I do not consider PR1546 closed just yet. What I mentioned in the PR was only two of ca. 140 Solaris failures. Most of them complain that llc cannot choose between C and MSIL output formats. The below prototypical patch corrects this type of failure. Is this the right way of handling it? Why does llc only fail on Solaris and not on Darwin? After I understood this problem I am happy to commit these 100+ files. Cheers, Gabor Index: /home/ggreif/llvm/test/CodeGen/Generic/2005-07-12-memcpy-i64-length.ll ==================================================================--- /home/ggreif/llvm/test/CodeGen/Generic/2005-07-12-memcpy-i64-length.ll (revision 38488) +++ /home/ggreif/llvm/test/CodeGen/Generic/2005-07-12-memcpy-i64-length.ll (working copy) @@ -1,4 +1,4 @@ -; RUN: llvm-upgrade < %s | llvm-as | llc +; RUN: llvm-upgrade < %s | llvm-as | llc -march=c ; Test that llvm.memcpy works with a i64 length operand on all targets.
On Wed, 11 Jul 2007, Gabor Greif wrote:> I do not consider PR1546 closed just yet. > What I mentioned in the PR was only two of > ca. 140 Solaris failures. > > Most of them complain that llc cannot choose between C and > MSIL output formats. > > The below prototypical patch corrects this type of failure. > > Is this the right way of handling it? > > Why does llc only fail on Solaris and not on Darwin? > > After I understood this problem I am happy to commit these 100+ files.I just committed a patch to the sparc backend. Please see if that helps. -Chris> Cheers, > > Gabor > > > Index: /home/ggreif/llvm/test/CodeGen/Generic/2005-07-12-memcpy-i64-length.ll > ==================================================================> --- /home/ggreif/llvm/test/CodeGen/Generic/2005-07-12-memcpy-i64-length.ll (revision 38488) > +++ /home/ggreif/llvm/test/CodeGen/Generic/2005-07-12-memcpy-i64-length.ll (working copy) > @@ -1,4 +1,4 @@ > -; RUN: llvm-upgrade < %s | llvm-as | llc > +; RUN: llvm-upgrade < %s | llvm-as | llc -march=c > ; Test that llvm.memcpy works with a i64 length operand on all targets. > > > > > > _______________________________________________ > LLVM Developers mailing list > LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev >-Chris -- http://nondot.org/sabre/ http://llvm.org/
Hi Chris!> > After I understood this problem I am happy to commit these 100+ files. > > I just committed a patch to the sparc backend. Please see if that helps. > > -ChrisIt *does* help! These: M ../llvm/test/CodeGen/PowerPC/2006-08-15-SelectionCrash.ll M ../llvm/test/CodeGen/PowerPC/or-addressing-mode.ll M ../llvm/test/CodeGen/PowerPC/2006-01-20-ShiftPartsCrash.ll I'll checkin soon, they should be analogous to your change. Then my problem list will have shrunk to: # of expected passes 1807 # of unexpected failures 15 # of expected failures 4 ...which is pretty good, already. Cheers, Gabor