[lfs-dev] [lfs-book] Apparent build race in e2fsprogs

Ken Moffat zarniwhoop at ntlworld.com
Sun Jun 5 14:14:01 PDT 2016


On Sun, Jun 05, 2016 at 03:23:42PM -0500, Bruce Dubbs wrote:
> Ken Moffat wrote:
> > 
> > If/when I hack that, I can take this.  I've had ff working fine with
> > both 46.0 and 46.0.1 by adding debug stuff (but the libs were still
> > stripped!), but I have not yet worked out which part of those changes
> > was the fix.  At the moment it crashes when it starts to play html5 -
> > 46.0 only crashed if I paused).
> 
> I'll see about throwing in the patch at my next build, but I prefer to wait
> a while until there are more packages to update.  Currently there is only an
> eudev update that merges systemd updates to udev and and a kernel point
> release.  I don't think of those as critical for -dev right now.
> 
I was assuming that I would get firefox working well, even on
youtube, and be able to sort out which of my build changes in the
debugging area was actually necessary.  If I had got to that point,
firefox would have to be flagged up as requiring -j1 without the
make changes.  With -j8, I can build it in under 13 and a half
minutes, -j1 will probably take a lot longer - and on a slow machine
it would be extremely tedious.

BUT : I cannot fix the youtube part of firefox.  I had it working
well in youtube with both 46.0 (forgot to change to 46.0.1 for that
build) and then 46.0.1.  But reducing my build changes to try to
establish which was/were important failed.  Now, I have reinstated
all three (comment out --enable-stripp and --enable-install-strip,
add -g to CXXFLAGS) but it now crashes within 3 seconds of starting
to play an html5 video.  So, for the moment testers can see the
patch and use it if they think it will help.

For me, eudev-3.2 seems to work fine (passed both its tests), as
does linux-4.6.1 (both for headers and the running kernel).  But I
agree neither are urgent.

ĸen
-- 
I had to walk fifteen miles to school, barefoot in the snow.  Uphill both ways.


More information about the lfs-dev mailing list