linux-kernel - Re: [PATCH] Revert "scripts/coccinelle: Find PTR

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: <aQJDIz-8Ow0OmczH@hovoldconsulting.com>
Date: Wed, 29 Oct 2025 17:38:59 +0100
From: Johan Hovold <johan@...nel.org>
To: Gal Pressman <gal@...dia.com>
Cc: Julia Lawall <Julia.Lawall@...ia.fr>,
	Nicolas Palix <nicolas.palix@...g.fr>,
	Jakub Kicinski <kuba@...nel.org>, Alexei Lazar <alazar@...dia.com>,
	Simon Horman <horms@...nel.org>, cocci@...ia.fr,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] Revert "scripts/coccinelle: Find PTR_ERR() to %pe
 candidates"

On Wed, Oct 29, 2025 at 03:59:50PM +0200, Gal Pressman wrote:
> On 29/10/2025 15:29, Johan Hovold wrote:
> > This reverts commit 57c49d2355729c12475554b4c51dbf830b02d08d.
> > 
> > Using "%pe" to print errnos is in no way mandated and a driver authors
> > may chose not to use it, for example, for consistency reasons.
> > 
> > Drop the recently added cocci script that has gotten the build bots to
> > send warning emails about perfectly valid code and which will likely
> > only result in churn and inconsistency.
> > 
> > Link: https://lore.kernel.org/all/aQHi4nUfIlcN1ac6@hovoldconsulting.com/
> > Signed-off-by: Johan Hovold <johan@...nel.org>
> 
> The test by no means mandates authors to use %pe, as the output says:
> WARNING: Consider using %pe to print PTR_ERR()
> 
> "Consider" :).

Right, but it's preceded by a big "WARNING".

> I would consider it best practice to use it, and a few drivers were
> converted thanks to this test.

Unlike the rest of the misc cocci scripts I skimmed, this one does not
guard against any bugs. Instead it's pushing for a subjective style
preference, which is just going to result in churn when the clean up
crew starts sending mindless conversions of individual printks.

By all means, use %pe for your drivers, but it should not be forced
upon the rest of us this way.

> If the issue is with automatic build bots, then maybe this test should
> be excluded from them, rather than deleted?

It's both; it's the noise the new warnings generate but also the coming
flood up patches to "fix" them. There are already some 40 commits or so
in linux-next referencing this script.

Johan