From: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
To: Neil Horman <nhorman@tuxdriver.com>
Cc: Network Development <netdev@vger.kernel.org>,
Matteo Croce <mcroce@redhat.com>,
"David S. Miller" <davem@davemloft.net>
Subject: Re: [PATCH v2 net] af_packet: Block execution of tasks waiting for transmit to complete in AF_PACKET
Date: Sat, 22 Jun 2019 22:21:31 -0400 [thread overview]
Message-ID: <CAF=yD-+8NDiL0dxM+eOFyiwi1ZhCW29dW--+VeEkssUaJqedWg@mail.gmail.com> (raw)
In-Reply-To: <CAF=yD-JC_r1vjitN1ccyvQ3DXiP9BNCwq9iiWU11cXznmhAY8Q@mail.gmail.com>
> > -static void __packet_set_status(struct packet_sock *po, void *frame, int status)
> > +static void __packet_set_status(struct packet_sock *po, void *frame, int status,
> > + bool call_complete)
> > {
> > union tpacket_uhdr h;
> >
> > @@ -381,6 +382,8 @@ static void __packet_set_status(struct packet_sock *po, void *frame, int status)
> > BUG();
> > }
> >
> > + if (po->wait_on_complete && call_complete)
> > + complete(&po->skb_completion);
>
> This wake need not happen before the barrier. Only one caller of
> __packet_set_status passes call_complete (tpacket_destruct_skb).
> Moving this branch to the caller avoids a lot of code churn.
>
> Also, multiple packets may be released before the process is awoken.
> The process will block until packet_read_pending drops to zero. Can
> defer the wait_on_complete to that one instance.
Eh no. The point of having this sleep in the send loop is that
additional slots may be released for transmission (flipped to
TP_STATUS_SEND_REQUEST) from another thread while this thread is
waiting.
Else, it would have been much simpler to move the wait below the send
loop: send as many packets as possible, then wait for all of them
having been released. Much clearer control flow.
Where to set and clear the wait_on_complete boolean remains. Integer
assignment is fragile, as the compiler and processor may optimize or
move simple seemingly independent operations. As complete() takes a
spinlock, avoiding that in the DONTWAIT case is worthwhile. But probably
still preferable to set when beginning waiting and clear when calling
complete.
next prev parent reply other threads:[~2019-06-23 2:22 UTC|newest]
Thread overview: 34+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-06-19 20:25 [PATCH net] af_packet: Block execution of tasks waiting for transmit to complete in AF_PACKET Neil Horman
2019-06-20 13:41 ` Willem de Bruijn
2019-06-20 14:01 ` Matteo Croce
2019-06-20 14:23 ` Neil Horman
2019-06-20 15:16 ` Willem de Bruijn
2019-06-20 16:14 ` Neil Horman
2019-06-20 16:18 ` Willem de Bruijn
2019-06-20 17:31 ` Neil Horman
2019-06-21 16:41 ` Neil Horman
2019-06-21 18:31 ` Willem de Bruijn
2019-06-21 19:18 ` Neil Horman
2019-06-21 20:06 ` Willem de Bruijn
2019-06-22 11:08 ` Neil Horman
2019-06-22 17:41 ` [PATCH v2 " Neil Horman
2019-06-23 2:12 ` Willem de Bruijn
2019-06-23 2:21 ` Willem de Bruijn [this message]
2019-06-23 11:40 ` Neil Horman
2019-06-23 14:39 ` Willem de Bruijn
2019-06-23 19:21 ` Neil Horman
2019-06-23 11:34 ` Neil Horman
2019-06-24 0:46 ` [PATCH v3 " Neil Horman
2019-06-24 18:08 ` Willem de Bruijn
2019-06-24 21:51 ` Neil Horman
2019-06-24 22:15 ` Willem de Bruijn
2019-06-25 11:02 ` Neil Horman
2019-06-25 13:37 ` Willem de Bruijn
2019-06-25 16:20 ` Neil Horman
2019-06-25 21:59 ` Willem de Bruijn
2019-06-25 21:57 ` [PATCH v4 " Neil Horman
2019-06-25 22:30 ` Willem de Bruijn
2019-06-26 10:54 ` Neil Horman
2019-06-26 15:05 ` Willem de Bruijn
2019-06-26 17:14 ` Neil Horman
2019-06-27 2:38 ` David Miller
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='CAF=yD-+8NDiL0dxM+eOFyiwi1ZhCW29dW--+VeEkssUaJqedWg@mail.gmail.com' \
--to=willemdebruijn.kernel@gmail.com \
--cc=davem@davemloft.net \
--cc=mcroce@redhat.com \
--cc=netdev@vger.kernel.org \
--cc=nhorman@tuxdriver.com \
/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).