Skip to contentSkip to conversation

Add test for per-NS NFS client statistics

PATCH In review Martin Doucha 3 participants 13 messages Aug 30, 2024 – Sep 3, 2024

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/2 Add test for per-NS NFS client statistics

Check that /proc/net/rpc/nfs file exists in nested network namespaces.

---

Only do the minimal check here. If the file exists in namespaces, nfsstat01.sh will take care of functional testing.

 runtest/net.nfs                              |  2 ++
 testcases/network/nfs/nfsstat01/nfsstat02.sh | 23 ++++++++++++++++++++
 2 files changed, 25 insertions(+)
 create mode 100755 testcases/network/nfs/nfsstat01/nfsstat02.sh
runtest/net.nfs +2 -0
index 929868a7b..7f84457bc 100644
@@ -116,6 +116,8 @@ nfsstat01_v40_ip6t nfsstat01.sh -6 -v 4 -t tcp
 nfsstat01_v41_ip6t nfsstat01.sh -6 -v 4.1 -t tcp
 nfsstat01_v42_ip6t nfsstat01.sh -6 -v 4.2 -t tcp
 
+nfsstat02 nfsstat02.sh
+
 fsx_v30_ip4u fsx.sh -v 3 -t udp
 fsx_v30_ip4t fsx.sh -v 3 -t tcp
 fsx_v40_ip4t fsx.sh -v 4 -t tcp
testcases/network/nfs/nfsstat01/nfsstat02.sh new file +23 -0
new file mode 100755
index 000000000..1e1bebe97
@@ -0,0 +1,23 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0-or-later
+# Copyright (c) 2024 SUSE LLC <[email protected]>
+
+TST_TESTFUNC="do_test"
+
+# PURPOSE:  Check that /proc/net/rpc/nfs exists in nested network namespaces
+do_test()
+{
+	local procfile="/proc/net/rpc/nfs"
+
+	if tst_rhost_run -c "test -e '$procfile'"; then
+		tst_res TPASS "$procfile exists in net namespaces"
+	else
+		tst_res TFAIL "$procfile missing in net namespaces"
+	fi
+}
+
+# Force use of nested net namespace
+unset RHOST
+
+. nfs_lib.sh
+tst_run
Signature · 5 lines
2.46.0


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#1 lore raw

PATCH 2/2 nfsstat01: Check that RPC stats don't leak between net namespaces

When the NFS server and client run on the same host in different net namespaces, check that RPC calls from the client namespace don't change RPC statistics in the root namespace.

---

I've initially tried to test both NFS and RPC client stats but it appears that NFS client stats are still shared across all namespaces. Only RPC client stats are separate for each net namespace. The kernel patchset[1] which introduced per-NS stats confirms that only RPC stats have been changed.

If NFS client stats should be separate for each namespace as well, let me know and I'll return the second set of NS checks in patch v2.

Tested on kernel v5.14 with Neil's backports.

[1] https://lore.kernel.org/linux-nfs/[email protected]/

 testcases/network/nfs/nfsstat01/nfsstat01.sh | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)
testcases/network/nfs/nfsstat01/nfsstat01.sh +16 -2
index 8d7202cf3..3379c4d46 100755
@@ -22,6 +22,7 @@ get_calls()
 	local name=$1
 	local field=$2
 	local nfs_f=$3
+	local netns=${4:-rhost}
 	local type="lhost"
 	local calls opt
 
@@ -30,7 +31,8 @@ get_calls()
 
 	if tst_net_use_netns; then
 		# In netns setup, rhost is the client
-		[ "$nfs_f" = "nfs" ] && [ $NS_STAT_RHOST -ne 0 ] && type="rhost"
+		[ "$nfs_f" = "nfs" ] && [ $NS_STAT_RHOST -ne 0 ] && \
+			type="$netns"
 	else
 		[ "$nfs_f" != "nfs" ] && type="rhost"
 	fi
@@ -64,13 +66,14 @@ get_calls()
 do_test()
 {
 	local client_calls server_calls new_server_calls new_client_calls
-	local client_field server_field
+	local client_field server_field root_calls new_root_calls
 	local client_v=$VERSION server_v=$VERSION
 
 	tst_res TINFO "checking RPC calls for server/client"
 
 	server_calls="$(get_calls rpc 2 nfsd)"
 	client_calls="$(get_calls rpc 2 nfs)"
+	root_calls="$(get_calls rpc 2 nfs lhost)"
 
 	tst_res TINFO "calls $server_calls/$client_calls"
 
@@ -79,6 +82,7 @@ do_test()
 
 	new_server_calls="$(get_calls rpc 2 nfsd)"
 	new_client_calls="$(get_calls rpc 2 nfs)"
+	new_root_calls="$(get_calls rpc 2 nfs lhost)"
 	tst_res TINFO "new calls $new_server_calls/$new_client_calls"
 
 	if [ "$new_server_calls" -le "$server_calls" ]; then
@@ -93,6 +97,16 @@ do_test()
 		tst_res TPASS "client RPC calls increased"
 	fi
 
+	if [ $NS_STAT_RHOST -ne 0 ]; then
+		tst_res TINFO "Root NS client RPC calls: $root_calls => $new_root_calls"
+
+		if [ $root_calls -ne $new_root_calls ]; then
+			tst_res TFAIL "RPC stats leaked between net namespaces"
+		else
+			tst_res TPASS "RPC stats stay within net namespaces"
+		fi
+	fi
+
 	tst_res TINFO "checking NFS calls for server/client"
 	case $VERSION in
 	2) client_field=13 server_field=13
Signature · 5 lines
2.46.0


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#2 lore raw

Re: [LTP] [PATCH 2/2] nfsstat01: Check that RPC stats don't leak between net namespaces

Quoted 11 lines · Martin Doucha

When the NFS server and client run on the same host in different net namespaces, check that RPC calls from the client namespace don't change RPC statistics in the root namespace.

---

I've initially tried to test both NFS and RPC client stats but it appears that NFS client stats are still shared across all namespaces. Only RPC client stats are separate for each net namespace. The kernel patchset[1] which introduced per-NS stats confirms that only RPC stats have been changed.

I believe that is correct, Josef changed only RPC counters. Which counters did you expect also would be containerized, exactly? Perhaps this issue should be raised on linux-nfs@vger, it could be considered to be another information leak.

Quoted 76 lines

If NFS client stats should be separate for each namespace as well, let me know and I'll return the second set of NS checks in patch v2.

Tested on kernel v5.14 with Neil's backports.

[1] https://lore.kernel.org/linux-nfs/[email protected]/

 testcases/network/nfs/nfsstat01/nfsstat01.sh | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)
testcases/network/nfs/nfsstat01/nfsstat01.sh +16 -2
index 8d7202cf3..3379c4d46 100755
@@ -22,6 +22,7 @@ get_calls()
 	local name=$1
 	local field=$2
 	local nfs_f=$3
+	local netns=${4:-rhost}
 	local type="lhost"
 	local calls opt
 
@@ -30,7 +31,8 @@ get_calls()
 
 	if tst_net_use_netns; then
 		# In netns setup, rhost is the client
-		[ "$nfs_f" = "nfs" ] && [ $NS_STAT_RHOST -ne 0 ] && type="rhost"
+		[ "$nfs_f" = "nfs" ] && [ $NS_STAT_RHOST -ne 0 ] && \
+			type="$netns"
 	else
 		[ "$nfs_f" != "nfs" ] && type="rhost"
 	fi
@@ -64,13 +66,14 @@ get_calls()
 do_test()
 {
 	local client_calls server_calls new_server_calls new_client_calls
-	local client_field server_field
+	local client_field server_field root_calls new_root_calls
 	local client_v=$VERSION server_v=$VERSION
 
 	tst_res TINFO "checking RPC calls for server/client"
 
 	server_calls="$(get_calls rpc 2 nfsd)"
 	client_calls="$(get_calls rpc 2 nfs)"
+	root_calls="$(get_calls rpc 2 nfs lhost)"
 
 	tst_res TINFO "calls $server_calls/$client_calls"
 
@@ -79,6 +82,7 @@ do_test()
 
 	new_server_calls="$(get_calls rpc 2 nfsd)"
 	new_client_calls="$(get_calls rpc 2 nfs)"
+	new_root_calls="$(get_calls rpc 2 nfs lhost)"
 	tst_res TINFO "new calls $new_server_calls/$new_client_calls"
 
 	if [ "$new_server_calls" -le "$server_calls" ]; then
@@ -93,6 +97,16 @@ do_test()
 		tst_res TPASS "client RPC calls increased"
 	fi
 
+	if [ $NS_STAT_RHOST -ne 0 ]; then
+		tst_res TINFO "Root NS client RPC calls: $root_calls => $new_root_calls"
+
+		if [ $root_calls -ne $new_root_calls ]; then
+			tst_res TFAIL "RPC stats leaked between net namespaces"
+		else
+			tst_res TPASS "RPC stats stay within net namespaces"
+		fi
+	fi
+
 	tst_res TINFO "checking NFS calls for server/client"
 	case $VERSION in
 	2) client_field=13 server_field=13
Signature · 1 line
2.46.0
Signature · 4 lines
Chuck Lever

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#3 lore raw

Re: [LTP] [PATCH 2/2] nfsstat01: Check that RPC stats don't leak between net namespaces

Hi all,

Quoted 5 lines · Martin Doucha
Quoted 3 lines

When the NFS server and client run on the same host in different net namespaces, check that RPC calls from the client namespace don't change RPC statistics in the root namespace.

Quoted 7 lines
Quoted 7 lines

---

I've initially tried to test both NFS and RPC client stats but it appears that NFS client stats are still shared across all namespaces. Only RPC client stats are separate for each net namespace. The kernel patchset[1] which introduced per-NS stats confirms that only RPC stats have been changed.

Yes, only RPC client stats needed to be fixed in LTP test.

Kind regards,
Petr

Quoted 81 lines

I believe that is correct, Josef changed only RPC counters. Which counters did you expect also would be containerized, exactly? Perhaps this issue should be raised on linux-nfs@vger, it could be considered to be another information leak.

Quoted 75 lines

If NFS client stats should be separate for each namespace as well, let me know and I'll return the second set of NS checks in patch v2.

Tested on kernel v5.14 with Neil's backports.

[1] https://lore.kernel.org/linux-nfs/[email protected]/

 testcases/network/nfs/nfsstat01/nfsstat01.sh | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)
testcases/network/nfs/nfsstat01/nfsstat01.sh +16 -2
index 8d7202cf3..3379c4d46 100755
@@ -22,6 +22,7 @@ get_calls()
 	local name=$1
 	local field=$2
 	local nfs_f=$3
+	local netns=${4:-rhost}
 	local type="lhost"
 	local calls opt
 
@@ -30,7 +31,8 @@ get_calls()
 
 	if tst_net_use_netns; then
 		# In netns setup, rhost is the client
-		[ "$nfs_f" = "nfs" ] && [ $NS_STAT_RHOST -ne 0 ] && type="rhost"
+		[ "$nfs_f" = "nfs" ] && [ $NS_STAT_RHOST -ne 0 ] && \
+			type="$netns"
 	else
 		[ "$nfs_f" != "nfs" ] && type="rhost"
 	fi
@@ -64,13 +66,14 @@ get_calls()
 do_test()
 {
 	local client_calls server_calls new_server_calls new_client_calls
-	local client_field server_field
+	local client_field server_field root_calls new_root_calls
 	local client_v=$VERSION server_v=$VERSION
 
 	tst_res TINFO "checking RPC calls for server/client"
 
 	server_calls="$(get_calls rpc 2 nfsd)"
 	client_calls="$(get_calls rpc 2 nfs)"
+	root_calls="$(get_calls rpc 2 nfs lhost)"
 
 	tst_res TINFO "calls $server_calls/$client_calls"
 
@@ -79,6 +82,7 @@ do_test()
 
 	new_server_calls="$(get_calls rpc 2 nfsd)"
 	new_client_calls="$(get_calls rpc 2 nfs)"
+	new_root_calls="$(get_calls rpc 2 nfs lhost)"
 	tst_res TINFO "new calls $new_server_calls/$new_client_calls"
 
 	if [ "$new_server_calls" -le "$server_calls" ]; then
@@ -93,6 +97,16 @@ do_test()
 		tst_res TPASS "client RPC calls increased"
 	fi
 
+	if [ $NS_STAT_RHOST -ne 0 ]; then
+		tst_res TINFO "Root NS client RPC calls: $root_calls => $new_root_calls"
+
+		if [ $root_calls -ne $new_root_calls ]; then
+			tst_res TFAIL "RPC stats leaked between net namespaces"
+		else
+			tst_res TPASS "RPC stats stay within net namespaces"
+		fi
+	fi
+
 	tst_res TINFO "checking NFS calls for server/client"
 	case $VERSION in
 	2) client_field=13 server_field=13
Signature · 1 line
2.46.0
Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#4 lore raw

Re: [PATCH 2/2] nfsstat01: Check that RPC stats don't leak between net namespaces

Quoted 10 lines

Hi all,

Quoted 7 lines · Martin Doucha
Quoted 6 lines

When the NFS server and client run on the same host in different net namespaces, check that RPC calls from the client namespace don't change RPC statistics in the root namespace.

---

Quoted 6 lines
Quoted 4 lines
Quoted 4 lines

I've initially tried to test both NFS and RPC client stats but it appears that NFS client stats are still shared across all namespaces. Only RPC client stats are separate for each net namespace. The kernel patchset[1] which introduced per-NS stats confirms that only RPC stats have been changed.

Yes, only RPC client stats needed to be fixed in LTP test.

OTOH there is also nfsd stats namespaced [2] as a second part of whole "Make nfs and nfsd stats visible in network ns" patchset. I suppose we would need to reverse client and server to detect this (IMHO worth of doing it).

Kind regards,
Petr

[2] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/log/?qt=author&q=4b14885411f74b2b0ce0eb2b39d0fffe54e5ca0d
[3] https://lore.kernel.org/linux-nfs/[email protected]/

Quoted 84 lines

Kind regards,
Petr

Quoted 81 lines

I believe that is correct, Josef changed only RPC counters. Which counters did you expect also would be containerized, exactly? Perhaps this issue should be raised on linux-nfs@vger, it could be considered to be another information leak.

Quoted 75 lines

If NFS client stats should be separate for each namespace as well, let me know and I'll return the second set of NS checks in patch v2.

Tested on kernel v5.14 with Neil's backports.

[1] https://lore.kernel.org/linux-nfs/[email protected]/

 testcases/network/nfs/nfsstat01/nfsstat01.sh | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)
testcases/network/nfs/nfsstat01/nfsstat01.sh +16 -2
index 8d7202cf3..3379c4d46 100755
@@ -22,6 +22,7 @@ get_calls()
 	local name=$1
 	local field=$2
 	local nfs_f=$3
+	local netns=${4:-rhost}
 	local type="lhost"
 	local calls opt
 
@@ -30,7 +31,8 @@ get_calls()
 
 	if tst_net_use_netns; then
 		# In netns setup, rhost is the client
-		[ "$nfs_f" = "nfs" ] && [ $NS_STAT_RHOST -ne 0 ] && type="rhost"
+		[ "$nfs_f" = "nfs" ] && [ $NS_STAT_RHOST -ne 0 ] && \
+			type="$netns"
 	else
 		[ "$nfs_f" != "nfs" ] && type="rhost"
 	fi
@@ -64,13 +66,14 @@ get_calls()
 do_test()
 {
 	local client_calls server_calls new_server_calls new_client_calls
-	local client_field server_field
+	local client_field server_field root_calls new_root_calls
 	local client_v=$VERSION server_v=$VERSION
 
 	tst_res TINFO "checking RPC calls for server/client"
 
 	server_calls="$(get_calls rpc 2 nfsd)"
 	client_calls="$(get_calls rpc 2 nfs)"
+	root_calls="$(get_calls rpc 2 nfs lhost)"
 
 	tst_res TINFO "calls $server_calls/$client_calls"
 
@@ -79,6 +82,7 @@ do_test()
 
 	new_server_calls="$(get_calls rpc 2 nfsd)"
 	new_client_calls="$(get_calls rpc 2 nfs)"
+	new_root_calls="$(get_calls rpc 2 nfs lhost)"
 	tst_res TINFO "new calls $new_server_calls/$new_client_calls"
 
 	if [ "$new_server_calls" -le "$server_calls" ]; then
@@ -93,6 +97,16 @@ do_test()
 		tst_res TPASS "client RPC calls increased"
 	fi
 
+	if [ $NS_STAT_RHOST -ne 0 ]; then
+		tst_res TINFO "Root NS client RPC calls: $root_calls => $new_root_calls"
+
+		if [ $root_calls -ne $new_root_calls ]; then
+			tst_res TFAIL "RPC stats leaked between net namespaces"
+		else
+			tst_res TPASS "RPC stats stay within net namespaces"
+		fi
+	fi
+
 	tst_res TINFO "checking NFS calls for server/client"
 	case $VERSION in
 	2) client_field=13 server_field=13
Signature · 1 line
2.46.0
#5 lore raw

Re: [LTP] [PATCH 2/2] nfsstat01: Check that RPC stats don't leak between net namespaces

In reply to Chuck Lever via ltp
Quoted 13 lines · Chuck Lever
Quoted 11 lines · Martin Doucha

When the NFS server and client run on the same host in different net namespaces, check that RPC calls from the client namespace don't change RPC statistics in the root namespace.

---

I've initially tried to test both NFS and RPC client stats but it appears that NFS client stats are still shared across all namespaces. Only RPC client stats are separate for each net namespace. The kernel patchset[1] which introduced per-NS stats confirms that only RPC stats have been changed.

Quoted 4 lines

I believe that is correct, Josef changed only RPC counters. Which counters did you expect also would be containerized, exactly? Perhaps this issue should be raised on linux-nfs@vger, it could be considered to be another information leak.

I tried to test the NFS client call counters, fields 13, 15 or 24 (depending on NFS version) in the "procX" line of /proc/net/rpc/nfs. These are the counters that the test already checks after RPC.

Although when I think about it some more, I'm not sure whether the NFS/RPC client statistics should be attached to network namespaces in the first place. AFAICT, processes from any network namespace can trigger client calls for both RPC and NFS as long as they can access the NFS mountpoint. Perhaps a mount namespace would be the more logical domain for counting per-NS statistics instead?

Signature · 11 lines
Martin Doucha   [email protected]
SW Quality Engineer
SUSE LINUX, s.r.o.
CORSO IIa
Krizikova 148/34
186 00 Prague 8
Czech Republic


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#6 lore raw

Re: [LTP] [PATCH 2/2] nfsstat01: Check that RPC stats don't leak between net namespaces

Quoted 17 lines · Martin Doucha
Quoted 12 lines · Chuck Lever
Quoted 11 lines · Martin Doucha

When the NFS server and client run on the same host in different net namespaces, check that RPC calls from the client namespace don't change RPC statistics in the root namespace.

---

I've initially tried to test both NFS and RPC client stats but it appears that NFS client stats are still shared across all namespaces. Only RPC client stats are separate for each net namespace. The kernel patchset[1] which introduced per-NS stats confirms that only RPC stats have been changed.

Quoted 8 lines
Quoted 4 lines

I believe that is correct, Josef changed only RPC counters. Which counters did you expect also would be containerized, exactly? Perhaps this issue should be raised on linux-nfs@vger, it could be considered to be another information leak.

I tried to test the NFS client call counters, fields 13, 15 or 24 (depending on NFS version) in the "procX" line of /proc/net/rpc/nfs. These are the counters that the test already checks after RPC.

Although when I think about it some more, I'm not sure whether the NFS/RPC client statistics should be attached to network namespaces in the first place. AFAICT, processes from any network namespace can trigger client calls for both RPC and NFS as long as they can access the NFS mountpoint. Perhaps a mount namespace would be the more logical domain for counting per-NS statistics instead?

Disclaimer: I'm not one of the NFS client maintainers, but only a very long time contributor to the Linux NFS implementation, so I can offer only a somewhat-educated opinion on this topic.

IIRC in a container, the RPC client is bound to the network namespace.

The statistics in /proc/self/mountstats are accrued to an individual mount. I think those are associated with the mount namespace. I could very well be wrong about that.

This is another topic that would be appropriate to bring to linux-nfs@ .

--
Chuck Lever

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#7 lore raw

Re: [LTP] [PATCH 2/2] nfsstat01: Check that RPC stats don't leak between net namespaces

Quoted 6 lines · Chuck Lever III

Disclaimer: I'm not one of the NFS client maintainers, but only a very long time contributor to the Linux NFS implementation, so I can offer only a somewhat-educated opinion on this topic.

IIRC in a container, the RPC client is bound to the network namespace.

I'm not a NFS expert either. Just thinking out loud because I've noticed that the nfsstat01 test triggers RPC calls from outside the client net namespace.

Signature · 11 lines
Martin Doucha   [email protected]
SW Quality Engineer
SUSE LINUX, s.r.o.
CORSO IIa
Krizikova 148/34
186 00 Prague 8
Czech Republic


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
Reviewed-by Tested-by on v1 2/2
#8 lore raw

Re: [LTP] [PATCH 2/2] nfsstat01: Check that RPC stats don't leak between net namespaces

In reply to Martin Doucha
Quoted 70 lines

When the NFS server and client run on the same host in different net namespaces, check that RPC calls from the client namespace don't change RPC statistics in the root namespace.

---

I've initially tried to test both NFS and RPC client stats but it appears that NFS client stats are still shared across all namespaces. Only RPC client stats are separate for each net namespace. The kernel patchset[1] which introduced per-NS stats confirms that only RPC stats have been changed.

If NFS client stats should be separate for each namespace as well, let me know and I'll return the second set of NS checks in patch v2.

Tested on kernel v5.14 with Neil's backports.

[1] https://lore.kernel.org/linux-nfs/[email protected]/

 testcases/network/nfs/nfsstat01/nfsstat01.sh | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)
testcases/network/nfs/nfsstat01/nfsstat01.sh +6 -2
index 8d7202cf3..3379c4d46 100755
@@ -22,6 +22,7 @@ get_calls()
 	local name=$1
 	local field=$2
 	local nfs_f=$3
+	local netns=${4:-rhost}
 	local type="lhost"
 	local calls opt
 
@@ -30,7 +31,8 @@ get_calls()
 
 	if tst_net_use_netns; then
 		# In netns setup, rhost is the client
-		[ "$nfs_f" = "nfs" ] && [ $NS_STAT_RHOST -ne 0 ] && type="rhost"
+		[ "$nfs_f" = "nfs" ] && [ $NS_STAT_RHOST -ne 0 ] && \
+			type="$netns"
 	else
 		[ "$nfs_f" != "nfs" ] && type="rhost"
 	fi
@@ -64,13 +66,14 @@ get_calls()
 do_test()
 {
 	local client_calls server_calls new_server_calls new_client_calls
-	local client_field server_field
+	local client_field server_field root_calls new_root_calls
 	local client_v=$VERSION server_v=$VERSION
 
 	tst_res TINFO "checking RPC calls for server/client"
 
 	server_calls="$(get_calls rpc 2 nfsd)"
 	client_calls="$(get_calls rpc 2 nfs)"
+	root_calls="$(get_calls rpc 2 nfs lhost)"
 
 	tst_res TINFO "calls $server_calls/$client_calls"
 
@@ -79,6 +82,7 @@ do_test()
 
 	new_server_calls="$(get_calls rpc 2 nfsd)"
 	new_client_calls="$(get_calls rpc 2 nfs)"
+	new_root_calls="$(get_calls rpc 2 nfs lhost)"
 	tst_res TINFO "new calls $new_server_calls/$new_client_calls"
 
 	if [ "$new_server_calls" -le "$server_calls" ]; then
@@ -93,6 +97,16 @@ do_test()
 		tst_res TPASS "client RPC calls increased"

testcases/network/nfs/nfsstat01/nfsstat01.sh do_test() · line 107

Quoted 10 lines
testcases/network/nfs/nfsstat01/nfsstat01.sh +8 -0
 	fi
 
+	if [ $NS_STAT_RHOST -ne 0 ]; then
+		tst_res TINFO "Root NS client RPC calls: $root_calls => $new_root_calls"
+
+		if [ $root_calls -ne $new_root_calls ]; then
+			tst_res TFAIL "RPC stats leaked between net namespaces"
+		else
+			tst_res TPASS "RPC stats stay within net namespaces"
+		fi

Maybe also add TCONF message? (can be added before merge)

else
tst_res TCONF "Not testing leak between root NS and net NS due old kernel"

Thanks for this test!

Kind regards,
Petr

Quoted 5 lines
testcases/network/nfs/nfsstat01/nfsstat01.sh +2 -0
+	fi
+
 	tst_res TINFO "checking NFS calls for server/client"
 	case $VERSION in
 	2) client_field=13 server_field=13
Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#9 lore raw

Re: [LTP] [PATCH 2/2] nfsstat01: Check that RPC stats don't leak between net namespaces

Hi,

Quoted 15 lines · Petr Vorel

testcases/network/nfs/nfsstat01/nfsstat01.sh do_test() · line 107

Quoted 12 lines
testcases/network/nfs/nfsstat01/nfsstat01.sh +8 -0
@@ -93,6 +97,16 @@ do_test()
  		tst_res TPASS "client RPC calls increased"
  	fi
 
+	if [ $NS_STAT_RHOST -ne 0 ]; then
+		tst_res TINFO "Root NS client RPC calls: $root_calls => $new_root_calls"
+
+		if [ $root_calls -ne $new_root_calls ]; then
+			tst_res TFAIL "RPC stats leaked between net namespaces"
+		else
+			tst_res TPASS "RPC stats stay within net namespaces"
+		fi

Maybe also add TCONF message? (can be added before merge)

Quoted 2 lines

else
tst_res TCONF "Not testing leak between root NS and net NS due old kernel"

I think the TCONF message doesn't make sense here. There are several cases where the new check will be skipped:
1) the NFS server runs on another machine ($RHOST is not empty)
2) the test is configure to ignore namespaces ($LTP_NFS_NETNS_USE_LO is not empty)
3) /proc/net/rpc/nfs doesn't exist in nested net namespaces

You want to print the TCONF message only in case #3. Let's keep the condition above simple.

Signature · 11 lines
Martin Doucha   [email protected]
SW Quality Engineer
SUSE LINUX, s.r.o.
CORSO IIa
Krizikova 148/34
186 00 Prague 8
Czech Republic


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#10 lore raw

Re: [LTP] [PATCH 2/2] nfsstat01: Check that RPC stats don't leak between net namespaces

Hi Martin,

Quoted 28 lines

Hi,

Quoted 15 lines · Petr Vorel

testcases/network/nfs/nfsstat01/nfsstat01.sh do_test() · line 107

Quoted 12 lines
testcases/network/nfs/nfsstat01/nfsstat01.sh +8 -0
@@ -93,6 +97,16 @@ do_test()
  		tst_res TPASS "client RPC calls increased"
  	fi
 
+	if [ $NS_STAT_RHOST -ne 0 ]; then
+		tst_res TINFO "Root NS client RPC calls: $root_calls => $new_root_calls"
+
+		if [ $root_calls -ne $new_root_calls ]; then
+			tst_res TFAIL "RPC stats leaked between net namespaces"
+		else
+			tst_res TPASS "RPC stats stay within net namespaces"
+		fi

Maybe also add TCONF message? (can be added before merge)

Quoted 2 lines

else
tst_res TCONF "Not testing leak between root NS and net NS due old kernel"

I think the TCONF message doesn't make sense here. There are several cases where the new check will be skipped:
1) the NFS server runs on another machine ($RHOST is not empty)
2) the test is configure to ignore namespaces ($LTP_NFS_NETNS_USE_LO is not empty)
3) /proc/net/rpc/nfs doesn't exist in nested net namespaces

Quoted 2 lines

You want to print the TCONF message only in case #3. Let's keep the condition above simple.

Fair enough. I merged just with adding kernel commit and patchset link in the first commit. Thanks!

NOTE: I noticed your point about mount namespaces (interesting I wonder if there will be some change), but merged as it reflects the current kernel code.

Kind regards,
Petr

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
Reviewed-by Tested-by on v1 1/2
#11 lore raw
In reply to Martin Doucha

Hi Martin,

Quoted 32 lines

Check that /proc/net/rpc/nfs file exists in nested network namespaces.

---

Only do the minimal check here. If the file exists in namespaces, nfsstat01.sh will take care of functional testing.

 runtest/net.nfs                              |  2 ++
 testcases/network/nfs/nfsstat01/nfsstat02.sh | 23 ++++++++++++++++++++
 2 files changed, 25 insertions(+)
 create mode 100755 testcases/network/nfs/nfsstat01/nfsstat02.sh
runtest/net.nfs +2 -0
index 929868a7b..7f84457bc 100644
@@ -116,6 +116,8 @@ nfsstat01_v40_ip6t nfsstat01.sh -6 -v 4 -t tcp
 nfsstat01_v41_ip6t nfsstat01.sh -6 -v 4.1 -t tcp
 nfsstat01_v42_ip6t nfsstat01.sh -6 -v 4.2 -t tcp
 
+nfsstat02 nfsstat02.sh
+
 fsx_v30_ip4u fsx.sh -v 3 -t udp
 fsx_v30_ip4t fsx.sh -v 3 -t tcp
 fsx_v40_ip4t fsx.sh -v 4 -t tcp
testcases/network/nfs/nfsstat01/nfsstat02.sh new file +0 -0
new file mode 100755
index 000000000..1e1bebe97

testcases/network/nfs/nfsstat01/nfsstat02.sh line 7

Quoted 7 lines
testcases/network/nfs/nfsstat01/nfsstat02.sh new file +7 -0
@@ -0,0 +1,23 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0-or-later
+# Copyright (c) 2024 SUSE LLC <[email protected]>
+
+TST_TESTFUNC="do_test"
+
+# PURPOSE:  Check that /proc/net/rpc/nfs exists in nested network namespaces

I would point here a commit which added it or a patchset.

Shell API does not have functionality to point out missing kernel git commit, but that might change. But even without it a comment is useful.

Maybe point out d47151b79e32 ("nfs: expose /proc/net/sunrpc/nfs in net namespaces") and whole patchset
https://lore.kernel.org/linux-nfs/[email protected]/

Tested on 6.8 and 6.11 mainline

Thanks for this test!

Kind regards,
Petr

Quoted 16 lines
testcases/network/nfs/nfsstat01/nfsstat02.sh new file +16 -0
+do_test()
+{
+	local procfile="/proc/net/rpc/nfs"
+
+	if tst_rhost_run -c "test -e '$procfile'"; then
+		tst_res TPASS "$procfile exists in net namespaces"
+	else
+		tst_res TFAIL "$procfile missing in net namespaces"
+	fi
+}
+
+# Force use of nested net namespace
+unset RHOST
+
+. nfs_lib.sh
+tst_run
#12 lore raw

Hi,

Quoted 11 lines · Petr Vorel

Hi Martin,

Quoted 1 line

+# PURPOSE: Check that /proc/net/rpc/nfs exists in nested network namespaces

I would point here a commit which added it or a patchset.

Shell API does not have functionality to point out missing kernel git commit, but that might change. But even without it a comment is useful.

Maybe point out d47151b79e32 ("nfs: expose /proc/net/sunrpc/nfs in net namespaces") and whole patchset
https://lore.kernel.org/linux-nfs/[email protected]/

Feel free to add the comment during merge.

Signature · 7 lines
Martin Doucha   [email protected]
SW Quality Engineer
SUSE LINUX, s.r.o.
CORSO IIa
Krizikova 148/34
186 00 Prague 8
Czech Republic