Date:   Mon, 25 Feb 2019 13:45:35 +0100
From:   Ard Biesheuvel <ard.biesheuvel@...aro.org>
To:     Marc Zyngier <marc.zyngier@....com>
Cc:     Amitkumar Karwar <amitkarwar@...il.com>,
        Enric Balletbo i Serra <enric.balletbo@...labora.com>,
        Ganapathi Bhat <gbhat@...vell.com>,
        Heiko Stuebner <heiko@...ech.de>,
        Kalle Valo <kvalo@...eaurora.org>,
        Nishant Sarmukadam <nishants@...vell.com>,
        Rob Herring <robh+dt@...nel.org>,
        Xinming Hu <huxinming820@...il.com>,
        Devicetree List <devicetree@...r.kernel.org>,
        "<netdev@...r.kernel.org>" <netdev@...r.kernel.org>,
        "<linux-wireless@...r.kernel.org>" <linux-wireless@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        linux-rockchip@...ts.infradead.org,
        "David S. Miller" <davem@...emloft.net>,
        linux-arm-kernel <linux-arm-kernel@...ts.infradead.org>
Subject: Re: [PATCH 0/4] mwifiex PCI/wake-up interrupt fixes

On Sun, 24 Feb 2019 at 15:08, Marc Zyngier <marc.zyngier@....com> wrote:
>
> For quite some time, I wondered why the PCI mwifiex device built in my
> Chromebook was unable to use the good old legacy interrupts. But as MSIs
> were working fine, I never really bothered investigating. I finally had a
> look, and the result isn't very pretty.
>
> On this machine (rk3399-based kevin), the wake-up interrupt is described as
> such:
>
> &pci_rootport {
>         mvl_wifi: wifi@0,0 {
>                 compatible = "pci1b4b,2b42";
>                 reg = <0x83010000 0x0 0x00000000 0x0 0x00100000
>                        0x83010000 0x0 0x00100000 0x0 0x00100000>;
>                 interrupt-parent = <&gpio0>;
>                 interrupts = <8 IRQ_TYPE_LEVEL_LOW>;
>                 pinctrl-names = "default";
>                 pinctrl-0 = <&wlan_host_wake_l>;
>                 wakeup-source;
>         };
> };
>
> Note how the interrupt is part of the properties directly attached to the
> PCI node. And yet, this interrupt has nothing to do with a PCI legacy
> interrupt, as it is attached to the wake-up widget that bypasses the PCIe RC
> altogether (Yay for the broken design!). This is in total violation of the
> IEEE Std 1275-1994 spec[1], which clearly documents that such interrupt
> specifiers describe the PCI device interrupts, and must obey the
> INT-{A,B,C,D} mapping. Oops!
>

The mapping of legacy PCIe INTx interrupts onto wired system
interrupts is a property of the PCIe host controller, not of a
particular PCIe device. So I would argue that the code is broken here
as well: it should never attempt to interpret 'interrupt' properties
at the PCI device level as having any bearing on how legacy interrupts
are routed.

> The net effect of the above is that Linux tries to do something vaguely
> sensible, and uses the same interrupt for both the wake-up widget and the
> PCI device. This doesn't work for two reasons: (1) the wake-up widget grabs
> the interrupt in exclusive mode, and (2) the PCI interrupt is still routed
> to the RC, leading to a screaming interrupt. This simply cannot work.
>
> To sort out this mess, we need to lift the confusion between the two
> interrupts. This is done by extending the DT binding to allow the wake-up
> interrupt to be described in a 'wake-up' subnode, sidestepping the issue
> completely. On my Chromebook, it now looks like this:
>
> &pci_rootport {
>         mvl_wifi: wifi@0,0 {
>                 compatible = "pci1b4b,2b42";
>                 reg = <0x83010000 0x0 0x00000000 0x0 0x00100000
>                        0x83010000 0x0 0x00100000 0x0 0x00100000>;
>                 pinctrl-names = "default";
>                 pinctrl-0 = <&wlan_host_wake_l>;
>                 wake-up {
>                         interrupt-parent = <&gpio0>;
>                         interrupts = <8 IRQ_TYPE_LEVEL_LOW>;
>                         wakeup-source;
>                 };
>         };
> };
>
> The driver is then updated to look for this subnode first, and fallback to
> the original, broken behaviour (spitting out a warning in the offending
> configuration).
>
> For good measure, there are two additional patches:
>
> - The wake-up interrupt requesting is horribly racy, and could lead to
>   unpredictable behaviours. Let's fix that properly.
>
> - A final patch implementing the above transformation for the whole
>   RK3399-based Chromebook range, which all use the same broken
>   configuration.
>
> With all that, I finally have PCI legacy interrupts working with the mwifiex
> driver on my Chromebook.
>
> [1] http://www.devicetree.org/open-firmware/bindings/pci/pci2_1.pdf
>
> Marc Zyngier (4):
>   dt-bindings/marvell-8xxx: Allow wake-up interrupt to be placed in a
>     separate node
>   mwifiex: Fetch wake-up interrupt from 'wake-up' subnode when it exists
>   mwifiex: Flag wake-up interrupt as IRQ_NOAUTOEN rather than disabling
>     it too late
>   arm64: dts: rockchip: gru: Move wifi wake-up interrupt into its own
>     subnode
>
>  .../bindings/net/wireless/marvell-8xxx.txt    | 23 ++++++++++++++++++-
>  .../dts/rockchip/rk3399-gru-chromebook.dtsi   |  8 ++++---
>  drivers/net/wireless/marvell/mwifiex/main.c   | 13 +++++++++--
>  3 files changed, 38 insertions(+), 6 deletions(-)
>
> --
> 2.20.1
>
>
> _______________________________________________
> linux-arm-kernel mailing list
> linux-arm-kernel@...ts.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Powered by blists - more mailing lists