[LTP] [PATCH v2 0/3] shell: Fixes for disabled IPv6

[LTP] [PATCH v2 0/3] shell: Fixes for disabled IPv6

[Petr Vorel]

Hi all,

changes v2->v3:
* Partly rewritten containers netns to use tst_net.sh. Further cleanup
  will be done, but changes are already big and for sake of
  working IPv6 is this enough.
  TODO: Test should be further cleanup to use tst_rhost_run over different
  LTP_NETNS and tst_ping instead of $tping.

* Use only $TST_NET_IPV6_ENABLED for both lhost and rhost. ATM when
  use TST_NET_SKIP_VARIABLE_INIT=1 it detects only lhost. This is useful
  for containers netns tests (and will be needed even after 

NOTE: no other shell tests which use only tst_test.sh need this
(if any needs in the future, can be migrated the same way as
netns_helper.sh currently).

C tests are ok, because they can check EAFNOSUPPORT (as done for SCTP
tests: https://patchwork.ozlabs.org/project/ltp/list/?series=284336)

NOTE: this is really v3, because I forget to add v2 in previous version
https://patchwork.ozlabs.org/project/ltp/list/?series=282638

Kind regards,
Petr

Petr Vorel (3):
  tst_net.sh: Fix for disabled IPv6
  broken_ip-nexthdr.sh: Check IPv6 support before forcing it
  netns: Rewrite to use tst_net.sh

 runtest/containers                            |  32 +--
 .../kernel/containers/netns/netns_breakns.sh  |  29 +--
 .../kernel/containers/netns/netns_comm.sh     |  53 +----
 .../kernel/containers/netns/netns_helper.sh   | 225 +++++++-----------
 testcases/lib/tst_net.sh                      |  71 +++++-
 .../stress/broken_ip/broken_ip-nexthdr.sh     |   1 +
 6 files changed, 169 insertions(+), 242 deletions(-)

-- 
2.35.1


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

[LTP] [PATCH v2 1/3] tst_net.sh: Fix for disabled IPv6

[Petr Vorel]

Tests failed in tst_init_iface even IPv4 only test was run.
Allow to init interfaces at least for IPv4.

We need IPv6 enabled on both endpoints to be usable,
unless test use TST_NET_SKIP_VARIABLE_INIT=1, i.e. use network library
functions but no test links (atm just library test tst_ipaddr_un.sh,
but netns_helper.sh will also use it). In this case only IPv6 on lhost
is tested.

Tests which use TST_IPV6=6 to force IPv6 (atm just broken_ip-nexthdr.sh)
need to be fixed, unless they use just tst_ipaddr_un() (e.g. library
test tst_ipaddr_un.sh).

Store result into $TST_NET_IPV6_ENABLED.

Signed-off-by: Petr Vorel <pvorel@suse.cz>
---
 testcases/lib/tst_net.sh | 71 +++++++++++++++++++++++++++++++++-------
 1 file changed, 60 insertions(+), 11 deletions(-)

diff --git a/testcases/lib/tst_net.sh b/testcases/lib/tst_net.sh
index 047686dc39..15fd595942 100644
--- a/testcases/lib/tst_net.sh
+++ b/testcases/lib/tst_net.sh
@@ -24,7 +24,9 @@ TST_IPV6_FLAG=${TST_IPV6_FLAG:-}
 tst_net_parse_args()
 {
 	case $1 in
-	6) TST_IPV6=6 TST_IPVER=6 TST_IPV6_FLAG="-6";;
+	6)  tst_net_require_ipv6
+	    TST_IPV6=6 TST_IPVER=6 TST_IPV6_FLAG="-6"
+	    ;;
 	*) [ "$TST_PARSE_ARGS_CALLER" ] && $TST_PARSE_ARGS_CALLER "$1" "$2";;
 	esac
 }
@@ -100,6 +102,32 @@ tst_brk_()
 	[ -z "$TST_USE_LEGACY_API" ] && tst_brk $@ || tst_brkm $@
 }
 
+
+tst_net_detect_ipv6()
+{
+	local type="${1:-lhost}"
+	local cmd='[ -f /proc/net/if_inet6 ]'
+	local ret
+
+	if [ "$type" = "lhost" ]; then
+		$cmd
+	else
+		tst_rhost_run -c "$cmd"
+	fi
+	ret=$?
+
+	if [ $ret -eq 0 ]; then
+		TST_NET_IPV6_ENABLED=1
+	else
+		tst_res TINFO "IPv6 disabled on $type"
+	fi
+}
+
+tst_net_require_ipv6()
+{
+	[ "$TST_NET_IPV6_ENABLED" = 1 ] || tst_brk_ TCONF "IPv6 disabled"
+}
+
 init_ltp_netspace()
 {
 	local pid
@@ -517,7 +545,9 @@ tst_init_iface()
 		ip link set $iface down || return $?
 		ip route flush dev $iface || return $?
 		ip addr flush dev $iface || return $?
-		sysctl -qw net.ipv6.conf.$iface.accept_dad=0 || return $?
+		if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
+			sysctl -qw net.ipv6.conf.$iface.accept_dad=0 || return $?
+		fi
 		ip link set $iface up
 		return $?
 	fi
@@ -529,7 +559,9 @@ tst_init_iface()
 	tst_rhost_run -c "ip link set $iface down" || return $?
 	tst_rhost_run -c "ip route flush dev $iface" || return $?
 	tst_rhost_run -c "ip addr flush dev $iface" || return $?
-	tst_rhost_run -c "sysctl -qw net.ipv6.conf.$iface.accept_dad=0" || return $?
+	if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
+		tst_rhost_run -c "sysctl -qw net.ipv6.conf.$iface.accept_dad=0" || return $?
+	fi
 	tst_rhost_run -c "ip link set $iface up"
 }
 
@@ -606,7 +638,9 @@ tst_restore_ipaddr()
 	local ret=0
 	local backup_tst_ipv6=$TST_IPV6
 	TST_IPV6= tst_add_ipaddr $type $link_num || ret=$?
-	TST_IPV6=6 tst_add_ipaddr $type $link_num || ret=$?
+	if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
+		TST_IPV6=6 tst_add_ipaddr $type $link_num || ret=$?
+	fi
 	TST_IPV6=$backup_tst_ipv6
 
 	return $ret
@@ -937,6 +971,9 @@ tst_default_max_pkt()
 	echo "$((mtu + mtu / 10))"
 }
 
+# detect IPv6 support on lhost for tests which don't use test links
+tst_net_detect_ipv6
+
 [ -n "$TST_PRINT_HELP" -o -n "$TST_NET_SKIP_VARIABLE_INIT" ] && return 0
 
