1. 20 Apr, 2020 3 commits
  2. 02 Apr, 2020 1 commit
    • Adam Dov's avatar
      prplmesh: move the load_steer_on_vaps to the UCI · 98ba3cfa
      Adam Dov authored
      The load_steer_on_vaps is required to be platform-specific. Currently,
      this is supported by compilation flags, but the platform-specific
      configuration should be supported using the UCI.
      Commit 129d4450 introduced the per radio
      hostap_iface_steer_vaps param to the prplmesh config, and the
      uci-defaults scripts adjustments.
      Construct the load_steer_on_vaps from the above-mentioned param read
      from UCI for all radios.
      Signed-off-by: Adam Dov's avatarAdam Dov <adamx.dov@intel.com>
  3. 30 Mar, 2020 1 commit
  4. 29 Mar, 2020 1 commit
    • Adam Dov's avatar
      iwlwav: prplmesh: config: add hostap_iface_steer_vaps to UCI · 129d4450
      Adam Dov authored
      Preparative Commit.
      Currently, the load_steer_on_vaps that is used by the optimal-path-task
      is read from the beerocks_controller.conf and is constructed differently
      for each platform according to compilation flags.
      Added the hostap_iface_steer_vaps param to each radio in the
      prplmesh config file. The load_steer_on_vaps will be constructed from
      the hostap_iface_steer_vaps of all radios in UCI.
      For RAX40, override the UCI defaults using the existing uci-defaults
      Signed-off-by: Adam Dov's avatarAdam Dov <adamx.dov@intel.com>
  5. 04 Mar, 2020 4 commits
    • Tomer Eliyahu's avatar
      prplmesh: use /etc/init.d/network reload · ead15818
      Tomer Eliyahu authored
      Using /etc/init.d/network restart causes a full restart for the network,
      including wireless interfaces which takes a long time (over 1 minute due
      to CAC).
      This especially hurts our CI which makes every test take over a minute
      longer due to this restart.
      The reason /etc/init.d/network restart is needed is only to manipulate
      the bridge - by adding or removing the wired interface and
      adding/removing ip addresses.
      As it appears, these operations are implemented in reload, so no need
      for full restart - therefore change all occurnces of /etc/init.d/network
      restart to /etc/init.d/network reload
      Signed-off-by: Tomer Eliyahu's avatarTomer Eliyahu <tomer.b.eliyahu@intel.com>
    • Tomer Eliyahu's avatar
      prplmesh: add certification_mode to init script · e9990651
      Tomer Eliyahu authored
      This commit refactors the prplmesh init script to support switching to
      certification mode with a single command:
      `/etc/init.d/prplmesh certification_mode controller` will switch to controller
      certification mode
      `/etc/init.d/prplmesh certification_mode agent` will switch to agent
      certification mode.
      Certification mode adds the ip address to the br-lan as
      an alias ip, and performs the necessary changes to the network and prplmesh
      configuration for certification as controller or agent.
      /etc/init.d/network restart is needed since when we move from gateway to
      repearer mode or vice versa, we manupulate the bridge by adding or
      removing the wired interface to/from it.
      Signed-off-by: Tomer Eliyahu's avatarTomer Eliyahu <tomer.b.eliyahu@intel.com>
    • Raphaël Mélotte's avatar
      Revert "hack: prplmesh: only depend on iwlwav-driver-uci if not using SDK" · acc744ea
      Raphaël Mélotte authored
      The purpose of commit 'edf5c63e' was
      to include "iwlwav-driver-uci" only if we're not building with the
      However, it turns out it doesn't include "iwlwav-driver-uci" even when
      building with the full prplWrt tree (so not the SDK).
      Since we plan to not use the SDK anymore, remove the check.
      A better solution would be to find a check that works as it was
      originally intended:
      - if the SDK is used, do not include "iwlwav-driver-uci"
      - if the SDK is NOT used, include "iwlwav-driver-uci"
    • Tomer Eliyahu's avatar
      dwpal: fieldValuesGet: remove all ' ' characters · 103796aa
      Tomer Eliyahu authored
      The fieldValuesGet function is used by DWPAL to parse lists of fields of
      the format "btm_supported=1", SupportedRates-2 4 11 ", "SSID=", etc.
      As part of parsing, the implementation adds ' ' characters to the buffer
      for handling special cases, and removes them at the end.
