[ldv-project] [net] wcn36xx: potential race condition

[Pavel Andrianov]

Hi!

There is a potential race condition in 
drivers/net/wireless/ath/wcn36xx/wcn36xx.ko. In wcn36xx_tx -> 
wcn36xx_start_tx -> wcn36xx_set_tx_data 
(http://lxr.free-electrons.com/source/drivers/net/wireless/ath/wcn36xx/txrx.c#L176) 
there is a read of sta_priv->bss_dpu_desc_index and 
sta_priv->bss_sta_index. In wcn36xx_bss_info_changed -> 
wcn36xx_smd_config_bss -> wcn36xx_smd_config_bss_rsp 
(http://lxr.free-electrons.com/source/drivers/net/wireless/ath/wcn36xx/smd.c#L1204) 
there is a write to the same fields. It  seems that the handlers may be 
called in parallel and inconsistent data may be obtained.
The same problem is with sta_priv->sta_index and 
sta_priv->sta_dpu_desc_index:
http://lxr.free-electrons.com/source/drivers/net/wireless/ath/wcn36xx/txrx.c#L181 

http://lxr.free-electrons.com/source/drivers/net/wireless/ath/wcn36xx/smd.c#L986 

Is it a real bug? Is it enough to add mutex_lock to wcn36xx_set_tx_data?

-- 
Pavel Andrianov
Linux Verification Center, ISPRAS
web: http://linuxtesting.org
e-mail: andrianov at ispras.ru