 # Management Link
@@ -971,8 +1008,13 @@ IPV6_RHOST="${IPV6_RHOST:-fd00:1:1:1::1/64}"
 if [ -z "$_tst_net_parse_variables" ]; then
 	eval $(tst_net_ip_prefix $IPV4_LHOST || echo "exit $?")
 	eval $(tst_net_ip_prefix -r $IPV4_RHOST || echo "exit $?")
-	eval $(tst_net_ip_prefix $IPV6_LHOST || echo "exit $?")
-	eval $(tst_net_ip_prefix -r $IPV6_RHOST || echo "exit $?")
+
+	tst_net_detect_ipv6 rhost
+
+	if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
+		eval $(tst_net_ip_prefix $IPV6_LHOST || echo "exit $?")
+		eval $(tst_net_ip_prefix -r $IPV6_RHOST || echo "exit $?")
+	fi
 fi
 
 [ -n "$TST_USE_NETNS" -a "$TST_INIT_NETNS" != "no" ] && init_ltp_netspace
@@ -981,19 +1023,26 @@ if [ -z "$_tst_net_parse_variables" ]; then
 	eval $(tst_net_iface_prefix $IPV4_LHOST || echo "exit $?")
 	eval $(tst_rhost_run -c 'tst_net_iface_prefix -r '$IPV4_RHOST \
 		|| echo "exit $?")
-	eval $(tst_net_iface_prefix $IPV6_LHOST || echo "exit $?")
-	eval $(tst_rhost_run -c 'tst_net_iface_prefix -r '$IPV6_RHOST \
-		|| echo "exit $?")
+
+	if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
+		eval $(tst_net_iface_prefix $IPV6_LHOST || echo "exit $?")
+		eval $(tst_rhost_run -c 'tst_net_iface_prefix -r '$IPV6_RHOST \
+			|| echo "exit $?")
+	fi
 
 	eval $(tst_net_vars $IPV4_LHOST/$IPV4_LPREFIX \
 		$IPV4_RHOST/$IPV4_RPREFIX || echo "exit $?")
-	eval $(tst_net_vars $IPV6_LHOST/$IPV6_LPREFIX \
-		$IPV6_RHOST/$IPV6_RPREFIX || echo "exit $?")
+
+	if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
+		eval $(tst_net_vars $IPV6_LHOST/$IPV6_LPREFIX \
+			$IPV6_RHOST/$IPV6_RPREFIX || echo "exit $?")
+	fi
 
 	tst_res_ TINFO "Network config (local -- remote):"
 	tst_res_ TINFO "$LHOST_IFACES -- $RHOST_IFACES"
 	tst_res_ TINFO "$IPV4_LHOST/$IPV4_LPREFIX -- $IPV4_RHOST/$IPV4_RPREFIX"
 	tst_res_ TINFO "$IPV6_LHOST/$IPV6_LPREFIX -- $IPV6_RHOST/$IPV6_RPREFIX"
+
 	export _tst_net_parse_variables="yes"
 fi
 
-- 
2.35.1


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

[LTP] [PATCH v2 2/3] broken_ip-nexthdr.sh: Check IPv6 support before forcing it

[Petr Vorel]

Signed-off-by: Petr Vorel <pvorel@suse.cz>
---
 testcases/network/stress/broken_ip/broken_ip-nexthdr.sh | 1 +
 1 file changed, 1 insertion(+)

diff --git a/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh b/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh
index ec6643af66..cb4a3dd399 100755
--- a/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh
+++ b/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh
@@ -11,6 +11,7 @@ TST_TESTFUNC="do_test"
 do_test()
 {
 	# not supported on IPv4
+	tst_net_require_ipv6
 	TST_IPV6=6
 	TST_IPVER=6
 
-- 
2.35.1


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

[LTP] [PATCH v2 3/3] netns: Rewrite to use tst_net.sh

[Petr Vorel]

Use TST_NET_SKIP_VARIABLE_INIT=1 thus we get network library
functionality but use it's own test links.

Using tst_net.sh provides:
* detection of disabled IPv6
* -6 support instead of specifying ipv4/ipv6
* ping6 detection
* use tst_ipaddr_un() to get unused addresses instead of basic ranges
  192.168.0.x / fd00::x

Also
* cleanup other parameter handling - instead of passing 7 parameters to
  the setup use as the default the modern variants (ip and netlink), use
  switches for other variants (ns_exec and ifconfig)
* use ROD to shorten code
* cleanup duplicate doc

TODO: Test should be further cleanup to use tst_rhost_run over different
LTP_NETNS and tst_ping instead of $tping (changes already big).

Signed-off-by: Petr Vorel <pvorel@suse.cz>
---
 runtest/containers                            |  32 +--
 .../kernel/containers/netns/netns_breakns.sh  |  29 +--
 .../kernel/containers/netns/netns_comm.sh     |  53 +----
 .../kernel/containers/netns/netns_helper.sh   | 225 +++++++-----------
 4 files changed, 108 insertions(+), 231 deletions(-)

diff --git a/runtest/containers b/runtest/containers
index eea7bfadb7..ab0b94f51a 100644
--- a/runtest/containers
+++ b/runtest/containers
@@ -25,22 +25,22 @@ mqns_04 mqns_04
 mqns_04_clone mqns_04 -clone
 