      The for loop removing the ' ' characters has a bug since it doesn't take
      into account the first ' ', so if an empty value is given such as
      "SSID=", the returned value will be ' '.
      So, fix it by adding a patch for it in dwpal.
      Signed-off-by: Tomer Eliyahu's avatarTomer Eliyahu <tomer.b.eliyahu@intel.com>
  6. 25 Feb, 2020 1 commit
  7. 19 Feb, 2020 1 commit
  8. 06 Feb, 2020 2 commits
    • Adam Dov's avatar
      iwlwav:prplmesh:files add prplmesh update config script · d340b644
      Adam Dov authored
      The default configuration of prplmesh in openWrt is aligned to the
      Turris Omnia and the GL.Inet platforms that make use of wlan0 and wlan1
      as the 2 radio interfaces.
      On the Rax40, on the other hand, the wlan0 and wlan1 are used
      The openWrt provides a feature for boot-time-default-configuration
      scripts that are automatically triggered as part of the boot sequence.
      The above-mentioned scripts must be installed to /etc/uci-defaults/ for
      the script to be triggered at boot.
      The sequence of the above-mentioned scripts is determined by the script
      file name: X-<rest-of-file-name>, where X=2 is triggered after files
      with X=1 and before files with X=3.
      The added script determines which platform is booted according to the
      /tmp/sysinfo/board_name. If GRX appears in the name, the platform uses
      wlan2 as the second radio, and configuration is updated accordingly.
      Signed-off-by: Adam Dov's avatarAdam Dov <adamx.dov@intel.com>
    • Adam Dov's avatar
      iwlwav:prplmesh:config: remove radio2 from prplmesh config · 0de8b6be
      Adam Dov authored
      The default count of radios is 2, and wlan0 and wlan1 are,
      respectively, the 2 interfaces used by default.
      Removed radio2 (third radio) from the default configuration.
      Signed-off-by: Adam Dov's avatarAdam Dov <adamx.dov@intel.com>
  9. 05 Feb, 2020 1 commit
    • Adam Dov's avatar
      feeds:prplmesh:config - add hostap_iface to config file · 56fb9228
      Adam Dov authored
      To provide better flexibility and per platform configuration, the
      slaves radio configuration is moved to the prplmesh configuration
      file - which is platform-specific.
      The prplmesh configuration file is shared between the following
      RAX40, Turris Omnia, GL.Inet
      As such, the new configuration is used as a default configuration. If a
      platform requires a different configuration - a separate feed should be
      created for that platform.
      RAX40 use interfaces wlan0 and wlan2
      Turris Omnia and GL.Inet use interfaces wlan0 and wlan1
      This change is a prerequisite for the prplmesh task of moving slave
      radio configuration from beerocks_agent.conf to the prplmesh config
      in UCI:
      Signed-off-by: Adam Dov's avatarAdam Dov <adamx.dov@intel.com>
  10. 30 Jan, 2020 1 commit
  11. 28 Jan, 2020 1 commit
  12. 16 Jan, 2020 2 commits
  13. 09 Jan, 2020 1 commit
    • Tomer Eliyahu's avatar
      prplmesh: Makefile adjustments and bump version · bf1f859c
      Tomer Eliyahu authored
      Makefile fixes/changes:
      - Remove PKG_REV which is not used.
      - Fix PKG_VERSION to include the PKG_SOURCE_VERSION - this is needed so
      the downloaded tar.gz will have a different name when we bump the feed
      - Change PKG_RELEASE to match the current prplmesh release (1.4.0)
      - Add REVISION CMAKE flag to CMAKE_OPTIONS since in OpenWRT the .git
      folder is deleted so the revision cannot be determined by prplmesh cmake
      build system.
      - Remove obsolete patch for FORTIFY_SOURCE - this was already fixed in
      prplMesh cmake build.
      - Bump prplMesh version to latesr - eeaac321
      Signed-off-by: Tomer Eliyahu's avatarTomer Eliyahu <tomer.b.eliyahu@intel.com>
  14. 08 Jan, 2020 1 commit
  15. 29 Dec, 2019 2 commits
  16. 23 Dec, 2019 3 commits
  17. 17 Dec, 2019 1 commit
  18. 16 Dec, 2019 2 commits
  19. 13 Dec, 2019 1 commit
  20. 11 Dec, 2019 7 commits
  21. 06 Dec, 2019 1 commit
  22. 03 Dec, 2019 2 commits