netdev - Re: [EXT] [PATCH net-next v2 1/3] net: ethernet: fec: Replace interrupt driven MDIO with polled IO

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]

Message-ID: <20200419003501.GE836632@lunn.ch>
Date:   Sun, 19 Apr 2020 02:35:01 +0200
From:   Andrew Lunn <andrew@...n.ch>
To:     Chris Healy <cphealy@...il.com>
Cc:     Andy Duan <fugang.duan@....com>,
        David Miller <davem@...emloft.net>,
        netdev <netdev@...r.kernel.org>,
        Florian Fainelli <f.fainelli@...il.com>,
        Heiner Kallweit <hkallweit1@...il.com>
Subject: Re: [EXT] [PATCH net-next v2 1/3] net: ethernet: fec: Replace
 interrupt driven MDIO with polled IO

On Sat, Apr 18, 2020 at 03:39:17PM -0700, Chris Healy wrote:
> I did some profiling using an oscilloscope with my NXP Vybrid based
> platform to see what different "sleep_us" values resulted in for start
> of MDIO to start of MDIO transaction times.  Here's what I found:
> 
> 0  - ~38us to ~40us
> 1  - ~48us to ~64us
> 2  - ~48us to ~64us
> 3  - ~48us to ~64us
> 4  - ~48us to ~64us
> 4  - ~48us to ~64us
> 5  - ~48us to ~64us
> 6  - ~48us to ~64us
> 7  - ~48us to ~64us
> 8  - ~48us to ~64us
> 9  - ~56us to ~88us
> 10 - ~56us to ~112us
> 
> Basically, with the "sleep_us" value set to 0, I would get the
> shortest inter transaction times with a very low variance.  Once I
> went to a non-zero value, the inter transaction time went up, as well
> as the variance, which I suppose makes sense....

Thanks for these numbers. They are suggesting that udelay() is quite
expensive, and is probably taking longer than we expect. I might
instrument the function to see what is happening.

	   Andrew