[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#638712: Re: Bug#638712: binNMUs for packages exporting (rest|save)(32|64\|)(fpr|gpr) symbols



On Monday 22 August 2011 19:32:28 Julien Cristau wrote:
> On Sun, Aug 21, 2011 at 13:50:19 +0200, Julien Cristau wrote:
> > On Sun, Aug 21, 2011 at 13:20:59 +0200, Clara Gnos wrote:
> > > Package: release.debian.org
> > > Severity: normal
> > > User: release.debian.org@packages.debian.org
> > > Usertags: binnmu
> > > 
> > > It was noticed in bug #638271 that some powerpc binaries would lead
> > > to
> > > unusable compilation results. The reason for that was that they
> > > export symbols that match the expression
> > > .*(rest|save)(32|64\|)(fpr|gpr).* and therefore would force the
> > > dynamic linker to make wrong decisions and crash before the actual
> > > program was started. This was fixed a long time ago in the gcc
> > > libraries http://gcc.gnu.org/ml/gcc-patches/2008-10/msg00264.html ,
> > > but some packages weren't recompiled since that time.
> > 
> > I'm working on a list too, will try to finish it later today so we can
> > compare...
> 
> So I restricted the search to shared libs under /usr/lib, and ended up
> with:
> 
[...]

Why do you think that only libraries are the problem? I thought also 
executables can export symbols which will be used by the dynamic linker to 
resolve unresolved symbols. So loading a library which needs that symbols 
would also cause a crash... (please correct me when I am wrong).

The first thing I noticed: I forgot to remove duplicated entries from my list 
(sry for that... sort|uniq would have been to easy) and that I include the 
\+b[0-9]* stuff (sed 's/+b[0-9]$//' would be useful too).

It looks like your list is a subset of mine. So it is at least a good start... 
but I am irritated that you don't have things like libopensync-plugin-
gnokii_0.22-1 (ok, it is not buildable right now... but it is still a 
problematic library/plugin) or libbz2-ruby_0.2.2-2

Attachment: difference.gz
Description: GNU Zip compressed data


Reply to: