mech_eap.git
8 years agotests: Fix autogo_pbc() waiting for WPS-PBC
Ilan Peer [Wed, 17 Jun 2015 13:24:53 +0000 (16:24 +0300)]
tests: Fix autogo_pbc() waiting for WPS-PBC

Wait on the global control interface to support
configurations that use a dedicated P2P Device interface.

Note that the group interface cannot be used, as the group
interface name is not saved since no group was created.

Signed-off-by: Ilan Peer <ilan.peer@intel.com>
8 years agotests: Use global and group control interface in test_p2p_autogo
Ilan Peer [Wed, 17 Jun 2015 13:24:50 +0000 (16:24 +0300)]
tests: Use global and group control interface in test_p2p_autogo

This is required when a dedicated P2P Device is used.

Signed-off-by: Ilan Peer <ilan.peer@intel.com>
8 years agotests: Modify reset() to use the global control interface
Ilan Peer [Wed, 17 Jun 2015 13:24:48 +0000 (16:24 +0300)]
tests: Modify reset() to use the global control interface

Change the reset() method to use the global control interface
for resetting P2P state and also add a call to P2P_FLUSH.

Signed-off-by: Ilan Peer <ilan.peer@intel.com>
8 years agotests: Use global interface for "P2P_SET"
Ben Rosenfeld [Wed, 17 Jun 2015 13:24:47 +0000 (16:24 +0300)]
tests: Use global interface for "P2P_SET"

Setting P2P per_sta_psk should be done using the global control
interface.

Signed-off-by: Ben Rosenfeld <ben.rosenfeld@intel.com>
8 years agotests: NFC static handover with invalid contents
Jouni Malinen [Thu, 18 Jun 2015 22:14:35 +0000 (01:14 +0300)]
tests: NFC static handover with invalid contents

This adds test coverage to p2p_procesS_nfc_connection_handover() error
paths. This is also a regression test case for a memory leak on two of
these error paths.

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agoP2P: Fix memory leak in p2p_process_nfc_connection_handover()
Ben Rosenfeld [Wed, 17 Jun 2015 13:16:35 +0000 (16:16 +0300)]
P2P: Fix memory leak in p2p_process_nfc_connection_handover()

p2p_process_nfc_connection_handover() allocates msg memory in the parser
and might return before memory is released if the received message is
not valid.

Signed-off-by: Ben Rosenfeld <ben.rosenfeld@intel.com>
8 years agotests: WPA2-Enterprise connection using EAP-FAST and OOM in PRF
Jouni Malinen [Thu, 18 Jun 2015 21:50:29 +0000 (00:50 +0300)]
tests: WPA2-Enterprise connection using EAP-FAST and OOM in PRF

This is a regression test case for a memory leak on a TLS PRF error
path. In addition, this provides more coverage for this error path.

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agoEAP-FAST peer: Stop immediately on key derivation failure
Jouni Malinen [Thu, 18 Jun 2015 21:49:01 +0000 (00:49 +0300)]
EAP-FAST peer: Stop immediately on key derivation failure

If key derivation fails, there is no point in trying to continue
authentication. In theory, this could happen if memory allocation during
TLS PRF fails.

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agoOpenSSL: Fix memory leak on an openssl_tls_prf() error path
Ben Rosenfeld [Wed, 17 Jun 2015 13:16:34 +0000 (16:16 +0300)]
OpenSSL: Fix memory leak on an openssl_tls_prf() error path

Free tmp_out before returning to prevent memory leak in case the second
memory allocation in openssl_tls_prf() fails. This is quite unlikely,
but at least theoretically possible memory leak with EAP-FAST.

Signed-off-by: Ben Rosenfeld <ben.rosenfeld@intel.com>
8 years agotests: More RADIUS testing coverage with CUI/User-Name in ACL response
Jouni Malinen [Thu, 18 Jun 2015 21:21:44 +0000 (00:21 +0300)]
tests: More RADIUS testing coverage with CUI/User-Name in ACL response

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agotests: Make ap_wps_ap_scan_2 more robust
Jouni Malinen [Thu, 18 Jun 2015 19:33:48 +0000 (22:33 +0300)]
tests: Make ap_wps_ap_scan_2 more robust

The test sequence "scan_and_bss_entry_removed ap_wps_ap_scan_2" resulted
in failure due to an old BSS entry remaining from the first test case to
the second and the WPS_PBC operation on a forced BSSID ending up picking
the incorrect BSS entry. Make this more robust by clearing the scan
results from cfg80211.

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agotests: Make ap_wps_pbc_overlap_2* less likely to cause issues
Jouni Malinen [Thu, 18 Jun 2015 18:59:05 +0000 (21:59 +0300)]
tests: Make ap_wps_pbc_overlap_2* less likely to cause issues

Both of these test cases were leaving out BSS entries with active PBC
mode at the end of the test. This could result in the next text case
failing, e.g., in "ap_wps_pbc_overlap_2ap grpform_ext_listen" and
"ap_wps_pbc_overlap_2sta grpform_ext_listen" sequences. Fix this by
flushing the scan results more carefully at the end of the PBC overlap
test cases.

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agotests: Print a command line to re-run failed test sequence
Jouni Malinen [Thu, 18 Jun 2015 17:44:59 +0000 (20:44 +0300)]
tests: Print a command line to re-run failed test sequence

This makes it easier to test whether a specific test sequence results in
a reproducible test failure.

