From: "Ananyev, Konstantin" <konstantin.ananyev@intel.com>
To: Matan Azrad <matan@mellanox.com>, Aaron Conole <aconole@redhat.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>,
"Yigit, Ferruh" <ferruh.yigit@intel.com>,
Marcelo Leitner <mleitner@redhat.com>,
Shahaf Shuler <shahafs@mellanox.com>,
Ori Kam <orika@mellanox.com>,
Thomas Monjalon <thomas@monjalon.net>
Subject: Re: [dpdk-dev] [PATCH] ethdev: move sanity checks to non-debug paths
Date: Mon, 23 Jul 2018 14:19:50 +0000 [thread overview]
Message-ID: <2601191342CEEE43887BDE71AB977258DF51B4D9@irsmsx105.ger.corp.intel.com> (raw)
In-Reply-To: <VI1PR0501MB2608D052D6902CB52C4952AED2560@VI1PR0501MB2608.eurprd05.prod.outlook.com>
> -----Original Message-----
> From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Matan Azrad
> Sent: Monday, July 23, 2018 1:14 PM
> To: Aaron Conole <aconole@redhat.com>
> Cc: dev@dpdk.org; Yigit, Ferruh <ferruh.yigit@intel.com>; Marcelo Leitner <mleitner@redhat.com>; Shahaf Shuler
> <shahafs@mellanox.com>; Ori Kam <orika@mellanox.com>; Thomas Monjalon <thomas@monjalon.net>
> Subject: Re: [dpdk-dev] [PATCH] ethdev: move sanity checks to non-debug paths
>
>
> Hi Aaron
> From: Aaron Conole
> > Sent: Monday, July 23, 2018 2:52 PM
> > To: Matan Azrad <matan@mellanox.com>
> > Cc: dev@dpdk.org; Ferruh Yigit <ferruh.yigit@intel.com>; Marcelo Leitner
> > <mleitner@redhat.com>; Shahaf Shuler <shahafs@mellanox.com>; Ori Kam
> > <orika@mellanox.com>; Thomas Monjalon <thomas@monjalon.net>
> > Subject: Re: [dpdk-dev] [PATCH] ethdev: move sanity checks to non-debug paths
> >
> > Matan Azrad <matan@mellanox.com> writes:
> >
> > > Hi Aaron
> > >
> > > From: Aaron Conole
> > >> These checks would have prevented a reported crash in the field. If
> > >> a user builds without ETHDEV_DEBUG, it should make their application
> > >> more stable, not less.
> > >>
> > >> Many of these functions immediately dereference arrays based on the
> > >> passed in values, so the sanity checks are quite important.
> > >>
> > >
> > > These functions are datapath functions.
> > > Do you really want to add more 3 checks + calculations per each burst call?
> > > Did you check the performance impact?
> > > I think that performance numbers must be added for the discussion of this
> > patch.
> >
> > I'll dig up performance numbers - but performance doesn't mean anything if
> > the application isn't running any longer due to crash.
>
> Yes, I understand your point, but think about that, if we are going to defend each user mistake it will cost a lot.
> For example in Tx path, Adding checks for each mbuf pointer and mbuf data validity will be very expensive.
>
> I think the best way is to check the common user mistakes in DEBUG mode to help for application debugging and that's it.
+1
The problem is that user provided an invalid input parameters.
Adding just extra checks inside data-path functions wouldn't solve it.
Konstantin
>
> > >> The logs are left as DEBUG only.
> > >>
> > >> Cc: Marcelo Leitner <mleitner@redhat.com>
> > >> Signed-off-by: Aaron Conole <aconole@redhat.com>
> > >> ---
> > >> lib/librte_ethdev/rte_ethdev.h | 29 +++++++++++++----------------
> > >> 1 file changed, 13 insertions(+), 16 deletions(-)
> > >>
> > >> diff --git a/lib/librte_ethdev/rte_ethdev.h
> > >> b/lib/librte_ethdev/rte_ethdev.h index f5f593b31..bfd6a3406 100644
> > >> --- a/lib/librte_ethdev/rte_ethdev.h
> > >> +++ b/lib/librte_ethdev/rte_ethdev.h
> > >> @@ -3805,15 +3805,16 @@ rte_eth_rx_burst(uint16_t port_id, uint16_t
> > >> queue_id,
> > >> struct rte_eth_dev *dev = &rte_eth_devices[port_id];
> > >> uint16_t nb_rx;
> > >>
> > >> -#ifdef RTE_LIBRTE_ETHDEV_DEBUG
> > >> RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
> > >> RTE_FUNC_PTR_OR_ERR_RET(*dev->rx_pkt_burst, 0);
> > >>
