netdev - RE: [RFC PATCH 0/4] More accurate PHC<->system clock synchronization

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 26 Oct 2018 16:40:59 +0000
From:   "Keller, Jacob E" <jacob.e.keller@...el.com>
To:     Miroslav Lichvar <mlichvar@...hat.com>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>
CC:     "intel-wired-lan@...ts.osuosl.org" <intel-wired-lan@...ts.osuosl.org>,
        Richard Cochran <richardcochran@...il.com>
Subject: RE: [RFC PATCH 0/4] More accurate PHC<->system clock synchronization

Hi Miroslav,

> -----Original Message-----
> From: Miroslav Lichvar [mailto:mlichvar@...hat.com]
> Sent: Friday, October 26, 2018 9:28 AM
> To: netdev@...r.kernel.org
> Cc: intel-wired-lan@...ts.osuosl.org; Richard Cochran <richardcochran@...il.com>;
> Keller, Jacob E <jacob.e.keller@...el.com>; Miroslav Lichvar <mlichvar@...hat.com>
> Subject: [RFC PATCH 0/4] More accurate PHC<->system clock synchronization
> 
> This series adds support for a more accurate synchronization between a
> PTP hardware clock and the system clock.
> 
> The first patch adds an extended version of the PTP_SYS_OFFSET ioctl,
> which returns three timestamps for each measurement. The idea is to
> shorten the interval between the system timestamps to contain just the
> reading of the lowest register of the PHC in order to reduce the error
> in the measured offset and give a better bound on the maximum error.
> 
> The other patches add support for the new ioctl to the e1000e, igb,
> and ixgbe driver. Tests with few different NICs in different machines
> (and PCIe slots) show that:
> - with an I219 (e1000e) the measured delay improved from 2500 to 1300 ns
>   and the error in the measured offset, when compared to cross
>   timestamping, was reduced by a factor of 5
> - with an I210 (igb) the delay improved from 5100 to 1700 ns
> - with an I350 (igb) the delay improved from 2300 to 750 ns
> - with an X550 (ixgbe) the delay improved from 1950 to 650 ns
> 

That is some very significant improvements! Excellent find.

> There is some duplication of code in the igb and ixgbe drivers, which I
> don't like very much, but I thought it's better than extending and
> wrapping the existing functions like in the e1000e driver. Also, mixing
> SYSTIM and "system time" in the code will probably be confusing.
> 

Yea...

> I wasn't able to find a better name for the ioctl, the structures, and
> the driver function. If anyone has suggestions, please let me know.
> 

I don't have any good suggestions yet. I'll reply after reviewing if I think of any.

Thanks,
Jake