[clangd] Don't clone SymbolSlab::Builder arenas when finalizing.

SymbolSlab::Builder has an arena to store strings of owned symbols, and
deduplicates them. build() copies all the strings and deduplicates them again!
This is potentially useful: we may have overwritten a symbol and
rendered some strings unreachable.

However in practice this is not the case. When testing on a variety of
files in LLVM (e.g. SemaExpr.cpp), the strings for the full preamble
index are 3MB and shrink by 0.4% (12KB). For comparison the serializde
preamble is >50MB.
There are also hundreds of smaller slabs (file sharding) that do not shrink at
all.

CPU time spent on this is significant (something like 3-5% of buildPreamble).
We're better off not bothering.

Differential Revision: https://reviews.llvm.org/D135231
This commit is contained in:
Sam McCall 2022-10-05 02:22:38 +02:00
parent 34004d2d03
commit 6ea83fc98f
1 changed files with 3 additions and 6 deletions

View File

@ -61,12 +61,9 @@ SymbolSlab SymbolSlab::Builder::build() && {
SortedSymbols.push_back(std::move(Entry.second));
llvm::sort(SortedSymbols,
[](const Symbol &L, const Symbol &R) { return L.ID < R.ID; });
// We may have unused strings from overwritten symbols. Build a new arena.
llvm::BumpPtrAllocator NewArena;
llvm::UniqueStringSaver Strings(NewArena);
for (auto &S : SortedSymbols)
own(S, Strings);
return SymbolSlab(std::move(NewArena), std::move(SortedSymbols));
// We may have unused strings from overwritten symbols.
// In practice, these are extremely small, it's not worth compacting.
return SymbolSlab(std::move(Arena), std::move(SortedSymbols));
}
llvm::raw_ostream &operator<<(llvm::raw_ostream &OS, const SymbolSlab &Slab) {