linux-kernel - Re: Questions: Should kernel panic when PCIe fatal error occurs?

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: <CAOSf1CEML1s0pOpi92By9Ms4nnq5onSX-d7wCY4jXXAviXO_qg@mail.gmail.com>
Date:   Mon, 25 Sep 2023 13:54:08 +1000
From:   "Oliver O'Halloran" <oohall@...il.com>
To:     David Laight <David.Laight@...lab.com>
Cc:     Bjorn Helgaas <helgaas@...nel.org>,
        Shuai Xue <xueshuai@...ux.alibaba.com>,
        "Rafael J. Wysocki" <rafael@...nel.org>,
        "gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>,
        Linux PCI <linux-pci@...r.kernel.org>,
        "mahesh@...ux.ibm.com" <mahesh@...ux.ibm.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-acpi@...r.kernel.org" <linux-acpi@...r.kernel.org>,
        "bp@...en8.de" <bp@...en8.de>,
        Baolin Wang <baolin.wang@...ux.alibaba.com>,
        Jonathan Cameron <Jonathan.Cameron@...wei.com>,
        "bhelgaas@...gle.com" <bhelgaas@...gle.com>,
        "james.morse@....com" <james.morse@....com>,
        "linuxppc-dev@...ts.ozlabs.org" <linuxppc-dev@...ts.ozlabs.org>,
        "lenb@...nel.org" <lenb@...nel.org>
Subject: Re: Questions: Should kernel panic when PCIe fatal error occurs?

On Fri, Sep 22, 2023 at 8:23 AM David Laight <David.Laight@...lab.com> wrote:
>
> > It would be nice if they worked the same, but I suspect that vendors
> > may rely on the fact that CPER_SEV_FATAL forces a restart/panic as
> > part of their system integrity story.
>
> The file system errors created by a panic (especially an NMI panic)
> could easily be more problematic than a failed PCIe data transfer.
> Evan a read that returned ~0u - which can be checked for.
>
> Panicking a system that is converting TDM telephony to RTP for the
> 911 emergency service because a PCIe cable/riser connecting one of the
> TDM board has become loose doesn't seem ideal.

For kernel native AER the default reaction to errors is
reset-and-reinit which probably isn't much better for your case.
Sounds like you would want a knob to suppress everything except error
reporting so you can handle it in userspace?

> (Or because the TDM board's fpga has decided it isn't going to respond
> to any accesses until the BARs are setup again...)
>
> The system can carry on with some TDM connections disabled - but that
> is ok because they are all duplicated in case a cable gets cuit.

Well that's a relief :)

Oliver