[OE-core] [oe][Patch] package.bbclass: fix host contamination warnings for source files

[Max Krummenacher]

Hi

Am Sonntag, den 03.04.2016, 22:45 +0100 schrieb Richard Purdie:
> On Sun, 2016-04-03 at 22:57 +0200, Max Krummenacher wrote:
> > Addresses https://bugzilla.yoctoproject.org/show_bug.cgi?id=8939
> > 
> > Source files deployed with the *-dbg packages are owned by the user
> > running bitbake leading to warnings as the one below.
> > 
> > WARNING: glibc-2.23-r0 do_package_qa: QA Issue: glibc: /glibc
> > -dbg/usr/src/debug/glibc/2.23-r0/git/include/resolv.h is owned by
> > uid
> > 1000, which is the same as the user running bitbake. This may be
> > due
> > to host contamination
> > glibc: /glibc-dbg/usr/src/debug/glibc/2.23
> > -r0/git/include/monetary.h
> > is owned by uid 1000, which is the same as the user running
> > bitbake.
> > This may be due to host contamination
> > glibc: /glibc-dbg/usr/src/debug/glibc/2.23-r0/git/include/locale.h
> > is
> > owned by uid 1000, which is the same as the user running bitbake.
> > This may be due to host contamination
> > ...
> > 
> > The files are copied as part of the do_package task.
> > The patch chowns all file in packages/usr/src after cpio copied
> > them
> > into the
> > package directory.
> > 
> > Signed-off-by: Max Krummenacher <max.krummenacher at toradex.com>
> > ---
> > 
> > 
> >  meta/classes/package.bbclass | 23 +++++++++++++++++++++++
> >  1 file changed, 23 insertions(+)
> > 
> > diff --git a/meta/classes/package.bbclass
> > b/meta/classes/package.bbclass
> > index bdbe96d..d9ef62c 100644
> > --- a/meta/classes/package.bbclass
> > +++ b/meta/classes/package.bbclass
> > @@ -362,6 +362,7 @@ def copydebugsources(debugsrcdir, d):
> >      # and copied to the destination here.
> >  
> >      import stat
> > +    import subprocess
> >  
> >      sourcefile = d.expand("${WORKDIR}/debugsources.list")
> >      if debugsrcdir and os.path.isfile(sourcefile):
> > @@ -410,6 +411,28 @@ def copydebugsources(debugsrcdir, d):
> >          if retval:
> >              bb.fatal("debugsrc symlink fixup failed with exit code
> > %s (cmd was %s)" % (retval, cmd))
> >  
> > +        # cpio --no-preserve-owner does not create the destination
> > files with
> > +        # owner root even when run under pseudo, chown them
> > explicitely.
> 
> How about passing --owner=0:0 to cpio? 
> 
> I'm a little worried about why I don't see this failure on my own
> local
> builds.
> 
> We have a few cases where things sometimes seem to work out and
> sometimes don't and I'd love to get to the bottom of how to reproduce
> it and to understand why its different for different people.

I finally got enough time to investigate further.

I found cpio -l (i.e. createing hardlinks) under pseudo does not set
the owner to root, neither width --no-preserve-owner nor width -
-owner=0:0.

The file ownership in yocto is corrected later with fs-perms.txt.
Angstrom does provide its own fs perms configuration which disables the
oe-core fs-perms.txt. But due to a bug the angstrom file is not active.
Patch sent to the angstrom ML.
http://article.gmane.org/gmane.linux.distributions.angstrom.devel/7856

Max

> Cheers,
> 
> Richard