Skip to contentSkip to conversation

knownnetworks: copy network_config on update

PATCH In review James Prestwood 2 participants 10 messages Jul 29, 2021

lore
Share & tools
lore
#0 lore raw

PATCH 1/9 knownnetworks: copy network_config on update

The network_config was not being copied to network_info when updated. This caused any new settings to be lost if the network configuration file was updated during runtime.

---
 src/knownnetworks.c | 2 ++
 1 file changed, 2 insertions(+)
src/knownnetworks.c +2 -0
index 09acdd3c..85648ead 100644
@@ -464,6 +464,8 @@ void known_network_update(struct network_info *network,
 	}
 
 	known_network_set_autoconnect(network, new->is_autoconnectable);
+
+	memcpy(&network->config, new, sizeof(struct network_config));
 }
 
 bool known_networks_foreach(known_networks_foreach_func_t function,
Signature · 1 line
2.31.1
#1 lore raw

PATCH 2/9 auto-t: hostapd: catch local socket remove exception

This happens if hostapd is killed ungracefully which is done in some tests.

---
 autotests/util/hostapd.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
autotests/util/hostapd.py +5 -1
index 4e694c84..6031b1dc 100644
@@ -118,7 +118,11 @@ class HostapdCLI:
             return
 
         self.ctrl_sock.close()
-        os.remove(self.local_ctrl)
+
+        try:
+            os.remove(self.local_ctrl)
+        except:
+            pass
 
         if self._hostapd_restarted:
             ctx.stop_process(ctx.hostapd.process, force)
Signature · 1 line
2.31.1
#2 lore raw

PATCH 3/9 test-runner: remove TLS debugging artifact

In reply to James Prestwood

When logging is enabled TLS debugging is turned on which creates a PEM file during runtime. There is no way for IWD itself to clean this up since its meant to be there for debugging.

---
 tools/test-runner | 4 ++++
 1 file changed, 4 insertions(+)
tools/test-runner +4 -0
index 98164f7a..fa3c434f 100755
@@ -1192,6 +1192,10 @@ def post_test(ctx, to_copy):
 				dbg("\n")
 				os.remove('/tmp/' + f)
 
+	# Special case for when logging is enabled
+	if os.path.isfile('/tmp/iwd-tls-debug-server-cert.pem'):
+		os.remove('/tmp/iwd-tls-debug-server-cert.pem')
+
 	allowed = ['phonesim.conf', 'certs', 'secrets', 'iwd']
 	for f in [f for f in os.listdir('/tmp') if f not in allowed]:
 		dbg("File %s was not cleaned up!" % f)
Signature · 1 line
2.31.1
#3 lore raw

PATCH 4/9 test-runner: set msize for --monitor

In reply to James Prestwood

This was added for logging but --monitor suffers the same warning without setting msize as well.

---
 tools/test-runner | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/test-runner +1 -1
index fa3c434f..2d4b6b5a 100755
@@ -1379,7 +1379,7 @@ def run_tests():
 				os.remove(f)
 	elif args.monitor:
 		parent = os.path.abspath(os.path.join(args.monitor, os.pardir))
-		mount('mondir', parent, '9p', 0, 'trans=virtio,version=9p2000.L')
+		mount('mondir', parent, '9p', 0, 'trans=virtio,version=9p2000.L,msize=10240')
 
 	if config.ctx.args.unit_tests is None:
 		run_auto_tests(config.ctx, args)
Signature · 1 line
2.31.1
#4 lore raw

PATCH 5/9 test-runner: increase RAM to 384MB

In reply to James Prestwood

Occationally python will fatally terminate trying to load a test using importlib with an out of memory exception. Increasing RAM allows reliable exection of all tests.

---
 tools/test-runner | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/test-runner +1 -1
index 2d4b6b5a..ff7d9586 100755
@@ -1585,7 +1585,7 @@ class Main:
 		if self.args.valgrind:
 			ram = 512
 		else:
-			ram = 256
+			ram = 384
 
 		print("Using %d cores, %d RAM for VM" % (smp, ram))
 
Signature · 1 line
2.31.1
#5 lore raw

PATCH 6/9 test-runner: don't reset dbus object on reset()

In reply to James Prestwood

Doing this is redundant since the dbus object is initialized anyways when a new test starts.

