From 9e51e0be59a8821061d5469be2d10d48b3b606d7 Mon Sep 17 00:00:00 2001
From: Jean Boussier <byroot@ruby-lang.org>
Date: Tue, 21 Feb 2023 12:38:25 +0100
Subject: [PATCH] Process.warmup: precompute strings coderange

This both save time for when it will be eventually needed,
and avoid mutating heap pages after a potential fork.

Instrumenting some large Rails app, I've witnessed up to
58% of String instances having their coderange still unknown.
---
 process.c                 | 25 ++++++++++++++++++++++++-
 test/ruby/test_process.rb | 12 ++++++++++++
 2 files changed, 36 insertions(+), 1 deletion(-)

diff --git a/process.c b/process.c
index 5761aea718d277..20360aceae4ba5 100644
--- a/process.c
+++ b/process.c
@@ -8535,6 +8535,27 @@ static VALUE rb_mProcUID;
 static VALUE rb_mProcGID;
 static VALUE rb_mProcID_Syscall;
 
+
+static int
+proc_warmup_each_object(void *vstart, void *vend, size_t stride, void *data)
+{
+    VALUE v = (VALUE)vstart;
+    for (; v != (VALUE)vend; v += stride) {
+        switch (BUILTIN_TYPE(v)) {
+          case T_STRING:
+            // precompute the string coderange. This both save time for when it will be
+            // eventually needed, and avoid mutating heap pages after a potential fork.
+            rb_enc_str_coderange(v);
+            break;
+          default:
+            break;
+        }
+    }
+
+    return 0;
+}
+
+
 /*
  *  call-seq:
  *     Process.warmup    -> true
@@ -8563,10 +8584,12 @@ proc_warmup(VALUE _)
     RB_VM_LOCK_ENTER();
     rb_gc_prepare_heap();
     RB_VM_LOCK_LEAVE();
+
+    rb_objspace_each_objects(proc_warmup_each_object, NULL);
+
     return Qtrue;
 }
 
-
 /*
  * Document-module: Process
  *
diff --git a/test/ruby/test_process.rb b/test/ruby/test_process.rb
index 6ca16733bbf014..12946dce669f46 100644
--- a/test/ruby/test_process.rb
+++ b/test/ruby/test_process.rb
@@ -2709,4 +2709,16 @@ def test_warmup_run_major_gc_and_compact
       assert_equal compact_count + 1, GC.stat(:compact_count)
     end;
   end
+
+  def test_warmup_precompute_string_coderange
+    assert_separately([], "#{<<~"begin;"}\n#{<<~'end;'}")
+    require 'objspace'
+    begin;
+      obj = "a" * 12
+      obj.force_encoding(Encoding::BINARY)
+      assert_includes(ObjectSpace.dump(obj), '"coderange":"unknown"')
+      Process.warmup
+      assert_includes(ObjectSpace.dump(obj), '"coderange":"7bit"')
+    end;
+  end
 end