On Sun, Nov 30, 2014 at 11:18:19PM +0100, Matteo Panella wrote:
> On 30/11/2014 22:30, Roger Leigh wrote:
> > Isn't the stripped one that would generally be used though when it comes
> > to the ABI? Unless I'm missing the intent here?
>
> You're right: at first I thought that -flto (unconditionally set by
> systemd's configure.ac when built with -O1 or higher) trampled over the
> DWARF sections during the build since I was unable to get an ABI dump
> using libudev1 and systemd-dbg, but after your message I ran eu-readelf
> -w /lib/x86_64-linux-gnu/libudev.so.1.5.0 and got all DWARF info nicely
> decoded.
>
> I strongly suspect there's a bug lurking in abi-dumper, so disregard the
> results I posted earlier.

I've certainly noticed that -flto can do some very strange things with
symbol tables, though I thought it was limited to mangling internal
symbols rather than public exports. I wouldn't be surprised if it has
other odd effects too.

-- 
  .''`.  Roger Leigh
 : :' :  Debian GNU/Linux    http://people.debian.org/~rleigh/
 `. `'   schroot and sbuild  http://alioth.debian.org/projects/buildd-tools
   `-    GPG Public Key      F33D 281D 470A B443 6756 147C 07B3 C8BC 4083 E800