netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [patch net] selftests: forwarding: fix race between packet receive and tc check
@ 2019-11-28 12:38 Jiri Pirko
  2019-11-30 20:24 ` David Miller
  0 siblings, 1 reply; 2+ messages in thread
From: Jiri Pirko @ 2019-11-28 12:38 UTC (permalink / raw)
  To: netdev; +Cc: davem, mlxsw

From: Jiri Pirko <jiri@mellanox.com>

It is possible that tc stats get checked before the packet we check for
actually arrived into the interface and accounted for.
Fix it by checking for the expected result in a loop until
timeout is reached (by default 1 second).

Fixes: 07e5c75184a1 ("selftests: forwarding: Introduce tc flower matching tests")
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 .../selftests/net/forwarding/tc_common.sh     | 39 +++++++++++++++----
 1 file changed, 31 insertions(+), 8 deletions(-)

diff --git a/tools/testing/selftests/net/forwarding/tc_common.sh b/tools/testing/selftests/net/forwarding/tc_common.sh
index d93589bd4d1d..64f652633585 100644
--- a/tools/testing/selftests/net/forwarding/tc_common.sh
+++ b/tools/testing/selftests/net/forwarding/tc_common.sh
@@ -3,16 +3,42 @@
 
 CHECK_TC="yes"
 
+# Can be overridden by the configuration file. See lib.sh
+TC_HIT_TIMEOUT=${TC_HIT_TIMEOUT:=1000} # ms
+
+__tc_check_packets()
+{
+	local id=$1
+	local handle=$2
+	local count=$3
+	local operator=$4
+
+	start_time="$(date -u +%s%3N)"
+	while true
+	do
+		cmd_jq "tc -j -s filter show $id" \
+		       ".[] | select(.options.handle == $handle) | \
+			    select(.options.actions[0].stats.packets $operator $count)" \
+		    &> /dev/null
+		ret=$?
+		if [[ $ret -eq 0 ]]; then
+			return $ret
+		fi
+		current_time="$(date -u +%s%3N)"
+		diff=$(expr $current_time - $start_time)
+		if [ "$diff" -gt "$TC_HIT_TIMEOUT" ]; then
+			return 1
+		fi
+	done
+}
+
 tc_check_packets()
 {
 	local id=$1
 	local handle=$2
 	local count=$3
 
-	cmd_jq "tc -j -s filter show $id" \
-	       ".[] | select(.options.handle == $handle) | \
-	              select(.options.actions[0].stats.packets == $count)" \
-	       &> /dev/null
+	__tc_check_packets "$id" "$handle" "$count" "=="
 }
 
 tc_check_packets_hitting()
@@ -20,8 +46,5 @@ tc_check_packets_hitting()
 	local id=$1
 	local handle=$2
 
-	cmd_jq "tc -j -s filter show $id" \
-	       ".[] | select(.options.handle == $handle) | \
-		      select(.options.actions[0].stats.packets > 0)" \
-	       &> /dev/null
+	__tc_check_packets "$id" "$handle" 0 ">"
 }
-- 
2.20.1


^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [patch net] selftests: forwarding: fix race between packet receive and tc check
  2019-11-28 12:38 [patch net] selftests: forwarding: fix race between packet receive and tc check Jiri Pirko
@ 2019-11-30 20:24 ` David Miller
  0 siblings, 0 replies; 2+ messages in thread
From: David Miller @ 2019-11-30 20:24 UTC (permalink / raw)
  To: jiri; +Cc: netdev, mlxsw

From: Jiri Pirko <jiri@resnulli.us>
Date: Thu, 28 Nov 2019 13:38:57 +0100

> From: Jiri Pirko <jiri@mellanox.com>
> 
> It is possible that tc stats get checked before the packet we check for
> actually arrived into the interface and accounted for.
> Fix it by checking for the expected result in a loop until
> timeout is reached (by default 1 second).
> 
> Fixes: 07e5c75184a1 ("selftests: forwarding: Introduce tc flower matching tests")
> Signed-off-by: Jiri Pirko <jiri@mellanox.com>

Applied and queued up for -stable, thanks Jiri.

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2019-11-30 20:24 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-11-28 12:38 [patch net] selftests: forwarding: fix race between packet receive and tc check Jiri Pirko
2019-11-30 20:24 ` David Miller

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).