2001, Jakub Jelinek wrote:
> 
> > Can anyone approve
> > http://sources.redhat.com/ml/binutils/2000-12/msg00328.html
> > or tell me what's wrong with it? Basically, readelf either does not work at
> > all with symbol versioning (if first vma is large) or shows wrong
> > information from time to time (even with first vma 0).
> 
> I noticed an inconsistency in the code printing verdefs.  The code you
> trimmed out used
> 
> 			  GET_DATA (offset + ivd.vd_aux, evda,
> 				    "version definition aux");
> 
> while the code left in uses
> 
> 		      GET_DATA (offset - ivd.vd_next + ivd.vd_aux,
> 				evda, "version def aux");
> 
> It looks to me you chose the correct one, but please double check this,
> then check it in.

The latter is correct (offset is incremented by ivd.vd_next in between) and
I think this is the reason why old readelf -a generates different result
than readelf -a with this patch:

The diff between the two outputs contains stuff like:
@@ -9935,7 +9936,7 @@ Version symbols section '.gnu.version' c
   038:   0 (*local*)       0 (*local*)       0 (*local*)       0 (*local*)
   03c:   0 (*local*)       0 (*local*)       7 (GLIBC_2.2)     2 (GLIBC_2.0)
   040:   2 (GLIBC_2.0)     2 (GLIBC_2.0)     2 (GLIBC_2.0)     2 (GLIBC_2.0)
-  044:   2 (GLIBC_2.0)     2 (GLIBC_2.1)     2 (GLIBC_2.0)     2 (GLIBC_2.0)
+  044:   2 (GLIBC_2.0)     2 (GLIBC_2.0)     2 (GLIBC_2.0)     2 (GLIBC_2.0)
   048:   3 (GLIBC_2.1)     3 (GLIBC_2.1)     2 (GLIBC_2.0)     2 (GLIBC_2.0)
   04c:   7 (GLIBC_2.2)     2 (GLIBC_2.0)     6 (GLIBC_2.1.3)   3 (GLIBC_2.1)
   050:   2 (GLIBC_2.0)     2 (GLIBC_2.0)     2 (GLIBC_2.0)     2 (GLIBC_2.0)
...

(- is readelf without this patch, + with that patch).

	Jakub