- Sep 25, 2008
-
-
Evan Cheng authored
With sse3 and when the source is a load or has multiple uses, favors movddup over shuffp*, pshufd, etc. Without sse3 or when the source is from a register, make use of movlhps llvm-svn: 56620
-
Dale Johannesen authored
meaning sse_regparm (i.e. float/double values go in XMM0 instead of ST0). Update documentation to reflect reality. llvm-svn: 56619
-
Dan Gohman authored
llvm-svn: 56608
-
Evan Cheng authored
llvm-svn: 56607
-
Evan Cheng authored
llvm-svn: 56600
-
Evan Cheng authored
Fix patterns for SSE4.1 move and sign extend instructions. Also add instructions which fold VZEXT_MOVL and VZEXT_LOAD. llvm-svn: 56594
-
Evan Cheng authored
llvm-svn: 56593
-
Anton Korobeynikov authored
llvm-svn: 56588
-
Anton Korobeynikov authored
llvm-svn: 56585
-
Anton Korobeynikov authored
llvm-svn: 56582
-
Anton Korobeynikov authored
llvm-svn: 56580
-
Anton Korobeynikov authored
llvm-svn: 56575
-
Anton Korobeynikov authored
llvm-svn: 56574
-
Anton Korobeynikov authored
llvm-svn: 56573
-
Anton Korobeynikov authored
llvm-svn: 56572
-
Anton Korobeynikov authored
llvm-svn: 56568
-
- Sep 24, 2008
-
-
Devang Patel authored
llvm-svn: 56527
-
Evan Cheng authored
Properly handle 'm' inline asm constraints. If a GV is being selected for the addressing mode, it requires the same logic for PIC relative addressing, etc. llvm-svn: 56526
-
Devang Patel authored
Do not check isDeclaration() in hasNote(). It is clients' responsibility. llvm-svn: 56524
-
Devang Patel authored
Function Notes also. Function notes are stored at index ~0. llvm-svn: 56511
-
- Sep 23, 2008
-
-
Dan Gohman authored
object. This will be needed to support debug info. llvm-svn: 56508
-
Dan Gohman authored
instruction opcodes are now numbered. No functionality change. llvm-svn: 56497
-
Dan Gohman authored
llvm-svn: 56495
-
Dan Gohman authored
X86ISelDAGToDAG.cpp and into X86InstrInfo.cpp. This will allow it to be reused by FastISel. llvm-svn: 56494
-
Evan Cheng authored
llvm-svn: 56483
-
Devang Patel authored
llvm-svn: 56477
-
Evan Cheng authored
llvm-svn: 56475
-
- Sep 22, 2008
-
-
Dale Johannesen authored
default. llvm-svn: 56471
-
Evan Cheng authored
llvm-svn: 56469
-
Oscar Fuentes authored
llvm-svn: 56453
-
Arnold Schwaighofer authored
Change the calling convention used when tail call optimization is enabled from CC_X86_32_TailCall to CC_X86_32_FastCC. llvm-svn: 56436
-
Oscar Fuentes authored
llvm-svn: 56419
-
- Sep 21, 2008
-
-
Chris Lattner authored
sarl $3, %ecx instead of: movl $3, %ecx sarl %cl, %edx This shrinks fast isel 176.gcc by about 2000 instructions (.3%) llvm-svn: 56413
-
- Sep 20, 2008
-
-
Chris Lattner authored
llvm-svn: 56391
-
Evan Cheng authored
llvm-svn: 56378
-
Evan Cheng authored
llvm-svn: 56377
-
Dan Gohman authored
llvm-svn: 56376
-
Dan Gohman authored
results in better code for globals. Also, unbreak the local CSE for GlobalValue stub loads. llvm-svn: 56371
-
- Sep 19, 2008
-
-
Dan Gohman authored
catches a fair number of common cases. Note that this currently causes Fast-ISel to leave behind lots of dead instructions. Those will be dealt with in subsequent commits. llvm-svn: 56320
-
- Sep 18, 2008
-
-
Dan Gohman authored
llvm-svn: 56311
-