On 8/26/2021 8:02 PM, Chengfeng Ye wrote:
> The lock pdata->phy_mutex may not be correctly released if the
> function return in these two branches, which may lead to
> deadlock if the lock is acquired again.
>
> Cc: cy...@connect.ust.hk
> Bugzilla ID: 776
>
> Signed-off-by: Chengfeng Ye <cy...@connect.ust.hk>
> ---
> drivers/net/axgbe/axgbe_phy_impl.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/net/axgbe/axgbe_phy_impl.c
> b/drivers/net/axgbe/axgbe_phy_impl.c
> index 02236ec192..cf39095ee4 100644
> --- a/drivers/net/axgbe/axgbe_phy_impl.c
> +++ b/drivers/net/axgbe/axgbe_phy_impl.c
> @@ -418,8 +418,10 @@ static int axgbe_phy_get_comm_ownership(struct
> axgbe_port *pdata)
> */
> pthread_mutex_lock(&pdata->phy_mutex);
>
> - if (phy_data->comm_owned)
> + if (phy_data->comm_owned) {
> + pthread_mutex_unlock(&pdata->phy_mutex);
> return 0;
> + }
>
> /* Clear the mutexes */
> XP_IOWRITE(pdata, XP_I2C_MUTEX, AXGBE_MUTEX_RELEASE);
> @@ -444,6 +446,7 @@ static int axgbe_phy_get_comm_ownership(struct axgbe_port
> *pdata)
> XP_IOWRITE(pdata, XP_MDIO_MUTEX, mutex_id);
>
> phy_data->comm_owned = 1;
> + pthread_mutex_unlock(&pdata->phy_mutex);
> return 0;
> }
>
>
'axgbe_phy_get_comm_ownership()' seems intended to keep the lock on success.
Please also check:
Fixes: c8c2296b5974 ("net/axgbe: fix double unlock")
Perhaps you can comment the 'axgbe_phy_get_comm_ownership()' function instead?
