[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <YZXEBXVc03uA904k@rocinante>
Date: Thu, 18 Nov 2021 04:09:57 +0100
From: Krzysztof Wilczyński <kw@...ux.com>
To: Dexuan-Linux Cui <dexuan.linux@...il.com>
Cc: Pali Rohár <pali@...nel.org>,
Koen Vandeputte <koen.vandeputte@...ymesh.com>,
Petr Štetiar <ynezz@...e.cz>,
Bjorn Helgaas <helgaas@...nel.org>,
Bjorn Helgaas <bhelgaas@...gle.com>, linux-pci@...r.kernel.org,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
Lorenzo Pieralisi <lorenzo.pieralisi@....com>,
Gregory Clement <gregory.clement@...tlin.com>,
Andrew Lunn <andrew@...n.ch>, Yinghai Lu <yinghai@...nel.org>,
Dexuan Cui <decui@...rosoft.com>,
Sasha Levin <sashal@...nel.org>
Subject: Re: PCI: Race condition in pci_create_sysfs_dev_files
[+CC Adding Sasha for visibility]
Hello,
I am terribly sorry for a very late reply!
[...]
> I think we're seeing the same issue with a Linux VM on Hyper-V.
Are you still seeing this issue happen? If so, does it happen often?
We haven't had a report from a system that runs as a hypervisor guest
before. I also assume this is x86 or x86_64 platform, correct?
> Here the kernel is
> https://git.launchpad.net/~canonical-kernel/ubuntu/+source/linux-azure/+git/bionic/tree/?h=Ubuntu-azure-5.4-5.4.0-1061.64_18.04.1
This is quite an old kernel, however the 5.4 is a long-term kernel, and
therefore we need to make sure that it would also work properly.
> '/devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A03:00/device:07/VMBUS:01/47505500-0003-0000-3130-444531334632/pci0003:00/0003:00:00.0/config'
Following the commit e1d3f3268b0e ("PCI/sysfs: Convert "config" to static
attribute"), the issue particularly with the "config" attribute should be
resolved, thus more modern kernel versions are no long suffer from the race
condition related to this specific and some other problematic attributes.
We quite likely have to back-port these changes to a number of older
long-term kernels. I am going to look into it.
Hopefully, once the patches land in upstream long-term kernels, then
distributions would be able to cherry-pick them and apply to their kernels
releases.
Krzysztof
Powered by blists - more mailing lists