linux-kernel - Re: [PATCH RESEND] power: supply: qcom_battmgr: abs() on POWER

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] [day] [month] [year] [list]

Message-ID: <ycz7kwtr5d6bnrjjfmauktq2s7vtwsunpg7nfunfywxu5uou34@gfgoeeyarh46>
Date: Fri, 28 Feb 2025 16:25:47 +0100
From: Anthony Ruhier <aruhier@...lbox.org>
To: Sebastian Reichel <sebastian.reichel@...labora.com>
Cc: Dmitry Baryshkov <dmitry.baryshkov@...aro.org>, 
	linux-arm-msm@...r.kernel.org, linux-pm@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH RESEND] power: supply: qcom_battmgr: abs() on POWER_NOW
 property

On Sat, Feb 15, 2025 at 04:08:25AM +0100, Sebastian Reichel wrote:
> Hi,
>
> There are other drivers reporting negative values as documented.
> Most of the embedded ones do this actually and there surely are
> (embedded) userspace programs relying on this by now. But the
> most used driver - generic ACPI battery - does not. That's why
> quite a few userspace tools handle it wrong without anyone
> noticing for quite some time. Fixing it to follow the ABI would
> obviously end up in a bunch of regression reports, so things are
> a bit messy :(
>
> > I think it is a problem of the 'acpi' tool. At least 'upower -d' uses
> > fabs internally since the initial commit in 2008.
>
> It's definitely sensible to fix the userspace tools. We can't change
> the documented ABI for current_now after that many years and while
> documentation for power_now is missing, it would be quite unexpected
> to have it behave differently than current_now. Also userspace
> tooling needs to handle current_now and power_now anyways. And we
> surely can't change the behaviour for all drivers reporting signed
> data. So let's keep qcom_battmgr as is. It follows the documented
> ABI and hopefully helps giving this more exposure (I'm typing this
> on a X1E laptop right now and can see your problem with waybar).
>
> But we should document the power_now property. It somehow fell
> through the cracks :)
>
> -- Sebastian

Hi,
As an update around this topic, I sent some patches in the different tools I'm
using to correctly handle negative values in current_now and power_now:

  * Waybar (included in release 0.12.0): https://github.com/Alexays/Waybar/pull/3942
  * Powertop (merged): https://github.com/fenrus75/powertop/pull/173
  * acpi-client (included in release 1.8): https://sourceforge.net/p/acpiclient/code/merge-requests/1/

It was quicker to get this merged than what I expected, which is good news!

There's probably other tools to fix, I just fixed the tools I'm using. I
encounter the issue on other tools, I'll send a patch.

--
Anthony Ruhier