Commit Graph

177 Commits

Author SHA1 Message Date
Brian Gaeke a1640d2427 I think this will handle double args.
llvm-svn: 16618
2004-09-30 19:44:32 +00:00
Brian Gaeke 7efd5fcb3a Mark the instructions that have delay slots with the hasDelaySlot flag.
Add some comments.

llvm-svn: 16611
2004-09-30 04:04:48 +00:00
Brian Gaeke 1f48ef2a9a Use TargetMachine::hasDelaySlot() instead of our old switch statement
to find instrs that have delay slots.

llvm-svn: 16610
2004-09-30 04:04:47 +00:00
Brian Gaeke fb1fd0cb5f Update list of shootout programs that should be working.
llvm-svn: 16595
2004-09-29 20:45:06 +00:00
Brian Gaeke ede068a8d3 Tell the target description that calls clobber registers O0...O5.
llvm-svn: 16594
2004-09-29 20:45:05 +00:00
Brian Gaeke 3a340dd7cd FITOD is spelled "fitod", not "fitos". Ouch.
llvm-svn: 16591
2004-09-29 19:59:07 +00:00
Brian Gaeke 47095da661 Don't use .quad to output double constants. The assembler must have a bug or
something, because the wrong bit patterns get output.

llvm-svn: 16590
2004-09-29 19:59:06 +00:00
Brian Gaeke 81f5e08573 Recognize FpMOVD as a move.
llvm-svn: 16586
2004-09-29 16:45:47 +00:00
Brian Gaeke c44da15813 add results
llvm-svn: 16579
2004-09-29 03:48:55 +00:00
Brian Gaeke 4023d4e191 Simplify copyConstantToRegister() for longs, using a pair of recursive calls.
Copy constant-pool entries' addresses into registers before loading out of them,
to avoid errors from the assembler.
Handle loading call args past the 6th one off the stack.
Add IMPLICIT_DEF pseudo-instrs for double and long arguments passed in register
pairs.
Use FpMOVD to copy doubles around instead of the horrible store-load thing we
were doing before.
Handle 'ret double' and 'ret long'.
Fix a bug in handling 'and/or/xor long'.

llvm-svn: 16577
2004-09-29 03:34:41 +00:00
Brian Gaeke eae5b17dba Fix bug recognizing moves: isMoveInstr should only treat ORs with %g0 as
moves, not all ORs.

llvm-svn: 16576
2004-09-29 03:28:15 +00:00
Brian Gaeke 069bb989e4 Use FpMOVD pseudo-instruction to move doubles around.
llvm-svn: 16575
2004-09-29 03:27:30 +00:00
Brian Gaeke da371d9cf3 Add new FpMOVD pseudo-instruction, used to move doubles around.
llvm-svn: 16574
2004-09-29 03:27:29 +00:00
Brian Gaeke 7f11dd4825 Fix double and long alignment.
Call the FPMover pass after register allocation.

llvm-svn: 16573
2004-09-29 03:26:27 +00:00
Brian Gaeke ead8e39832 Put quotes around argument to .section directive.
llvm-svn: 16572
2004-09-29 03:25:40 +00:00
Brian Gaeke 457180c597 Add createSparcV8FPMoverPass().
llvm-svn: 16571
2004-09-29 03:25:39 +00:00
Brian Gaeke ce0ba4857b Pass which converts FpMOVD (double move pseudoinstructions) to pairs
of FMOVS instrs.

llvm-svn: 16570
2004-09-29 03:24:34 +00:00
Misha Brukman d409d38151 SparcV8 int regs are not only 32-bits in width, but they are 32-bit aligned!
llvm-svn: 16526
2004-09-27 18:22:18 +00:00
Misha Brukman a68d76ebfa Fix the copy-pasto that Brian noticed: V8 int regs are 32-bits wide, not 64.
llvm-svn: 16518
2004-09-26 21:07:43 +00:00
Misha Brukman 379a05b52a Use the V8/V9 shared register file description
llvm-svn: 16485
2004-09-22 21:48:50 +00:00
Misha Brukman 3c08658d56 Combine the F2 and F3 instruction classes into one file for simplicity
llvm-svn: 16484
2004-09-22 21:38:42 +00:00
Misha Brukman efce1ef9c3 Fix file header path
llvm-svn: 16483
2004-09-22 21:29:12 +00:00
Misha Brukman 009aaf9968 Prettify formatting of the file, adjust paths to making V8 a subdir of Sparc
llvm-svn: 16482
2004-09-22 20:09:29 +00:00
Misha Brukman d55f854a5d V8 is now a subdirectory of Sparc; adjust paths accordingly
llvm-svn: 16481
2004-09-22 20:08:52 +00:00
Chris Lattner 247c29e4b0 Changes to make this work with Jason's patch. I checked this by hand, but
would appreciate if others would also look at this to make sure I didn't
botch something obvious

