2011/10/1 Henrik Nordström <henrik(a)henriknordstrom.net>:
We just ran into the same issue on armv7hl and found that mentioned
patch have finally been integrated mainline. We will be pulling in
binutils-2.21.51.0.6-9.fc15 to armv7hl to address this.
It seems to be a very very thin line where it triggers. We have two
identical gcc with only cosmetic specfile differences where one triggers
it while building ghc while the other do not.
Great. I'm not surprised that its needed. We will put the same version
into armv5 stage3 then we'll be in sync on that.
Just curious, how did you identify that what you faced is the same issue?
In armv7hl we just found that the GCC we have which we thought were
just
with a minor patch actually diverge significantly from mainline. Seems
the gcc-4.6.1-4.fc15 we have is not at all gcc-4.6.1-4.fc15 but rather a
git version after -3 but before -4 and with quite a lot of arm hackery,
no changelog notices and no tag in Release. This needs to be
investigated and cleaned up as we don't know yet how much of those
changes are in 4.6.1-9 or which of those changes should even be there.
I'd be interested in a summary of the changes, maybe I can make some
suggestions based on my experience with F14 and with 4.6.1-9.fc15.
armv5tel gcc build progressing nicely - should be finished today.
Daniel