---
 tools/test-runner | 3 ---
 1 file changed, 3 deletions(-)
tools/test-runner +0 -3
index ff7d9586..cb97657e 100755
@@ -593,9 +593,6 @@ class Namespace:
 
 		self.radios = []
 
-		if self.name == "root":
-			self._bus = dbus.bus.BusConnection(address_or_type=self.dbus_address)
-
 		for p in list(self.processes):
 			print("Killing process %s" % p.name)
 			p.kill()
Signature · 1 line
2.31.1
#6 lore raw

PATCH 7/9 auto-t: remove unneeded hidden networks

In reply to James Prestwood

testScan was creating 10 separate hidden networks which sometimes bogged down hostapd to the point that it would not start up in time before test-runner's timeouts fired. This appeared to be due to hostapd needing to create 10 separate interfaces which would sometimes fail with -ENFILE.

The test itself only needed two separate networks, so instead the additional 8 can be completely removed.

---
 autotests/testScan/hw.conf                | 11 +++---
 autotests/testScan/periodic_scan_test.py  | 12 ++-----
 autotests/testScan/requested_scan_test.py |  8 -----
 autotests/testScan/ssid_hidden.conf       | 41 -----------------------
 autotests/testScan/ssid_hidden_0.conf     |  4 +++
 autotests/testScan/ssid_hidden_1.conf     |  4 +++
 autotests/testScan/ssid_hidden_2.open     |  3 --
 autotests/testScan/ssid_hidden_3.open     |  3 --
 autotests/testScan/ssid_hidden_4.open     |  3 --
 autotests/testScan/ssid_hidden_5.open     |  3 --
 autotests/testScan/ssid_hidden_6.open     |  3 --
 autotests/testScan/ssid_hidden_7.open     |  3 --
 autotests/testScan/ssid_hidden_8.open     |  3 --
 autotests/testScan/ssid_hidden_9.open     |  3 --
 14 files changed, 16 insertions(+), 88 deletions(-)
 delete mode 100644 autotests/testScan/ssid_hidden.conf
 create mode 100644 autotests/testScan/ssid_hidden_0.conf
 create mode 100644 autotests/testScan/ssid_hidden_1.conf
 delete mode 100644 autotests/testScan/ssid_hidden_2.open
 delete mode 100644 autotests/testScan/ssid_hidden_3.open
 delete mode 100644 autotests/testScan/ssid_hidden_4.open
 delete mode 100644 autotests/testScan/ssid_hidden_5.open
 delete mode 100644 autotests/testScan/ssid_hidden_6.open
 delete mode 100644 autotests/testScan/ssid_hidden_7.open
 delete mode 100644 autotests/testScan/ssid_hidden_8.open
 delete mode 100644 autotests/testScan/ssid_hidden_9.open
autotests/testScan/hw.conf +6 -5
index b2464baf..660c874e 100644
@@ -1,9 +1,10 @@
 [SETUP]
-num_radios=5
+num_radios=6
 start_iwd=0
 
 [HOSTAPD]
-rad0=ssid_hidden.conf
-rad1=ssid_open.conf
-rad2=ssid_8021x.conf
-rad3=ssid_psk.conf
+rad0=ssid_hidden_0.conf
+rad1=ssid_hidden_1.conf
+rad2=ssid_open.conf
+rad3=ssid_8021x.conf
+rad4=ssid_psk.conf
autotests/testScan/periodic_scan_test.py +2 -10
index 3252f79f..439b844d 100644
@@ -12,11 +12,7 @@ class Test(unittest.TestCase):
         'ssid_open': False,
         'ssid_psk': False,
         'ssid_8021x': False,
-        'ssid_hidden_5': False,
-        'ssid_hidden_6': False,
-        'ssid_hidden_7': False,
-        'ssid_hidden_8': False,
-        'ssid_hidden_9': False,
+        'ssid_hidden_1': False,
     }
 
     def set_network(self, ssid):
@@ -61,11 +57,7 @@ class Test(unittest.TestCase):
 
     @classmethod
     def setUpClass(cls):
-        IWD.copy_to_storage('ssid_hidden_5.open')
-        IWD.copy_to_storage('ssid_hidden_6.open')
-        IWD.copy_to_storage('ssid_hidden_7.open')
-        IWD.copy_to_storage('ssid_hidden_8.open')
-        IWD.copy_to_storage('ssid_hidden_9.open')
+        IWD.copy_to_storage('ssid_hidden_1.open')
 
     @classmethod
     def tearDownClass(cls):
autotests/testScan/requested_scan_test.py +0 -8
index 3858b4f5..69297f27 100644
@@ -15,10 +15,6 @@ class Test(unittest.TestCase):
         'ssid_psk': False,
         'ssid_8021x': False,
         'ssid_hidden_0': False,
-        'ssid_hidden_1': False,
-        'ssid_hidden_2': False,
-        'ssid_hidden_3': False,
-        'ssid_hidden_4': False,
     }
 
     def set_network(self, ssid):
@@ -63,10 +59,6 @@ class Test(unittest.TestCase):
     @classmethod
     def setUpClass(cls):
         IWD.copy_to_storage('ssid_hidden_0.open')
-        IWD.copy_to_storage('ssid_hidden_1.open')
-        IWD.copy_to_storage('ssid_hidden_2.open')
-        IWD.copy_to_storage('ssid_hidden_3.open')
-        IWD.copy_to_storage('ssid_hidden_4.open')
 
     @classmethod
     def tearDownClass(cls):
autotests/testScan/ssid_hidden.conf deleted +0 -41

41 lines not on this page · raw message

autotests/testScan/ssid_hidden_0.conf new file +4 -0
new file mode 100644
index 00000000..5d15e360
@@ -0,0 +1,4 @@
+ssid=ssid_hidden_0
+hw_mode=g
+channel=1
+ignore_broadcast_ssid=1
autotests/testScan/ssid_hidden_1.conf new file +4 -0
new file mode 100644
index 00000000..cc11803f
@@ -0,0 +1,4 @@
+ssid=ssid_hidden_1
+hw_mode=g
+channel=1
+ignore_broadcast_ssid=1
autotests/testScan/ssid_hidden_2.open deleted +0 -3
deleted file mode 100644
index 28622594..00000000
@@ -1,3 +0,0 @@
-[Settings]
-AutoConnect=false
-Hidden=true
autotests/testScan/ssid_hidden_3.open deleted +0 -3
deleted file mode 100644
index 28622594..00000000
@@ -1,3 +0,0 @@
-[Settings]
-AutoConnect=false
-Hidden=true
autotests/testScan/ssid_hidden_4.open deleted +0 -3
deleted file mode 100644
index 28622594..00000000
@@ -1,3 +0,0 @@
-[Settings]
-AutoConnect=false
-Hidden=true
autotests/testScan/ssid_hidden_5.open deleted +0 -3
deleted file mode 100644
index 28622594..00000000
@@ -1,3 +0,0 @@
-[Settings]
-AutoConnect=false
-Hidden=true
autotests/testScan/ssid_hidden_6.open deleted +0 -3
deleted file mode 100644
index 28622594..00000000
@@ -1,3 +0,0 @@
-[Settings]
-AutoConnect=false
-Hidden=true
autotests/testScan/ssid_hidden_7.open deleted +0 -3
deleted file mode 100644
index 28622594..00000000
@@ -1,3 +0,0 @@
-[Settings]
-AutoConnect=false
-Hidden=true
autotests/testScan/ssid_hidden_8.open deleted +0 -3
deleted file mode 100644
index 28622594..00000000
@@ -1,3 +0,0 @@
-[Settings]
-AutoConnect=false
-Hidden=true
autotests/testScan/ssid_hidden_9.open deleted +0 -3
deleted file mode 100644
index 28622594..00000000
@@ -1,3 +0,0 @@
-[Settings]
-AutoConnect=false
-Hidden=true
Signature · 1 line
2.31.1
#7 lore raw