 netns_netlink netns_netlink
-netns_breakns_ns_exec_ipv4_netlink netns_breakns.sh ns_exec ipv4 netlink
-netns_breakns_ns_exec_ipv6_netlink netns_breakns.sh ns_exec ipv6 netlink
-netns_breakns_ns_exec_ipv4_ioctl netns_breakns.sh ns_exec ipv4 ioctl
-netns_breakns_ns_exec_ipv6_ioctl netns_breakns.sh ns_exec ipv6 ioctl
-netns_breakns_ip_ipv4_netlink netns_breakns.sh ip ipv4 netlink
-netns_breakns_ip_ipv6_netlink netns_breakns.sh ip ipv6 netlink
-netns_breakns_ip_ipv4_ioctl netns_breakns.sh ip ipv4 ioctl
-netns_breakns_ip_ipv6_ioctl netns_breakns.sh ip ipv6 ioctl
-netns_comm_ns_exec_ipv4_netlink netns_comm.sh ns_exec ipv4 netlink
-netns_comm_ns_exec_ipv6_netlink netns_comm.sh ns_exec ipv6 netlink
-netns_comm_ns_exec_ipv4_ioctl netns_comm.sh ns_exec ipv4 ioctl
-netns_comm_ns_exec_ipv6_ioctl netns_comm.sh ns_exec ipv6 ioctl
-netns_comm_ip_ipv4_netlink netns_comm.sh ip ipv4 netlink
-netns_comm_ip_ipv6_netlink netns_comm.sh ip ipv6 netlink
-netns_comm_ip_ipv4_ioctl netns_comm.sh ip ipv4 ioctl
-netns_comm_ip_ipv6_ioctl netns_comm.sh ip ipv6 ioctl
+netns_breakns_ip_ipv4_netlink netns_breakns.sh
+netns_breakns_ip_ipv6_netlink netns_breakns.sh -6
+netns_breakns_ip_ipv4_ioctl netns_breakns.sh -I
+netns_breakns_ip_ipv6_ioctl netns_breakns.sh -6I
+netns_breakns_ns_exec_ipv4_netlink netns_breakns.sh -e
+netns_breakns_ns_exec_ipv6_netlink netns_breakns.sh -6e
+netns_breakns_ns_exec_ipv4_ioctl netns_breakns.sh -eI
+netns_breakns_ns_exec_ipv6_ioctl netns_breakns.sh -6eI
+netns_comm_ip_ipv4_netlink netns_comm.sh
+netns_comm_ip_ipv6_netlink netns_comm.sh -6
+netns_comm_ip_ipv4_ioctl netns_comm.sh -I
+netns_comm_ip_ipv6_ioctl netns_comm.sh -6I
+netns_comm_ns_exec_ipv4_netlink netns_comm.sh -e
+netns_comm_ns_exec_ipv6_netlink netns_comm.sh -6e
+netns_comm_ns_exec_ipv4_ioctl netns_comm.sh -eI
+netns_comm_ns_exec_ipv6_ioctl netns_comm.sh -6eI
 netns_sysfs netns_sysfs.sh
 
 shmnstest_none shmnstest none
diff --git a/testcases/kernel/containers/netns/netns_breakns.sh b/testcases/kernel/containers/netns/netns_breakns.sh
index 1ce5d37efd..64810b711e 100755
--- a/testcases/kernel/containers/netns/netns_breakns.sh
+++ b/testcases/kernel/containers/netns/netns_breakns.sh
@@ -1,23 +1,9 @@
 #!/bin/sh
 # SPDX-License-Identifier: GPL-2.0-or-later
+# Copyright (c) 2022 Petr Vorel <pvorel@suse.cz>
 # Copyright (c) Köry Maincent <kory.maincent@bootlin.com> 2020
 # Copyright (c) 2015 Red Hat, Inc.
 #
-# SYNOPSIS:
-# netns_breakns.sh <NS_EXEC_PROGRAM> <IP_VERSION> <COMM_TYPE>
-#
-# OPTIONS:
-#	* NS_EXEC_PROGRAM (ns_exec|ip)
-#		Program which will be used to enter and run other commands
-#		inside a network namespace.
-#	* IP_VERSION (ipv4|ipv6)
-#		Version of IP. (ipv4|ipv6)
-#	* COMM_TYPE (netlink|ioctl)
-#		Communication type between kernel and user space
-#		for basic setup: enabling and assigning IP addresses
-#		to the virtual ethernet devices. (Uses 'ip' command for netlink
-#		and 'ifconfig' for ioctl.)
-#
 # Tests communication with ip (uses netlink) and ifconfig (uses ioctl)
 # over a device which is not visible from the current network namespace.
 #
@@ -26,21 +12,8 @@
 # 1. using netlink (ip command).
 # 2. using ioctl (ifconfig command).
 
-TST_POS_ARGS=3
-TST_SETUP=do_setup
-TST_TESTFUNC=do_test
 . netns_helper.sh
 
-PROG=$1
-IP_VER=$2
-COM_TYPE=$3
-
-do_setup()
-{
-	netns_setup $PROG $IP_VER $COM_TYPE "192.168.0.2" "192.168.0.3" "fd00::2" "fd00::3"
-	tst_res TINFO "NS interaction: $PROG | devices setup: $COM_TYPE"
-}
-
 do_test()
 {
 	EXPECT_FAIL $NS_EXEC $NS_HANDLE0 $NS_TYPE ip address add $IP1/$NETMASK dev veth1
diff --git a/testcases/kernel/containers/netns/netns_comm.sh b/testcases/kernel/containers/netns/netns_comm.sh
index ccb8b47b11..3fdbf13b59 100755
--- a/testcases/kernel/containers/netns/netns_comm.sh
+++ b/testcases/kernel/containers/netns/netns_comm.sh
@@ -1,23 +1,9 @@
 #!/bin/sh
 # SPDX-License-Identifier: GPL-2.0-or-later
+# Copyright (c) 2022 Petr Vorel <pvorel@suse.cz>
 # Copyright (c) Köry Maincent <kory.maincent@bootlin.com> 2020
 # Copyright (c) 2015 Red Hat, Inc.
 #
-# SYNOPSIS:
-# netns_comm.sh <NS_EXEC_PROGRAM> <IP_VERSION> <COMM_TYPE>
-#
-# OPTIONS:
-#       * NS_EXEC_PROGRAM (ns_exec|ip)
-#               Program which will be used to enter and run other commands
-#               inside a network namespace.
-#       * IP_VERSION (ipv4|ipv6)
-#               Version of IP. (ipv4|ipv6)
-#	* COMM_TYPE (netlink|ioctl)
-#		Communication type between kernel and user space
-#		for basic setup: enabling and assigning IP addresses
-#		to the virtual ethernet devices. (Uses 'ip' command for netlink
-#		and 'ifconfig' for ioctl.)
-#
 # Tests that a separate network namespace can configure and communicate
 # over the devices it sees. Tests are done using netlink(7) ('ip' command)
 # or ioctl(2) ('ifconfig' command) for controlling devices.
@@ -29,41 +15,22 @@
 #   3. communication over the lo (localhost) device in a separate
 #      network namespace
 
-TST_POS_ARGS=3
-TST_SETUP=do_setup
-TST_TESTFUNC=do_test
 . netns_helper.sh
 
-PROG=$1
-IP_VER=$2
-COM_TYPE=$3
-
-do_setup()
-{
-	netns_setup $PROG $IP_VER $COM_TYPE "192.168.0.2" "192.168.0.3" "fd00::2" "fd00::3"
-	tst_res TINFO "NS interaction: $PROG | devices setup: $COM_TYPE"
-}
-
 do_test()
 {
-	EXPECT_PASS $NS_EXEC $NS_HANDLE0 $NS_TYPE $tping -q -c2 -I veth0 $IP1 1>/dev/null
+	local ip_lo="127.0.0.1"
+	[ "$TST_IPV6" ] && ip_lo="::1"
 
+	EXPECT_PASS $NS_EXEC $NS_HANDLE0 $NS_TYPE $tping -q -c2 -I veth0 $IP1 1>/dev/null
 	EXPECT_PASS $NS_EXEC $NS_HANDLE1 $NS_TYPE $tping -q -c2 -I veth1 $IP0 1>/dev/null
 
