1. 18 Jan, 2016 3 commits
  2. 09 Nov, 2015 28 commits
  3. 25 Sep, 2015 9 commits
    • Kang Xu's avatar
      Release 4.5.20.020 · 4b141611
      Kang Xu authored
      Change-Id: Ie42923cd476f6f4ea49b13e42b200dfc69273812
      CRs-Fixed: 710831
      4b141611
    • Abhishek Singh's avatar
      qcacld: Set the hdd state to not connected if disconnect timeout. · bbc45ec8
      Abhishek Singh authored
      If disconnect command is queued in SME and same time the connect
      is in progress which failed eventualy, driver will not call the
      disconnect handler leading to the wait for disconnect handler
      timeout. This timeout will leave the hdd state as disconneting
      and thus the next connect commands will always try to disconnect
      thinking that disconnect is still in progress leading to connect
      return with failure.
      With this fix hdd will set the hdd state from disconnecting to
      not connected even if disconnect timeout.
      
      Change-Id: I349bfe129d476894bc60615477818245cb131a9b
      CRs-Fixed: 832623
      bbc45ec8
    • Srinivas Girigowda's avatar
      qcacld: Set connState to not connected, if disconnect cmd is not queued · d9ed7945
      Srinivas Girigowda authored
      The analysis of this issue is, if the function sme_RoamDisconnect returns
      eHAL_STATUS_CMD_NOT_QUEUED (which means command not queued because there
      is no connection). At this point state m/c is set to Disconnecting and
      we are starting the disconnect timer which timesout and returns -EETIMEDOUT
      and hence connState is not set to eConnectionState_NotConnected.
      
      When the connection state "Connecting", it can move to either "Connected"
      or "Disconnected". Here since there is a connection failure, we should be
      moving to "Disconnected". Upon connection failure and hdd disconnect in
      progress, we don't move to "Not connected" state and state is stuck at
      "Connection in progress". Hence scan failure.
      
      To fix this issue, if sme_RoamDisconnect returns eHAL_STATUS_CMD_NOT_QUEUED
      move the state to "Not connected"
      
      Change-Id: I174eda7eec679f5861946d04c78ad1e76a17975e
      CRs-Fixed: 728064
      d9ed7945
    • Chandrasekaran, Manishekar's avatar
      qcacld: Abort scan for SSID if deauth is received from supplicant. · 007ee031
      Chandrasekaran, Manishekar authored
      During connection if AP is not found in scan list, driver tries
      to scan for the ssid (eCsrScanForSsid) and if AP is found,
      the connect cmd is started.
      
      During this scan if Disconnect comes from supplicant, as there is
      no ongoing connect command and sme is not in connected state, the
      disconnect command is dropped in SME and hdd is indicated success.
      This leads to HDD waiting on the event disconnect_comp_var and
      connState to eConnectionState_Disconnecting.
      
      After this if supplicant send connect and as HDD is in
      disconnecting state it will again wait for disconnect_comp_var
      and return failure after timeout and this will continue and user
      will not able to connect.
      
      If disconnect is received during eCsrScanForSsid abort/dequeue
      the eCsrScanForSsid in SME and in HDD set the connState to
      eConnectionState_NotConnected and set disconnect_comp_var
      complete.
      
      Change-Id: I1da89065e1e59bcf08806179f85e3d9d00d5828e
      CRs-Fixed: 689890
      007ee031
    • kaliu's avatar
      qcacld-2.0: segment mdns resp txt content by '/' · c12d5029
      kaliu authored
      When sap respond mdns request, the txt content should be segment
      by '/', otherwise all content will be sent in one segment.
      
      Change-Id: I64ab0062ced3fd028ceb0da7a3a2cb737bf726cb
      CRs-Fixed: 903983
      c12d5029
    • Kang Xu's avatar
      Release 4.5.20.19 · 4fa18384
      Kang Xu authored
      Change-Id: Ib840822e581a6495c99b65c1531de3a83ef09a32
      CRs-Fixed: 710831
      4fa18384
    • Kai Liu's avatar
      qcacld-2.0: align psk with 4 bytes in wmi WMI_SAP_OFL_ENABLE_CMDID · d712a344
      Kai Liu authored
      In sap auth offload enabled case, if psk isn't 4 bytes aligned,
      tlv check of WMI_SAP_OFL_ENABLE_CMDID will fail.
      
      Change-Id: Icd6c437f8d7705262e8853b9c348f950b04790b5
      CRs-Fixed: 903134
      d712a344
    • Kang Xu's avatar
      Release 4.5.20.18 · cbb47867
      Kang Xu authored
      Change-Id: Ie9551936a3ec0d2ce10a13b3188f09e0f9ccce9e
      CRs-Fixed: 710831
      cbb47867
    • Chandrasekaran, Manishekar's avatar
      qcacld: Clean-up remain on channel enqueue/dequeue mechanism · e31f9dec
      Chandrasekaran, Manishekar authored
      Clean-up enqueue/dequeue of remain on channel mechanism in
      the driver:
      
      (a) Instead of looping through all the elements in the roc
      queue and processing them back to back, process the roc
      requests, one at a time during dequeue to avoid drop of any
      roc request
      (b) Ensure that the roc completion is indicated at the right
      instance using 'is_roc_inprogress' flag so that the any new
      roc request can be immediately processed without being
      queued
      (c) Remove redundant prints from roc dequeue function
      (d) Ensure that 'hdd_remain_on_chan_cancel_in_progress'
      flag is considered during the clean up of roc context
      (e) Ensure that any new roc request is not queued while an
      roc is in progress. This will ensure that the new roc will
      not get dropped immediately
      (f) Ensure that the supplicant is notified when an roc
      is dropped within the driver
      
      Change-Id: I93ad8e5bbcecf0389efc26ae8d741826f4cf6018
      CRs-Fixed: 895394
      e31f9dec