From: Johannes Berg <johannes@sipsolutions.net> To: Michal Kazior <michal.kazior@tieto.com> Cc: Eric Dumazet <eric.dumazet@gmail.com>, Neal Cardwell <ncardwell@google.com>, linux-wireless <linux-wireless@vger.kernel.org>, Network Development <netdev@vger.kernel.org>, Eyal Perry <eyalpe@dev.mellanox.co.il> Subject: Re: Throughput regression with `tcp: refine TSO autosizing` Date: Tue, 24 Feb 2015 11:24:55 +0100 [thread overview] Message-ID: <1424773495.2192.23.camel@sipsolutions.net> (raw) In-Reply-To: <CA+BoTQnxo-x94uVfErZO43jTowVcEMEt+7xDzoAGiR8fRDv=ug@mail.gmail.com> (sfid-20150212_084824_976893_36D95ADD) On Thu, 2015-02-12 at 08:48 +0100, Michal Kazior wrote: > > Good point. I was actually thinking about it. I can try cooking a > > patch unless you want to do it yourself :-) > > I've taken a look into this. The most obvious place to add the > timestamp for each packet would be ieee80211_tx_info (i.e. the > skb->cb[48]). The problem is it's very tight there. Even squeezing 2 > bytes (allowing up to 64ms of tx completion delay which I'm worried > won't be enough) will be troublesome. Some drivers already use every > last byte of their allowance on 64bit archs (e.g. ar5523 uses entire > 40 bytes of driver_data). Couldn't we just repurpose the existing skb->tstamp field for this, as long as the skb is fully contained within the wireless layer? Actually, it looks like we can't, since I guess timestamping options can be turned on on any socket. > I wonder if it's okay to bump skb->cb to 56 bytes to avoid the cascade > of changes required to implement the tx completion delay accounting? I have no doubt that would be rejected :) johannes
WARNING: multiple messages have this Message-ID (diff)
From: Johannes Berg <johannes-cdvu00un1VgdHxzADdlk8Q@public.gmane.org> To: Michal Kazior <michal.kazior-++hxYGjEMp0AvxtiuMwx3w@public.gmane.org> Cc: Eric Dumazet <eric.dumazet-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>, Neal Cardwell <ncardwell-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>, linux-wireless <linux-wireless-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>, Network Development <netdev-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>, Eyal Perry <eyalpe-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb@public.gmane.org> Subject: Re: Throughput regression with `tcp: refine TSO autosizing` Date: Tue, 24 Feb 2015 11:24:55 +0100 [thread overview] Message-ID: <1424773495.2192.23.camel@sipsolutions.net> (raw) In-Reply-To: <CA+BoTQnxo-x94uVfErZO43jTowVcEMEt+7xDzoAGiR8fRDv=ug-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org> (sfid-20150212_084824_976893_36D95ADD) On Thu, 2015-02-12 at 08:48 +0100, Michal Kazior wrote: > > Good point. I was actually thinking about it. I can try cooking a > > patch unless you want to do it yourself :-) > > I've taken a look into this. The most obvious place to add the > timestamp for each packet would be ieee80211_tx_info (i.e. the > skb->cb[48]). The problem is it's very tight there. Even squeezing 2 > bytes (allowing up to 64ms of tx completion delay which I'm worried > won't be enough) will be troublesome. Some drivers already use every > last byte of their allowance on 64bit archs (e.g. ar5523 uses entire > 40 bytes of driver_data). Couldn't we just repurpose the existing skb->tstamp field for this, as long as the skb is fully contained within the wireless layer? Actually, it looks like we can't, since I guess timestamping options can be turned on on any socket. > I wonder if it's okay to bump skb->cb to 56 bytes to avoid the cascade > of changes required to implement the tx completion delay accounting? I have no doubt that would be rejected :) johannes -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html