-	case "$IP_VER" in
-	ipv4) ip_lo="127.0.0.1" ;;
-	ipv6) ip_lo="::1" ;;
-	esac
-	case "$COM_TYPE" in
-	netlink)
-		$NS_EXEC $NS_HANDLE0 $NS_TYPE ip link set dev lo up || \
-			tst_brk TBROK "enabling lo device failed"
-		;;
-	ioctl)
-		$NS_EXEC $NS_HANDLE0 $NS_TYPE ifconfig lo up || \
-			tst_brk TBROK "enabling lo device failed"
-		;;
-	esac
+	if [ "$COMM_TYPE" = "netlink" ]; then
+		ROD $NS_EXEC $NS_HANDLE0 $NS_TYPE ip link set dev lo up
+	else
+		ROD $NS_EXEC $NS_HANDLE0 $NS_TYPE ifconfig lo up
+	fi
+
 	EXPECT_PASS $NS_EXEC $NS_HANDLE0 $NS_TYPE $tping -q -c2 -I lo $ip_lo 1>/dev/null
 }
 
diff --git a/testcases/kernel/containers/netns/netns_helper.sh b/testcases/kernel/containers/netns/netns_helper.sh
index d7d4d2c2de..4b6ad23009 100755
--- a/testcases/kernel/containers/netns/netns_helper.sh
+++ b/testcases/kernel/containers/netns/netns_helper.sh
@@ -1,37 +1,32 @@
 #!/bin/sh
 # SPDX-License-Identifier: GPL-2.0-or-later
+# Copyright (c) 2022 Petr Vorel <pvorel@suse.cz>
 # Copyright (c) Linux Test Project, 2014-2021
 # Copyright (c) 2015 Red Hat, Inc.
 
-TST_CLEANUP=netns_ns_exec_cleanup
 TST_NEEDS_ROOT=1
 TST_NEEDS_CMDS="ip ping"
 TST_NEEDS_DRIVERS="veth"
-. tst_test.sh
 
-# Set to 1 only for test cases using ifconfig (ioctl).
-USE_IFCONFIG=0
+TST_OPTS="eI"
+TST_PARSE_ARGS="netns_parse_args"
+TST_USAGE=netns_usage
+TST_SETUP=netns_setup
+TST_CLEANUP=netns_ns_exec_cleanup
+TST_TESTFUNC=do_test
 
-# Variables which can be used in test cases (set by netns_setup() function):
+TST_NET_SKIP_VARIABLE_INIT=1
 
-# Use in test cases to execute commands inside a namespace. Set to 'ns_exec' or
-# 'ip netns exec' command according to NS_EXEC_PROGRAM argument specified in
-# netns_setup() function call.
-NS_EXEC=
+# from tst_net_vars.c
+IPV4_NET16_UNUSED="10.23"
+IPV6_NET32_UNUSED="fd00:23"
+. tst_net.sh
 
 # Set to "net" for ns_create/ns_exec as their options requires
 # to specify a namespace type. Empty for ip command.
 NS_TYPE=
 
-# IP addresses of veth0 (IP0) and veth1 (IP1) devices (ipv4/ipv6 variant
-# is determined according to the IP_VERSION argument specified in netns_setup()
-# function call.
-IP0=
-IP1=
-NETMASK=
-
-# 'ping' or 'ping6' according to the IP_VERSION argument specified
-# in netns_setup() function call.
+# 'ping' or 'ping6'
 tping=
 
 # Network namespaces handles for manipulating and executing commands inside
@@ -46,107 +41,61 @@ NS_HANDLE1=
 # ifconfig <device> $IFCONF_IN6_ARG IP/NETMASK
 IFCONF_IN6_ARG=
 
-# Sets up global variables which can be used in test cases (documented above),
-# creates two network namespaces and a pair of virtual ethernet devices, each
-# device in one namespace. Each device is then enabled and assigned an IP
-# address according to the function parameters. IFCONF_IN6_ARG variable is set
-# only if ipv6 variant of test case is used (determined by IP_VERSION argument).
-#
-# SYNOPSIS:
-# netns_setup <NS_EXEC_PROGRAM> <IP_VERSION> <COMM_TYPE> <IP4_VETH0>
-#             <IP4_VETH1> <IP6_VETH0> <IP6_VETH1>
-#
-# OPTIONS:
-#	* NS_EXEC_PROGRAM (ns_exec|ip)
-#		Program which will be used to enter and run other commands
-#		inside a network namespace.
-#	* IP_VERSION (ipv4|ipv6)
-#		Version of IP. (ipv4|ipv6)
-#	* COMM_TYPE (netlink|ioctl)
-#		Communication type between kernel and user space
-#		for enabling and assigning IP addresses to the virtual
-#		ethernet devices. Uses 'ip' command for netlink and 'ifconfig'
-#		for ioctl. (If set to ioctl, function also checks the existance
-#		of the 'ifconfig' command.)
-#	* IP4_VETH0, IP4_VETH1
-#		IPv4 addresses for veth0 and veth1 devices.
-#	* IP6_VETH0, IP6_VETH1
-#		IPv6 addresses for veth0 and veth1 devices.
-#
-# On success function returns, on error tst_brk is called and TC is terminated.
+# Program which will be used to enter and run other commands inside a network namespace.
+# (ns_exec|ip)
+NS_EXEC="ip"
+
+# Communication type between kernel and user space for basic setup: enabling and
+# assigning IP addresses to the virtual ethernet devices. (Uses 'ip' command for
+# netlink and 'ifconfig' for ioctl.)
+# (netlink|ioctl)
+COMM_TYPE="netlink"
+
+netns_parse_args()
+{
+	case $1 in
+	e) NS_EXEC="ns_exec" ;;
+	I) COMM_TYPE="ioctl"; tst_require_cmds ifconfig ;;
+	esac
+}
+
+netns_usage()
+{
+	echo "usage: $0 [ -e ] [ -t ]"
+	echo "OPTIONS"
+	echo "-e use ns_exec instead of ip"
+	echo "-I test ioctl (with ifconfig) instead of netlink (with ip)"
+}
+
 netns_setup()
 {
-	case "$1" in
-	ns_exec)
+	if [ "$NS_EXEC" = "ip" ]; then
+		netns_ip_setup
+		TST_CLEANUP=netns_ip_cleanup
+	else
 		setns_check
 		[ $? -eq 32 ] && tst_brk TCONF "setns not supported"
 
