linux-kernel - Re: [PATCH v2] PCI: Document PCIE BUS MPS parameters

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: <CAErSpo6suRwkRoShnNUKxxHt4D4gFnMVqesoZO3t-aC8tvOwnA@mail.gmail.com>
Date:	Thu, 24 Jan 2013 21:57:55 -0700
From:	Bjorn Helgaas <bhelgaas@...gle.com>
To:	Yijing Wang <wangyijing@...wei.com>
Cc:	Rob Landley <rob@...dley.net>, linux-doc@...r.kernel.org,
	linux-kernel@...r.kernel.org, linux-pci@...r.kernel.org,
	Andrew Murray <andrew.murray@....com>,
	Joe Lawrence <Joe.Lawrence@...atus.com>,
	Randy Dunlap <rdunlap@...radead.org>,
	Hanjun Guo <guohanjun@...wei.com>, jiang.liu@...wei.com,
	Jon Mason <jdmason@...zu.us>
Subject: Re: [PATCH v2] PCI: Document PCIE BUS MPS parameters

[+cc Jon, can you make sure this documentation is accurate?]

On Wed, Jan 23, 2013 at 7:58 PM, Yijing Wang <wangyijing@...wei.com> wrote:
> v0->v1: Update MPS parameters as non-arch and add MRRS
>                 description into pcie_bus_perf parameter suggested
>                 by Andrew Murray.
> v1->v2: Update some semantic problems and add MPS and MRRS
>                 explanation suggested by Joe Lawrence and Randy Dunlap.
>
> Document PCIE BUS MPS parameters pcie_bus_tune_off, pcie_bus_safe,
> pcie_bus_peer2peer, pcie_bus_perf into Documentation/kernel-parameters.txt.
> These parameters were introduced by Jon Mason <mason@...i.com> at
> commit 5f39e6705 and commit b03e7495a8.
>
> Signed-off-by: Yijing Wang <wangyijing@...wei.com>
> ---
>  Documentation/kernel-parameters.txt |   16 ++++++++++++++++
>  1 files changed, 16 insertions(+), 0 deletions(-)
>
> diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
> index 363e348..0bb279f 100644
> --- a/Documentation/kernel-parameters.txt
> +++ b/Documentation/kernel-parameters.txt
> @@ -2227,6 +2227,22 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
>                                 This sorting is done to get a device
>                                 order compatible with older (<= 2.4) kernels.
>                 nobfsort        Don't sort PCI devices into breadth-first order.
> +               pcie_bus_tune_off       Disable PCIe MPS (Max Payload Size)
> +                               turning and using the BIOS-configured MPS defaults.

s/turning/tuning/
s/using/use/

> +               pcie_bus_safe   Use the smallest common denominator MPS
> +                               of the entire tree below a root complex for every
> +                               device on that fabric. Can avoid inconsistent MPS
> +                               problem caused by hotplug.

I'm not sure about this.  "smallest common denominator" is colloquial
and not exact.  I think you probably mean "the smallest supported MPS
of any device below a root complex."

I'm also not sure how this will avoid MPS problems caused by hotplug.
I think you have to assume that a hot-added device may support only a
128B MPS, which means the only way to guarantee that you can use that
device is to set MPS=128 for any device that will send packets to the
hot-added device.  Or we could reconfigure things at the time of the
hot-add, but I don't think we do that today (except for the hot-added
device itself).

> +               pcie_bus_perf   Configure pcie device MPS to the largest

Capitalize "PCIe" consistently.  Better yet, drop it completely since
it's obvious that we're only talking about PCIe here.

> +                               allowable MPS based on its parent bus. Also set
> +                               MRRS (Max Read Request Size) to the largest supported
> +                               value (no larger than the MPS that the device or bus
> +                               can support) for Max performance.

s/Max/best/

> +               pcie_bus_peer2peer      Make the system-wide MPS the smallest
> +                               possible value (128B). This configuration could prevent
> +                               peer to peer DMA transmission from working by having
> +                               the MPS on one root port different than the MPS on
> +                               another.

I think the idea is that pcie_bus_peer2peer would *allow* peer-to-peer
DMA to work, which is the opposite of what you wrote.  Maybe something
like this would be better:

  Set every device's MPS to 128B, which every device is guaranteed to support.
  This configuration allows peer-to-peer DMA between any pair of devices,
  possibly at the cost of reduced performance.

>                 cbiosize=nn[KMG]        The fixed amount of bus space which is
>                                 reserved for the CardBus bridge's IO window.
>                                 The default value is 256 bytes.
> --
> 1.7.1
>
>
--
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/