linux-kernel - Re: [PATCH 23/23] x86, PCI: add pcibios_root

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]

Date:	Wed, 7 Mar 2012 21:27:47 -0700
From:	Bjorn Helgaas <bhelgaas@...gle.com>
To:	Yinghai Lu <yinghai@...nel.org>
Cc:	Bjorn Helgaas <bjorn.helgaas@...il.com>,
	Jesse Barnes <jbarnes@...tuousgeek.org>, x86 <x86@...nel.org>,
	linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org,
	Jan Beulich <JBeulich@...e.com>,
	Mauro Carvalho Chehab <mchehab@...hat.com>
Subject: Re: [PATCH 23/23] x86, PCI: add pcibios_root_rescan

On Wed, Mar 7, 2012 at 5:58 PM, Yinghai Lu <yinghai@...nel.org> wrote:
> On Wed, Mar 7, 2012 at 3:32 PM, Bjorn Helgaas <bhelgaas@...gle.com> wrote:
>>
>> What's the reason we need to discover those CPU devices in the first place?
>>
>> I don't think the current policy of blindly probing for things should
>> be continued forever.
>
> Do you want to revert follow commit?
>
> BTW, edac never work on nehalem-ex and westmere-ex.

Oh, yes, now I remember.  Apparently the i7core-edac driver relies on
these devices.  We had a discussion about this a year ago:

    http://www.spinics.net/lists/linux-pci/msg10426.html

[+cc Jan and Mauro]

Personally, I think we should put a stake in the ground and say "For
machines newer than 2013, Linux will not blindly probe for PCI
devices.  If you want EDAC functionality, make sure your BIOS exposes
the appropriate PCI host bridges."  If OEMs do care about EDAC, it's a
simple BIOS change to do this.

I certainly don't think we need to add hotplug or bus rescan
functionality to cover this case.  This sort of stuff makes
maintenance MUCH harder because we have to worry about all these
corner cases.

Bjorn

> commit a3170c1f924ce2565c4e160b9b095e65c03b2dc6
> Author: Jan Beulich <JBeulich@...ell.com>
> Date:   Wed Feb 23 10:08:10 2011 +0000
>
>    x86/PCI: derive pcibios_last_bus from ACPI MCFG
>
>    On various newer Intel systems the PCI bus(ses) the non-core devices
>    live on aren't getting announced by ACPI except through the bus range
>    covered by mmconfig. At least the i7core-edac driver depends on these
>    devices getting detected.
>
>    Mauro, could you check whether with this change the Xeon 55xx hack in
>    that driver can go away altogether, and with it the bogus exporting of
>    pcibios_scan_specific_bus()?
>
>    Signed-off-by: Jan Beulich <jbeulich@...ell.com>
>    Cc: Mauro Carvalho Chehab <mchehab@...hat.com>
>    Cc: Aristeu Sergio <arozansk@...hat.com>
>    Signed-off-by: Jesse Barnes <jbarnes@...tuousgeek.org>
>
> diff --git a/arch/x86/pci/mmconfig-shared.c b/arch/x86/pci/mmconfig-shared.c
> index e282886..750c346 100644
> --- a/arch/x86/pci/mmconfig-shared.c
> +++ b/arch/x86/pci/mmconfig-shared.c
> @@ -606,6 +606,16 @@ static void __init __pci_mmcfg_init(int early)
>        if (list_empty(&pci_mmcfg_list))
>                return;
>
> +       if (pcibios_last_bus < 0) {
> +               const struct pci_mmcfg_region *cfg;
> +
> +               list_for_each_entry(cfg, &pci_mmcfg_list, list) {
> +                       if (cfg->segment)
> +                               break;
> +                       pcibios_last_bus = cfg->end_bus;
> +               }
> +       }
> +
>        if (pci_mmcfg_arch_init())
>                pci_probe = (pci_probe & ~PCI_PROBE_MASK) | PCI_PROBE_MMCONF;
>        else {
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/