netdev - ethtool SFF-8472 issue

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

Date:   Thu, 17 Oct 2019 08:16:38 -0400
From:   Joe Buehler <aspam@....net>
To:     linville@...driver.com, netdev@...r.kernel.org
Subject: ethtool SFF-8472 issue

In sfpdiag.c the sff8472_calibration routine does not look correct for
optical RX power.

The code is currently calculating this:

power = a + b*x + c*x + d*x // x is the raw A/D reading

First problem is that there is another term (SFF_A2_CAL_RXPWR4 in
sfpdiag.c) in SFF-8472 that has been dropped.

The more serious issue is that this computation is not correct.  The
SFF-8472 document could be a little clearer -- which is probably what
caused this -- but I think what it intends is a least-squares polynomial
fit to convert from A/D reading to RX power.

So the formula should be:

power = a + b *x + c*x*x + d*x*x*x + e*x*x*x*x

Joe Buehler