DPDK patches and discussions
 help / color / mirror / Atom feed
From: Ferruh Yigit <ferruh.yigit@intel.com>
To: David Marchand <david.marchand@redhat.com>
Cc: dev <dev@dpdk.org>, dpdk stable <stable@dpdk.org>,
	"A.McLoughlin" <aideen.mcloughlin@intel.com>
Subject: Re: [dpdk-dev] [PATCH 2/3] net/pcap: fix transmit return count in error conditions
Date: Thu, 25 Jul 2019 12:01:26 +0100	[thread overview]
Message-ID: <ebf18ea8-0035-ce40-f522-5c82d20615dd@intel.com> (raw)
In-Reply-To: <CAJFAV8wRsQD6HsmTLw_NHZxOzes_NUHG15f3cUjcu-62LYwgsQ@mail.gmail.com>

On 7/25/2019 8:40 AM, David Marchand wrote:
> On Wed, Jul 24, 2019 at 8:36 PM Ferruh Yigit <ferruh.yigit@intel.com> wrote:
>>
>> On 7/24/2019 12:54 PM, David Marchand wrote:
>>> When a packet cannot be transmitted, the driver is supposed to free this
>>> packet and report it as handled.
>>> This is to prevent the application from retrying to send the same packet
>>> and ending up in a liveloop since the driver will never manage to send
>>> it.
>>>
>>> Fixes: 49a0a2ffd5db ("net/pcap: fix possible mbuf double freeing")
>>> Fixes: 6db141c91e1f ("pcap: support jumbo frames")
>>> CC: stable@dpdk.org
>>>
>>> Signed-off-by: David Marchand <david.marchand@redhat.com>
>>> ---
>>>  drivers/net/pcap/rte_eth_pcap.c | 18 ++++++++++--------
>>>  1 file changed, 10 insertions(+), 8 deletions(-)
>>>
>>> diff --git a/drivers/net/pcap/rte_eth_pcap.c b/drivers/net/pcap/rte_eth_pcap.c
>>> index 470867d..5e5aab7 100644
>>> --- a/drivers/net/pcap/rte_eth_pcap.c
>>> +++ b/drivers/net/pcap/rte_eth_pcap.c
>>> @@ -354,7 +354,8 @@ struct pmd_devargs_all {
>>>                                       mbuf->pkt_len,
>>>                                       RTE_ETHER_MAX_JUMBO_FRAME_LEN);
>>>
>>> -                             break;
>>> +                             rte_pktmbuf_free(mbuf);
>>> +                             continue;
>>
>> +1
>> Very recently 'rte_pktmbuf_free()' was moved because it wasn't compatible with
>> return value, but this looks better, to free the mbuf and record it as error.
>>
>>>                       }
>>>               }
>>>
>>> @@ -373,7 +374,7 @@ struct pmd_devargs_all {
>>>       dumper_q->tx_stat.bytes += tx_bytes;
>>>       dumper_q->tx_stat.err_pkts += nb_pkts - num_tx;
>>>
>>> -     return num_tx;
>>> +     return nb_pkts;
>>>  }
>>>
>>>  /*
>>> @@ -439,14 +440,15 @@ struct pmd_devargs_all {
>>>                                       mbuf->pkt_len,
>>>                                       RTE_ETHER_MAX_JUMBO_FRAME_LEN);
>>>
>>> -                             break;
>>> +                             rte_pktmbuf_free(mbuf);
>>> +                             continue;
>>>                       }
>>>               }
>>>
>>> -             if (unlikely(ret != 0))
>>> -                     break;
>>> -             num_tx++;
>>> -             tx_bytes += mbuf->pkt_len;
>>> +             if (ret == 0) {
>>> +                     num_tx++;
>>> +                     tx_bytes += mbuf->pkt_len;
>>> +             }
>>
>> I don't know this part, this is in 'eth_pcap_tx()' which writes packets to the
>> interfaces.
>>
>> if 'pcap_sendpacket()' fails this doesn't mean packet can't be sent and may
>> cause a liveloop. Why not keep the existing behavior and let application to decide?
> 
> The manual is not clear to me.
> Do we really have temporary situations where retries are fine?

Same here, I am not that clear, I assume these situations exists, specially
taking into account nobody complaining about existing implementation.

> and if
> so, can we differentiate them from things like incorrect permissions
> etc...
> 
> If we can't, dropping is safer.
> 

May not differentiate them, because all we are getting is the fail from API
without reason.

But overall instead of driving freeing a packet, it feels better to leave this
decision to application unless driver needs to free it.

Other changes in the patchset is fixing defects in driver, only this part
changes behavior without a clear defect, I am leaning to keep old behavior.

  reply	other threads:[~2019-07-25 11:01 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-07-24 11:54 [dpdk-dev] [PATCH 0/3] Multiseg fixes for pcap pmd David Marchand
2019-07-24 11:54 ` [dpdk-dev] [PATCH 1/3] net/pcap: fix Rx with small buffers David Marchand
2019-07-24 18:28   ` Ferruh Yigit
2019-07-24 11:54 ` [dpdk-dev] [PATCH 2/3] net/pcap: fix transmit return count in error conditions David Marchand
2019-07-24 18:36   ` Ferruh Yigit
2019-07-25  7:40     ` David Marchand
2019-07-25 11:01       ` Ferruh Yigit [this message]
2019-07-24 11:54 ` [dpdk-dev] [PATCH 3/3] net/pcap: fix concurrent multiseg packet transmits David Marchand
2019-07-24 18:43   ` Ferruh Yigit
2019-07-25  8:18   ` [dpdk-dev] [dpdk-stable] " David Marchand
2019-07-25 11:07     ` Ferruh Yigit
2019-07-25 12:04 ` [dpdk-dev] [PATCH v2 0/3] Multiseg fixes for pcap pmd David Marchand
2019-07-25 12:04   ` [dpdk-dev] [PATCH v2 1/3] net/pcap: fix Rx with small buffers David Marchand
2019-07-25 12:04   ` [dpdk-dev] [PATCH v2 2/3] net/pcap: fix transmit return count in error conditions David Marchand
2019-07-25 14:43     ` Ferruh Yigit
2019-07-25 12:04   ` [dpdk-dev] [PATCH v2 3/3] net/pcap: fix concurrent multiseg packet transmits David Marchand
2019-07-25 12:05     ` [dpdk-dev] [dpdk-stable] " David Marchand
2019-07-25 14:47     ` [dpdk-dev] " Ferruh Yigit
2019-07-25 19:24 ` [dpdk-dev] [PATCH v3 0/3] Multiseg fixes for pcap pmd David Marchand
2019-07-25 19:24   ` [dpdk-dev] [PATCH v3 1/3] net/pcap: fix Rx with small buffers David Marchand
2019-07-25 19:24   ` [dpdk-dev] [PATCH v3 2/3] net/pcap: fix transmit return count in error conditions David Marchand
2019-07-25 19:24   ` [dpdk-dev] [PATCH v3 3/3] net/pcap: fix concurrent multiseg packet transmits David Marchand
2019-07-25 22:36   ` [dpdk-dev] [PATCH v3 0/3] Multiseg fixes for pcap pmd Ferruh Yigit

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=ebf18ea8-0035-ce40-f522-5c82d20615dd@intel.com \
    --to=ferruh.yigit@intel.com \
    --cc=aideen.mcloughlin@intel.com \
    --cc=david.marchand@redhat.com \
    --cc=dev@dpdk.org \
    --cc=stable@dpdk.org \
    /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).