On Wed, Aug 16, 2017 at 05:19:28PM +0300, Matan Azrad wrote:
> The corrupted code couldn't recognize that all sub devices
> were not ready for tx traffic when failsafe PMD was trying
> to switch device because of an unreachable condition using.
>
> Hence, the current tx sub device variable was not updated
> correctly.
>
> The fix removed the unreachable branch and added new one
> in the right place respecting the original intent.
>
> Fixes: ebea83f899d8 ("net/failsafe: add plug-in support")
> Fixes: 598fb8aec6f6 ("net/failsafe: support device removal")
> Cc: sta...@dpdk.org
>
> Signed-off-by: Matan Azrad <ma...@mellanox.com>
> Acked-by: Gaetan Rivet <gaetan.ri...@6wind.com>
> ---
> drivers/net/failsafe/failsafe_private.h | 9 +++++----
> 1 file changed, 5 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/net/failsafe/failsafe_private.h
> b/drivers/net/failsafe/failsafe_private.h
> index 0361cf4..ef646db 100644
> --- a/drivers/net/failsafe/failsafe_private.h
> +++ b/drivers/net/failsafe/failsafe_private.h
> @@ -334,7 +334,7 @@ fs_switch_dev(struct rte_eth_dev *dev,
> } else if ((txd && txd->state < req_state) ||
> txd == NULL ||
> txd == banned) {
> - struct sub_device *sdev;
> + struct sub_device *sdev = NULL;
Good catch, actually this makes me think that the FOREACH_SUBDEV_STATE
macro is not following the usual tailq API (which sets the iterator to
NULL upon terminating). This can throw-off the unsuspecting writer.
I will fix this soon™. In the meantime, sdev should be initialized to
NULL.
> uint8_t i;
>
> /* Using acceptable device */
> @@ -346,9 +346,10 @@ fs_switch_dev(struct rte_eth_dev *dev,
> PRIV(dev)->subs_tx = i;
> break;
> }
> - } else if (txd && txd->state < req_state) {
> - DEBUG("No device ready, deactivating tx_dev");
> - PRIV(dev)->subs_tx = PRIV(dev)->subs_tail;
> + if (i >= PRIV(dev)->subs_tail || sdev == NULL) {
> + DEBUG("No device ready, deactivating tx_dev");
> + PRIV(dev)->subs_tx = PRIV(dev)->subs_tail;
> + }
> } else {
> return;
> }
> --
> 2.7.4
>
--
Gaëtan Rivet
6WIND