next prev parent reply other threads:[~2015-02-24 10:25 UTC|newest] Thread overview: 107+ messages / expand[flat|nested] mbox.gz Atom feed top 2015-01-29 11:48 Throughput regression with `tcp: refine TSO autosizing` Michal Kazior 2015-01-29 13:14 ` Eric Dumazet 2015-01-30 10:29 ` Arend van Spriel 2015-01-30 13:19 ` Eric Dumazet 2015-01-30 13:19 ` Eric Dumazet 2015-01-30 13:47 ` Arend van Spriel 2015-01-30 14:37 ` Eric Dumazet [not found] ` <CAA93jw5fqhz0Hiw74L2GXgtZ9JsMg+NtYydKxKzGDrvQcZn4hA@mail.gmail.com> [not found] ` <CAA93jw7b0E9jjQYXrEPzjLLC9j8xNC0TFYXpWVtgFameJaNBdw@mail.gmail.com> [not found] ` <1422741065.199624134@apps.rackspace.com> [not found] ` <CAPp0ZBb2nkA6Y0s=W0kw=zvyn0wi0NMBRsBCw_xcD61ScOmgQg@mail.gmail.com> [not found] ` <CAA_e5Z46Bu+zZZFzf_ejzA35Gw3g1_OG85yv6yd7MpbwZcE-nw@mail.gmail.com> 2015-02-01 8:45 ` Fwd: " Dave Taht 2015-02-01 10:47 ` Jonathan Morton 2015-02-01 14:43 ` dpreed 2015-02-01 23:34 ` Andrew McGregor 2015-02-02 4:04 ` [Cerowrt-devel] " Avery Pennarun 2015-02-02 4:04 ` Avery Pennarun 2015-02-02 15:25 ` Jim Gettys 2015-02-02 4:21 ` Avery Pennarun 2015-02-02 7:07 ` David Lang 2015-02-02 7:07 ` David Lang 2015-01-30 13:39 ` Michal Kazior 2015-01-30 13:39 ` Michal Kazior 2015-01-30 14:40 ` Eric Dumazet 2015-01-30 14:40 ` Eric Dumazet 2015-02-02 10:27 ` Michal Kazior 2015-02-02 10:27 ` Michal Kazior 2015-02-02 18:52 ` Eric Dumazet 2015-02-02 21:25 ` Ben Greear 2015-02-02 23:06 ` Eric Dumazet 2015-02-02 23:06 ` Eric Dumazet 2015-02-03 9:00 ` Michal Kazior 2015-02-03 9:00 ` Michal Kazior 2015-02-03 1:18 ` Eric Dumazet 2015-02-03 1:18 ` Eric Dumazet 2015-02-03 11:50 ` Michal Kazior 2015-02-03 14:27 ` Eric Dumazet 2015-02-03 14:27 ` Eric Dumazet 2015-02-03 15:03 ` Eric Dumazet 2015-02-03 15:03 ` Eric Dumazet 2015-02-04 11:35 ` Michal Kazior 2015-02-04 11:57 ` Eric Dumazet 2015-02-04 11:57 ` Eric Dumazet 2015-02-04 12:22 ` Michal Kazior 2015-02-04 12:38 ` Eric Dumazet 2015-02-04 12:53 ` Michal Kazior 2015-02-04 12:55 ` Johannes Berg 2015-02-04 13:16 ` Eric Dumazet 2015-02-04 13:29 ` Eric Dumazet 2015-02-04 21:11 ` Eric Dumazet 2015-02-04 21:11 ` Eric Dumazet 2015-02-05 6:46 ` Michal Kazior 2015-02-05 6:46 ` Michal Kazior 2015-02-05 13:03 ` Eric Dumazet 2015-02-05 13:03 ` Eric Dumazet 2015-02-05 8:38 ` Michal Kazior 2015-02-05 12:57 ` Eric Dumazet 2015-02-05 13:19 ` Eric Dumazet 2015-02-05 13:33 ` Eric Dumazet 2015-02-05 13:33 ` Eric Dumazet 2015-02-05 13:44 ` Michal Kazior 2015-02-05 14:41 ` Eric Dumazet 2015-02-05 14:41 ` Eric Dumazet 2015-02-05 17:10 ` Eric Dumazet 2015-02-06 9:42 ` Michal Kazior 2015-02-06 13:40 ` Eric Dumazet 2015-02-06 13:40 ` Eric Dumazet 2015-02-06 13:53 ` Eric Dumazet 2015-02-06 13:53 ` Eric Dumazet 2015-02-06 14:09 ` Michal Kazior 2015-02-09 13:47 ` Michal Kazior 2015-02-09 15:11 ` Eric Dumazet 2015-02-10 10:33 ` Michal Kazior 2015-02-10 12:54 ` Eric Dumazet 2015-02-10 13:05 ` Eric Dumazet 2015-02-10 13:14 ` Eric Dumazet 2015-02-11 8:33 ` Michal Kazior 2015-02-11 13:17 ` Eric Dumazet 2015-02-11 13:17 ` Eric Dumazet 2015-02-12 7:16 ` Michal Kazior 2015-02-12 7:16 ` Michal Kazior 2015-02-10 14:19 ` Johannes Berg 2015-02-10 15:09 ` Eric Dumazet 2015-02-11 8:57 ` Michal Kazior 2015-02-11 8:57 ` Michal Kazior 2015-02-12 7:48 ` Michal Kazior 2015-02-12 8:33 ` Dave Taht 2015-02-24 10:24 ` Johannes Berg [this message] 2015-02-24 10:24 ` Johannes Berg 2015-02-24 10:30 ` Johannes Berg 2015-02-24 10:59 ` Johannes Berg 2015-03-31 11:08 ` Johannes Berg 2015-02-06 14:10 ` Eric Dumazet 2015-02-06 14:31 ` David Laight 2015-02-06 14:31 ` David Laight 2015-02-06 15:02 ` Eric Dumazet 2015-02-06 15:02 ` Eric Dumazet 2015-02-06 14:08 ` Michal Kazior 2015-02-06 14:08 ` Michal Kazior 2015-02-06 14:35 ` Eric Dumazet 2015-02-06 14:35 ` Eric Dumazet 2015-02-06 17:48 ` Rick Jones 2015-02-06 17:48 ` Rick Jones 2015-02-05 14:48 ` Eric Dumazet 2015-02-05 14:48 ` Eric Dumazet 2015-02-06 9:39 ` Nicolas Cavallari 2015-02-06 9:39 ` Nicolas Cavallari 2015-02-05 19:50 ` Dave Taht 2015-02-06 9:57 ` Michal Kazior 2015-02-06 9:57 ` Michal Kazior 2015-02-03 8:44 ` Michal Kazior
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=1424773495.2192.23.camel@sipsolutions.net \ --to=johannes@sipsolutions.net \ --cc=eric.dumazet@gmail.com \ --cc=eyalpe@dev.mellanox.co.il \ --cc=linux-wireless@vger.kernel.org \ --cc=michal.kazior@tieto.com \ --cc=ncardwell@google.com \ --cc=netdev@vger.kernel.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: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.