Re: Next upload?
On Wed, 2008-07-09 at 19:43 -0700, Russ Allbery wrote:
> On Wed, 2008-07-09 at 22:52 +0100, Adam D. Barratt wrote:
>
> > Never one to resist the urge to tinker, I've come up with a (not so
> > sneaky) solution that munges the output of readelf in to something that
> > resembles the original objdump info - at least sufficiently for
> > lintian's uses.
>
> Hm, do we lose anything in readelf output that we would have gotten in
> objdump? If readelf works when objdump doesn't, maybe we should just
> switch entirely over to readelf so that we don't have to run both of them
> the way that we do now.
The honest answer from my point of view is "I'm not sure". I know that
the output produced by my script currently loses some, because I only
made the munging intelligent enough to satisfy the uses that we're
currently making of the data - for instance, the symbol listing from
objdump looks like
00000000000756fc g DF .text 0000000000000064 GLIBCXX_3.4
_ZNSt11logic_errorC1ERKSs
whereas my script produces
00 XX .text 000000 GLIBCXX_3.4 _ZNSt11logic_errorC1ERKSs
(afaict, we only use the third, fifth and sixth fields in this case).
I'd suspect that all of the information that objdump provides is
available via readelf, but in any case objdump's output is much less
work :-) (the symbol number -> version and section number -> name
mappings are automagically done, for instance).
Adam
Reply to: