My recent restart related work has made this race more likely
to happen and we've now noticed it, but it seems that it was
always possible. The race is that the add stream work can be
scheduled just before a restart is scheduled and then execute
before the restart, accessing the device while it's doing the
restart and not accessible.
To fix this, check if the device is restarting and abort the
work in that case. Reschedule it after the restart as well.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Miri Korenblit <miriam.rachel.korenblit@intel.com>
Link: https://patch.msgid.link/20241228223206.9c30af039b4d.I1a32936776f8ba5e83dda0a68ffc2722d9d37950@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
IWL_INFO(mvm, "restart completed\n");
iwl_trans_finish_sw_reset(mvm->trans);
+
+ /* no need to lock, adding in parallel would schedule too */
+ if (!list_empty(&mvm->add_stream_txqs))
+ schedule_work(&mvm->add_stream_wk);
}
void iwl_mvm_mac_reconfig_complete(struct ieee80211_hw *hw,
struct iwl_mvm *mvm = container_of(wk, struct iwl_mvm,
add_stream_wk);
- mutex_lock(&mvm->mutex);
+ guard(mvm)(mvm);
+
+ /* will reschedule to run after restart */
+ if (test_bit(IWL_MVM_STATUS_HW_RESTART_REQUESTED, &mvm->status) ||
+ test_bit(IWL_MVM_STATUS_IN_HW_RESTART, &mvm->status))
+ return;
iwl_mvm_inactivity_check(mvm, IWL_INVALID_STA);
iwl_mvm_mac_itxq_xmit(mvm->hw, txq);
local_bh_enable();
}
-
- mutex_unlock(&mvm->mutex);
}
static int iwl_mvm_reserve_sta_stream(struct iwl_mvm *mvm,