 		NS_TYPE="net"
 		netns_ns_exec_setup
 		TST_CLEANUP=netns_ns_exec_cleanup
-		;;
-	ip)
-		netns_ip_setup
-		TST_CLEANUP=netns_ip_cleanup
-		;;
-	*)
-		tst_brk TBROK \
-		"first argument must be a program used to enter a network namespace (ns_exec|ip)"
-		;;
-	esac
-
-	case "$3" in
-	netlink)
-		;;
-	ioctl)
-		USE_IFCONFIG=1
-		tst_require_cmds ifconfig
-		;;
-	*)
-		tst_brk TBROK \
-		"third argument must be a comm. type between kernel and user space (netlink|ioctl)"
-		;;
-	esac
-
-	if [ -z "$4" ]; then
-		tst_brk TBROK "fourth argument must be the IPv4 address for veth0"
-	fi
-	if [ -z "$5" ]; then
-		tst_brk TBROK "fifth argument must be the IPv4 address for veth1"
-	fi
-	if [ -z "$6" ]; then
-		tst_brk TBROK "sixth argument must be the IPv6 address for veth0"
-	fi
-	if [ -z "$7" ]; then
-		tst_brk TBROK "seventh argument must be the IPv6 address for veth1"
 	fi
 
-	case "$2" in
-	ipv4)
-		IP0=$4
-		IP1=$5
-		tping="ping"
-		NETMASK=24
-		;;
-	ipv6)
+	IP0=$(tst_ipaddr_un -c 1)
+	IP1=$(tst_ipaddr_un -c 2)
+
+	if [ "$TST_IPV6" ]; then
 		IFCONF_IN6_ARG="inet6 add"
-		IP0=$6
-		IP1=$7
-
-		if tst_cmd_available ping6; then
-			tping="ping6"
-		else
-			tping="ping -6"
-			tst_res_ TINFO "ping6 binary/symlink is missing, using workaround. Please, report missing ping6 to your distribution."
-		fi
 		NETMASK=64
-		;;
-	*)
-		tst_brk TBROK "second argument must be an ip version (ipv4|ipv6)"
-		;;
-	esac
+	else
+		NETMASK=24
+	fi
+
+	tping=ping$TST_IPV6
 
 	netns_set_ip
+
+	tst_res TINFO "testing netns over $COMM_TYPE with $NS_EXEC $PROG"
 }
 
 # Sets up NS_EXEC to use 'ns_exec', creates two network namespaces and stores
@@ -198,23 +147,18 @@ netns_ip_setup()
 	ip netns del $NS_HANDLE0 2>/dev/null
 	ip netns del $NS_HANDLE1 2>/dev/null
 
-	ip netns add $NS_HANDLE0 || \
-		tst_brk TBROK "unable to create a new network namespace"
-	ip netns add $NS_HANDLE1 || \
-		tst_brk TBROK "unable to create a new network namespace"
-
-	$NS_EXEC $NS_HANDLE0 ip link add veth0 type veth peer name veth1 || \
-		tst_brk TBROK "unable to create veth pair devices"
+	ROD ip netns add $NS_HANDLE0
+	ROD ip netns add $NS_HANDLE1
 
-	$NS_EXEC $NS_HANDLE0 ip link set veth1 netns $NS_HANDLE1 || \
-		tst_brk TBROK "unable to add device veth1 to the separate network namespace"
+	ROD $NS_EXEC $NS_HANDLE0 ip link add veth0 type veth peer name veth1
+	ROD $NS_EXEC $NS_HANDLE0 ip link set veth1 netns $NS_HANDLE1
 }
 
 # Enables virtual ethernet devices and assigns IP addresses for both
 # of them (IPv4/IPv6 variant is decided by netns_setup() function).
 netns_set_ip()
 {
-	[ "$NS_EXEC" ] || tst_brk TBROK "netns_setup() function must be called first"
+	local cmd="ip"
 
 	# This applies only for ipv6 variant:
 	# Do not accept Router Advertisements (accept_ra) and do not use
@@ -223,35 +167,28 @@ netns_set_ip()
 	# there is no other host with the same address, the address is
 	# considered to be "tentative" (attempts to bind() to the address fail
 	# with EADDRNOTAVAIL) which may cause problems for tests using ipv6.
-	echo 0 | $NS_EXEC $NS_HANDLE0 $NS_TYPE \
-		tee /proc/sys/net/ipv6/conf/veth0/accept_dad \
-		/proc/sys/net/ipv6/conf/veth0/accept_ra >/dev/null
-	echo 0 | $NS_EXEC $NS_HANDLE1 $NS_TYPE \
-		tee /proc/sys/net/ipv6/conf/veth1/accept_dad \
-		/proc/sys/net/ipv6/conf/veth1/accept_ra >/dev/null
-
-	case $USE_IFCONFIG in
-	1)
-		$NS_EXEC $NS_HANDLE0 $NS_TYPE ifconfig veth0 $IFCONF_IN6_ARG $IP0/$NETMASK ||
-			tst_brk TBROK "adding address to veth0 failed"
-		$NS_EXEC $NS_HANDLE1 $NS_TYPE ifconfig veth1 $IFCONF_IN6_ARG $IP1/$NETMASK ||
-			tst_brk TBROK "adding address to veth1 failed"
-		$NS_EXEC $NS_HANDLE0 $NS_TYPE ifconfig veth0 up ||
-			tst_brk TBROK "enabling veth0 device failed"
-		$NS_EXEC $NS_HANDLE1 $NS_TYPE ifconfig veth1 up ||
-			tst_brk TBROK "enabling veth1 device failed"
-		;;
-	*)
-		$NS_EXEC $NS_HANDLE0 $NS_TYPE ip address add $IP0/$NETMASK dev veth0 ||
-			tst_brk TBROK "adding address to veth0 failed"
-		$NS_EXEC $NS_HANDLE1 $NS_TYPE ip address add $IP1/$NETMASK dev veth1 ||
-			tst_brk TBROK "adding address to veth1 failed"
-		$NS_EXEC $NS_HANDLE0 $NS_TYPE ip link set veth0 up ||
-			tst_brk TBROK "enabling veth0 device failed"
-		$NS_EXEC $NS_HANDLE1 $NS_TYPE ip link set veth1 up ||
-			tst_brk TBROK "enabling veth1 device failed"
-		;;
-	esac
+	if [ "$TST_IPV6" ]; then
+		echo 0 | $NS_EXEC $NS_HANDLE0 $NS_TYPE \
+			tee /proc/sys/net/ipv6/conf/veth0/accept_dad \
+			/proc/sys/net/ipv6/conf/veth0/accept_ra >/dev/null
+		echo 0 | $NS_EXEC $NS_HANDLE1 $NS_TYPE \
+			tee /proc/sys/net/ipv6/conf/veth1/accept_dad \
+			/proc/sys/net/ipv6/conf/veth1/accept_ra >/dev/null
+	fi
+
+	[ "$COMM_TYPE" = "ioctl" ] && cmd="ifconfig"
+
+	if [ "$COMM_TYPE" = "netlink" ]; then
+		ROD $NS_EXEC $NS_HANDLE0 $NS_TYPE ip address add $IP0/$NETMASK dev veth0
+		ROD $NS_EXEC $NS_HANDLE1 $NS_TYPE ip address add $IP1/$NETMASK dev veth1
+		ROD $NS_EXEC $NS_HANDLE0 $NS_TYPE ip link set veth0 up
+		ROD $NS_EXEC $NS_HANDLE1 $NS_TYPE ip link set veth1 up
+	else
+		ROD $NS_EXEC $NS_HANDLE0 $NS_TYPE ifconfig veth0 $IFCONF_IN6_ARG $IP0/$NETMASK
+		ROD $NS_EXEC $NS_HANDLE1 $NS_TYPE ifconfig veth1 $IFCONF_IN6_ARG $IP1/$NETMASK
+		ROD $NS_EXEC $NS_HANDLE0 $NS_TYPE ifconfig veth0 up
+		ROD $NS_EXEC $NS_HANDLE1 $NS_TYPE ifconfig veth1 up
+	fi
 }
 
