llvm-svn: 9085

a crappy form of post-order traversal which really does not belong here.  While
we are at it, improve documentation and use a vector instead of a stack.

This improves the post dominator analysis pass by ~5%, and probably also helps
other passes as well.

llvm-svn: 9084

llvm-svn: 9083

llvm-svn: 9082

llvm-svn: 9081

llvm-svn: 9080

llvm-svn: 9079

a lot of virtual method dispatch overhead.

llvm-svn: 9078

llvm-svn: 9077

llvm-svn: 9076

llvm-svn: 9075

* Fix a nasty initializer ordering bug.  Any only-CFG passes which registered
  themselves before the CFGOnlyAnalysis vector initialized got forgotten and
  thus got invalidated and recomputed.

In particular, in my compiled version of gccas, the Loop information pass was
being recomputed unnecessarily.

llvm-svn: 9074

llvm-svn: 9073

llvm-svn: 9072

llvm-svn: 9071

llvm-svn: 9070

llvm-svn: 9069

llvm-svn: 9068

llvm-svn: 9067

Only transform call sites in a setjmp'ing function which are reachable from
the setjmp.  If the call dominates the setjmp (for example), the called
function cannot longjmp to the setjmp.

This dramatically reduces the number of invoke instructions created in some
large testcases.

llvm-svn: 9066

have a SINGLE backedge.  This is useful to, for example, the -indvars pass.

This implements testcase LoopSimplify/single-backedge.ll and closes PR#34

llvm-svn: 9065

llvm-svn: 9064

llvm-svn: 9063

llvm-svn: 9062

llvm-svn: 9061

llvm-svn: 9060

llvm-svn: 9059

llvm-svn: 9058

llvm-svn: 9057

allows GCCAS to only run it once.

llvm-svn: 9056

llvm-svn: 9055

llvm-svn: 9054

* Print floating point values using C99 hexadecimal style FP if possible.
  This increases the number of floating point constants that may be emitted
  inline, and improves precision for global variable initializers which
  can not be emitted in integer form.

This fixes the Olden/Power benchmark with the CBE!!!!

llvm-svn: 9052

llvm-svn: 9050

* Fix isFPCSafeToPrint to find more constants safe to print, which it was
  failing because ftostr was padding with leading space characters.
* Scan the entire module for global constants instead of each function at a
  time.  This has the advantage of allowing us to emit constants at global
  scope instead of function scope.  This speeds FP programs quite a bit.

llvm-svn: 9048

 try to understand it; maybe this will help someone else do so too.
Default CVSRootDir to $CVSROOT first, then the path in ~vadve only if $CVSROOT
 was not set.
Checkout with -z3 if we might be checking out over ssh/rsh.
Check the filename of gnuplot to see if it is executable; if not, as a
 last resort, try just plain "gnuplot", hoping it is in the path somewhere.
Remove a little extra whitespace.

llvm-svn: 9046

Linux. This is consistent with what FreeBSD and Solaris both want.
This makes the JIT work on FreeBSD 5.1-RELEASE. Whee.

llvm-svn: 9045

I don't remember precisely what bug this fixed, but I remember that
it fixed a bug. Sorry. (I think one of these args had a Funny Shell Character
in it, or a space, or something, in my last build, and this seemed
like the obvious fix.)

llvm-svn: 9044

exception header.  This is the final missing piece from the PR#27 puzzle.

llvm-svn: 9043

causing the build of lib/Target/X86 to fail.

llvm-svn: 9042