From: Frank Sehr <fsehr@whamcloud.com>
Date: Tue, 10 Sep 2024 23:30:21 +0000 (-0700)
Subject: LU-18208 lnet: Server VM crashed: unable to handle
X-Git-Tag: 2.15.91~3
X-Git-Url: https://git.whamcloud.com/gitweb?a=commitdiff_plain;h=dca487cd2dd6c0a805928bd22f49ceb7328df814;p=fs%2Flustre-release.git

LU-18208 lnet: Server VM crashed: unable to handle

Revert "LU-18160 lnet: ensure lnetctl ping completes in a finite time"
t seems like the patch for LU-18160 introduced crashes. Maybe the
change from wait_for_complete from timeout to interupt. Reverting that
patch solved the problem.

This reverts commit 1666840bb06bbeeb35b2f9a51f9235c36886a3c6.

Test-Parameters: trivial testlist=sanity
Signed-off-by: Frank Sehr <fsehr@whamcloud.com>
Change-Id: Ie48185eb973eee65df2810d7acf940cf6981b83e
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56322
Reviewed-by: Timothy Day <timday@amazon.com>
Reviewed-by: Serguei Smirnov <ssmirnov@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
---

diff --git a/lnet/lnet/api-ni.c b/lnet/lnet/api-ni.c
index 2053f1b..91f993f 100644
--- a/lnet/lnet/api-ni.c
+++ b/lnet/lnet/api-ni.c
@@ -10019,6 +10019,7 @@ static int lnet_ping(struct lnet_processid *id, struct lnet_nid *src_nid,
 	u32 *st;
 	int nob;
 	int rc;
+	int rc2;
 
 	genradix_init(&plist->lgpl_list);
 
@@ -10056,22 +10057,27 @@ static int lnet_ping(struct lnet_processid *id, struct lnet_nid *src_nid,
 	init_completion(&pd.completion);
 
 	rc = LNetMDBind(&md, LNET_UNLINK, &pd.mdh);
-	if (rc) {
+	if (rc != 0) {
 		CERROR("Can't bind MD: %d\n", rc);
 		goto fail_ping_buffer_decref;
 	}
 
 	rc = LNetGet(src_nid, pd.mdh, id, LNET_RESERVED_PORTAL,
 		     LNET_PROTO_PING_MATCHBITS, 0, false);
-	if (rc)
-		LASSERT(!LNetMDUnlink(pd.mdh));
+	if (rc != 0) {
+		/* Don't CERROR; this could be deliberate! */
+		rc2 = LNetMDUnlink(pd.mdh);
+		LASSERT(rc2 == 0);
 
-	/* Ensure completion in finite time... */
-	wait_for_completion_interruptible_timeout(&pd.completion,
-						  timeout);
+		/* NB must wait for the UNLINK event below... */
+	}
 
-	if (!pd.pd_unlinked)
+	/* Ensure completion in finite time... */
+	wait_for_completion_timeout(&pd.completion, timeout);
+	if (!pd.pd_unlinked) {
 		LNetMDUnlink(pd.mdh);
+		wait_for_completion(&pd.completion);
+	}
 
 	if (!pd.replied) {
 		rc = pd.rc ?: -EIO;