netdev - Re: [RFC ethtool 6/6] netlink: add support for standard stats

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] [day] [month] [year] [list]

Message-ID: <20210417124422.587df060@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
Date:   Sat, 17 Apr 2021 12:44:22 -0700
From:   Jakub Kicinski <kuba@...nel.org>
To:     Ido Schimmel <idosch@...sch.org>
Cc:     netdev@...r.kernel.org, idosch@...dia.com, mkubecek@...e.cz
Subject: Re: [RFC ethtool 6/6] netlink: add support for standard stats

On Sat, 17 Apr 2021 22:22:57 +0300 Ido Schimmel wrote:
> > > So you will have something like:
> > > 
> > > ETHTOOL_A_STATS_GRP_HIST_BKT_UNITS_BYTES  
> > 
> > Histogram has two dimensions, what's the second dimension for bytes?
> > Time? Packet arrival?  
> 
> Not sure what you mean. Here you are counting how many Rx/Tx packets are
> between N to M bytes in length. I meant to add two attributes. One that
> tells user space that you are counting Rx/Tx packets and the second that
> N to M are in bytes.

Ah, these were not part of the same enum, I get it now.

I thought maybe bytes were trying to cater to the queue length use case
and I wasn't sure how that histogram is constructed.

> But given your comment below about this histogram probably being a one
> time thing, I think maybe staying with the current attributes is OK.
> There is no need to over-engineer it if we don't see ourselves adding
> new histograms.
> 
> Anyway, these histograms are under ETHTOOL_A_STATS_GRP that should give
> user space all the context about what is being counted.