#1387 Change SIOCETHTOOL ioctl to use ETHTOOL_GLINKSETTINGS instead of obsolete ETHTOOL_GSET

[Vasile Laurentiu Stanimir]

Change SIOCETHTOOL ioctl to use ETHTOOL_GLINKSETTINGS instead of obsolete ETHTOOL_GSET.
Changes needed to avoid kernel warnings: "warning: 'snmpd' uses legacy ethtool link settings API, link modes are only partially reported"
Version: 5.8
Tested on Linux, kernel 4.12.29 x86_64

[Vasile Laurentiu Stanimir]

Patch update for older kernels without support for ETHTOOL_GLINKSETTINGS

[Bart Van Assche]

A modified version of this patch has been applied on the v5.8 and master branches. Please retest.

[Sam Tannous]

These patches in V5-8-patches break the ifSpeed and ifHighSpeed functionality:

09da7e2 Fix CentOS6 compile witch does not define ETHTOOL_GLINKSETTINGS
ec9f614 CHANGES: snmpd: Use ETHTOOL_GLINKSETTINGS when available

I compiled latest patches from this branch and tested on a recently upgraded Debian Buster
VM server:

Latest code:

root@buster1:/home/stannous# ./sbin/snmpd -v

NET-SNMP version:  5.8
Web:               http://www.net-snmp.org/
Email:             net-snmp-coders@lists.sourceforge.net

root@buster1:/home/stannous# /sbin/ethtool enp0s3
Settings for enp0s3:
    Supported ports: [ TP ]
    Supported link modes:   10baseT/Half 10baseT/Full 
                            100baseT/Half 100baseT/Full 
                            1000baseT/Full 
    Supported pause frame use: No
    Supports auto-negotiation: Yes
    Supported FEC modes: Not reported
    Advertised link modes:  10baseT/Half 10baseT/Full 
                            100baseT/Half 100baseT/Full 
                            1000baseT/Full 
    Advertised pause frame use: No
    Advertised auto-negotiation: Yes
    Advertised FEC modes: Not reported
    Speed: 1000Mb/s
    Duplex: Full
    Port: Twisted Pair
    PHYAD: 0
    Transceiver: internal
    Auto-negotiation: on
    MDI-X: off (auto)
    Supports Wake-on: umbg
    Wake-on: d
    Current message level: 0x00000007 (7)
                   drv probe link
    Link detected: yes
root@buster1:/home/stannous# 
root@buster1:/home/stannous# 
root@buster1:/home/stannous# which snmpwalk
root@buster1:/home/stannous# snmpwalk -v2c -cpublic localhost 1 | grep Speed
IF-MIB::ifSpeed.1 = Gauge32: 10000000
IF-MIB::ifSpeed.2 = Gauge32: 0
HOST-RESOURCES-MIB::hrSWRunParameters.20433 = STRING: "Speed"
IF-MIB::ifHighSpeed.1 = Gauge32: 10
IF-MIB::ifHighSpeed.2 = Gauge32: 0

Where previously, using the default (5.7.3) snmpd, I had these results:

root@buster1:/home/stannous# which snmpd 
root@buster1:/home/stannous# /usr/sbin/snmpd -v

NET-SNMP version:  5.7.3
Web:               http://www.net-snmp.org/
Email:             net-snmp-coders@lists.sourceforge.net

root@buster1:/home/stannous# 
root@buster1:/home/stannous# snmpwalk -v2c -cpublic localhost 1 | grep Speed
IF-MIB::ifSpeed.1 = Gauge32: 10000000
IF-MIB::ifSpeed.2 = Gauge32: 1000000000
HOST-RESOURCES-MIB::hrSWRunParameters.20448 = STRING: "Speed"
IF-MIB::ifHighSpeed.1 = Gauge32: 10
IF-MIB::ifHighSpeed.2 = Gauge32: 1000
root@buster1:/home/stannous# 

Can you please either fix the ETHTOOL_GLINKSETTINGS patch or just revert it?
I looked at the code briefly but couldn't see why it was failing. It seems that the
ioctl call with ETHTOOL_GLINKSETTINGS returns 0 (in my small sample code) but
doesn't retrieve the actual speed (it's 0) so that the ETHTOOL_GSET call is never made.

Thanks,
Sam

[Sam Tannous]

When looking over the ethtool code, it appears that multiple ioctl calls may be required.
Why, I do not know.....This small patch works:

--Regards,
Sam

[Bart Van Assche]

I think that patch could cause a stack overflow so I have checked in a slightly different patch. Please retest.

[Sam Tannous]

That works fine. And no stack smashing. Thanks, Bart.

--Sam