search for: defaultanalysisusage

Displaying 3 results from an estimated 3 matches for "defaultanalysisusage".

2008 May 09
0
[LLVMdev] [PATCH] Split LoopUnroll pass into mechanism and policy
...ass that performs loop unrolling, should depend on the LoopSimplify, LCSSA and LoopInfo passes. Also, the UnrollLoop code always preservers the latter two passes. This means a large part of the getAnalysisUsage for every unroll pass will be copy-pasted. This could be solved by adding an extra "defaultAnalysisUsage()" function to Utils/UnrollLoop.cpp or something similar, but I'm not sure that's very pretty. Any suggestions? Additionally, after loop unrolling, the following check is made: // Update the loop information for this loop. // If we completely unrolled the loop, remove it from the...
2008 May 07
8
[LLVMdev] [PATCH] Split LoopUnroll pass into mechanism and policy
Hello Matthijs, Separating mechanism from policy is a good thing for the LoopUnroll pass. Instead of moving the policy to a subclass though, I think it'd be better to move the mechanism, the unrollLoop function, out to be a standalone utility function, with the LoopInfo object passed in explicitly. FoldBlockIntoPredecessor would also be good to make into a standalone utility function, since
2008 May 09
3
[LLVMdev] [PATCH] Split LoopUnroll pass into mechanism and policy
...depend > on the > LoopSimplify, LCSSA and LoopInfo passes. Also, the UnrollLoop code > always > preservers the latter two passes. This means a large part of the > getAnalysisUsage for every unroll pass will be copy-pasted. This > could be > solved by adding an extra "defaultAnalysisUsage()" function to > Utils/UnrollLoop.cpp or something similar, but I'm not sure that's > very pretty. > Any suggestions? I'm not sure that's very pretty either :-}. My sense is that it's better to just duplicate those few lines. The unrollLoop function verifies isL...