- Nov 05, 2011
-
-
-
Eli Friedman authored
Enhanced vzeroupper insertion pass that avoids inserting vzeroupper where it is unnecessary through local analysis. Patch from Bruno Cardoso Lopes, with some additional changes. I'm going to wait for any review comments and perform some additional testing before turning this on by default. llvm-svn: 143750
-
Chad Rosier authored
llvm-svn: 143749
-
Chad Rosier authored
in a 16-bit immediate. However, for the shorter non-legal types (i.e., i1, i8, i16) we should not sign-extend. This prevents us from materializing things such as 'true' (i.e., i1 1). llvm-svn: 143743
-
- Nov 04, 2011
-
-
Chad Rosier authored
llvm-svn: 143739
-
Bruno Cardoso Lopes authored
llvm-svn: 143738
-
Daniel Dunbar authored
for target definitions. llvm-svn: 143731
-
Daniel Dunbar authored
added a layer of indirection with no value (not even conciseness). llvm-svn: 143727
-
Rafael Espindola authored
point back in the file in the included testcase. Fixes PR11300. llvm-svn: 143726
-
Dan Gohman authored
llvm-svn: 143719
-
Eli Friedman authored
llvm-svn: 143712
-
Benjamin Kramer authored
llvm-svn: 143695
-
Craig Topper authored
llvm-svn: 143683
-
Evan Cheng authored
llvm-svn: 143675
-
Chad Rosier authored
llvm-svn: 143670
-
Chad Rosier authored
llvm-svn: 143669
-
Daniel Dunbar authored
the end of it is dead.", which appears to break bootstrapping LLVM. llvm-svn: 143668
-
- Nov 03, 2011
-
-
Dan Gohman authored
across calls, and only check for nested dependences on the special call-sequence-resource register. llvm-svn: 143660
-
Pete Cooper authored
llvm-svn: 143646
-
Dan Bailey authored
llvm-svn: 143640
-
Daniel Dunbar authored
llvm-svn: 143634
-
Pete Cooper authored
Only currently done if the later store is writing to a power of 2 address or has the same alignment as the earlier store as then its likely to not break up large stores into smaller ones Fixes <rdar://problem/10140300> llvm-svn: 143630
-
Craig Topper authored
llvm-svn: 143612
-
Chad Rosier authored
llvm-svn: 143603
-
Pete Cooper authored
Treat objc selector reference globals as invariant so that MachineLICM can hoist them out of loops. Fixes <rdar://problem/6027699> llvm-svn: 143600
-
Chad Rosier authored
compatibility in the BitcodeReader. llvm-svn: 143598
-
Lang Hames authored
llvm-svn: 143594
-
Chad Rosier authored
llvm-svn: 143591
-
Bill Wendling authored
into the function. Reflect that here so that the array will be placed next to the SP. <rdar://problem/10128329> llvm-svn: 143590
-
- Nov 02, 2011
-
-
Lang Hames authored
llvm-svn: 143582
-
Nick Lewycky authored
implied by directory index zero. llvm-svn: 143570
-
Michael J. Spencer authored
llvm-svn: 143562
-
Michael J. Spencer authored
llvm-svn: 143561
-
Chad Rosier authored
llvm-svn: 143559
-
Owen Anderson authored
Fix the issue that r143552 was trying to address the _right_ way. One-register lists are legal on LDM/STM instructions, but we should not print the PUSH/POP aliases when they appear. This fixes round tripping on this instruction. llvm-svn: 143557
-
Kevin Enderby authored
Thanks to Nick for spotting this! llvm-svn: 143556
-
Owen Anderson authored
The rules disallowing single-register reglist operands only apply to the POP alias, not to LDM/STM instructions. Revert r143552. llvm-svn: 143553
-
Owen Anderson authored
Register list operands are not allowed to contain only a single register. Alternate encodings are used in that case. llvm-svn: 143552
-
Benjamin Kramer authored
llvm-svn: 143549
-
Chad Rosier authored
llvm-svn: 143547
-