PATCH 8/9 auto-t: add regular 8021x AKM to FT-FILS tests

In reply to James Prestwood

After some code changes the FT-FILS AKM was no longer selectable inside network_can_connect_bss. This normally shouldn't matter since station ends up selecting the AKM explicitly, including passing the fils_hint, but since the autotests only included FT-FILS AKMs this caused the transition to fail with no available BSS's.

To fix this the standard 8021x AKM was added to the hostapd configs. This allows these BSS's to be selected when attempting to roam, but since FT-FILS is the only other AKM it will be used for the actual transition.

---
 autotests/testFT-FILS-SHA256/ft-eap-ccmp-2.conf | 2 +-
 autotests/testFT-FILS-SHA384/ft-eap-ccmp-2.conf | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
autotests/testFT-FILS-SHA256/ft-eap-ccmp-2.conf +1 -1
index 24266ea9..585f1ca1 100644
@@ -9,7 +9,7 @@ nas_identifier=dummy2
 wpa=2
 # Can support WPA-EAP and FT-EAP (space separated list) at the same
 # time but we want to force FT
-wpa_key_mgmt=FT-FILS-SHA256
+wpa_key_mgmt=WPA-EAP FT-FILS-SHA256
 wpa_pairwise=CCMP
 ieee8021x=1
 
autotests/testFT-FILS-SHA384/ft-eap-ccmp-2.conf +1 -1
index 8af73cbc..a6bdf641 100644
@@ -9,7 +9,7 @@ nas_identifier=dummy2
 wpa=2
 # Can support WPA-EAP and FT-EAP (space separated list) at the same
 # time but we want to force FT
-wpa_key_mgmt=FT-FILS-SHA384
+wpa_key_mgmt=WPA-EAP FT-FILS-SHA384
 wpa_pairwise=CCMP
 ieee8021x=1
 
Signature · 1 line
2.31.1
#8 lore raw

PATCH 9/9 test-runner: refactor process output code

In reply to James Prestwood

The Process class requires the ability to write out any processes output to stdout, logging, or an explicit file, as well as store it inside python for processing by test utilities. To accomplish this each process was given a temporary file to write to, and that file had an IO watch set on it. Any data that was written was then read, and re-written out to where it needed to go. This ended up being very buggy and quite complex due to needing to mess with read/write pointers inside the file.

Popen already creates pipes to stdout if told, and they are accessable via the p.stdout. Its then as simple as setting an IO watch on that pipe and keeping the same code for reading out new data and writing it to any files we want. This greatly reduces the complexity.

---
 tools/test-runner | 77 +++++++++++++++++------------------------------
 1 file changed, 27 insertions(+), 50 deletions(-)
tools/test-runner +27 -50
index cb97657e..75e46444 100755
@@ -175,24 +175,12 @@ class Process:
 		self.io_watch = None
 		self.cleanup = cleanup
 		self.verbose = False
-
-		if not namespace:
-			self.output_name = '/tmp/%s-out' % self.name
-		else:
-			self.output_name = '/tmp/%s-%s-out' % (self.name, namespace)
+		self.out = ''
 
 		if namespace:
 			self.args = ['ip', 'netns', 'exec', namespace]
 			self.args.extend(args)
 
-		#
-		# For simplicity all processes will write to a temp file
-		# (/tmp/<name>-out). If any verbose options are required this file
-		# will get an IO watch and write out any bytes to the needed FDs.
-		#
-		self.stdout = open(self.output_name, 'a+')
-		self.io_position = self.stdout.tell()
-
 		if ctx:
 			# Verbose requested, add stdout/stderr to write FD list.
 			# This will end up always returning true if logging is
@@ -228,18 +216,18 @@ class Process:
 				os.fchown(f.fileno(), int(ctx.args.log_uid), int(ctx.args.log_gid))
 				self.write_fds.append(f)
 
