From 884859369f81c91cc45627f3889062092b6e4055 Mon Sep 17 00:00:00 2001
From: Matt Rohrer <matt@prognostikos.com>
Date: Thu, 12 Sep 2024 11:09:09 +0200
Subject: [PATCH] Prevent server crash by restarting child

Before this change, if Errno::EINVAL is thrown when sending IO to the
child the server would crash. Now we catch the exception and, assuming
the child has problems, we start a new child.
---
 CHANGELOG.md                      | 2 ++
 lib/spring/application_manager.rb | 2 +-
 2 files changed, 3 insertions(+), 1 deletion(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 1de64a19..0a193b13 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,7 @@
 ## Next Release
 
+* Fixed a bug that would crash the server if sending IO to the child failed
+
 ## 4.3.0
 
 * Fix reloading issue in Ruby 3.3.
diff --git a/lib/spring/application_manager.rb b/lib/spring/application_manager.rb
index 6ac016d8..ce8c63bc 100644
--- a/lib/spring/application_manager.rb
+++ b/lib/spring/application_manager.rb
@@ -42,7 +42,7 @@ def with_child
         if alive?
           begin
             yield
-          rescue Errno::ECONNRESET, Errno::EPIPE
+          rescue Errno::ECONNRESET, Errno::EPIPE, Errno::EINVAL
             # The child has died but has not been collected by the wait thread yet,
             # so start a new child and try again.
             log "child dead; starting"