On 08/12/2019 07:31 AM, Sun Ke wrote:
> When try to get the lock failed, before return, execute the
> nbd_config_put() to decrease the nbd->config_refs.
>
> If the nbd->config_refs is added but not decreased. Then will not
> execute nbd_clear_sock() in nbd_config_put(). bd->task_setup will
> not be cleared away. Finally, print"Device being setup by another
> task" in nbd_add_sock() and nbd device can not be reused.
>
> Fixes: 8f3ea35929a0 ("nbd: handle unexpected replies better")
> Signed-off-by: Sun Ke <sunke32@xxxxxxxxxx>
> ---
> drivers/block/nbd.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c
> index e21d2de..a69a90a 100644
> --- a/drivers/block/nbd.c
> +++ b/drivers/block/nbd.c
> @@ -357,8 +357,10 @@ static enum blk_eh_timer_return nbd_xmit_timeout(struct request *req,
> }
> config = nbd->config;
>
> - if (!mutex_trylock(&cmd->lock))
> + if (!mutex_trylock(&cmd->lock)) {
> + nbd_config_put(nbd);
> return BLK_EH_RESET_TIMER;
> + }
>
> if (config->num_connections > 1) {
> dev_err_ratelimited(nbd_to_dev(nbd),
>
I just sent the same patch
https://www.spinics.net/lists/linux-block/msg43718.html
here
https://www.spinics.net/lists/linux-block/msg43715.html
so it looks good to me.
Reviewed-by: Mike Christie <mchristi@xxxxxxxxxx>
