forked from OSchip/llvm-project
92dadc0bca
Summary: Currently bitcasting constants from f64 to v2i32 is done by storing the value to the stack and then loading it again. This is not necessary, but seems to happen because v2i32 is a valid type for Sparc V8. If it had not been legal, we would have gotten help from the type legalizer. This patch tries to do the same work as the legalizer would have done by bitcasting the floating point constant and splitting the value up into a vector of two i32 values. Reviewers: venkatra, jyknight Reviewed By: jyknight Subscribers: glaubitz, fedor.sergeev, jrtc27, llvm-commits Differential Revision: https://reviews.llvm.org/D49219 llvm-svn: 340723 |
||
---|---|---|
.. | ||
AsmParser | ||
Disassembler | ||
InstPrinter | ||
MCTargetDesc | ||
TargetInfo | ||
CMakeLists.txt | ||
DelaySlotFiller.cpp | ||
LLVMBuild.txt | ||
LeonFeatures.td | ||
LeonPasses.cpp | ||
LeonPasses.h | ||
README.txt | ||
Sparc.h | ||
Sparc.td | ||
SparcAsmPrinter.cpp | ||
SparcCallingConv.td | ||
SparcFrameLowering.cpp | ||
SparcFrameLowering.h | ||
SparcISelDAGToDAG.cpp | ||
SparcISelLowering.cpp | ||
SparcISelLowering.h | ||
SparcInstr64Bit.td | ||
SparcInstrAliases.td | ||
SparcInstrFormats.td | ||
SparcInstrInfo.cpp | ||
SparcInstrInfo.h | ||
SparcInstrInfo.td | ||
SparcInstrVIS.td | ||
SparcMCInstLower.cpp | ||
SparcMachineFunctionInfo.cpp | ||
SparcMachineFunctionInfo.h | ||
SparcRegisterInfo.cpp | ||
SparcRegisterInfo.h | ||
SparcRegisterInfo.td | ||
SparcSchedule.td | ||
SparcSubtarget.cpp | ||
SparcSubtarget.h | ||
SparcTargetMachine.cpp | ||
SparcTargetMachine.h | ||
SparcTargetObjectFile.cpp | ||
SparcTargetObjectFile.h | ||
SparcTargetStreamer.h |
README.txt
To-do ----- * Keep the address of the constant pool in a register instead of forming its address all of the time. * We can fold small constant offsets into the %hi/%lo references to constant pool addresses as well. * When in V9 mode, register allocate %icc[0-3]. * Add support for isel'ing UMUL_LOHI instead of marking it as Expand. * Emit the 'Branch on Integer Register with Prediction' instructions. It's not clear how to write a pattern for this though: float %t1(int %a, int* %p) { %C = seteq int %a, 0 br bool %C, label %T, label %F T: store int 123, int* %p br label %F F: ret float undef } codegens to this: t1: save -96, %o6, %o6 1) subcc %i0, 0, %l0 1) bne .LBBt1_2 ! F nop .LBBt1_1: ! T or %g0, 123, %l0 st %l0, [%i1] .LBBt1_2: ! F restore %g0, %g0, %g0 retl nop 1) should be replaced with a brz in V9 mode. * Same as above, but emit conditional move on register zero (p192) in V9 mode. Testcase: int %t1(int %a, int %b) { %C = seteq int %a, 0 %D = select bool %C, int %a, int %b ret int %D } * Emit MULX/[SU]DIVX instructions in V9 mode instead of fiddling with the Y register, if they are faster. * Codegen bswap(load)/store(bswap) -> load/store ASI * Implement frame pointer elimination, e.g. eliminate save/restore for leaf fns. * Fill delay slots * Use %g0 directly to materialize 0. No instruction is required.