linux-kernel - Re: [PATCH 6/9] firmware: print firmware name on fallback path

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>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20180508002055.GY27853@wotan.suse.de>
Date:   Tue, 8 May 2018 00:20:55 +0000
From:   "Luis R. Rodriguez" <mcgrof@...nel.org>
To:     Andres Rodriguez <andresx7@...il.com>
Cc:     "Luis R. Rodriguez" <mcgrof@...nel.org>,
        linux-kernel@...r.kernel.org, gregkh@...uxfoundation.org,
        alexdeucher@...il.com, christian.koenig@....com,
        kvalo@...eaurora.org, arend.vanspriel@...adcom.com,
        linux-wireless@...r.kernel.org, ath10k@...ts.infradead.org,
        hdegoede@...hat.com, Kees Cook <keescook@...omium.org>,
        Mimi Zohar <zohar@...ux.vnet.ibm.com>
Subject: Re: [PATCH 6/9] firmware: print firmware name on fallback path

On Fri, May 04, 2018 at 10:57:26PM -0400, Andres Rodriguez wrote:
> On 2018-05-03 07:42 PM, Luis R. Rodriguez wrote:
> > On Mon, Apr 23, 2018 at 04:12:02PM -0400, Andres Rodriguez wrote:
> > > Previously, one could assume the firmware name from the preceding
> > > message: "Direct firmware load for {name} failed with error %d".
> > > 
> > > However, with the new firmware_request_nowarn() entrypoint, the message
> > > outlined above will not always be printed.
> > 
> > I though the whole point was to not print an error message. What if
> > we want later to disable this error message? This would prove a bit
> > pointless.
> > 
> > Let's discuss the exact semantics desired here. Why would only the
> > fallback be desirable here?
> > 
> > Andres, Kalle?
> > 
> > After we address this I'll address resubmitting this lat patch
> > along with the last one. For now I'll skip it.
> 
> You are correct. I initially thought it would be useful to know that the
> usermode fallback was being triggered. And for that message to be useful we
> would need a fw name.
> 
> But now that you point it out, this behaviour is inconsistent with the
> _nowarn() definition. We shouldn't have a message in the first place.
> 
> So it might be better to instead have:
> 
> if (!(opt_flags & FW_OPT_NO_WARN) )
>     dev_warn(device, "Falling back to user helper\n");
> 
> No need to add the firmware name, cause we either:
>     a) FW_OPT_NO_WARN is set and no messages are printed, or
>     b) FW_OPT_NO_WARN is not set and we get both messages.
> 
> Yay, nay?

I welcome such a new warning but not for any of the reasons stated.

It make sense if FW_OPT_NO_WARN is not set and only because the fallback
mechanism can fail for a slew of different firmware files, and just getting
informed a failure with a fallback occurred does not tell us for which file it
failed for.

I'll add such a patch to my queue and send it off soon prior to your own
new API nowarn call.

  Luis