| Message ID | 20190915200837.196283-1-alexander@wetzel-home.de |
|---|---|
| Headers | show |
| Series | Support seamless PTK rekeys with Extended Key ID | expand |
On Sun, Sep 15, 2019 at 10:08:20PM +0200, Alexander Wetzel wrote: > The patch series consist of basically five parts: > > 01-hostapd: Provide correct keyid to wpa_send_eapol() > 02-nl80211: Migrate to current netlink key message format As an initial step, I applied those two now. > 03-Driver: Add key_type and Extended Key ID driver flag > 04-Add new argument key_type to all set_key() functions > 05-hostapd: Set the correct key_type for key installs > 06-wpa_supplicant: Set the correct key_type for key installs And those I have went through and have couple of pending questions (or well, already answered questions, but I have not yet fully reviewed the responses). I put my current work versions here so that you can see how exactly I was thinking of changing these: https://w1.fi/p/set-key/ It would be good to consider those as baseline for any additional changes. I have not yet went through all the details in the following patches, so no changes proposed for those as of now.
> I put my current work versions here so that you can see how exactly I > was thinking of changing these: > > https://w1.fi/p/set-key/ > > It would be good to consider those as baseline for any additional > changes. I have not yet went through all the details in the following > patches, so no changes proposed for those as of now. > As mentioned I already have some more fixes pending. Now a quick test run on Thursday was generally looking good but showed something unexpected in the logs. I had no time to follow up on that, yet... (The new TKIP only AP test was working, but I did trip over an unexpected log message, indicating something may be off or I messed something up with the test run.) It does not look like I will be able to look into that till probably Monday evening. So I just rebased my working tree yours and exported the WIP patches: https://www.awhome.eu/index.php/s/o5m9Q29n37zbEga You can at least see now how I want to change handle_extended_key_id() They probably still have some minor errors and the commit messages are crap. Alexander
Remark: Most of the text here and the other commends are taken 1:1 from the v3 of the series and is identical to what was in v4 and v5 with only minor updates. Extended Key ID for Individually Addressed Frames has been added to the IEEE 802.11 - 2012 standard to allow STAs to rekey the pairwise key without interruption for ongoing data traffic. A good and easy to follow introduction to Extended Key ID can be found here: https://mentor.ieee.org/802.11/dcn/10/11-10-0313-01-000m-rekeying-protocol-fix.ppt In a nutshell Extended Key ID is just allowing to also use keyid 1 for pairwise keys: We can now have two pairwise keys in parallel, allowing to switch between them comparable how it's already done for GTK and IGTK keys. All implementations of IEEE 802.11 without Extended Key ID only use the keyid 0 for unicast MPDUs. Since each STA can only have a single unicast key installed at any time there is no time frame when the old or the new key can be used. When rekeying the unicast key under load MPDU losses are therefore unavoidable: Each STA will get MPDUs encrypted with a key which is either no longer installed or is not installed, yet. In both cases the remote STA will be unable to decrypt the MPDUs and drop them. (Since only the decryption fails the frames are still acknowledged and are truly lost. It's up to e.g. tcp to retransmit them.) Especially on high speed connections this can easily affect a few dozen MPDUs, drastically throttle the speed for tcp due or cause drastic audio/video glitches with udp transmissions. (Most implementations are even not able to rekeys correctly, making matters worse. But that's not really relevant for this patch series.) This patch series implements Extended Key ID for both hostapd and wpa_supplicant to allow unicast key rekeys without MPDU loss and/or impact to the performance. While there are still ways to extend or improve this patch set it's already quite complex and I believe this to be good enough to be merged in the current state. Or in other words: It's high time to get some feedback on this big patch set instead of adding more to it. Since Extended Key ID breaks the long-standing agreement of only one pairwise key (keyid 0) and also has to postpone the Tx usage of a newly installed key this can only work with drivers actively supporting Extended Key ID. So far Extended Key ID is only supported for linux kernels >=5.2 and cards only providing SW crypto. This patch series has been developed using hwsim (works out of the box with a kernel >= 5.2.12) and - not yet official supported - iwlwifi cards (supporting HW crypto). (All iwlwifi cards prior to the current 22000 series are working and probably will officially start to support Extended Key ID with one of the next linux kernels.) Some early versions of this series also have been tested with ath9, but that required special mac80211 workarounds which have not been merged upstream and abandoned for the time being. The patch series consist of basically five parts: 01-hostapd: Provide correct keyid to wpa_send_eapol() 02-nl80211: Migrate to current netlink key message format 03-Driver: Add key_type and Extended Key ID driver flag 04-Add new argument key_type to all set_key() functions 05-hostapd: Set the correct key_type for key installs 06-wpa_supplicant: Set the correct key_type for key installs 07-drivers: Migrate drivers from set_tx to key_type API 08-nl80211: Switch to the new key_type API & cleanup 09-nl80211,wpa_supplicant: Drop outdated tdls hack 1) The first 9 patches are addressing issues not directly tied to the Extended Key ID support. More details are directly in the individual patch descriptions, but they are primarily related to deprecating the set_tx "boolean" used in various set_key() functions and replace it with something more flexible which also can handle Extended Key ID key installs. 