i40e: synchronize nvmupdate command and adminq subtask
authorSudheer Mogilappagari <sudheer.mogilappagari@intel.com>
Wed, 12 Jul 2017 09:46:07 +0000 (05:46 -0400)
committerJeff Kirsher <jeffrey.t.kirsher@intel.com>
Fri, 25 Aug 2017 21:52:52 +0000 (14:52 -0700)
commit2bf01935ec5362aee6ff9ffc2476043af321aa42
treeccd7653782c567b8b8a0762ffc3fae379a249737
parent06b2decd924891b6c7570a91f91e11a5a8fed421
i40e: synchronize nvmupdate command and adminq subtask

During NVM update, state machine gets into unrecoverable state because
i40e_clean_adminq_subtask can get scheduled after the admin queue
command but before other state variables are updated. This causes
incorrect input to i40e_nvmupd_check_wait_event and state transitions
don't happen.

This issue existed before but surfaced after commit 373149fc99a0
("i40e: Decrease the scope of rtnl lock")

This fix adds locking around admin queue command and update of
state variables so that adminq_subtask will have accurate information
whenever it gets scheduled.

Signed-off-by: Sudheer Mogilappagari <sudheer.mogilappagari@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
drivers/net/ethernet/intel/i40e/i40e_nvm.c