linux-kernel - Re: [PATCH 11/13] PCI: Add has_mem64 for struct host

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: <20170508132504.GA18996@bhelgaas-glaptop.roam.corp.google.com>
Date:   Mon, 8 May 2017 08:25:04 -0500
From:   Bjorn Helgaas <helgaas@...nel.org>
To:     Christian König <deathsimple@...afone.de>
Cc:     Yinghai Lu <yinghai@...nel.org>,
        Bjorn Helgaas <bhelgaas@...gle.com>,
        David Miller <davem@...emloft.net>,
        Benjamin Herrenschmidt <benh@...nel.crashing.org>,
        Wei Yang <weiyang@...ux.vnet.ibm.com>,
        Khalid Aziz <khalid.aziz@...cle.com>,
        linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 11/13] PCI: Add has_mem64 for struct host_bridge

On Mon, May 08, 2017 at 10:54:55AM +0200, Christian König wrote:
> Am 05.05.2017 um 01:04 schrieb Bjorn Helgaas:
> >I *think* this will be broken by the current implementation of
> >Christian's patch to enable a 64-bit host bridge window:
> >
> >   https://lkml.kernel.org/r/1493890270-1188-5-git-send-email-deathsimple@vodafone.de
> >
> >because pci_register_host_bridge() runs before we scan the bus, and
> >Christian's patch adds a quirk that runs when we enumerate the AMD
> >host bridge device.
> >
> >If we apply this and Christian's patch, I think we could end up with
> >a host bridge window above 4G, but with bridge->has_mem64 not set.
> Yes, indeed. I can adjust my patch, but I would prefer not to do so.
> 
> I don't completely understand the background of this change, but
> from what I know how the BIOS (at least on X86) allocates resources
> it doesn't sounds correct to me.
> 
> Maybe we just need a Sparc specific quirk here instead of changing
> the common logic?

There's nothing in Yinghai's patch that's conceptually Sparc-specific,
so I would prefer not to artificially tie it to Sparc.

One possibility would be to compute has_mem64 when we need it instead
of caching it.