Improve decomp throughput.
authorPat Gavlin <pagavlin@microsoft.com>
Thu, 8 Sep 2016 14:27:36 +0000 (07:27 -0700)
committerPat Gavlin <pagavlin@microsoft.com>
Thu, 8 Sep 2016 16:09:36 +0000 (09:09 -0700)
commita00bfab08779c6bb1cf515f54fff069470f1be12
tree71adb776536f132df45743a298953c94227c07c4
parent1188790f1d12dc4caa2c20a7cadb8b2012b65368
Improve decomp throughput.

Don't find uses unless necessary. This saves about 50 milliseconds while
crossgen'ing mscorlib.

Commit migrated from https://github.com/dotnet/coreclr/commit/db2281ccbcc610ca255d174a3d3e5cf287776089
src/coreclr/src/jit/decomposelongs.cpp
src/coreclr/src/jit/decomposelongs.h