On Sun, 16 Oct 2011 20:03:24 +0200, Karel Gardas <[email protected]> wrote: > > Also I'm quite out of idea what was your intention on line above so I'm > not able to fix this myself. > Doh, yes. You are right. Fixed. I also fixed a serious oversight in my target address calculation.
> Also another simple fix is needed for compilation: > > --- a/rts/Linker.c > +++ b/rts/Linker.c > @@ -4388,7 +4388,7 @@ do_Elf_Rel_relocations ( ObjectCode* oc, char* ehdrC, > StgWord16 *lower = (StgWord16 *)(P + 2); > int sign = (*upper >> 10) & 1; > int j1, j2, i1, i2; > - int is_bl = lower & (1<<12); > + int is_bl = *lower & (1<<12); > Fixed. > > When you said you were unable to get the LLVM backend running on Thumb, > > what exactly did you mean? I've found your blog post Did you find that > > things segfaulted? > > Yes, simply speaking ghc-stage2 segfaulted on me and I've not invested > the time to find out why. This is task for future... > Alright. Not to complicate things, but I realized shortly after I finished my last reply that perhaps it would be quicker to cross-compile LLVM. I did this and it just finished. Given the binaries are 1.2GB, I'll need to wait until I'm on campus tomorrow to upload them to the ARM. That being said, this could be a much quicker way to test configurations. I've also read that recently someone was working on cross-compilation support for GHC. I've emailed the author to see whether this was ever merged. Again, this would likely make this an order of magnitude quicker for me. Anyways, I'll try building a non-Thumb GHC next. How do I configure the build system to accomplish this? Cheers, - Ben _______________________________________________ Cvs-ghc mailing list [email protected] http://www.haskell.org/mailman/listinfo/cvs-ghc
