RE: [PATCH] hv_utils: Add validation for untrusted Hyper-V values

From: Stephen Hemminger
Date: Wed Jul 29 2020 - 12:22:23 EST


Ok at least use the ratelimit form of kernel logging.

Netdev_err_ratelimited...

-----Original Message-----
From: Andres Beltran <lkmlabelt@xxxxxxxxx>
Sent: Wednesday, July 29, 2020 9:10 AM
To: Stephen Hemminger <sthemmin@xxxxxxxxxxxxx>
Cc: KY Srinivasan <kys@xxxxxxxxxxxxx>; Haiyang Zhang <haiyangz@xxxxxxxxxxxxx>; Wei Liu <wei.liu@xxxxxxxxxx>; linux-hyperv@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; Michael Kelley <mikelley@xxxxxxxxxxxxx>; Andrea Parri <parri.andrea@xxxxxxxxx>; Saruhan Karademir <skarade@xxxxxxxxxxxxx>
Subject: Re: [PATCH] hv_utils: Add validation for untrusted Hyper-V values

On Tue, Jul 28, 2020 at 5:04 PM Stephen Hemminger
<sthemmin@xxxxxxxxxxxxx> wrote:
>
> You may want to use one of the macros that prints this once only.
> This is a "should never happen" type error, so if something goes wrong it might happens so much that journal/syslog would get overloaded.

Certainly, printing error messages once would be ideal if we were only
dealing with Linux kernel bugs. But under the assumption that Hyper-V
can send bogus values at any time, I think it would be better to print
error messages every time so that we are aware of malicious/erroneous
data sent by the host.