This module/tool was built years ago against a kernel-version I don't remember.
Don't assume this to build against recent kernels.
Anyway, I decided to dump it here.
The original README follows:
----------
ifstatfake provides an interface to modify the statistics of the kernel's
netdevice drivers, which can be read from '/proc/net/dev' or are printed by
ifconfig.
This package contains a kernel-module and a bash-script.
When the module is loaded, it creates /proc/net/ifstatfake, a write-only
pseudo-file. Based on commands written to this file the module will change
the statistics.
The bashscript provides another interface to the file in the proc-filesystem
with userspace syntax-checking of the commands and automatical loading of
the module. It's purpose is to save some typing and avoid the necessity for
searching the system's logs for errors when something went wrong.
The syntax of the commands is the same, for both writing directly to
/proc/net/ifstatfake and the ifstatfake-shellscript:
IFACE rx|tx FIELD AMOUNT
Where the arguments are:
IFACE The name of the interface to change the statistics,
i.e. eth0
rx, tx Sets whether the statistics in context of received
(rx) or sent (tx) data should be changed
FIELD Which field should be changed. Possible values are:
- bytes
- packets
- errs
- drop
- fifo
- frame (only with 'rx' set)
- compressed
- multicast (only with 'rx' set)
- colls (only with 'tx' set)
- carrier (only with 'tx' set)
AMOUNT A number the FIELD should be set to or a number to be
added or substracted from the current value with a
trailing '=', '+' or '-' to specify if the value
should be set, added or substracted
For example the (corresponding to each other) commands:
echo "eth0 rx drop +20" >> /proc/net/ifstatfake
and
ifstatfake eth0 rx drop +20
would increment the number of dropped, incoming ethernet-frames of eth0 by
20.
Further information can be found in 'man 5 ifstatfake' and
'man 8 ifstatfake'.