Re: [PATCH v3 2/3] hwrng: add Rockchip SoC hwrng driver

From: Dragan Simic
Date: Sun Jun 23 2024 - 01:41:44 EST


Hello Uwe,

On 2024-06-23 02:20, Uwe Kleine-König wrote:
On Sat, Jun 22, 2024 at 10:45:22PM +0200, Dragan Simic wrote:
On 2024-06-22 22:26, Heiko Stübner wrote:
> Am Samstag, 22. Juni 2024, 12:29:33 CEST schrieb Dragan Simic:
> > On 2024-06-22 00:16, Uwe Kleine-König wrote:
> > > On 6/21/24 20:13, Dragan Simic wrote:
> > >> On 2024-06-21 11:57, Krzysztof Kozlowski wrote:
> > >>> On 21/06/2024 03:25, Daniel Golle wrote:
> > >>>> + dev_info(&pdev->dev, "Registered Rockchip hwrng\n");
> > >>>
> > >>> Drop, driver should be silent on success.
> > >>
> [...]
> So really this message should be dropped or at least as Uwe suggests
> made a dev_dbg.

As a note, "dmesg --level=err,warn", for example, is rather useful
when it comes to filtering the kernel messages to see only those that
are signs of a trouble.

IMHO it's a bit sad, that there is such a long thread about something so
trivial, but I want to make a few points:

- not all dmesg implementations support this:

root@machine:~ dmesg --level=err,warn
dmesg: unrecognized option '--level=err,warn'
BusyBox v1.36.1 () multi-call binary.

Usage: dmesg [-cr] [-n LEVEL] [-s SIZE]

Print or control the kernel ring buffer

-c Clear ring buffer after printing
-n LEVEL Set console logging level
-s SIZE Buffer size
-r Print raw message buffer

- Your argument that the output of this dev_info can easily be ignored
is a very weak reason to keep it.

- I personally consider it hard sometimes to accept feedback if I think
it's wrong and there is a good reason to do it the way I want it.
But there are now three people opposing your position, who brought
forward (IMHO) good reasons and even a constructive alternative was
presented. Please stay open minded while weighting the options.
The questions to ask here include:

- How many people care for this message? During every boot? Is it
maybe enough for these people to check in /sys if the device
probed successfully? Or maybe even the absence of an error message
is enough?
- How many people get this message and don't care about the
presented information? How many people are even annoyed by it?
- Is the delay and memory usage introduced by this message justified
then, even if it's small?

I'm sorry if my responses caused any inconvenience. I find most of your
points totally valid, but there are a couple of them I could continue
arguing about. Though, as you also pointed out, my opinion has been
already outvoted, so I'll remain silent.