Write relocation sections contiguously.

Linkers normally read all the relocations upfront to compute the references
between sections. Putting them together is a bit more cache friendly.

I benchmarked linking a Release+Asserts clang with gold on a vm. I tried all
4 combinations of --gc-sections/no --gc-section hot and cold cache.

I cleared the cache with

echo 3 > /proc/sys/vm/drop_caches

and warmed it up by running the link once before timing the subsequent ones.

With cold cache and --gc-sections the time goes from

1.86130781665 +- 0.01713126697463843 seconds
to
1.82370735105 +- 0.014127522318814516 seconds

With cold cache and no --gc-sections the time goes from

1.6087245435500002 +- 0.012999066825178644 seconds
to
1.5687122041500001 +- 0.013145850126026619 seconds

With hot cache and no --gc-sections the time goes from

0.926200939 ( +-  0.33% ) seconds
to
0.907200079 ( +-  0.31% ) seconds

With hot cache and gc sections the time goes from

1.183038049 ( +-  0.34% ) seconds
to
1.147355862 ( +-  0.39% ) seconds

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@235165 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Rafael Espindola
2015-04-17 08:11:38 +00:00
parent 5023f151a7
commit b4b0b34181
5 changed files with 18 additions and 11 deletions

View File

@ -943,6 +943,8 @@ void ELFObjectWriter::computeIndexMap(MCAssembler &Asm,
SectionIndexMap[&Section] = Index++;
}
std::vector<const MCSectionELF *> RelSections;
for (MCAssembler::iterator it = Asm.begin(),
ie = Asm.end(); it != ie; ++it) {
const MCSectionData &SD = *it;
@ -956,9 +958,14 @@ void ELFObjectWriter::computeIndexMap(MCAssembler &Asm,
if (MCSectionData *RelSD = createRelocationSection(Asm, SD)) {
const MCSectionELF *RelSection =
static_cast<const MCSectionELF *>(&RelSD->getSection());
SectionIndexMap[RelSection] = Index++;
RelSections.push_back(RelSection);
}
}
// Put relocation sections close together. The linker reads them
// first, so this improves cache locality.
for (const MCSectionELF * Sec: RelSections)
SectionIndexMap[Sec] = Index++;
}
void ELFObjectWriter::computeSymbolTable(