> > >> if (queue_id >= dev->data->nb_rx_queues) {
> > >> +#ifdef RTE_LIBRTE_ETHDEV_DEBUG
> > >> RTE_ETHDEV_LOG(ERR, "Invalid RX queue_id=%u\n",
> > queue_id);
> > >> +#endif
> > >> return 0;
> > >> }
> > >> -#endif
> > >> +
> > >> nb_rx = (*dev->rx_pkt_burst)(dev->data->rx_queues[queue_id],
> > >> rx_pkts, nb_pkts);
> > >>
> > >> @@ -3928,14 +3929,12 @@ rte_eth_rx_descriptor_status(uint16_t
> > >> port_id, uint16_t queue_id,
> > >> struct rte_eth_dev *dev;
> > >> void *rxq;
> > >>
> > >> -#ifdef RTE_LIBRTE_ETHDEV_DEBUG
> > >> RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV); -#endif
> > >> dev = &rte_eth_devices[port_id];
> > >> -#ifdef RTE_LIBRTE_ETHDEV_DEBUG
> > >> +
> > >> if (queue_id >= dev->data->nb_rx_queues)
> > >> return -ENODEV;
> > >> -#endif
> > >> +
> > >> RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_descriptor_status, -
> > >> ENOTSUP);
> > >> rxq = dev->data->rx_queues[queue_id];
> > >>
> > >> @@ -3985,14 +3984,12 @@ static inline int
> > >> rte_eth_tx_descriptor_status(uint16_t port_id,
> > >> struct rte_eth_dev *dev;
> > >> void *txq;
> > >>
> > >> -#ifdef RTE_LIBRTE_ETHDEV_DEBUG
> > >> RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV); -#endif
> > >> dev = &rte_eth_devices[port_id];
> > >> -#ifdef RTE_LIBRTE_ETHDEV_DEBUG
> > >> +
> > >> if (queue_id >= dev->data->nb_tx_queues)
> > >> return -ENODEV;
> > >> -#endif
> > >> +
> > >> RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->tx_descriptor_status, -
> > >> ENOTSUP);
> > >> txq = dev->data->tx_queues[queue_id];
> > >>
> > >> @@ -4071,15 +4068,15 @@ rte_eth_tx_burst(uint16_t port_id, uint16_t
> > >> queue_id, {
> > >> struct rte_eth_dev *dev = &rte_eth_devices[port_id];
> > >>
> > >> -#ifdef RTE_LIBRTE_ETHDEV_DEBUG
> > >> RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, 0);
> > >> RTE_FUNC_PTR_OR_ERR_RET(*dev->tx_pkt_burst, 0);
> > >>
> > >> if (queue_id >= dev->data->nb_tx_queues) {
> > >> +#ifdef RTE_LIBRTE_ETHDEV_DEBUG
> > >> RTE_ETHDEV_LOG(ERR, "Invalid TX queue_id=%u\n",
> > queue_id);
> > >> +#endif
> > >> return 0;
> > >> }
> > >> -#endif
> > >>
> > >> #ifdef RTE_ETHDEV_RXTX_CALLBACKS
> > >> struct rte_eth_rxtx_callback *cb = dev->pre_tx_burst_cbs[queue_id];
> > >> @@ -4160,23 +4157,23 @@ rte_eth_tx_prepare(uint16_t port_id, uint16_t
> > >> queue_id, {
> > >> struct rte_eth_dev *dev;
> > >>
> > >> -#ifdef RTE_LIBRTE_ETHDEV_DEBUG
> > >> if (!rte_eth_dev_is_valid_port(port_id)) {
> > >> +#ifdef RTE_LIBRTE_ETHDEV_DEBUG
> > >> RTE_ETHDEV_LOG(ERR, "Invalid TX port_id=%u\n", port_id);
> > >> +#endif
> > >> rte_errno = -EINVAL;
> > >> return 0;
> > >> }
> > >> -#endif
> > >>
> > >> dev = &rte_eth_devices[port_id];
> > >>
> > >> -#ifdef RTE_LIBRTE_ETHDEV_DEBUG
> > >> if (queue_id >= dev->data->nb_tx_queues) {
> > >> +#ifdef RTE_LIBRTE_ETHDEV_DEBUG
> > >> RTE_ETHDEV_LOG(ERR, "Invalid TX queue_id=%u\n",
> > queue_id);
> > >> +#endif
> > >> rte_errno = -EINVAL;
> > >> return 0;
> > >> }
> > >> -#endif
> > >>
> > >> if (!dev->tx_pkt_prepare)
> > >> return nb_pkts;
> > >> --
> > >> 2.14.3
next prev parent reply other threads:[~2018-07-23 14:20 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-07-22 11:21 Aaron Conole
2018-07-23 6:12 ` Matan Azrad
2018-07-23 11:51 ` Aaron Conole
2018-07-23 12:13 ` Matan Azrad
2018-07-23 14:19 ` Ananyev, Konstantin [this message]
2018-07-23 20:07 ` Andrew Rybchenko
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=2601191342CEEE43887BDE71AB977258DF51B4D9@irsmsx105.ger.corp.intel.com \
--to=konstantin.ananyev@intel.com \
--cc=aconole@redhat.com \
--cc=dev@dpdk.org \
--cc=ferruh.yigit@intel.com \
--cc=matan@mellanox.com \
--cc=mleitner@redhat.com \
--cc=orika@mellanox.com \
--cc=shahafs@mellanox.com \
--cc=thomas@monjalon.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).