Log - 4ee451de366474b9c228b4e5fa573795a715216d - platform/external/llvm

4ee451d Remove attribution from file headers, per discussion on llvmdev. by Chris Lattner · 17 years ago
6894942 The physical register + virtual register joining requirement was much too strict. by Evan Cheng · 17 years ago
22f07ff Switch over to MachineLoopInfo. by Evan Cheng · 17 years ago
dfb1561 Add an option to control this heuristic tweak so I can test it. by Evan Cheng · 17 years ago
5ef3a04 Fix for PR1831: if all defs of an interval are re-materializable, then it's a preferred spill candiate. by Evan Cheng · 17 years ago
c3fc7d9 Replace the odd kill# hack with something less fragile. by Evan Cheng · 17 years ago
81a0382 Live interval splitting: by Evan Cheng · 17 years ago
c498b02 Clean up sub-register implementation by moving subReg information back to by Evan Cheng · 17 years ago
f2fbca6 Refactor some code. by Evan Cheng · 17 years ago
8fc9a10 First step towards moving the coalescer to priority_queue based machinery. by Evan Cheng · 17 years ago
a461c4d Move SimpleRegisterCoalescing.h to lib/CodeGen since there is now a common by Evan Cheng · 17 years ago
5031fd2 Skip over deleted val#'s. by Evan Cheng · 17 years ago
0547bab - Coalesce extract_subreg when both intervals are relatively small. by Evan Cheng · 17 years ago
4ae31a5 Really fix PR1734. Carefully track which register uses are sub-register uses by by Evan Cheng · 17 years ago
95f0ab6 One more extract_subreg coalescing bug fix. by Evan Cheng · 17 years ago
dcd8f78 Fix PR1734. by Evan Cheng · 17 years ago
8b0b874 Code clean up. by Evan Cheng · 17 years ago
f5c7359 Fix PR1729: watch out for val# with no def. by Evan Cheng · 17 years ago
3472925 When coalescing an EXTRACT_SUBREG and the dst register is a physical register, by Evan Cheng · 17 years ago
52c7ff7 Restrict EXTRACT_SUBREG coalescing to avoid negative performance impact. by Evan Cheng · 17 years ago
32dfbea EXTRACT_SUBREG coalescing support. The coalescer now treats EXTRACT_SUBREG like by Evan Cheng · 17 years ago
be8db24 Bad choice of variable name. by Evan Cheng · 17 years ago
5edc57f Fix an extremely stupid bug that prevented first round of coalescing (physical registers only) from happening. by Evan Cheng · 17 years ago
92dfe20 Remove isReg, isImm, and isMBB, and change all their users to use by Dan Gohman · 17 years ago
2c17c4d by David Greene · 17 years ago
f3bb2e6 Use pool allocator for all the VNInfo's to improve memory access locality. This reduces coalescing time on siod Mac OS X PPC by 35%. Also remove the back ptr from VNInfo to LiveInterval and other tweaks. by Evan Cheng · 17 years ago
3430135 More tweaks to improve compile time. by Evan Cheng · 17 years ago
fadfb5b std::map -> DenseMap for slight compile time benefit. by Evan Cheng · 17 years ago
c14b144 Use std::map instead of a (potentially very sparse) array to track val# defined by copy from the other live range. Minor compile time win when number of val# is large. by Evan Cheng · 17 years ago
7ecb38b Change LiveRange so it keeps a pointer to the VNInfo rather than an index. by Evan Cheng · 17 years ago
1a66f0a Recover most of the compile time regression due to recent live interval changes. by Evan Cheng · 17 years ago
bc165e4 Fix some kill info update bugs; add hidden option -disable-rematerialization to turn off remat for debugging. by Evan Cheng · 17 years ago
dc5294f Fix for PR1596: AdjustCopiesBackFrom() should conservatively check if any of its sub-registers may overlap with the interval of the copy that's being coalesced. by Evan Cheng · 17 years ago
430a7b0 Kill info update bugs. by Evan Cheng · 17 years ago
549f27d3 Re-implement trivial rematerialization. This allows def MIs whose live intervals that are coalesced to be rematerialized. by Evan Cheng · 17 years ago
ccb36a4 No need to remove dead range from soon-to-be-dead live interval. Its val# may be out of whack. by Evan Cheng · 17 years ago
4f8ff16 Code to maintain kill information during register coalescing. by Evan Cheng · 17 years ago
b31e91c unbreak the build by Chris Lattner · 17 years ago
a86e347 Bug fix. ~1U marks the val# dead. by Evan Cheng · 17 years ago
8df7860 - Each val# can have multiple kills. by Evan Cheng · 17 years ago
a8d94f1 - LiveInterval value#'s now have 3 components: def instruction #, by Evan Cheng · 17 years ago
e76fad2 shorten this name by Chris Lattner · 17 years ago
3b1f55e simpleregistercoalescing -> regcoalescing. It's too long for me to handle. by Evan Cheng · 17 years ago
c674a92 Don't assume that only Uses can be kills. Defs are marked as kills initially by Dan Gohman · 17 years ago
273288c Add comment. by Evan Cheng · 17 years ago
e510b3a fix typos by Gabor Greif · 17 years ago
2513330 Factor live variable analysis so it does not do register coalescing by David Greene · 17 years ago