From: Aurelien DARRAGON <adarragon@haproxy.com>
Date: Tue, 23 Jul 2024 17:18:32 +0000 (+0200)
Subject: MINOR: sink: distinguish between hard and soft close in _sink_forward_io_handler()
X-Git-Tag: v3.1-dev4~6
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=c6ab0e14e277063925ffa352b25bb096043262f8;p=thirdparty%2Fhaproxy.git

MINOR: sink: distinguish between hard and soft close in _sink_forward_io_handler()

Aborting the socket on soft-stop is not the same as aborting it due to
unexpected error. As such, let's leverage the granularity offered by
sedesc flags to better reflect the situation: abort during soft-stop is
handled as a soft close thanks to EOI+EOS flags, while abort due to
unexpected error is handled as hard error thanks to ERROR+EOS flags.

Thanks to this change, hard error will always emit RST packet even if
the NOLINGER option wasn't set on the socket.
---

diff --git a/src/sink.c b/src/sink.c
index 18c519a7d3..d935f49a82 100644
--- a/src/sink.c
+++ b/src/sink.c
@@ -361,7 +361,7 @@ static void _sink_forward_io_handler(struct appctx *appctx,
 
 	/* if stopping was requested, close immediately */
 	if (unlikely(stopping))
-		goto close;
+		goto soft_close;
 
 	/* if the connection is not established, inform the stream that we want
 	 * to be notified whenever the connection completes.
@@ -377,7 +377,7 @@ static void _sink_forward_io_handler(struct appctx *appctx,
 	if (appctx != sft->appctx) {
 		/* FIXME: is this even supposed to happen? */
 		HA_SPIN_UNLOCK(SFT_LOCK, &sft->lock);
-		goto close;
+		goto hard_close;
 	}
 
 	MT_LIST_DELETE(&appctx->wait_entry);
@@ -404,8 +404,11 @@ out:
 	co_skip(sc_oc(sc), sc_oc(sc)->output);
 	return;
 
-close:
+soft_close:
 	se_fl_set(appctx->sedesc, SE_FL_EOS|SE_FL_EOI);
+	return;
+hard_close:
+	se_fl_set(appctx->sedesc, SE_FL_EOS|SE_FL_ERROR);
 }
 
 /*