 netns_ns_exec_cleanup()
-- 
2.35.1


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH v2 1/3] tst_net.sh: Fix for disabled IPv6

[Cyril Hrubis]

Hi!
> Signed-off-by: Petr Vorel <pvorel@suse.cz>
> ---
>  testcases/lib/tst_net.sh | 71 +++++++++++++++++++++++++++++++++-------
>  1 file changed, 60 insertions(+), 11 deletions(-)
> 
> diff --git a/testcases/lib/tst_net.sh b/testcases/lib/tst_net.sh
> index 047686dc39..15fd595942 100644
> --- a/testcases/lib/tst_net.sh
> +++ b/testcases/lib/tst_net.sh
> @@ -24,7 +24,9 @@ TST_IPV6_FLAG=${TST_IPV6_FLAG:-}
>  tst_net_parse_args()
>  {
>  	case $1 in
> -	6) TST_IPV6=6 TST_IPVER=6 TST_IPV6_FLAG="-6";;
> +	6)  tst_net_require_ipv6
> +	    TST_IPV6=6 TST_IPVER=6 TST_IPV6_FLAG="-6"
> +	    ;;
>  	*) [ "$TST_PARSE_ARGS_CALLER" ] && $TST_PARSE_ARGS_CALLER "$1" "$2";;
>  	esac
>  }
> @@ -100,6 +102,32 @@ tst_brk_()
>  	[ -z "$TST_USE_LEGACY_API" ] && tst_brk $@ || tst_brkm $@
>  }
>  
> +
> +tst_net_detect_ipv6()
> +{
> +	local type="${1:-lhost}"
> +	local cmd='[ -f /proc/net/if_inet6 ]'
> +	local ret
> +
> +	if [ "$type" = "lhost" ]; then
> +		$cmd
> +	else
> +		tst_rhost_run -c "$cmd"
> +	fi
> +	ret=$?
> +
> +	if [ $ret -eq 0 ]; then
> +		TST_NET_IPV6_ENABLED=1
> +	else
> +		tst_res TINFO "IPv6 disabled on $type"
> +	fi
> +}
> +
> +tst_net_require_ipv6()
> +{
> +	[ "$TST_NET_IPV6_ENABLED" = 1 ] || tst_brk_ TCONF "IPv6 disabled"
> +}

Shouldn't we detect support for IPv6 on both lhost and rhost in the
tst_net_detect_ipv6() function? I would say that it's pretty obvious
that we need the support on both machines.

>  init_ltp_netspace()
>  {
>  	local pid
> @@ -517,7 +545,9 @@ tst_init_iface()
>  		ip link set $iface down || return $?
>  		ip route flush dev $iface || return $?
>  		ip addr flush dev $iface || return $?
> -		sysctl -qw net.ipv6.conf.$iface.accept_dad=0 || return $?
> +		if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
> +			sysctl -qw net.ipv6.conf.$iface.accept_dad=0 || return $?
> +		fi
>  		ip link set $iface up
>  		return $?
>  	fi
> @@ -529,7 +559,9 @@ tst_init_iface()
>  	tst_rhost_run -c "ip link set $iface down" || return $?
>  	tst_rhost_run -c "ip route flush dev $iface" || return $?
>  	tst_rhost_run -c "ip addr flush dev $iface" || return $?
> -	tst_rhost_run -c "sysctl -qw net.ipv6.conf.$iface.accept_dad=0" || return $?
> +	if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
> +		tst_rhost_run -c "sysctl -qw net.ipv6.conf.$iface.accept_dad=0" || return $?
> +	fi
>  	tst_rhost_run -c "ip link set $iface up"
>  }
>  
> @@ -606,7 +638,9 @@ tst_restore_ipaddr()
>  	local ret=0
>  	local backup_tst_ipv6=$TST_IPV6
>  	TST_IPV6= tst_add_ipaddr $type $link_num || ret=$?
> -	TST_IPV6=6 tst_add_ipaddr $type $link_num || ret=$?
> +	if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
> +		TST_IPV6=6 tst_add_ipaddr $type $link_num || ret=$?
> +	fi
>  	TST_IPV6=$backup_tst_ipv6
>  
>  	return $ret
> @@ -937,6 +971,9 @@ tst_default_max_pkt()
>  	echo "$((mtu + mtu / 10))"
>  }
>  
> +# detect IPv6 support on lhost for tests which don't use test links
> +tst_net_detect_ipv6
> +
>  [ -n "$TST_PRINT_HELP" -o -n "$TST_NET_SKIP_VARIABLE_INIT" ] && return 0
>  
>  # Management Link
> @@ -971,8 +1008,13 @@ IPV6_RHOST="${IPV6_RHOST:-fd00:1:1:1::1/64}"
>  if [ -z "$_tst_net_parse_variables" ]; then
>  	eval $(tst_net_ip_prefix $IPV4_LHOST || echo "exit $?")
>  	eval $(tst_net_ip_prefix -r $IPV4_RHOST || echo "exit $?")
> -	eval $(tst_net_ip_prefix $IPV6_LHOST || echo "exit $?")
> -	eval $(tst_net_ip_prefix -r $IPV6_RHOST || echo "exit $?")
> +
> +	tst_net_detect_ipv6 rhost
> +
> +	if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
> +		eval $(tst_net_ip_prefix $IPV6_LHOST || echo "exit $?")
> +		eval $(tst_net_ip_prefix -r $IPV6_RHOST || echo "exit $?")
> +	fi
>  fi
>  
>  [ -n "$TST_USE_NETNS" -a "$TST_INIT_NETNS" != "no" ] && init_ltp_netspace
> @@ -981,19 +1023,26 @@ if [ -z "$_tst_net_parse_variables" ]; then
>  	eval $(tst_net_iface_prefix $IPV4_LHOST || echo "exit $?")
>  	eval $(tst_rhost_run -c 'tst_net_iface_prefix -r '$IPV4_RHOST \
>  		|| echo "exit $?")
> -	eval $(tst_net_iface_prefix $IPV6_LHOST || echo "exit $?")
> -	eval $(tst_rhost_run -c 'tst_net_iface_prefix -r '$IPV6_RHOST \
> -		|| echo "exit $?")
> +
> +	if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
> +		eval $(tst_net_iface_prefix $IPV6_LHOST || echo "exit $?")
> +		eval $(tst_rhost_run -c 'tst_net_iface_prefix -r '$IPV6_RHOST \
> +			|| echo "exit $?")
> +	fi
>  
>  	eval $(tst_net_vars $IPV4_LHOST/$IPV4_LPREFIX \
>  		$IPV4_RHOST/$IPV4_RPREFIX || echo "exit $?")
> -	eval $(tst_net_vars $IPV6_LHOST/$IPV6_LPREFIX \
> -		$IPV6_RHOST/$IPV6_RPREFIX || echo "exit $?")
> +
> +	if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
> +		eval $(tst_net_vars $IPV6_LHOST/$IPV6_LPREFIX \
> +			$IPV6_RHOST/$IPV6_RPREFIX || echo "exit $?")
> +	fi
>  
>  	tst_res_ TINFO "Network config (local -- remote):"
>  	tst_res_ TINFO "$LHOST_IFACES -- $RHOST_IFACES"
>  	tst_res_ TINFO "$IPV4_LHOST/$IPV4_LPREFIX -- $IPV4_RHOST/$IPV4_RPREFIX"
>  	tst_res_ TINFO "$IPV6_LHOST/$IPV6_LPREFIX -- $IPV6_RHOST/$IPV6_RPREFIX"
> +
>  	export _tst_net_parse_variables="yes"
>  fi
>  
> -- 
> 2.35.1
> 

