Auto merge of rust-lang#129714 - saethlin:half-a-recursion, r=compiler-errors

Use a reduced recursion limit in the MIR inliner's cycle breaker

This probably papers over rust-lang#128887, but primarily I'm opening this PR because multiple compiler people have thought about making this change which probably means it's a good idea.

r? compiler-errors

Author: bors

compiler/rustc_mir_transform/src/inline/cycle.rs

@@ -135,6 +135,14 @@ pub(crate) fn mir_callgraph_reachable<'tcx>(
         }
         false
     }
+    // FIXME(-Znext-solver): Remove this hack when trait solver overflow can return an error.
+    // In code like that pointed out in #128887, the type complexity we ask the solver to deal with
+    // grows as we recurse into the call graph. If we use the same recursion limit here and in the
+    // solver, the solver hits the limit first and emits a fatal error. But if we use a reduced
+    // limit, we will hit the limit first and give up on looking for inlining. And in any case,
+    // the default recursion limits are quite generous for us. If we need to recurse 64 times
+    // into the call graph, we're probably not going to find any useful MIR inlining.
+    let recursion_limit = tcx.recursion_limit() / 2;
     process(
         tcx,
         param_env,
@@ -143,7 +151,7 @@ pub(crate) fn mir_callgraph_reachable<'tcx>(
         &mut Vec::new(),
         &mut FxHashSet::default(),
         &mut FxHashMap::default(),
-        tcx.recursion_limit(),
+        recursion_limit,
     )
 }