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agotests: P2PS with large number of services in Probe Request/Response
Jouni Malinen [Thu, 18 Jun 2015 15:13:30 +0000 (18:13 +0300)]
tests: P2PS with large number of services in Probe Request/Response

This verifies capability to split the response into multiple P2P
elements.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agotests: Extend P2PS service seek test to cover multiple services
Jouni Malinen [Thu, 18 Jun 2015 14:02:36 +0000 (17:02 +0300)]
tests: Extend P2PS service seek test to cover multiple services

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agotests: P2PS wildcard SD probe
Max Stepanov [Wed, 10 Jun 2015 08:43:49 +0000 (11:43 +0300)]
tests: P2PS wildcard SD probe

Verify that advertiser returns 'org.wi-fi.wfds' wildcard in a Probe
Response frame if at least one P2PS advertisement is present.

Signed-off-by: Max Stepanov <Max.Stepanov@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agoP2PS: Fix Probe Response frame building in error cases
Jouni Malinen [Thu, 18 Jun 2015 14:24:52 +0000 (17:24 +0300)]
P2PS: Fix Probe Response frame building in error cases

org.wi-fi.wfds service is not a replacement for non-WFA service matches.
Do not try to replace the results with that if there is not sufficient
room for the response. Instead, reply with all the matching services
that fit into the message. org.wi-fi.wfds is the first entry in the list
(if matching request/service is present), so it won't get overridden by
other services.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2PS: Fix org.wi-fi.wfds matching when building the response
Jouni Malinen [Thu, 18 Jun 2015 14:23:09 +0000 (17:23 +0300)]
P2PS: Fix org.wi-fi.wfds matching when building the response

The service hash for org.wi-fi.wfds is supposed to match only if the
device has a WFA defined org.wi-fi.wfds.* service. Verify that before
adding org.wi-fi.wfds to the response.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2PS: Add more debug prints for service info building
Jouni Malinen [Thu, 18 Jun 2015 14:21:23 +0000 (17:21 +0300)]
P2PS: Add more debug prints for service info building

This makes the debug log much more helpful for figuring out results from
service hash matching.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2PS: Remove unnecessary service hash filtering from p2p_reply_probe()
Jouni Malinen [Thu, 18 Jun 2015 14:18:54 +0000 (17:18 +0300)]
P2PS: Remove unnecessary service hash filtering from p2p_reply_probe()

Probe Response building is already doing service matching and there is
no need to do this in both places, so simplify the p2p_reply_probe()
implementation.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2PS: Do not ignore other hashes if org.wi-fi.wfds hash is included
Jouni Malinen [Thu, 18 Jun 2015 14:15:45 +0000 (17:15 +0300)]
P2PS: Do not ignore other hashes if org.wi-fi.wfds hash is included

When doing initial processing of Probe Request frame service hashes, the
previous implementation dropped all other hash values if a hash for
org.wi-fi.wfds was included. This is not correct, since that is not a
full wildcard of all services (it only matches WFA defined
org.wi-fi.wfds.* services).

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2PS: Fix service hash matching for org.wi-fi.wfds
Jouni Malinen [Thu, 18 Jun 2015 14:11:27 +0000 (17:11 +0300)]
P2PS: Fix service hash matching for org.wi-fi.wfds

This "wildcard" match is for WFA specified org.wi-fi.wfds.* services,
not for all services. Verify that there is a really matching service
being advertised instead of assuming this "wildcard" matches if any
services are advertised.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2PS: Fix p2p_find handling to allow "wildcard" with other hash values
Jouni Malinen [Thu, 18 Jun 2015 14:07:47 +0000 (17:07 +0300)]
P2PS: Fix p2p_find handling to allow "wildcard" with other hash values

The org.wi-fi.wfds "wildcard" is not a full wildcard of all service
names and as such, it must not remove other service name hash values
from the Probe Request frames.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2PS: Verify service name length in P2P_FIND command
Jouni Malinen [Thu, 18 Jun 2015 14:04:16 +0000 (17:04 +0300)]
P2PS: Verify service name length in P2P_FIND command

p2ps_gen_hash() has a limit on service names based on the temporary
buffer from stack. Verify that the service name from the local P2P_FIND
command is short enough to fix into that buffer.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2PS: Fix P2P_FIND seek parameter parsing
Jouni Malinen [Thu, 18 Jun 2015 14:01:02 +0000 (17:01 +0300)]
P2PS: Fix P2P_FIND seek parameter parsing

Only the first seek=<service name> parameter was accepted from the
P2P_FIND command. Fix this to go through all seek parameters to
construct the list of service hash values to seek.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2PS: Add a wildcard with other advertised service info
Max Stepanov [Wed, 10 Jun 2015 08:43:48 +0000 (11:43 +0300)]
P2PS: Add a wildcard with other advertised service info

Quoting P2PS specification: "If multiple Service Hash values are
included in the Probe Request frame, then the ASP shall find a match for
each Service Hash, and it shall send a Probe Response frame with the
information listed in this section for all matched Service Hashes." This
commit changes handling of wildcard hash matching by adding a
wildcard 'org.wi-fi.wfds' info together with the other hash matches.

Signed-off-by: Max Stepanov <Max.Stepanov@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agoP2PS: Re-factor p2p_buf_add_service_instance function
Max Stepanov [Wed, 10 Jun 2015 08:43:47 +0000 (11:43 +0300)]
P2PS: Re-factor p2p_buf_add_service_instance function

Add auxiliary functions to write a single advertised service info record
into a wpabuf and to find P2PS wildcard hash in a received hash
attribute. Re-factor p2p_buf_add_service_instance() function to allow
adding new wildcard types in future commits.

Signed-off-by: Max Stepanov <Max.Stepanov@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agowpa_cli: Fix process termination in wpa_cli action mode case
Jouni Malinen [Thu, 18 Jun 2015 10:48:21 +0000 (13:48 +0300)]
wpa_cli: Fix process termination in wpa_cli action mode case

Commit 4be9f27595f313773612d2fe534c013dfedfe679 ('wpa_cli: Use eloop
during connection attempts in interactive mode') did not take into
account the needs for signal processing in action mode. eloop_run() was
not called in this case and the internal select() loop would block eloop
processing anyway and prevent clean shutdown. Fix this by using eloop
for action mode operations.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoClear allocated debug message buffers explicitly
Jouni Malinen [Wed, 17 Jun 2015 13:30:34 +0000 (16:30 +0300)]
Clear allocated debug message buffers explicitly

When hostapd or wpa_supplicant is run in debug more with key material
prints allowed (-K on the command line), it is possible for passwords
and keying material to show up in debug prints. Since some of the debug
cases end up allocating a temporary buffer from the heap for processing
purposes, a copy of such password may remain in heap. Clear these
temporary buffers explicitly to avoid causing issues for hwsim test
cases that verify contents of memory against unexpected keys.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoClear control interface command explicitly from stack
Jouni Malinen [Wed, 17 Jun 2015 13:29:33 +0000 (16:29 +0300)]
Clear control interface command explicitly from stack

The control interface commands may include passwords or other private
key material, so clear it explicitly from memory as soon as the
temporary buffer is not needed anymore.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2P: Fix group interface addition failure properly for concurrent case
Jouni Malinen [Mon, 15 Jun 2015 20:34:11 +0000 (23:34 +0300)]
P2P: Fix group interface addition failure properly for concurrent case

It was possible for a P2P group formation failure to result in a
concurrent station mode operation getting disconnected in the specific
error case where group interface addition fails after a successful GO
Negotiation. Fix this by skipping the wpas_p2p_group_delete() call in
this specific case since the group interface does not exists anymore at
the point wpas_group_formation_completed() gets called.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2PS: Refactor p2p_data::query_hash and p2p_data::query_count use
Max Stepanov [Wed, 10 Jun 2015 08:43:46 +0000 (11:43 +0300)]
P2PS: Refactor p2p_data::query_hash and p2p_data::query_count use

Avoid using p2p_data::query_hash for both Probe Request frame processing
and for hashes specified by p2p_find. It's resolved by use of local
query_hash and query_count variables in p2p_reply_probe().

Since p2p_data::query_hash is used only for seek hash values rename
p2p_data::query_hash to p2ps_seek_hash.

Delete p2p_data::query_count since it's not needed anymore.

Signed-off-by: Max Stepanov <Max.Stepanov@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agotests: VENDOR_ELEM_ADD for various frames
Jouni Malinen [Mon, 15 Jun 2015 11:15:30 +0000 (14:15 +0300)]
tests: VENDOR_ELEM_ADD for various frames

This increases testing coverage for VENDOR_ELEM mechanism by explicitly
verifying that the requested element gets added to each of the supported
frame types.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2P: Add vendor elements into Invitation Response frames
Sunil Dutt [Sun, 14 Jun 2015 11:55:56 +0000 (17:25 +0530)]
P2P: Add vendor elements into Invitation Response frames

Commit 86bd36f0d5b3d359075c356d68977b4d2e7c9f71 ('Add generic mechanism
for adding vendor elements into frames') introduced a mechanism to add
vendor elements into various frames, but missed the addition to the
Invitation Response frame. This commit addresses the same.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2PS: Delete p2ps_svc_found from struct p2p_data
Max Stepanov [Wed, 10 Jun 2015 08:43:45 +0000 (11:43 +0300)]
P2PS: Delete p2ps_svc_found from struct p2p_data

This variable is used locally only in the p2p_reply_probe() function.
The value of this variable is valid only in the context of the single
Probe Request message handling and doesn't make much sense in p2p
context.

Signed-off-by: Max Stepanov <Max.Stepanov@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agoP2PS: Add a function to free a PD context
Max Stepanov [Wed, 10 Jun 2015 08:43:40 +0000 (11:43 +0300)]
P2PS: Add a function to free a PD context

Free a PD context with a function encapsulating both os_free() call and
setting a PD context pointer to NULL.

Signed-off-by: Max Stepanov <Max.Stepanov@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agotests: Add p2ps_cli_probe and p2ps_go_probe tests
Andrei Otcheretianski [Wed, 10 Jun 2015 08:43:39 +0000 (11:43 +0300)]
tests: Add p2ps_cli_probe and p2ps_go_probe tests

Test P2PS GO and CLI discoverability on group operating channel.
In order to implement these tests, refactor p2ps_connect_p2ps_method
and test_p2ps_connect_adv_go_pin_method to reuse the code for
connection establishment. Also change p2ps_exact_seek so it will
allow getting Probe Response frames from several peers.

Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agotests: Fix inconsistent indentation
Andrei Otcheretianski [Wed, 10 Jun 2015 08:43:38 +0000 (11:43 +0300)]
tests: Fix inconsistent indentation

Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agotests: Fix p2ps_connect_adv_go_pin_method case
Andrei Otcheretianski [Wed, 10 Jun 2015 08:43:38 +0000 (11:43 +0300)]
tests: Fix p2ps_connect_adv_go_pin_method case

test_p2ps_connect_adv_go_pin_method() expects that
p2ps_provision_keypad_method() returns P2PS-PROV-DONE with details
needed for a connection. However, this event was overridden which
resulted in an incorrect test flow skipping the connection
establishement. The test would pass, however, without really trying to
connect. Fix this by returning the correct event.

Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agoP2P: Prefer direct Probe Response frames over GO's client list
Andrei Otcheretianski [Wed, 10 Jun 2015 08:43:37 +0000 (11:43 +0300)]
P2P: Prefer direct Probe Response frames over GO's client list

A P2P Client may be discoverable and reply to Probe Request frames,
while at the same time the P2P GO would also be discoverable and include
the P2P Client information in the P2P Group Info attribute of the Probe
Response frames.

If a seeker constantly hears the Probe Response frames from a P2P Client
and then from the GO, but handles them in the opposite order (due to
scan results ordering), the more valuable Probe Response frame from the
P2P Client will be ignored. Fix this by defining a threshold (1 second)
during which the direct Probe Response frame will be preferred over the
information acquired from the GO and will not be considered as old.

Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agoP2P: Use more precise device timestamping for group clients
Andrei Otcheretianski [Wed, 10 Jun 2015 08:43:36 +0000 (11:43 +0300)]
P2P: Use more precise device timestamping for group clients

When adding group clients to the P2P peer list, use the driver provided
BSS entry timestamp instead of the current time. Otherwise, the time
comparison which is made in p2p_add_device() doesn't make sense.

Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agoP2P: Specify frequency when sending Probe Response frame
Andrei Otcheretianski [Wed, 10 Jun 2015 08:43:35 +0000 (11:43 +0300)]
P2P: Specify frequency when sending Probe Response frame

If the RX frequency of the Probe Request frame is known, specify it when
sending the Probe Response frame. This is needed when the Probe Request
frame is received on another virtual interface, for example, when a GO
or P2PS client are discoverable on the group operating channel.

Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agodrivers: Add freq parameter to send_mlme() function
Andrei Otcheretianski [Wed, 10 Jun 2015 08:43:34 +0000 (11:43 +0300)]
drivers: Add freq parameter to send_mlme() function

Change send_mlme() API to allow sending management frames on a specific
channel, overriding the internal driver decision.

Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agoP2P: Fix return value of p2p_reply_probe() and p2p_probe_req_rx()
Max Stepanov [Wed, 10 Jun 2015 08:43:33 +0000 (11:43 +0300)]
P2P: Fix return value of p2p_reply_probe() and p2p_probe_req_rx()

Return P2P_PREQ_PROCESSED instead of P2P_PREQ_NOT_PROCESSED on
a successful Probe Request frame handling in p2p_reply_probe().

Verify a return value of p2p_reply_probe() in p2p_probe_req_rx()
and continue a pending invitation/connection flow only if the
Probe Request frame is from an expected P2P peer.

Signed-off-by: Max Stepanov <Max.Stepanov@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agoP2PS: Enable Probe Request frame processing by P2P Client
Max Stepanov [Wed, 10 Jun 2015 08:43:32 +0000 (11:43 +0300)]
P2PS: Enable Probe Request frame processing by P2P Client

1. Add global p2p_cli_probe property to enable/disable Probe Request
frame RX reporting for connected P2P Clients. The property can be set to
0 - disable or 1 - enable. The default value is 0.

2. Enable Probe Request frame RX reporting for P2P Client on
WPA_COMPLETED state if p2p_cli_probe property is set to 1. Disable it
when an interface state is changing to any other state.

3. Don't cancel Probe Request frame RX reporting on wpa_stop_listen for
a connected P2P Client handling Probe Request frames.

Signed-off-by: Max Stepanov <Max.Stepanov@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agoP2P: Add rx_freq parameter to Probe Request frame handler
Max Stepanov [Wed, 10 Jun 2015 08:43:31 +0000 (11:43 +0300)]
P2P: Add rx_freq parameter to Probe Request frame handler

In some cases, Probe Request frames can be received by a peer not only
on a listen channel. In this case an additional rx_freq parameter
explitly contains a Probe Request frame RX frequency. In case rx_freq is
set to 0, a Probe Request frame RX channel is assumed to be our own
listen channel (p2p->cfg->channel).

Signed-off-by: Max Stepanov <Max.Stepanov@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
8 years agoP2P: Update target GO Device Address from BSS entry during join
Jouni Malinen [Sun, 14 Jun 2015 20:14:26 +0000 (23:14 +0300)]
P2P: Update target GO Device Address from BSS entry during join

It is possible for P2P_CONNECT-join command to be issued on a GO's P2P
Interface Address before the P2P peer entry is available to map this
into the GO's P2P Device Address. This could result in the join
operation failing to continue after receiving PD Response due to the
address mismatch (source address = P2P Device Address while only the P2P
Interface Address is known). Fix this by updating the pending join P2P
Device Address based on the BSS entry, if needed.

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agotests: IBSS on 5 GHz band
Jouni Malinen [Sun, 14 Jun 2015 16:35:19 +0000 (19:35 +0300)]
tests: IBSS on 5 GHz band

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agotests: WNM BSS Transition Management in different countries
Jouni Malinen [Sun, 14 Jun 2015 15:11:14 +0000 (18:11 +0300)]
tests: WNM BSS Transition Management in different countries

This increases testing coverage for ieee802_chan_to_freq*() functions.

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agoRemove duplicated country code from operating class lists
Jouni Malinen [Sun, 14 Jun 2015 14:40:10 +0000 (17:40 +0300)]
Remove duplicated country code from operating class lists

CA country code was included mistakenly (copy-paste..) in cn_op_class_cc
while it was supposed to be included only in us_op_class_cc. In
practice, this did not result in incorrect operation due to the
us_op_class_cc list being checked first. Anyway, better fix
cn_op_class_cc to avoid confusion here.

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agotests: Additional ieee802_11_parse_elems() module test coverage
Jouni Malinen [Sun, 14 Jun 2015 13:47:29 +0000 (16:47 +0300)]
tests: Additional ieee802_11_parse_elems() module test coverage

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agotests: Module test for gas.c corner cases in gas_anqp_set_len()
Jouni Malinen [Sun, 14 Jun 2015 13:27:21 +0000 (16:27 +0300)]
tests: Module test for gas.c corner cases in gas_anqp_set_len()

This increases code coverage for gas.c testing to cover areas that
cannot be reached with pure hwsim test cases.

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agotests: GAS/ANQP query and OOM
Jouni Malinen [Sun, 14 Jun 2015 13:17:33 +0000 (16:17 +0300)]
tests: GAS/ANQP query and OOM

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agotests: Verify correct VLAN operation in multi-BSS multi-VLAN case
Michael Braun [Mon, 27 Apr 2015 07:08:04 +0000 (09:08 +0200)]
tests: Verify correct VLAN operation in multi-BSS multi-VLAN case

This adds hwsim test ap_vlan_iface_cleanup_multibss. It connects two
stations in different BSS but the same hostapd process. First both
stations are in VLAN 1, then they get reauthenticated into VLAN 2. Due
to the ordering of the stations moving around, this test checks that
bridge and tagged interface referencing counting is done globally, such
that the tagged interface is not removed too early and no bridge is
left over.

Signed-off-by: Michael Braun <michael-dev@fami-braun.de>
8 years agoFix removal of tagged interface and bridge when multiple BSS share them
Michael Braun [Mon, 27 Apr 2015 07:08:03 +0000 (09:08 +0200)]
Fix removal of tagged interface and bridge when multiple BSS share them

Currently, if multiple bss share are bridge and tagged vlan interface,
only the first instance of struct hostapd_vlan for this vlanid will have
the DVLAN_CLEAN_VLAN flag added. Thus, when this instance is removed,
the tagged vlan interface will be removed from bridge, thought other bss
might still need it. Similarily, the bridge will be left over, as the
does not have zero ports when the first instance of a struct
hostapd_vlan is freed.

This patch fixes this by having a global (per process) reference counter
for dynamic tagged vlan and dynamically created bridge interfaces, so
they are only removed after all local users are freed. (struct
hapd_interfaces *)->vlan_priv is used to hold src/ap/vlan_init.c global
per-process data like drv_priv does; right now this is only used for the
interface reference counting, but could get extended when needed. Then
possibly some vlan_global_init / vlan_global_deinit should be added, but
this is not required right now.

Additionally, vlan->configured is checked to avoid reference counter
decreasing before vlan_newlink increased them.

In order to avoid race conditions, vlan_dellink is called explicitly
after hostapd_vlan_if_remove. Otherwise there would be a short timeframe
between hostapd_vlan_if_remove and vlan_dellink during which the struct
hostapd_vlan still exists, so ap_sta_bind_vlan would try to attach
stations to it.

Signed-off-by: Michael Braun <michael-dev@fami-braun.de>
8 years agoCombine multiple function calls to a single statement
Purushottam Kushwaha [Fri, 12 Jun 2015 11:36:27 +0000 (11:36 +0000)]
Combine multiple function calls to a single statement

This cleans up p2p_ctrl_group_add() to share a single call to
wpas_p2p_group_add().

Signed-off-by: Purushottam Kushwaha <p.kushwaha@samsung.com>
8 years agotests: D-Bus Reconnect command
Jouni Malinen [Sun, 14 Jun 2015 08:38:14 +0000 (11:38 +0300)]
tests: D-Bus Reconnect command

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agodbus: Add Reconnect command to D-Bus Interface
MAYANK HAARIT [Fri, 12 Jun 2015 11:27:19 +0000 (11:27 +0000)]
dbus: Add Reconnect command to D-Bus Interface

This has same behavior as the ctrl_iface RECONNECT command.

Signed-off-by: Mayank Haarit <mayank.h@samsung.com>
Signed-off-by: Purushottam Kushwaha <p.kushwaha@samsung.com>
8 years agodoc: Update D-Bus GONegotiationRequest Signal: add device_go_intent
MAYANK HAARIT [Fri, 12 Jun 2015 07:26:10 +0000 (07:26 +0000)]
doc: Update D-Bus GONegotiationRequest Signal: add device_go_intent

This new parameter was added by commit
aa2b12562b6e70fe48c7ce4abebf094206a9332a ('P2P: Add GO Intent of
connecting device in GO Negotiation Request event').

Signed-off-by: Rahul Jain <rahul.jain@samsung.com>
Signed-off-by: Mayank Haarit <mayank.h@samsung.com>
8 years agotests: Make flush_scan_cache() more robust
Jouni Malinen [Fri, 12 Jun 2015 18:28:14 +0000 (21:28 +0300)]
tests: Make flush_scan_cache() more robust

There seem to be cases where flush_scan_cache() was unable to clear all
BSS entries due to a hidden SSID BSS (SSID length 0) showing up again
from cfg80211 BSS table. Check for this and run the flush operation
again if any entries remain.

This fixes an issue where the following hwsim test case sequence
resulted in the last test case failing due to the old BSS entry from the
first test case being in place and showing unexpected flags information:
ap_hs20_min_bandwidth_home_hidden_ssid_in_scan_res
ap_hs20_remediation_required
ap_mixed_security

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agotests: P2P GO and 5 GHz channels 165 (allowed) and 169 (disallowed) in US
Jouni Malinen [Fri, 12 Jun 2015 17:54:01 +0000 (20:54 +0300)]
tests: P2P GO and 5 GHz channels 165 (allowed) and 169 (disallowed) in US

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2P: Add Operating class 125 for P2P supported channels
Amr BEN ABDESSALEM [Thu, 2 Apr 2015 16:22:58 +0000 (18:22 +0200)]
P2P: Add Operating class 125 for P2P supported channels

Add operating class 125 (channels 149..169) to the list of P2P supported
channels. This allows the 5 GHz channels 161 and 169 to be used for P2P
GO when those channels are allowed for AP mode in the current regulatory
domain.

Signed-off-by: Amr BEN ABDESSALEM <amrx.ben.abdessalem@intel.com>
8 years agotests: Pending GAS query interrupted with deinit
Jouni Malinen [Wed, 10 Jun 2015 23:40:13 +0000 (02:40 +0300)]
tests: Pending GAS query interrupted with deinit

These are regression test cases for a segmentation fault issue (use of
freed memory) where interface removal happened while a gas-query item
was pending.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoGAS: Remove all radio works before calling gas_query_deinit()
Ilan Peer [Mon, 1 Jun 2015 10:38:10 +0000 (13:38 +0300)]
GAS: Remove all radio works before calling gas_query_deinit()

Remove all gas-query radio works before calling gas_query_deinit()
as gas_query_deinit() flow frees the query context, which might
be later be accessed from the radio work callback (and result
with unexpected behavior, e.g., segmentation fault).

Signed-off-by: Ilan Peer <ilan.peer@intel.com>
8 years agotests: Create a P2P persistent group with ADD_NETWORK
Jouni Malinen [Wed, 10 Jun 2015 20:49:01 +0000 (23:49 +0300)]
tests: Create a P2P persistent group with ADD_NETWORK

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2P: Fix persistent group profile on manual disabled=2 change
Jouni Malinen [Wed, 10 Jun 2015 20:58:21 +0000 (23:58 +0300)]
P2P: Fix persistent group profile on manual disabled=2 change

ssid->p2p_persistent_group was left to its old value when changing a
network profile to/from P2P persistent group type (disabled=2). This
could result in unexpected behavior when using an incomplete persistent
group profile. This was mainly visible through D-Bus network profile
handling where a persistent group would not be unregistered insome
cases.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agotests: D-Bus WPS/PBC operation and signal for PBC overlap
Jouni Malinen [Wed, 10 Jun 2015 16:41:28 +0000 (19:41 +0300)]
tests: D-Bus WPS/PBC operation and signal for PBC overlap

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoD-Bus: Add WPS pbc-overlap Event
Avichal Agarwal [Fri, 5 Jun 2015 12:46:51 +0000 (18:16 +0530)]
D-Bus: Add WPS pbc-overlap Event

This sends an Event D-Bus signal with name "pbc-overlap" for
WPS-EVENT-OVERLAP.

Signed-off-by: Avichal Agarwal <avichal.a@samsung.com>
8 years agoP2P: Add ModelNumber and SerialNumber info into D-Bus peer interface
Amit Khatri [Wed, 10 Jun 2015 10:06:09 +0000 (10:06 +0000)]
P2P: Add ModelNumber and SerialNumber info into D-Bus peer interface

Signed-off-by: Amit Khatri <amit.khatri@samsung.com>
Signed-off-by: Rahul Jain <rahul.jain@samsung.com>
8 years agoP2P: Add ModelName info into D-Bus peer interface
Amit Khatri [Wed, 10 Jun 2015 10:06:09 +0000 (10:06 +0000)]
P2P: Add ModelName info into D-Bus peer interface

Signed-off-by: Amit Khatri <amit.khatri@samsung.com>
Signed-off-by: Rahul Jain <rahul.jain@samsung.com>
8 years agodoc: Add missing ListenChannel to P2P information
Amit Khatri [Wed, 10 Jun 2015 10:06:09 +0000 (10:06 +0000)]
doc: Add missing ListenChannel to P2P information

Signed-off-by: Amit Khatri <amit.khatri@samsung.com>
Signed-off-by: Rahul Jain <rahul.jain@samsung.com>
8 years agoDocument the wpa_msg_cb "global" parameter
Jouni Malinen [Fri, 6 Feb 2015 10:08:28 +0000 (12:08 +0200)]
Document the wpa_msg_cb "global" parameter

Instead of an int variable with magic values 0, 1, 2, use an enum that
gives clearer meaning to the values now that the original boolean type
global argument is not really a boolean anymore.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoDo not set own_disconnect_req flag if not connected
Jouni Malinen [Wed, 10 Jun 2015 12:40:35 +0000 (15:40 +0300)]
Do not set own_disconnect_req flag if not connected

This fixes some issues where a disconnection event may get ignored if an
earlier operation to remove or disable a network resulted in an attempt
to disconnect event though no connection was in place.
wpa_s->current_ssid != NULL alone is not sufficient to determine that
there will be a driver event notifying completion of such disconnection
request. Set own_disconnect_req to 1 only if wpa_s->wpa_state is also
indicating that there is a connection or an attempt to complete one.