-- 
Cyril Hrubis
chrubis@suse.cz

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH v2 1/3] tst_net.sh: Fix for disabled IPv6

[Petr Vorel]

Hi Cyril,

...
> > +tst_net_detect_ipv6()
> > +{
> > +	local type="${1:-lhost}"
> > +	local cmd='[ -f /proc/net/if_inet6 ]'
> > +	local ret
> > +
> > +	if [ "$type" = "lhost" ]; then
> > +		$cmd
> > +	else
> > +		tst_rhost_run -c "$cmd"
> > +	fi
> > +	ret=$?
> > +
> > +	if [ $ret -eq 0 ]; then
> > +		TST_NET_IPV6_ENABLED=1
> > +	else
> > +		tst_res TINFO "IPv6 disabled on $type"
> > +	fi
> > +}
> > +
> > +tst_net_require_ipv6()
> > +{
> > +	[ "$TST_NET_IPV6_ENABLED" = 1 ] || tst_brk_ TCONF "IPv6 disabled"
> > +}

> Shouldn't we detect support for IPv6 on both lhost and rhost in the
> tst_net_detect_ipv6() function? I would say that it's pretty obvious
> that we need the support on both machines.
Yes, that's done - see my explanations below. BTW in previous version I had 2
variables, but Alexey suggested to use just one variable
https://lore.kernel.org/ltp/29051e9c-f4c2-9840-7f84-8c9f49efba93@bell-sw.com/

> >  init_ltp_netspace()
> >  {
> >  	local pid
> > @@ -517,7 +545,9 @@ tst_init_iface()
> >  		ip link set $iface down || return $?
> >  		ip route flush dev $iface || return $?
> >  		ip addr flush dev $iface || return $?
> > -		sysctl -qw net.ipv6.conf.$iface.accept_dad=0 || return $?
> > +		if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
> > +			sysctl -qw net.ipv6.conf.$iface.accept_dad=0 || return $?
> > +		fi
> >  		ip link set $iface up
> >  		return $?
> >  	fi
> > @@ -529,7 +559,9 @@ tst_init_iface()
> >  	tst_rhost_run -c "ip link set $iface down" || return $?
> >  	tst_rhost_run -c "ip route flush dev $iface" || return $?
> >  	tst_rhost_run -c "ip addr flush dev $iface" || return $?
> > -	tst_rhost_run -c "sysctl -qw net.ipv6.conf.$iface.accept_dad=0" || return $?
> > +	if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
> > +		tst_rhost_run -c "sysctl -qw net.ipv6.conf.$iface.accept_dad=0" || return $?
> > +	fi
> >  	tst_rhost_run -c "ip link set $iface up"
> >  }

> > @@ -606,7 +638,9 @@ tst_restore_ipaddr()
> >  	local ret=0
> >  	local backup_tst_ipv6=$TST_IPV6
> >  	TST_IPV6= tst_add_ipaddr $type $link_num || ret=$?
> > -	TST_IPV6=6 tst_add_ipaddr $type $link_num || ret=$?
> > +	if [ "$TST_NET_IPV6_ENABLED" = 1 ]; then
> > +		TST_IPV6=6 tst_add_ipaddr $type $link_num || ret=$?
> > +	fi
> >  	TST_IPV6=$backup_tst_ipv6

> >  	return $ret
> > @@ -937,6 +971,9 @@ tst_default_max_pkt()
> >  	echo "$((mtu + mtu / 10))"
> >  }

> > +# detect IPv6 support on lhost for tests which don't use test links
> > +tst_net_detect_ipv6
NOTE Here we detect on lhost - for netns based tests (third patch: https://lore.kernel.org/ltp/20220204194648.32165-4-pvorel@suse.cz/
Test uses TST_NET_SKIP_VARIABLE_INIT=1, because currently it does not use
tst_rhost_run (will be used in further cleanups - I wanted to do as minimal
changes as possible).

> > +
> >  [ -n "$TST_PRINT_HELP" -o -n "$TST_NET_SKIP_VARIABLE_INIT" ] && return 0

> >  # Management Link
> > @@ -971,8 +1008,13 @@ IPV6_RHOST="${IPV6_RHOST:-fd00:1:1:1::1/64}"
> >  if [ -z "$_tst_net_parse_variables" ]; then
> >  	eval $(tst_net_ip_prefix $IPV4_LHOST || echo "exit $?")
> >  	eval $(tst_net_ip_prefix -r $IPV4_RHOST || echo "exit $?")
> > -	eval $(tst_net_ip_prefix $IPV6_LHOST || echo "exit $?")
> > -	eval $(tst_net_ip_prefix -r $IPV6_RHOST || echo "exit $?")
> > +
> > +	tst_net_detect_ipv6 rhost
And here we detect on rhost for all network tests

...

