[lld] r288757 - Inline MergeInputSection::getData().

[Rafael Avila de Espindola via llvm-commits]

Rui Ueyama via llvm-commits <llvm-commits at lists.llvm.org> writes:

> Author: ruiu
> Date: Mon Dec  5 20:19:30 2016
> New Revision: 288757
>
> URL: http://llvm.org/viewvc/llvm-project?rev=288757&view=rev
> Log:
> Inline MergeInputSection::getData().
>
> This change seems to make LLD 0.6% faster when linking Clang with
> debug info. I don't want us to have lots of local optimizations,
> but this function is very hot, and the improvement is small but
> not negligible, so I think it's worth doing.

Not sure I can reproduce the speedup:

firefox-O0
  master 4.274663107
  patch  4.272991445 1.00039121585x faster
firefox
  master 7.003041086
  patch  6.994833418 1.00117339006x faster
firefox-gc
  master 7.175516297
  patch  7.169051579 1.00090175359x faster
chromium
  master 4.060833657
  patch  4.065375513 1.00111845409x slower
chromium fast
  master 2.028954515
  patch  2.031154844 1.00108446443x slower
the gold plugin
  master 0.343704186
  patch  0.344360484 1.00190948504x slower
clang
  master 0.590592532
  patch  0.592174102 1.00267793769x slower
llvm-as
  master 0.033631152
  patch  0.033628637 1.00007478745x faster
the gold plugin fsds
  master 0.369853882
  patch  0.370788961 1.00252823898x slower
clang fsds
  master 0.669831177
  patch  0.671908308 1.00310097689x slower
llvm-as fsds
  master 0.033604414
  patch  0.033585695 1.00055735038x faster
scylla
  master 3.198505154
  patch  3.184586272 1.00437070339x faster

Cheers,
Rafael