llvm-svn: 12895

llvm-svn: 12894

llvm-svn: 12893

llvm-svn: 12892

llvm-svn: 12891

LoopSimplify was not updating dominator frontiers correctly in some cases.

llvm-svn: 12890

llvm-svn: 12889

llvm-svn: 12888

llvm-svn: 12887

llvm-svn: 12886

llvm-svn: 12885

This is fairly straight-forward, but was a real nightmare to get just
perfect.  aarg.  :)

llvm-svn: 12884

consists mostly of changing sloppy K&R C code to slightly more
disciplined K&R C code, and doing the usual things to shut gcc up.

llvm-svn: 12877

llvm-svn: 12876

llvm-svn: 12875

llvm-svn: 12874

llvm-svn: 12873

llvm-svn: 12872

llvm-svn: 12871

Update testcase to illustrate the coalescing problem. The previous one did not work because of a fix in the x86 instruction selector.

llvm-svn: 12870

llvm-svn: 12869

llvm-svn: 12868

intrinsics.

Modified llvm.readio and llvm.writeio to use pointers to memory instead
of integers.  This should take care of problems such as different pointer
sizes, casting integers to pointers, weird architectural pointer types, etc.

Re-worded the description of llvm.readio and llvm.writeio so that it should
be more clear as to why they should be used over regular loads/stores for
I/O.

llvm-svn: 12867

llvm-svn: 12866

llvm-svn: 12865

llvm-svn: 12864

llvm-svn: 12863

llvm-svn: 12862

execute other CallGraphSCCPasses after the inliner without crashing.

llvm-svn: 12861

SCC passes much more useful.  In particular, this should fix the incredibly
stupid missed inlining opportunities that the inliner suffered from.

llvm-svn: 12860

llvm-svn: 12859

llvm-svn: 12858

llvm-svn: 12857

llvm-svn: 12856

llvm-svn: 12855

I have unsaved emacs buffers, geeze...

llvm-svn: 12854

llvm-svn: 12853

of the fucom[p][p] instructions.  This allows us to code generate this function

bool %test(double %X, double %Y) {
        %C = setlt double %Y, %X
        ret bool %C
}

... into:

test:
        fld QWORD PTR [%ESP + 4]
        fld QWORD PTR [%ESP + 12]
        fucomip %ST(1)
        fstp %ST(0)
        setb %AL
        movsx %EAX, %AL
        ret

where before we generated:

test:
        fld QWORD PTR [%ESP + 4]
        fld QWORD PTR [%ESP + 12]
        fucompp
**      fnstsw
**      sahf
        setb %AL
        movsx %EAX, %AL
        ret

The two marked instructions (which are the ones eliminated) are very bad,
because they serialize execution of the processor.  These instructions are
available on the PPRO and later, but since we already use cmov's we aren't
losing any portability.

I retained the old code for the day when we decide we want to support back
to the 386.

llvm-svn: 12852

llvm-svn: 12851

llvm-svn: 12850