1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-23 11:13:28 +01:00
llvm-mirror/lib/Target/Sparc
Alex Bradbury da4afcc85c [Sparc][NFC] Clean up SelectCC lowering
The ARM, BPF, MSP430, Sparc and Mips backends all use a similar code sequence 
for lowering SelectCC. As pointed out by @reames in D29937, this code isn't 
particularly clear and in most of these backends doesn't actually match the 
comments. This patch makes the code sequence clearer for the Sparc backend 
through better variable naming and more accurate comments (e.g. we are 
inserting triangle control flow, _not_ diamond). There is no functional 
change.

Differential Revision: https://reviews.llvm.org/D37194

llvm-svn: 312713
2017-09-07 11:30:55 +00:00
..
AsmParser Sort the remaining #include lines in include/... and lib/.... 2017-06-06 11:49:48 +00:00
Disassembler Sort the remaining #include lines in include/... and lib/.... 2017-06-06 11:49:48 +00:00
InstPrinter
MCTargetDesc Delete Default and JITDefault code models 2017-08-03 02:16:21 +00:00
TargetInfo
CMakeLists.txt
DelaySlotFiller.cpp [LegacyPassManager] Remove TargetMachine constructors 2017-05-18 17:21:13 +00:00
LeonFeatures.td [SPARC] Clean up the support for disabling fsmuld and fmuls instructions. 2017-07-20 20:09:11 +00:00
LeonPasses.cpp [SPARC] Clean up the support for disabling fsmuld and fmuls instructions. 2017-07-20 20:09:11 +00:00
LeonPasses.h [SPARC] Clean up the support for disabling fsmuld and fmuls instructions. 2017-07-20 20:09:11 +00:00
LLVMBuild.txt
README.txt
Sparc.h [LegacyPassManager] Remove TargetMachine constructors 2017-05-18 17:21:13 +00:00
Sparc.td [SPARC] Clean up the support for disabling fsmuld and fmuls instructions. 2017-07-20 20:09:11 +00:00
SparcAsmPrinter.cpp Sort the remaining #include lines in include/... and lib/.... 2017-06-06 11:49:48 +00:00
SparcCallingConv.td
SparcFrameLowering.cpp
SparcFrameLowering.h
SparcInstr64Bit.td
SparcInstrAliases.td
SparcInstrFormats.td
SparcInstrInfo.cpp
SparcInstrInfo.h
SparcInstrInfo.td [SPARC] Clean up the support for disabling fsmuld and fmuls instructions. 2017-07-20 20:09:11 +00:00
SparcInstrVIS.td
SparcISelDAGToDAG.cpp
SparcISelLowering.cpp [Sparc][NFC] Clean up SelectCC lowering 2017-09-07 11:30:55 +00:00
SparcISelLowering.h [SelectionDAG] Use KnownBits struct in DAG's computeKnownBits and simplifyDemandedBits 2017-04-28 05:31:46 +00:00
SparcMachineFunctionInfo.cpp
SparcMachineFunctionInfo.h
SparcMCInstLower.cpp Sort the remaining #include lines in include/... and lib/.... 2017-06-06 11:49:48 +00:00
SparcRegisterInfo.cpp
SparcRegisterInfo.h
SparcRegisterInfo.td [SPARC] Support 'f' and 'e' inline asm constraints. 2017-05-12 15:59:10 +00:00
SparcSchedule.td [Sparc] Remove execute permissions from non-executable text files 2017-05-17 11:05:20 +00:00
SparcSubtarget.cpp [SPARC] Clean up the support for disabling fsmuld and fmuls instructions. 2017-07-20 20:09:11 +00:00
SparcSubtarget.h [SPARC] Clean up the support for disabling fsmuld and fmuls instructions. 2017-07-20 20:09:11 +00:00
SparcTargetMachine.cpp Delete Default and JITDefault code models 2017-08-03 02:16:21 +00:00
SparcTargetMachine.h Delete Default and JITDefault code models 2017-08-03 02:16:21 +00:00
SparcTargetObjectFile.cpp [Solaris] emit .init_array instead of .ctors on Solaris (Sparc/x86) 2017-06-21 20:36:32 +00:00
SparcTargetObjectFile.h [Solaris] emit .init_array instead of .ctors on Solaris (Sparc/x86) 2017-06-21 20:36:32 +00:00
SparcTargetStreamer.h

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.