This showed up in a failure, e.g., when running the hwsim test case
scan_int followed by ap_vlan_wpa2_psk_radius_required where the latter
ended up not processing a connection failure event and getting stuck not
trying to run a new scan and connection attempt.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agowpa_gui: Port to Qt5
Jason A. Donenfeld [Fri, 8 May 2015 21:01:01 +0000 (23:01 +0200)]
wpa_gui: Port to Qt5

This adds support for Qt5 while still preserving support for building
with Qt4. The same source code builds with both Qt versions now. The
functionality should be identical.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
8 years agoeapol_test: Support IPv6 for authentication server
Ondřej Caletka [Thu, 21 May 2015 14:24:56 +0000 (16:24 +0200)]
eapol_test: Support IPv6 for authentication server

This allows testing RADIUS servers over IPv6.

Signed-off-by: Ondřej Caletka <ondrej@caletka.cz>
8 years agoFix pairwise cipher suite bitfields to the driver in mixed mode
Jouni Malinen [Wed, 25 Mar 2015 09:41:26 +0000 (11:41 +0200)]
Fix pairwise cipher suite bitfields to the driver in mixed mode

Commit 95b6bca66de9dc0a2ddd9164ec052a7d5f58804b ('Add rsn_pairwise bits
to set_ieee8021x() driver_ops') modified cipher configuration to use
unconditionally wpa_pairwise | rsn_pairwise. While that works for many
cases, it does not handle the case of dynamic configuration changes over
the control interface where wpa_pairwise or rsn_pairwise values may not
get cleared when the wpa parameter is modified. Fix this inconsistency
by configuring the driver with only the bits that are valid for the
currently enabled WPA/WPA2 version(s).

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoRemove [MU-BEAMFORMEE] option from hostapd vht_capab parameter
Jouni Malinen [Wed, 10 Jun 2015 10:41:54 +0000 (13:41 +0300)]
Remove [MU-BEAMFORMEE] option from hostapd vht_capab parameter

The standard hardcodes the MU Beamformee Capable subfield is hardcoded
to 0 when transmitting by an AP, so there is no need to provide a
configuration parameter for setting this to one.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoFix spelling of initialize in a comment and an error message
Jouni Malinen [Tue, 9 Jun 2015 19:03:53 +0000 (22:03 +0300)]
Fix spelling of initialize in a comment and an error message

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2P: Add GO Intent of connecting device in GO Negotiation Request event
MAYANK HAARIT [Tue, 2 Jun 2015 05:47:33 +0000 (05:47 +0000)]
P2P: Add GO Intent of connecting device in GO Negotiation Request event

Add GO Intent information of connecting device in GO Negotiation Request
event which will help applications to decide its own GO intent value in
advance and can avoid failure cases when both devices use GO Intent 15
depending on application requirement.

Signed-off-by: Mayank Haarit <mayank.h@samsung.com>
8 years agotests: Allow D-Bus GONegotiationRequest signal to include extra argument
Jouni Malinen [Sat, 6 Jun 2015 15:15:23 +0000 (18:15 +0300)]
tests: Allow D-Bus GONegotiationRequest signal to include extra argument

This is needed to allow the GO Intent argument to be added without
breaking these test cases.

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agodoc: Fix a typo in D-Bus API document
Amit Khatri [Wed, 3 Jun 2015 11:19:27 +0000 (16:49 +0530)]
doc: Fix a typo in D-Bus API document

Signed-off-by: Amit Khatri <amit.khatri@samsung.com>
Signed-off-by: Rahul Jain <rahul.jain@samsung.com>
8 years agoSet Acct-Session-Id from os_get_random() instead of os_get_time()
Alan T. DeKok [Thu, 4 Jun 2015 20:53:06 +0000 (16:53 -0400)]
Set Acct-Session-Id from os_get_random() instead of os_get_time()

So that systems with bad clocks will send random session IDs,
instead of always ones starting at the same second.

If os_get_random() isn't available, use os_get_time(). But also
mix in now.tv_usec, so that the accounting session ID is more
likely to be globally and temporally unique.

Signed-off-by: Alan DeKok <aland@freeradius.org>
8 years agoOpenSSL: Fix build iwth OpenSSL 0.9.8
MAYANK HAARIT [Wed, 3 Jun 2015 05:39:34 +0000 (05:39 +0000)]
OpenSSL: Fix build iwth OpenSSL 0.9.8

The OPENSSL_VERSION_NUMBER < 0x00909000L case of
openssl_get_keyblock_size() had not been kept in sync with the cleanup
changes.

Signed-off-by: Mayank Haarit <mayank.h@samsung.com>
8 years agotests: More WPA2-EAP + PMF test coverage
Jouni Malinen [Sat, 6 Jun 2015 14:13:21 +0000 (17:13 +0300)]
tests: More WPA2-EAP + PMF test coverage

Signed-off-by: Jouni Malinen <j@w1.fi>
8 years agotests: Control interface behavior on scan SSID parameter
Jouni Malinen [Fri, 5 Jun 2015 12:32:08 +0000 (15:32 +0300)]
tests: Control interface behavior on scan SSID parameter

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoAdd support to request a scan with specific SSIDs
Krishna Vamsi [Thu, 21 May 2015 13:59:31 +0000 (19:29 +0530)]
Add support to request a scan with specific SSIDs

