Re: [dpdk-dev] [PATCH] ethdev: fix checking Rx/Tx queue status

["Zhang, Qi Z" <qi.z.zhang@intel.com>]

> -----Original Message-----
> From: Ananyev, Konstantin
> Sent: Friday, May 11, 2018 5:51 PM
> To: Zhang, Qi Z <qi.z.zhang@intel.com>; Wu, Yanglong
> <yanglong.wu@intel.com>; dev@dpdk.org
> Cc: Lu, Wenzhuo <wenzhuo.lu@intel.com>; Wu, Yanglong
> <yanglong.wu@intel.com>
> Subject: RE: [dpdk-dev] [PATCH] ethdev: fix checking Rx/Tx queue status
> 
> 
> 
> > -----Original Message-----
> > From: Zhang, Qi Z
> > Sent: Friday, May 11, 2018 10:24 AM
> > To: Ananyev, Konstantin <konstantin.ananyev@intel.com>; Wu, Yanglong
> > <yanglong.wu@intel.com>; dev@dpdk.org
> > Cc: Lu, Wenzhuo <wenzhuo.lu@intel.com>; Wu, Yanglong
> > <yanglong.wu@intel.com>
> > Subject: RE: [dpdk-dev] [PATCH] ethdev: fix checking Rx/Tx queue
> > status
> >
> >
> >
> > > -----Original Message-----
> > > From: Ananyev, Konstantin
> > > Sent: Friday, May 11, 2018 4:39 PM
> > > To: Wu, Yanglong <yanglong.wu@intel.com>; dev@dpdk.org
> > > Cc: Zhang, Qi Z <qi.z.zhang@intel.com>; Lu, Wenzhuo
> > > <wenzhuo.lu@intel.com>; Wu, Yanglong <yanglong.wu@intel.com>
> > > Subject: RE: [dpdk-dev] [PATCH] ethdev: fix checking Rx/Tx queue
> > > status
> > >
> > >
> > >
> > > > -----Original Message-----
> > > > From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Yanglong Wu
> > > > Sent: Friday, May 11, 2018 9:22 AM
> > > > To: dev@dpdk.org
> > > > Cc: Zhang, Qi Z <qi.z.zhang@intel.com>; Lu, Wenzhuo
> > > > <wenzhuo.lu@intel.com>; Wu, Yanglong <yanglong.wu@intel.com>
> > > > Subject: [dpdk-dev] [PATCH] ethdev: fix checking Rx/Tx queue
> > > > status
> > > >
> > > > Relax the check for queue setup, since some device may not update
> > > > queue states during dev_stop.
> > > >
> > > > Fixes: cac923cfea47 ("ethdev: support runtime queue setup")
> > > > Signed-off-by: Yanglong Wu <yanglong.wu@intel.com>
> > > > ---
> > > >  lib/librte_ethdev/rte_ethdev.c | 10 ++++++----
> > > >  1 file changed, 6 insertions(+), 4 deletions(-)
> > > >
> > > > diff --git a/lib/librte_ethdev/rte_ethdev.c
> > > > b/lib/librte_ethdev/rte_ethdev.c index a357ee09f..34379642a 100644
> > > > --- a/lib/librte_ethdev/rte_ethdev.c
> > > > +++ b/lib/librte_ethdev/rte_ethdev.c
> > > > @@ -1479,8 +1479,9 @@ rte_eth_rx_queue_setup(uint16_t port_id,
> > > uint16_t rx_queue_id,
> > > >  			RTE_ETH_DEV_CAPA_RUNTIME_RX_QUEUE_SETUP))
> > > >  		return -EBUSY;
> > > >
> > > > -	if (dev->data->rx_queue_state[rx_queue_id] !=
> > > > -		RTE_ETH_QUEUE_STATE_STOPPED)
> > > > +	if (dev->data->dev_started &&
> > > > +		(dev->data->rx_queue_state[rx_queue_id] !=
> > > > +			RTE_ETH_QUEUE_STATE_STOPPED))
> > > >  		return -EBUSY;
> > > >
> > > >  	rxq = dev->data->rx_queues;
> > > > @@ -1611,8 +1612,9 @@ rte_eth_tx_queue_setup(uint16_t port_id,
> > > uint16_t tx_queue_id,
> > > >  			RTE_ETH_DEV_CAPA_RUNTIME_TX_QUEUE_SETUP))
> > > >  		return -EBUSY;
> > > >
> > > > -	if (dev->data->tx_queue_state[tx_queue_id] !=
> > > > -		RTE_ETH_QUEUE_STATE_STOPPED)
> > > > +	if (dev->data->dev_started &&
> > > > +		(dev->data->tx_queue_state[tx_queue_id] !=
> > > > +			RTE_ETH_QUEUE_STATE_STOPPED))
> > > >  		return -EBUSY;
> > > >
> > > >  	txq = dev->data->tx_queues;
> > > > --
> > >
> > > That would break runtime queue setup.
> >
> > Em... would you explain more, I can't figure out why this break runtime
> queue setup so far.
> 
> Ah yes, it was me not reading code properly.
> So yep - you right  it wouldn't break things.
> 
> >
> > > Why not fix offending PMDs instead?
> >
> > Ixgbe, i40e have this issue, not sure what about others.
> > probably we should fix at rte_eth_dev_stop to save all PMD?
> 
> Probably yes, that's a good thing to do in long term anyway.

I'm thinking if it's better to move all queue state change into ethdev layer.
So there could be a patch to update queue states properly at dev_start/dev_stop/queue_start/queue_stop and maintainer help them self to cleanup PMD, and I think our proposed fix should be part of this.
But for now, can we just keep current simple fix?

Regards
Qi

> 
> >
> > >
> > > > 2.11.0
> 
> Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>