linux-kernel - Re: [Regression] [PCI/ASPM] [ASUS PN51] Reboot on resume attempt (bisect done; commit found)

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: <20240105155100.GA1861423@bhelgaas>
Date: Fri, 5 Jan 2024 09:51:00 -0600
From: Bjorn Helgaas <helgaas@...nel.org>
To: Michael Schaller <michael@...aller.de>
Cc: Kai-Heng Feng <kai.heng.feng@...onical.com>, bhelgaas@...gle.com,
	linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org,
	regressions@...ts.linux.dev, macro@...am.me.uk,
	ajayagarwal@...gle.com, sathyanarayanan.kuppuswamy@...ux.intel.com,
	gregkh@...uxfoundation.org, hkallweit1@...il.com,
	michael.a.bottini@...ux.intel.com, johan+linaro@...nel.org
Subject: Re: [Regression] [PCI/ASPM] [ASUS PN51] Reboot on resume attempt
 (bisect done; commit found)

On Fri, Jan 05, 2024 at 12:18:32PM +0100, Michael Schaller wrote:
> On 05.01.24 04:25, Kai-Heng Feng wrote:
> > Just wondering, does `echo 0 > /sys/power/pm_asysnc` help?
> 
> Yes, `echo 0 | sudo tee /sys/power/pm_async` does indeed also result in a
> working resume. I've tested this on kernel 6.6.9 (which still has commit
> 08d0cc5f3426). I've also attached the relevant dmesg output of the
> suspend/resume cycle in case this helps.

Thanks for testing that!

> Furthermore does this mean that commit 08d0cc5f3426 isn't at fault but
> rather that we are dealing with a timing issue?

PCI does have a few software timing requirements, mostly related to
reset and power state (D0/D3cold).  ASPM has some timing parameters,
too, but I think they're all requirements on the hardware, not on
software.

Adding an arbitrary delay anywhere shouldn't break anything, and other
than those few required situations, it shouldn't fix anything either.

Bjorn