[PATCH] D125495: [Inline][Remark] Annotate inline pass name with link phase information for analysis.
David Li via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Tue May 31 12:54:05 PDT 2022
davidxl added a comment.
I suggest splitting out the LTOPhase passing interface changes into a different patch. That one is more mechanical. After which this patch can be greatly simplified.
================
Comment at: llvm/include/llvm/Analysis/InlineAdvisor.h:43
+enum class InlineAdvisorContext : int {
+ EarlyInliner,
----------------
Is it possible to define sample loader inliner here too ?
================
Comment at: llvm/include/llvm/Analysis/InlineAdvisor.h:44
+enum class InlineAdvisorContext : int {
+ EarlyInliner,
+ CGSCCInliner,
----------------
AlwaysInline should probably be defined here too.
================
Comment at: llvm/include/llvm/Analysis/InlineAdvisor.h:197
+ /// may not be the accurate reason.
+ /// For example, a inline decision in postlink CGSCC pass could be a result
+ /// of applying heuristics, but could be a result of mandatory inline as
----------------
a inline --> an inline
================
Comment at: llvm/include/llvm/Analysis/InlineAdvisor.h:198
+ /// For example, a inline decision in postlink CGSCC pass could be a result
+ /// of applying heuristics, but could be a result of mandatory inline as
+ /// well.
----------------
If we split out AlwaysInline, then it will be less ambiguous.
================
Comment at: llvm/include/llvm/Passes/PassBuilder.h:220
+ buildModuleOptimizationPipeline(OptimizationLevel Level,
+ ThinOrFullLTOPhase LTOPhase);
----------------
This change can be split out into an independent patch.
================
Comment at: llvm/lib/Analysis/InlineAdvisor.cpp:568
+ return "replay-sample-profile-inline";
+ case (InlineAdvisorContext::UnspecifiedInlinerContext):
+ return DEBUG_TYPE;
----------------
In what case it is unspecified?
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D125495/new/
https://reviews.llvm.org/D125495
More information about the llvm-commits
mailing list