Support a request to scan specific SSIDs given by user with the SCAN
command. The SSID list can be suffixed to the scan command as follows.
For example, if SSIDs "ABC" and "abc123" need to be specifically
scanned, the command should be "SCAN ssid 414243 ssid 616263313233". The
value of the SSID is passed in hexadecimal representation.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agotests: wps_priority in configuration and WPS behavior
Jouni Malinen [Thu, 4 Jun 2015 11:04:16 +0000 (14:04 +0300)]
tests: wps_priority in configuration and WPS behavior

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoWPS: Allow the priority for the WPS networks to be configured
Sunil Dutt [Mon, 25 May 2015 13:12:24 +0000 (18:42 +0530)]
WPS: Allow the priority for the WPS networks to be configured

This commit adds a configurable parameter (wps_priority) to specify the
priority for the networks derived through WPS connection.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agowpa_supplicant: Remove trailing whitespace
Anton Nayshtut [Sun, 16 Nov 2014 15:28:53 +0000 (17:28 +0200)]
wpa_supplicant: Remove trailing whitespace

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agohostapd: Make sure band selection does not result in NULL dereference
Jouni Malinen [Mon, 1 Jun 2015 15:56:19 +0000 (18:56 +0300)]
hostapd: Make sure band selection does not result in NULL dereference

Explicitly check for iface->current_mode before dereferencing it. While
this case may not happen in practice, it is better for the setup
functions to be more careful when doing the initial band selection.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2P: Add manufacturer info into D-Bus peer interface
Amit Khatri [Thu, 28 May 2015 12:19:46 +0000 (17:49 +0530)]
P2P: Add manufacturer info into D-Bus peer interface

Allow the Manufacturer information for a found P2P peer device to be
fetched through the D-Bus interface similarly to p2p_peer <mac address>
on ctrl interface.

Signed-off-by: Amit Khatri <amit.khatri@samsung.com>
Signed-off-by: Rahul Jain <rahul.jain@samsung.com>
Signed-off-by: Ashutosh Kaushik <k.ashutosh@samsung.com>
8 years agotests: D-Bus P2P Cancel
Jouni Malinen [Fri, 29 May 2015 18:13:31 +0000 (21:13 +0300)]
tests: D-Bus P2P Cancel

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2P: Add P2P Cancel method over D-Bus interface
Nishant Chaprana [Tue, 26 May 2015 05:05:47 +0000 (10:35 +0530)]
P2P: Add P2P Cancel method over D-Bus interface

This patch adds P2P Cancel method over
fi.w1.wpa_supplicant1.Interface.P2PDevice dbus interface.

Signed-off-by: Nishant Chaprana <n.chaprana@samsung.com>
8 years agotests: D-Bus WPS Cancel operation
Jouni Malinen [Fri, 29 May 2015 18:00:02 +0000 (21:00 +0300)]
tests: D-Bus WPS Cancel operation

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoWPS: Add WPS Cancel method over D-Bus interface
Nishant Chaprana [Tue, 26 May 2015 04:08:39 +0000 (09:38 +0530)]
WPS: Add WPS Cancel method over D-Bus interface

Signed-off-by: Nishant Chaprana <n.chaprana@samsung.com>
8 years agoP2P: Fix a potential memory leak in a P2P+NFC corner case
Purushottam Kushwaha [Fri, 29 May 2015 09:20:41 +0000 (09:20 +0000)]
P2P: Fix a potential memory leak in a P2P+NFC corner case

Avoid memory leak due to previous allocation for ssid->ssid. It does not
look like this is hit in normal cases, but at might be possible for the
SSID to get set if the peer is present in previous scan results and WPS
code ends up copying the SSID from there.

Signed-off-by: Purushottam Kushwaha <p.kushwaha@samsung.com>
8 years agoAdd libwpa_client build option to use a dynamic library
Krishna Vamsi [Fri, 8 May 2015 22:32:36 +0000 (15:32 -0700)]
Add libwpa_client build option to use a dynamic library

Add support to compile libwpa_client.so. This can be used by external
programs to interact with the wpa_supplicant control interface. Also
rename the static version of this library to libwpa_client.a to be
consistent with the name used previous in Android builds.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agotests: P2P group re-invocation with invitation response getting lost
Jouni Malinen [Fri, 29 May 2015 10:34:22 +0000 (13:34 +0300)]
tests: P2P group re-invocation with invitation response getting lost

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoP2P: Restart group formation timer upon receiving new Inv Req
Krishna Vamsi [Thu, 28 May 2015 14:17:41 +0000 (19:47 +0530)]
P2P: Restart group formation timer upon receiving new Inv Req

A new Invitation Request might be received on a just started persistent
group if the previous Invitation Response sent isn't received at the
peer (GO of the persistent group). When the peer sends an Invitation
Request again, treat it as start of group formation and restart the
group formation timer at this point of time.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
8 years agoPostpone updating of wpa_s->current_bss till association event
Sunil Dutt [Thu, 28 May 2015 15:11:12 +0000 (20:41 +0530)]
Postpone updating of wpa_s->current_bss till association event

In the case of driver-based BSS selection, a bssid_hint in the connect
request might not result in to the selection of the same BSS. Thus,
postpone the updation of the wpa_s->current_bss till the association
event is received unless the BSSID is forced. This fixes issues where
wpa_s->current_bss may end up being updated to point to the BSS that
wpa_supplicant provided as a hint for a roaming case, but then not
restored if the driver decides to "return" to the current BSS instead.
This could result in some operations (e.g., WNM BSS TM response) not
working properly due to incorrect BSS being identified in
wpa_s->current_bss.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>