[llvm] [AMDGPU] Graph-based Module Splitting Rewrite (PR #104763)

[Matt Arsenault via llvm-commits]

================
@@ -278,351 +253,1042 @@ static bool canBeIndirectlyCalled(const Function &F) {
                            /*IgnoreCastedDirectCall=*/true);
 }
 
-/// When a function or any of its callees performs an indirect call, this
-/// takes over \ref addAllDependencies and adds all potentially callable
-/// functions to \p Fns so they can be counted as dependencies of the function.
+//===----------------------------------------------------------------------===//
+// Graph-based Module Representation
+//===----------------------------------------------------------------------===//
+
+/// AMDGPUSplitModule's view of the source Module, as a graph of all components
+/// that can be split into different modules.
 ///
-/// This is needed due to how AMDGPUResourceUsageAnalysis operates: in the
-/// presence of an indirect call, the function's resource usage is the same as
-/// the most expensive function in the module.
-/// \param M    The module.
-/// \param Fns[out] Resulting list of functions.
-static void addAllIndirectCallDependencies(const Module &M,
-                                           DenseSet<const Function *> &Fns) {
-  for (const auto &Fn : M) {
-    if (canBeIndirectlyCalled(Fn))
-      Fns.insert(&Fn);
+/// The most trivial instance of this graph is just the CallGraph of the module,
+/// but it is not guaranteed that the graph is strictly equal to the CFG. It
----------------
arsenm wrote:

Call graph is not "CFG"

https://github.com/llvm/llvm-project/pull/104763