Kind regards,
Petr

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH v2 2/3] broken_ip-nexthdr.sh: Check IPv6 support before forcing it

[Cyril Hrubis]

Hi!
> diff --git a/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh b/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh
> index ec6643af66..cb4a3dd399 100755
> --- a/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh
> +++ b/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh
> @@ -11,6 +11,7 @@ TST_TESTFUNC="do_test"
>  do_test()
>  {
>  	# not supported on IPv4
> +	tst_net_require_ipv6
>  	TST_IPV6=6
>  	TST_IPVER=6

I was looking at the code if we can simply instead do:

[ "$TST_IPVER" == 6 ] && tst_net_require_ipv6

in the test library, but it looks like the parameters are parsed in the
tst_test.sh in the tst_run() function. Which means that the TST_IPVER is
not actually set until the test starts and the library has no way of
knowing the variable value beforehand.

I guess that we can actually move the option parsing code in the
tst_test.sh so that it happens just right after the script is sourced,
which would make things much easier as the TST_IPVER would end up being
defined in the tst_network.sh and we coud simply use the statement above
without any further hacks like this patch adds.

As a side effect we could clean up the test option parsing code since we
do actually have two different getopts loop in the tst_test.sh library
and as far as I can tell we can do just with one.

-- 
Cyril Hrubis
chrubis@suse.cz

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH v2 2/3] broken_ip-nexthdr.sh: Check IPv6 support before forcing it

[Petr Vorel]

Hi Cyril,

first, thanks a lot for a review of network shell tests.

> Hi!
> > diff --git a/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh b/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh
> > index ec6643af66..cb4a3dd399 100755
> > --- a/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh
> > +++ b/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh
> > @@ -11,6 +11,7 @@ TST_TESTFUNC="do_test"
> >  do_test()
> >  {
> >  	# not supported on IPv4
> > +	tst_net_require_ipv6
> >  	TST_IPV6=6
> >  	TST_IPVER=6

> I was looking at the code if we can simply instead do:

> [ "$TST_IPVER" == 6 ] && tst_net_require_ipv6
BTW this code was requested by Alexey, originally I suggested different way
to declare TST_NET_IPV{4,6}_ONLY=1 in the test:

https://lore.kernel.org/ltp/20210714140716.1568-3-pvorel@suse.cz/

> in the test library, but it looks like the parameters are parsed in the
> tst_test.sh in the tst_run() function. Which means that the TST_IPVER is
> not actually set until the test starts and the library has no way of
> knowing the variable value beforehand.
Yes.

> I guess that we can actually move the option parsing code in the
> tst_test.sh so that it happens just right after the script is sourced,
> which would make things much easier as the TST_IPVER would end up being
> defined in the tst_network.sh and we coud simply use the statement above
> without any further hacks like this patch adds.
Interesting. Originally I thought I'd need IPv6 check for netns_*.sh tests, but
in the end I realized that better will be if they use tst_net.sh with forcing
netns only.

Sure, I can have look into putting -6 handling into tst_test.sh. It's just a bit
strange to me that tst_test.sh parses it, but it's actually used elsewhere.

> As a side effect we could clean up the test option parsing code since we
> do actually have two different getopts loop in the tst_test.sh library
> and as far as I can tell we can do just with one.
When I looked last time into this and it looked to me that both are needed.
The second one (which is actually run the first - ":hi:$TST_OPTS" is for
ignoring errors when -h is set. But maybe these warnings ("Invalid number of
positional parameters:" and "Unexpected positional arguments" might be really
possible to handle inside of tst_run().

Kind regards,
Petr

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH v2 2/3] broken_ip-nexthdr.sh: Check IPv6 support before forcing it

[Petr Vorel]

Hi Cyril,

> Hi!
> > diff --git a/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh b/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh
> > index ec6643af66..cb4a3dd399 100755
> > --- a/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh
> > +++ b/testcases/network/stress/broken_ip/broken_ip-nexthdr.sh
> > @@ -11,6 +11,7 @@ TST_TESTFUNC="do_test"
> >  do_test()
> >  {
> >  	# not supported on IPv4
> > +	tst_net_require_ipv6
> >  	TST_IPV6=6
> >  	TST_IPVER=6

> I was looking at the code if we can simply instead do:

> [ "$TST_IPVER" == 6 ] && tst_net_require_ipv6

> in the test library, but it looks like the parameters are parsed in the
> tst_test.sh in the tst_run() function. Which means that the TST_IPVER is
> not actually set until the test starts and the library has no way of
> knowing the variable value beforehand.

> I guess that we can actually move the option parsing code in the
> tst_test.sh so that it happens just right after the script is sourced,
> which would make things much easier as the TST_IPVER would end up being
> defined in the tst_network.sh and we coud simply use the statement above
> without any further hacks like this patch adds.

> As a side effect we could clean up the test option parsing code since we
> do actually have two different getopts loop in the tst_test.sh library
> and as far as I can tell we can do just with one.

I tried to implement this [1] (or [2] if I force push), and using getopts the
end of the script only would require also to move loading '. tst_test.sh' at the
end of *all* the test scripts because setup functions needs to be loaded before
sourcing tst_test.sh. Do we want this?

Kind regards,
Petr

[1] https://github.com/pevik/ltp/commit/9d77e5d964751ea3f0c8d22c4265c0041e2ebbc9
[2] https://github.com/pevik/ltp/commits/tst_test.sh/cleanup-getopts

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH v2 2/3] broken_ip-nexthdr.sh: Check IPv6 support before forcing it

[Cyril Hrubis]

Hi!
> I tried to implement this [1] (or [2] if I force push), and using getopts the
> end of the script only would require also to move loading '. tst_test.sh' at the
> end of *all* the test scripts because setup functions needs to be loaded before
> sourcing tst_test.sh. Do we want this?

Sounds reasonable. I would go even further and put the line that sources
the tst_test.sh to be last line in the test.

-- 
Cyril Hrubis
chrubis@suse.cz

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH v2 2/3] broken_ip-nexthdr.sh: Check IPv6 support before forcing it

[Petr Vorel]

> Hi!
> > I tried to implement this [1] (or [2] if I force push), and using getopts the
> > end of the script only would require also to move loading '. tst_test.sh' at the
> > end of *all* the test scripts because setup functions needs to be loaded before
> > sourcing tst_test.sh. Do we want this?

> Sounds reasonable. I would go even further and put the line that sources
> the tst_test.sh to be last line in the test.
+1, I'll also update docs examples and mention it.

Kind regards,
Petr

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp