diff mbox series

[4/8] tests: Disable SPP A-MSDU in several tests

Message ID 20250107125149.1820266-4-andrei.otcheretianski@intel.com
State Deferred
Headers show
Series [1/8] defs: Add RSNXE SPP A-MSDU capable definition | expand

Commit Message

Andrei Otcheretianski Jan. 7, 2025, 12:51 p.m. UTC 
From: Daniel Gabay <daniel.gabay@intel.com>

1. Some of the tests under test_ap_psk/test_ap_ft/test_ocv, do not
   expect the use of RSNXE or expecting hardcoded RSNXE value.
   Enabling SPP A-MSDU by default in a future patch that implements
   this feature may lead to failures. To prevent this, disable
   SPP A-MSDU.

2. test_ap_wds_sta and test_ap_pmf_toggle uses
   reassoc_same_bss_optim=1. After reassociation, the 4-way handshake
   fails due to the absence of RSNXE. This appears to be a general
   bug. Until a fix is implemented, disable SPP A-MSDU to prevent
   unexpected failures.

Signed-off-by: Daniel Gabay <daniel.gabay@intel.com>
---
 tests/hwsim/test_ap_ft.py     | 2 ++
 tests/hwsim/test_ap_params.py | 4 ++++
 tests/hwsim/test_ap_pmf.py    | 4 ++++
 tests/hwsim/test_ap_psk.py    | 2 ++
 tests/hwsim/test_ocv.py       | 2 ++
 5 files changed, 14 insertions(+)

Comments

Jouni Malinen Jan. 26, 2025, 5:35 p.m. UTC | #1 
On Tue, Jan 07, 2025 at 02:51:45PM +0200, Andrei Otcheretianski wrote:
> 1. Some of the tests under test_ap_psk/test_ap_ft/test_ocv, do not
>    expect the use of RSNXE or expecting hardcoded RSNXE value.
>    Enabling SPP A-MSDU by default in a future patch that implements
>    this feature may lead to failures. To prevent this, disable
>    SPP A-MSDU.

I applied other patches in the series (i.e., 1-3, 5-8) with some cleanup
and fixes. However, I ended up disabling SPP A-MSDU support in hostapd
by default since I do not want to force known interoperability issues
with deployed STAs having issues with the RSNXE to occur in the default
configuration. Furthermore, there has been very limited interoperability
testing of SPP A-MSDU so far, so I'd also want that to change before
promoting this functionality significantly.

With that change in the spp_amsdu default, this patch 4 is not needed.

> 2. test_ap_wds_sta and test_ap_pmf_toggle uses
>    reassoc_same_bss_optim=1. After reassociation, the 4-way handshake
>    fails due to the absence of RSNXE. This appears to be a general
>    bug. Until a fix is implemented, disable SPP A-MSDU to prevent
>    unexpected failures.

I did not see these test cases failing when enabling SPP A-MSDU support.
diff mbox series

Patch

diff --git a/tests/hwsim/test_ap_ft.py b/tests/hwsim/test_ap_ft.py
index bbbe8cf221..be0498bbf0 100644
--- a/tests/hwsim/test_ap_ft.py
+++ b/tests/hwsim/test_ap_ft.py
@@ -52,6 +52,8 @@  def ft_params(rsn=True, ssid=None, passphrase=None, mobility_domain="a1b2"):
     params["r0_key_lifetime"] = "10000"
     params["pmk_r1_push"] = "1"
     params["reassociation_deadline"] = "1000"
+    # disable spp_amsdu (may affect some tests by add/modify unexpected RSNXE)
+    params['spp_amsdu'] = '0'
     return params
 
 def ft_params1a(rsn=True, ssid=None, passphrase=None):
diff --git a/tests/hwsim/test_ap_params.py b/tests/hwsim/test_ap_params.py
index 1eeebbd4ed..a8553e597f 100644
--- a/tests/hwsim/test_ap_params.py
+++ b/tests/hwsim/test_ap_params.py
@@ -296,6 +296,10 @@  def test_ap_wds_sta(dev, apdev):
     params = hostapd.wpa2_params(ssid=ssid, passphrase=passphrase)
     params['wds_sta'] = "1"
     params['wds_bridge'] = "wds-br0"
+    # FIXME: there is a bug when using reassoc_same_bss_optim=1 which
+    # cause the RSNXE to be omitted in the 4-way-handshake after reassoc,
+    # disable spp_amsdu until fixed.
+    params["spp_amsdu"] = "0"
     hapd = hostapd.add_ap(apdev[0], params)
 
     try:
diff --git a/tests/hwsim/test_ap_pmf.py b/tests/hwsim/test_ap_pmf.py
index 331ac393ba..7da15f27a2 100644
--- a/tests/hwsim/test_ap_pmf.py
+++ b/tests/hwsim/test_ap_pmf.py
@@ -842,6 +842,10 @@  def _test_ap_pmf_toggle(dev, apdev):
     params["ieee80211w"] = "1"
     params["assoc_sa_query_max_timeout"] = "1"
     params["assoc_sa_query_retry_timeout"] = "1"
+    # FIXME: there is a bug when using reassoc_same_bss_optim=1 which
+    # cause the RSNXE to be omitted in the 4-way-handshake after reassoc,
+    # disable spp_amsdu until fixed.
+    params["spp_amsdu"] = "0"
     hapd = hostapd.add_ap(apdev[0], params)
     Wlantest.setup(hapd)
     wt = Wlantest()
diff --git a/tests/hwsim/test_ap_psk.py b/tests/hwsim/test_ap_psk.py
index 84b08d32b0..100a4aafb8 100644
--- a/tests/hwsim/test_ap_psk.py
+++ b/tests/hwsim/test_ap_psk.py
@@ -1527,6 +1527,8 @@  def eapol_test(apdev, dev, wpa2=True, ieee80211w=0):
         params = hostapd.wpa_params(ssid=ssid)
     params['wpa_psk'] = psk
     params['ieee80211w'] = str(ieee80211w)
+    # disable spp_amsdu (may affect some tests by add/modify unexpected RSNXE)
+    params['spp_amsdu'] = '0'
     hapd = hostapd.add_ap(apdev, params)
     hapd.request("SET ext_eapol_frame_io 1")
     dev.request("SET ext_eapol_frame_io 1")
diff --git a/tests/hwsim/test_ocv.py b/tests/hwsim/test_ocv.py
index 32af729313..c4b3db4461 100644
--- a/tests/hwsim/test_ocv.py
+++ b/tests/hwsim/test_ocv.py
@@ -374,6 +374,7 @@  class APConnection:
                                           passphrase=self.passphrase))
         params["wpa_pairwise_update_count"] = "10"
         params["ocv"] = ap_ocv
+        params['spp_amsdu'] = "0"
         try:
             self.hapd = hostapd.add_ap(apdev, params)
         except Exception as e:
@@ -748,6 +749,7 @@  class STAConnection:
         params.update(hostapd.wpa2_params(ssid=self.ssid,
                                           passphrase=self.passphrase))
         params['wpa_pairwise_update_count'] = "10"
+        params['spp_amsdu'] = "0"
 
         try:
             self.hapd = hostapd.add_ap(apdev, params)