makes the code faster.

llvm-svn: 81220

llvm-svn: 81215

instruction to insert before can be end(). getDebugLoc on
end() returns an invalid value, therefore use the debug
loc of the call instruction, and give it to InsertLabel.

llvm-svn: 81207

llvm-svn: 81206

When remat'ing and destination virtual register has a sub-register index. Make sure the sub-register class matches the register class of the remat'ed instruction definition register class.

llvm-svn: 81204

llvm-svn: 81203

llvm-svn: 81202

llvm-svn: 81201

llvm-svn: 81199

in the .td files.  This gets us down to 18 failures in codegen/x86 with the
new asmprinter.

llvm-svn: 81198

subreg32 modifiers.

llvm-svn: 81196

the problem with subreg32 modifiers.  This gets all of Olden working
with the new asmprinter.

llvm-svn: 81195

performance.

llvm-svn: 81193

depth first order, so it wouldn't process unreachable blocks.
When compiling at -O0, late dead block elimination isn't done
and the bad instructions got to isel.

llvm-svn: 81187

llvm-svn: 81186

extractelement operations into a bitcast of the pointer,
then a gep, then a scalar load.  Disable this when the vector
only has one element, because it leads to infinite loops in
instcombine (PR4908).

This transformation seems like a really bad idea to me, as it
will likely disable CSE of vector load/stores etc and can be
better done in the code generator when profitable.  This
goes all the way back to the first days of packed types,
r25299 specifically.

I'll let those people who care about the performance of vector
code decide what to do with this.

llvm-svn: 81185

so "Assert1(isa<>); cast<>" is a valid idiom.

Actually check the PHI node's odd-numbered operands for BasicBlock-ness, like
the comment said.

llvm-svn: 81182

null in the case of an empty struct, so don't try to call getNumValues
on it.

llvm-svn: 81180

Make the verifier more robust by avoiding unprotected cast<> calls. Notably,
Assert1(isa<>); cast<> is not safe as Assert1 does not terminate the program.

llvm-svn: 81179

context for the newly created operations.
 
Patch by Jakub Staszak!

llvm-svn: 81175

This fixes PR4905

llvm-svn: 81174

llvm-svn: 81173

llvm-svn: 81172

from floating-point to integer first, and bitcast the result
back to floating-point. Previously, this test was passing by
falling back to SelectionDAG lowering. The resulting code isn't
as nice, but it's correct and CodeGen now stays on the fast path.

llvm-svn: 81171

compile-time constant integers or that are out of bounds for their
corresponding static array types. These can cause aliasing that
GlobalOpt assumes won't happen.

llvm-svn: 81165

an aggregate store overlapping a different aggregate store, despite
the stores having distinct addresses.

llvm-svn: 81164

is missing the inbounds flag. This is slightly conservative, but it
avoids problems with two constants pointing to the same address but
getting distinct entries in the Memory DenseMap.

llvm-svn: 81163

into a getelementptr ConstantExpr.

llvm-svn: 81162

llvm-svn: 81161

llvm-svn: 81157

llvm-svn: 81156

equivalent to SIG_IGN.

llvm-svn: 81144

 - Hopefully this unbreaks some llvm-gcc bootstraps.

llvm-svn: 81141

llvm-svn: 81138

dependent. Hopefully, this will fix PR3801.

llvm-svn: 81132

when the new method gives the same result as the original
(as far as I can see).  This will hopefully pacify icc.

llvm-svn: 81131

llvm-svn: 81128

llvm-svn: 81127

llvm-svn: 81126

llvm-svn: 81124