-			#
-			# Only add an IO watch for long running processes. If
-			# the process is being waited for, the log/outfile bits
-			# will be handled after the process exists.
-			#
-			if self.write_fds != [] and not wait and not check or self.verbose:
-				self.io_watch = GLib.io_add_watch(self.stdout, GLib.IO_IN,
-								self.io_callback)
-
-		self.pid = subprocess.Popen(self.args, stdout=self.stdout, stderr=self.stdout,
+		self.pid = subprocess.Popen(self.args, stdout=subprocess.PIPE,
+						stderr=subprocess.STDOUT,
 						env=env, cwd=os.getcwd())
 
+		# Set as non-blocking so read() in the IO callback doesn't block forever
+		fl = fcntl.fcntl(self.pid.stdout, fcntl.F_GETFL)
+		fcntl.fcntl(self.pid.stdout, fcntl.F_SETFL, fl | os.O_NONBLOCK)
+
+		if not wait and not check:
+			self.io_watch = GLib.io_add_watch(self.pid.stdout, GLib.IO_IN |
+						GLib.IO_HUP | GLib.IO_ERR, self.io_callback)
+
 		print("Starting process {}".format(self.pid.args))
 
 		if not wait and not check:
@@ -249,12 +237,7 @@ class Process:
 		self.killed = True
 		self.ret = self.pid.returncode
 
-		self.stdout.seek(self.io_position)
-		self.out = self.stdout.read()
-		self.stdout.seek(0, 2)
-
-		if len(self.write_fds) > 0:
-			self.process_io(self.stdout)
+		self.process_io(self.pid.stdout)
 
 		self.write_fds = []
 
@@ -263,21 +246,16 @@ class Process:
 			raise subprocess.CalledProcessError(returncode=self.ret, cmd=self.args)
 
 	def process_io(self, source):
-		#
-		# The file will have already been written to, meaning the seek
-		# position points to EOF. This is why the position is saved so
-		# we can seek to where we were last time, read data, and seek
-		# back to EOF.
-		#
-		source.seek(self.io_position)
 		data = source.read()
 
-		self.io_position += len(data)
-		source.seek(self.io_position)
-
-		if len(data) == 0:
+		if not data:
 			return True
 
+		data = data.decode('utf-8')
+
+		# Save data away in case the caller needs it (e.g. list_sta)
+		self.out += data
+
 		for f in self.write_fds:
 			f.write(data)
 			f.flush()
@@ -294,10 +272,6 @@ class Process:
 	def __del__(self):
 		print("Del process %s" % self.args)
 
-		os.remove(self.output_name)
-
-		self.stdout.close()
-
 		if not self.killed:
 			self.kill()
 
@@ -312,19 +286,22 @@ class Process:
 		else:
 			self.pid.terminate()
 
-		self.pid.wait(timeout=15)
-		self.pid = None
+		try:
+			self.pid.wait(timeout=15)
+		except:
+			dbg("Process %s did not complete in 15 seconds!" % self.name)
+			self.pid.kill()
 
 		if self.ctx and self in self.ctx.processes:
 			self.ctx.processes.remove(self)
 
-		self.ctx = None
-
-		self.process_io(self.stdout)
-
 		if self.cleanup:
 			self.cleanup()
 
+		self.process_io(self.pid.stdout)
+
+		self.ctx = None
+		self.pid = None
 		self.write_fds = []
 
 		if self.io_watch:
Signature · 1 line
2.31.1
#9 lore raw
In reply to James Prestwood

Hi James,

Quoted 6 lines · James Prestwood

The network_config was not being copied to network_info when updated. This caused any new settings to be lost if the network configuration file was updated during runtime.

---
  src/knownnetworks.c | 2 ++
  1 file changed, 2 insertions(+)

All applied, thanks.

Regards,
-Denis