10-hostapd: Add support for Extended Key ID 11-wpa_supplicant: AP Extended Key ID support 12-tests: Extended Key ID tests 2) The patches 10-12 are handling the Extended Key ID support for hostapd/wpa_supplicant and modify/amend to the existing tests to be able to work correctly with and without Extended Key ID support. 13-hostapd: FILS Extended Key ID support 14-wpa_supplicant: FILS Extended Key ID support 3) Patches 13 and 14 are trying to close a gap not covered in the official standard. I'm far from sure that this is the correct way and hope we can have a discussion here how to best handle FILS and probably also FT. (The standard does not have any guidance for that, too.) 15-nl80211: Extended Key ID support 4) The patch 15 is adding Extended Key ID support to nl80211, basically activating all the previous Extended Key ID patches when you are using linux 5.2 or greater with a compatible card. As of now all cards using mac80211 and NOT providing the "set_key()" callback are compatible. There is also a patch allowing to use it with most iwlwifi cards: https://patchwork.kernel.org/patch/11101659/ There is an API bug in 5.2 kernels < 5.2.12, preventing unicast keys using keyid 1 to be installed for FILS and FT handshakes. This patch set therefore requires a kernel >= 5.2.12. 16-Drop set_tx from all set_key() functions 17-hostapd: Extended Key ID stress test 5) The last two patches, (16+17) are kind of optional. Patch 16 is finishing the work the first 9 patches started. Since redesigning the set_tx API is the most invasive part of the patch series it is sometimes interesting to see how the old set_tx key calls would have acted. Patch 17 is so far just a proof that everything still works when we use keyid 1 at the initial connect instead of the more traditional keyid 0. I have some ideas to extend that, allowing hostapd to detect broken STAs claiming to be compatible with Extended Key ID - due to broken capability announcement - and still work with APs supporting it. (At least the Samsung Galaxy Tab S3 is setting the RSN capabilities wrong and chances are it's not the only one...) Besides using the hostapd tests I've also verified the function with a real AP using a patched iwlwifi driver with HW crypto offload. The integration of Extended Key ID support in hostapd/wpa_supplicant tries to maintain full backward compatibility with kernels not supporting Extended Key ID. While this is foreseen in the standard for normal operation it was more challenging for the built-in test cases. Instead of basically doubling all test cases and execute them with and without Extended Key ID support I've just modified all test cases to be able to run with and without Extended Key ID support. All tests will per default use Extended Key ID when supported or silently fall back to the "classical" way to do things. The downside here is, that the same tests executed on a system supporting Extended Key ID will test something slightly different than a system not supporting it... At the moment there are just four special Extended Key ID tests to verify it's being used and we interact correctly with STAs not supporting it. To make it a bit simpler to see what the patches are changing I've uploaded you some files containing multiple full test runs: 1) 21dc1627f-orig.tgz Tests run with a unpatched hostapd (git version 21dc1627f) 2) 21dc1627f-patched.tgz Test run with hwsim supporting Extended Key ID and all patches of the of this series applied. 3) 21dc1627f-patched-legacy.tgz All patches of this series applied. With Extended Key ID detection sabotaged to force classical rekeys. Each file is roughly 190MB big and can be downloaded here: https://www.awhome.eu/index.php/s/jDQfpZmDGwsLPXA (The test runs from the older patch sets are also available there) All tests were executed running wt-2019-09-04 from wireless-testing. (The kernel has two additional patched on top of the official version to use Extended Key ID with iwlwifi, and a debug patch with printk's informing me e.g. when installing, deleting or activating any PTK. Most of the failed tests are related to a wrong Python call to AES encrypt data, missing one argument. (I'm using Python 3.6.9) All failing tests seem to be unrelated to the changes here and can be reproduced without the patches. I've added two extra files in each test directory: - failed.log All tests failed in this run. With a comment when it worked when trying the test a second time. (But without adding the output from the