[llvm] [LTO] Sort DefinedGlobals in LTO cache key (PR #210025)

[Nikita Popov via llvm-commits]

https://github.com/nikic updated https://github.com/llvm/llvm-project/pull/210025

>From f910fe8307c52177d78d98cbad27675215fb26d0 Mon Sep 17 00:00:00 2001
From: Nikita Popov <npopov at redhat.com>
Date: Thu, 16 Jul 2026 12:29:43 +0200
Subject: [PATCH 1/2] [LTO] Sort DefinedGlobals in LTO cache key

Sort the DefindeGlobals by GUID when computing the LTO cache key,
to avoid making the cache key dependent on the insertion order.
This fixes large compile-time regressions in rust incremental builds.

Alternatively one could make collectDefinedGVSummariesPerModule()
work on the sortedRange(), but as that is also used in other places,
it probably makes sense to sort locally.

The issue was introduced in 760bb06cb3cd98832f1e4e7a4eaebd98b66d2bdd.
---
 llvm/lib/LTO/LTO.cpp | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/llvm/lib/LTO/LTO.cpp b/llvm/lib/LTO/LTO.cpp
index b0fd1d77a2d66..4863ab2973948 100644
--- a/llvm/lib/LTO/LTO.cpp
+++ b/llvm/lib/LTO/LTO.cpp
@@ -306,7 +306,10 @@ std::string llvm::computeLTOCacheKey(
 
   // Include the hash for the linkage type to reflect internalization and weak
   // resolution, and collect any used type identifier resolutions.
-  for (auto &GS : DefinedGlobals) {
+  SmallVector<std::pair<GlobalValue::GUID, GlobalValueSummary *>>
+      SortedDefinedGlobals(DefinedGlobals.begin(), DefinedGlobals.end());
+  llvm::sort(SortedDefinedGlobals, llvm::less_first());
+  for (auto &GS : SortedDefinedGlobals) {
     GlobalValue::LinkageTypes Linkage = GS.second->linkage();
     Hasher.update(
         ArrayRef<uint8_t>((const uint8_t *)&Linkage, sizeof(Linkage)));

>From ad4dd2c4e18555fcdb373fb821312052b01e6305 Mon Sep 17 00:00:00 2001
From: Nikita Popov <npopov at redhat.com>
Date: Thu, 16 Jul 2026 12:40:36 +0200
Subject: [PATCH 2/2] Add test

---
 .../X86/Inputs/cache-defined-globals-order.ll |  6 +++++
 .../X86/cache-defined-globals-order.ll        | 27 +++++++++++++++++++
 2 files changed, 33 insertions(+)
 create mode 100644 llvm/test/ThinLTO/X86/Inputs/cache-defined-globals-order.ll
 create mode 100644 llvm/test/ThinLTO/X86/cache-defined-globals-order.ll

diff --git a/llvm/test/ThinLTO/X86/Inputs/cache-defined-globals-order.ll b/llvm/test/ThinLTO/X86/Inputs/cache-defined-globals-order.ll
new file mode 100644
index 0000000000000..5a527c9bae924
--- /dev/null
+++ b/llvm/test/ThinLTO/X86/Inputs/cache-defined-globals-order.ll
@@ -0,0 +1,6 @@
+target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+define linkonce_odr void @shared29() {
+  ret void
+}
diff --git a/llvm/test/ThinLTO/X86/cache-defined-globals-order.ll b/llvm/test/ThinLTO/X86/cache-defined-globals-order.ll
new file mode 100644
index 0000000000000..84f71f20f9e6d
--- /dev/null
+++ b/llvm/test/ThinLTO/X86/cache-defined-globals-order.ll
@@ -0,0 +1,27 @@
+; RUN: rm -rf %t && mkdir -p %t
+; RUN: opt -module-hash -module-summary %s -o %t/t.bc
+; RUN: opt -module-hash -module-summary %S/Inputs/cache-defined-globals-order.ll -o %t/a.bc
+
+; Tests that the LTO cache key is insensitive to the order of the modules.
+; The linkonce_odr function @shared29 is defined in both t.bc and a.bc, so
+; it gets a different position in the combined index depending on which
+; module is processed first, which results in a different insertion order
+; for DefinedGlobals.
+
+; RUN: llvm-lto2 run -cache-dir %t/cache -o %t.o %t/t.bc %t/a.bc -r=%t/t.bc,main,plx -r=%t/t.bc,shared29,plx -r=%t/a.bc,shared29,lx
+; RUN: ls %t/cache | count 2
+
+; RUN: llvm-lto2 run -cache-dir %t/cache -o %t.o %t/a.bc %t/t.bc -r=%t/t.bc,main,plx -r=%t/t.bc,shared29,plx -r=%t/a.bc,shared29,lx
+; RUN: ls %t/cache | count 2
+
+target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+define void @main() {
+  call void @shared29()
+  ret void
+}
+
+define linkonce_odr void @shared29() {
+  ret void
+}