llvm-svn: 31152

llvm-svn: 31151

a place more useful.  In particular, if we can put them in a place where code
will be able to fall into it, do so.  Otherwise, put it in a place it can fall
through into a successor.  Otherwise, if preventing a fallthrough, move to the
end of the function, out of the way.

This deletes several hundred unconditional branches from spass.

llvm-svn: 31149

llvm-svn: 31148

passes llvm-gcc bootstrap.

llvm-svn: 31146

llvm-svn: 31145

llvm-svn: 31140

llvm-svn: 31138

llvm-svn: 31137

llvm-svn: 31136

llvm-svn: 31135

Prolangs-C/agrep and SCCP/2006-10-23-IPSCCP-Crash.ll

llvm-svn: 31132

        movl 32(%esp), %eax
        cmpl $1, %eax
        je LBB1_1       #bb
LBB1_4: #entry
        cmpl $2, %eax
        je LBB1_2       #bb2
        jmp LBB1_3      #UnifiedReturnBlock
LBB1_1: #bb

notice that we would miss the fall through and emit this instead:

        movl 32(%esp), %eax
        cmpl $2, %eax
        je LBB1_2       #bb2
LBB1_4: #entry
        cmpl $1, %eax
        jne LBB1_3      #UnifiedReturnBlock
LBB1_1: #bb

llvm-svn: 31130

llvm-svn: 31129

llvm-svn: 31128

property is added by running through the list of uses of the value and
adding resolved properties to the property set.

llvm-svn: 31126

llvm-svn: 31125

llvm-svn: 31124

llvm-svn: 31123

llvm-svn: 31122

llvm-svn: 31121

This will allow (after llvm-gcc linkage fix) more dummy tests using
libstdc++ to pass on mingw :)

llvm-svn: 31120

llvm-svn: 31119

optimization opportunity pointed out by Chris Lattner.

llvm-svn: 31118

opportunity pointed out by Andrew Lewycky.

llvm-svn: 31115

transformation.  This speeds up a C++ app 2.25x.

llvm-svn: 31113

has a different prototype than the one #included from <setjmp.h>. This
patch fixes siod and a number of other test cases on Linux that were
failing the CBE because of this _setjmp issue.

llvm-svn: 31112

llvm-svn: 31111

llvm-svn: 31110

llvm-svn: 31109

llvm-svn: 31108

llvm-svn: 31107

1. Better document what is going on here.
2. Only hack on one branch per iteration, making the results less conservative.
3. Handle the problematic case by marking edges executable instead of by
   playing with value lattice states.  This is far less pessimistic, and fixes
   SCCP/ipsccp-gvar.ll.

llvm-svn: 31106

llvm-svn: 31102

This provides stuff like:

        cmpw cr0, r15, r29
        mr r14, r15
-       bge cr0, LBB3_111       ;bb656
-       b LBB3_90       ;bb501
+       blt cr0, LBB3_90        ;bb501
 LBB3_111:      ;bb656
        lwz r18, 68(r1)

which is particularly good for dispatch group formation.

llvm-svn: 31101

        call L_strcmp$stub
        testl %eax, %eax
-       jne LBB26_208   #cond_true6020
-       jmp LBB26_227   #bb7119
+       je LBB26_227    #bb7119
 LBB26_208:     #cond_true6020
        movl $l31_str14, 4(%esp)

        testl %eax, %eax
-       jne LBB26_704   #cond_true13042
-       jmp LBB26_713   #bb13151
+       je LBB26_713    #bb13151
 LBB26_704:     #cond_true13042
        movl $_str52, 4(%esp)

        cmpl 76(%ecx), %eax
-       jge LBB26_1628  #cond_false63.i.i
-       jmp LBB26_1769  #_Z8makeGridP13mrSurfaceListidiidd.exit.i
+       jl LBB26_1769   #_Z8makeGridP13mrSurfaceListidiidd.exit.i
 LBB26_1628:    #cond_false63.i.i
        movl $0, 48964(%esp)

llvm-svn: 31100

llvm-svn: 31099

  jle FOO
  jmp BAR
BAR:

into:

  jle FOO
BAR:

... whoa!

llvm-svn: 31098

llvm-svn: 31097

llvm-svn: 31096