rerun.) - script.log The output of the actual test run. Version history: Patch version 6: - Added the "missing" patch number 1. Should have been part of V3 but somehow it never got exported from my local tree... - Reverted the incorrect fix for max keyid in the patch "wpa_supplicant: AP Extended Key ID support" - Moved code disabling Extended Key ID for tkip to a better location in patch "hostapd: Add support for Extended Key ID" - Tweaked Extended Key ID configuration handling in "hostapd: Add support for Extended Key ID": It's now automatically disabled - Some minor tweaks to comments and output and a bit more invasive rewording of commit messages Patch version 5: - undo incorrect conversion from "int key_type" to "enum key_type key_type" (this is also an independent variable) - fixed multiple broken "enum key_type" which must be "enum key_type key_type" - undo pointless variable move in "wpa_supplicant: FILS Extended Key ID support" which did break build tests. - again minor fixes to commit messages Patch version 4: - Fixed sequence number key queries (undo invalid switch to new API names.) - get rid of some forgotten "int key_flag" references and switched them over to the correct "enum key_flag key_flag" - Dropped the workaround patch enabling FILS and FT key installs with current released kernels (API is being updated upstream) - removed some stray newlines - reworded Extended Key ID description in wpa_supplicant.conf - minor updates to commit messages - Overhauled and updated commit remarks (nothing critical) Patch version 3: - use key_type (enum) instead of key_flag (bit array) - migrate set_tx to the new key_type API and clean up related workarounds and bugs. - also support Extended Key ID for OSEN, FT and FILS - wpa_supplicant is now checking RSN capabilities instead of the presence of the KeyID KDE to enable Extended Key ID support to follow the wording of the standard - detect and prevent dropping or adding Extended Key ID support on rekey. (Initial connection decides if we can use it.) - Don't use Extended Key ID for TKIP. (Not allowed in the standard.) - also flip KeyIDs for WPA_REAUTH and WPA_REAUTH_EAPOL - fixed many "test-only" bugs (both tests and test only code) - split fixes and extensions in different patches - dropped "GET drv_flags" and handle it differently - better description of the patches Patch version 2: - typo fix in commit message Patch version 1: Initial release Alexander Wetzel (17): hostapd: Provide correct keyid to wpa_send_eapol() nl80211: Migrate to current netlink key message format Driver: Add key_type and Extended Key ID driver flag Add new argument key_type to all set_key() functions hostapd: Set the correct key_type for key installs wpa_supplicant: Set the correct key_type for key installs drivers: Migrate drivers from set_tx to key_type API nl80211: Switch to the new key_type API & cleanup nl80211,wpa_supplicant: Drop outdated tdls hack hostapd: Add support for Extended Key ID wpa_supplicant: AP Extended Key ID support tests: Extended Key ID tests hostapd: FILS Extended Key ID support wpa_supplicant: FILS Extended Key ID support nl80211: Extended Key ID support Drop set_tx from all set_key() functions hostapd: Extended Key ID stress test hostapd/config_file.c | 2 + hostapd/ctrl_iface.c | 68 +++++++----- hostapd/hostapd.conf | 10 ++ src/ap/ap_config.c | 10 ++ src/ap/ap_config.h | 1 + src/ap/ap_drv_ops.c | 9 +- src/ap/ap_drv_ops.h | 5 +- src/ap/hostapd.c | 17 +-- src/ap/hs20.c | 2 + src/ap/ieee802_11.c | 6 +- src/ap/ieee802_1x.c | 14 +-- src/ap/wpa_auth.c | 110 ++++++++++++++----- src/ap/wpa_auth.h | 4 +- src/ap/wpa_auth_ft.c | 13 ++- src/ap/wpa_auth_glue.c | 38 ++++++- src/ap/wpa_auth_i.h | 3 + src/ap/wpa_auth_ie.c | 50 ++++++++- src/common/wpa_common.c | 1 + src/common/wpa_common.h | 9 ++ src/drivers/driver.h | 33 ++++-- src/drivers/driver_atheros.c | 10 +- src/drivers/driver_bsd.c | 13 +-- src/drivers/driver_hostap.c | 9 +- src/drivers/driver_ndis.c | 25 +++-- src/drivers/driver_nl80211.c | 164 ++++++++++++++++------------- src/drivers/driver_nl80211_capa.c | 4 + src/drivers/driver_openbsd.c | 5 +- src/drivers/driver_privsep.c | 12 +-- src/drivers/driver_wext.c | 27 ++--- src/drivers/driver_wext.h | 4 +- src/rsn_supp/tdls.c | 9 +- src/rsn_supp/wpa.c | 152 ++++++++++++++++++++++---- src/rsn_supp/wpa.h | 9 +- src/rsn_supp/wpa_ft.c | 17 +-- src/rsn_supp/wpa_i.h | 12 ++- src/rsn_supp/wpa_ie.c | 11 ++ src/rsn_supp/wpa_ie.h | 1 + tests/hwsim/hostapd.py | 19 +++- tests/hwsim/test_ap_ciphers.py | 2 +- tests/hwsim/test_ap_eap.py | 20 +++- tests/hwsim/test_ap_psk.py | 124 ++++++++++++++++++++-- tests/hwsim/test_ocv.py | 13 ++- tests/hwsim/test_rrm.py | 7 +- tests/hwsim/wpasupplicant.py | 7 +- wlantest/bss.c | 4 +- wlantest/rx_eapol.c | 4 + wpa_supplicant/config.c | 2 + wpa_supplicant/config_file.c | 1 + wpa_supplicant/config_ssid.h | 10 ++ wpa_supplicant/ctrl_iface.c | 36 ++++--- wpa_supplicant/driver_i.h | 23 ++-- wpa_supplicant/ibss_rsn.c | 19 ++-- wpa_supplicant/mesh_mpm.c | 12 +-- wpa_supplicant/mesh_rsn.c | 14 +-- wpa_supplicant/preauth_test.c | 2 +- wpa_supplicant/wpa_cli.c | 3 +- wpa_supplicant/wpa_supplicant.c | 38 +++++-- wpa_supplicant/wpa_supplicant.conf | 5 + wpa_supplicant/wpas_glue.c | 24 +++-- 59 files changed, 947 insertions(+), 331 deletions(-)