[lld] 0d55d0b - [ELF] Parallelize ICF's section sort. NFC (#223216)

[via llvm-commits]

Author: Fangrui Song
Date: 2026-09-13T06:58:12Z
New Revision: 0d55d0bf81ce2e0d6d3c46b7835efb788bde634a

URL: https://github.com/llvm/llvm-project/commit/0d55d0bf81ce2e0d6d3c46b7835efb788bde634a
DIFF: https://github.com/llvm/llvm-project/commit/0d55d0bf81ce2e0d6d3c46b7835efb788bde634a.diff

LOG: [ELF] Parallelize ICF's section sort. NFC (#223216)

All candidate sections are sorted by a serial stable_sort, which is a
scalability bottleneck. Switch to parallel sort using Schwartzian
transform. Linking Chromium with --icf=all (--threads=8) drops from
4.47s to 3.18s.

Added: 
    

Modified: 
    lld/ELF/ICF.cpp

Removed: 
    


################################################################################
diff  --git a/lld/ELF/ICF.cpp b/lld/ELF/ICF.cpp
index b64f49ec8b0bd..62c42bdbb550d 100644
--- a/lld/ELF/ICF.cpp
+++ b/lld/ELF/ICF.cpp
@@ -510,9 +510,16 @@ template <class ELFT> void ICF<ELFT>::run() {
 
   // From now on, sections in Sections vector are ordered so that sections
   // in the same equivalence class are consecutive in the vector.
-  llvm::stable_sort(sections, [](const InputSection *a, const InputSection *b) {
-    return a->eqClass[0] < b->eqClass[0];
+  SmallVector<uint64_t, 0> keys(sections.size());
+  parallelFor(0, sections.size(), [&](size_t i) {
+    keys[i] = uint64_t(sections[i]->eqClass[0]) << 32 | i;
   });
+  parallelSort(keys.begin(), keys.end());
+  SmallVector<InputSection *, 0> sorted;
+  sorted.reserve(keys.size());
+  for (uint64_t k : keys)
+    sorted.push_back(sections[uint32_t(k)]);
+  sections = std::move(sorted);
 
   // Compare static contents and assign unique equivalence class IDs for each
   // static content. Use a base offset for these IDs to ensure no overlap with