linux-kernel - Re: [PATCH v2] of: overlay: log the error cause on resolver failure

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: <40fdf0f2-85a4-9f84-6994-a59b7b56cec4@lucaceresoli.net>
Date:   Thu, 27 Feb 2020 09:11:13 +0100
From:   Luca Ceresoli <luca@...aceresoli.net>
To:     Frank Rowand <frowand.list@...il.com>, devicetree@...r.kernel.org,
        Geert Uytterhoeven <geert+renesas@...der.be>
Cc:     Pantelis Antoniou <pantelis.antoniou@...sulko.com>,
        Rob Herring <robh+dt@...nel.org>, linux-kernel@...r.kernel.org,
        Geert Uytterhoeven <geert@...ux-m68k.org>
Subject: Re: [PATCH v2] of: overlay: log the error cause on resolver failure

Hi Frank,

On 26/02/20 04:53, Frank Rowand wrote:
> On 2/25/20 10:45 AM, Luca Ceresoli wrote:
>> For some of its error paths, of_resolve_phandles() only logs a very generic
>> error which does not help much in finding the origin of the problem:
>>
>>   OF: resolver: overlay phandle fixup failed: -22
>>
>> Add error messages for all the error paths that don't have one. Now a
>> specific message is always emitted, thus also remove the generic catch-all
>> message emitted before returning.
>>
>> For example, in case a DT overlay has a fixup node that is not present in
>> the base DT __symbols__, this error is now logged:
>>
>>   OF: resolver: node gpio9 not found in base DT, fixup failed
>>
>> Signed-off-by: Luca Ceresoli <luca@...aceresoli.net>
>> Cc: Geert Uytterhoeven <geert@...ux-m68k.org>
>> ---
>>
>> I don't know in detail the meaning of the adjust_local_phandle_references()
>> and update_usages_of_a_phandle_reference() error paths, thus I have put
>> pretty generic messages. Any suggestion on better wording would be welcome.
> 
> If you have not read the code to understand what the meaning of
> the errors are, you should not be suggesting changes to the error
> messages.
> 
> Only one of the issues detected as errors can possibly be something
> other than an error either in the resolver.c code or the dtc
> compiler -- a missing symbol in the live devicetree.  This may
> be because of failing to compile the base devicetree without
> symbols, depending on a symbol from another overlay where the
> other overlay has not been applied, or depending on a symbol
> from another overlay where the other overlay is applied but
> the overlay was not compiled with symbols.  (Not meant to be
> an exhaustive list, but it might be.)  Thus the missing
> symbol problem might be fixable without a fix to kernel
> code.  The error message philosophy for overlay related
> errors is to minimize error messages that help diagnose
> the precise cause of a kernel code bug, with the intent
> of keeping the code more compact and readable.  When a
> bug occurs, debugging messages can be added for the
> debug session.

Got it, sorry about that.

> Following this philosophy, only the message in the second
> patch chunk is ok.

Then I think you can apply the v1 patch which only contains the message
about the problem I experienced, and which was caused by an incorrect DTO:

https://patchwork.ozlabs.org/patch/1243987/

Just ignore the note saying the patch is not for mainline, it's wrong.

-- 
Luca