[Buildroot] Analysis of build failures

Baruch Siach baruch at tkos.co.il
Mon Nov 11 10:20:50 UTC 2013


Hi Thomas,

On Mon, Nov 11, 2013 at 11:15:46AM +0100, Thomas Petazzoni wrote:
> On Mon, 11 Nov 2013 12:10:53 +0200, Baruch Siach wrote:
> > On Mon, Nov 11, 2013 at 10:58:30AM +0100, Thomas Petazzoni wrote:
> > > On Sun, 10 Nov 2013 21:33:16 -0800, Chris Zankel wrote:
> > > > (I'm a bit surprised to get all those errors as binutils currently doesn't 
> > > > build. Are you using an older toolchain, or are these packages not 
> > > > dependent on the final binutils build?)
> > > 
> > > The binutils issues you're seeing are while building native binutils
> > > for the target, not when building cross-binutils for the cross
> > > compilation toolchain. Building the cross-binutils works ok, because
> > > they are built against the build machine C library, which is glibc and
> > > therefore has obstack stuff.
> > 
> > Why do we need binutils on target?
> 
> $ git grep "select BR2_PACKAGE_BINUTILS"
> package/dropwatch/Config.in:	select BR2_PACKAGE_BINUTILS
> package/oprofile/Config.in:	select BR2_PACKAGE_BINUTILS
> 
> So we need it for dropwatch and oprofile. In fact, what is needed is
> not the full binutils (i.e the binutils programs), but the libbfd
> library which is part of the binutils tarball.

This is strange. More than half of xtensa autobuilds used to fail on target 
binutils before Chris' obstack fix. Does this mean that all failing configs 
had dropwatch or oprofile enabled?

baruch

-- 
     http://baruch.siach.name/blog/                  ~. .~   Tk Open Systems
=}------------------------------------------------ooO--U--Ooo------------{=
   - baruch at tkos.co.il - tel: +972.2.679.5364, http://www.tkos.co.il -


More information about the buildroot mailing list