llvm-svn: 16312
2004-09-13 21:32:03 +00:00
Misha Brukman 04bbe6fa25 Renamed file to SparcV8ISelSimple.cpp
llvm-svn: 16267
2004-09-10 18:51:12 +00:00
Brian Gaeke acafb39cc3 This file does not need <iostream>, I think.
llvm-svn: 16245
2004-09-08 04:10:52 +00:00
Brian Gaeke c6e81f0fde Back to compiling land for v8
llvm-svn: 16138
2004-09-02 02:37:43 +00:00
Chris Lattner b7ddc73b45 Convert bytes to bits in alignment
llvm-svn: 15971
2004-08-21 20:09:46 +00:00
Chris Lattner 26971f4bc2 V8 never used the instrselectorgenerator
llvm-svn: 15791
2004-08-15 23:07:40 +00:00
Chris Lattner a0b38d3cb1 Code insertion methods now return void instead of an int.
llvm-svn: 15780
2004-08-15 22:15:11 +00:00
Chris Lattner 98de1d7795 These methods no longer take a TargetRegisterClass* operand.
llvm-svn: 15774
2004-08-15 21:56:44 +00:00
Chris Lattner b08d2a2a11 Update to no longer take MF as an argument
llvm-svn: 15748
2004-08-14 22:57:22 +00:00
Misha Brukman 8f951e579b Remove ClassPrefix variable as it's no longer used.
llvm-svn: 15586
2004-08-09 19:13:29 +00:00
Misha Brukman 0eb9cee698 The (future) SparcV8 JIT would do well to have a class prefix.
llvm-svn: 15583
2004-08-09 18:13:09 +00:00
Brian Gaeke a1a0643e38 Update the To-Do list according to my notes + assertions
llvm-svn: 15535
2004-08-06 04:38:13 +00:00
Chris Lattner 50c6df4ba0 getValues is gone
llvm-svn: 15494
2004-08-04 17:27:27 +00:00
Chris Lattner 4504b95801 I'm pretty sure that ba is branch always, which is a barrier. Brg should
check this :)

llvm-svn: 15357
2004-07-31 02:24:37 +00:00
Chris Lattner 51ebdec5c0 I think that V8 should coallesce registers, don't you?
llvm-svn: 15192
2004-07-25 06:19:04 +00:00
Chris Lattner 9ce34cb323 CPR fixes
llvm-svn: 14960
2004-07-18 07:26:17 +00:00
Brian Gaeke 135b870f72 Add a class for pseudo-instructions. Use it.
Add IMPLICIT_USE and IMPLICIT_DEF, a la X86.

llvm-svn: 14884
2004-07-16 10:32:10 +00:00
Brian Gaeke a57943154e Add what will eventually be the TSFlags. Big switch(opcode) statements are bad.
llvm-svn: 14883
2004-07-16 10:31:59 +00:00
Brian Gaeke f486726c56 Add special handling for pseudo-instructions (print them as comments).
llvm-svn: 14882
2004-07-16 10:31:47 +00:00
Brian Gaeke 97b71c9fa9 Add to-do list.
llvm-svn: 14881
2004-07-16 10:31:36 +00:00
Brian Gaeke 5ce1408537 Do IMPLICIT_DEFs on incoming args' hard regs, to avoid confusing the regalloc.
Support single-fp incoming args.
Support single-fp outgoing args ('call' operands).
Support double-fp return values.

llvm-svn: 14880
2004-07-16 10:31:25 +00:00
Chris Lattner c2726e63ed Delete the allocate*TargetMachine function, which is now dead .
The shared command line options are now in a header that makes sense.

llvm-svn: 14756
2004-07-11 04:17:10 +00:00
Chris Lattner 73eb9436ee Make these format a bit nicer
llvm-svn: 14747
2004-07-11 03:27:42 +00:00
Chris Lattner 4266f8017d Auto-registrate target
llvm-svn: 14745
2004-07-11 02:48:49 +00:00
Brian Gaeke 8165863a9d Support setcc on fp values.
llvm-svn: 14687
2004-07-08 09:08:35 +00:00
Brian Gaeke 7b4722e62c Add floating-point branches and compares. Compares don't complete
until the next cycle, and there's no interlock, so they effectively
have a delay slot.

llvm-svn: 14686
2004-07-08 09:08:22 +00:00