[PATCH cygport 1/2] postinst: Never remove an existing .gnu_debuglink

[Corinna Vinschen]

On Feb  1 20:33, Jon Turney wrote:
> On 01/02/2022 19:22, Corinna Vinschen wrote:
> > Hi Jon,
> > 
> > On Feb  1 17:25, Jon Turney wrote:
> > > Be more careful not to remove an existing .gnu_debuglink, even if we
> > > think this package has no useful debug symbols.
> > > 
> > > (Some versions of 'llvm-objdump -l' fail to find line number info even
> > > though it's there.  Don't break a package which manages it's own debug
> > > symbols (e.g. cygwin) when that happens.)
> > > ---
> > >   lib/src_postinst.cygpart | 28 ++++++++++++++++++----------
> > >   1 file changed, 18 insertions(+), 10 deletions(-)
> > > 
> > > diff --git a/lib/src_postinst.cygpart b/lib/src_postinst.cygpart
> > > index d8bb226..e29b2cb 100644
> > > --- a/lib/src_postinst.cygpart
> > > +++ b/lib/src_postinst.cygpart
> > > @@ -1051,23 +1051,31 @@ __prepstrip() {
> > >   			lines=$(${objdump} -d -l "${exe}" 2>/dev/null | sed -ne "s|.*\(/usr/src/debug/${PF}/.*\):[0-9]*$|\1|gp" | sort -u | tee -a ${T}/.dbgsrc.out | wc -l);
> > 
> > Shouldn't lines be computed *after* the new check for .gnu_deb?  After
> > all, it's still pretty time-consuming and if the .gnu_deb check kicks in
> > it's never tested...
> 
> The objdump invocation has the side effect of creating ${T}/.dbgsrc.out,
> which is later used to determine what source files to put into the debuginfo
> package.

Ah, that explains it☝️