MIME-Version: 1.0
References: <20230227064800.340833-1-korantwork@gmail.com>
In-Reply-To: <20230227064800.340833-1-korantwork@gmail.com>
From:   Xinghui Li <korantwork@gmail.com>
Date:   Mon, 27 Feb 2023 14:57:50 +0800
Message-ID: <CAEm4hYU0tnq2epaCUAeq=vpsfAfX2Sr_Z7cJa_-fHsjhc2epUw@mail.gmail.com>
Subject: Re: [PATCH] PCI:vmd: add the module param to adjust msi mode
To:     nirmal.patel@linux.intel.com, kbusch@kernel.org,
        jonathan.derrick@linux.dev, lpieralisi@kernel.org
Cc:     linux-pci@vger.kernel.org, linux-kernel@vger.kernel.org,
        Xinghui Li <korantli@tencent.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

<korantwork@gmail.com> =E4=BA=8E2023=E5=B9=B42=E6=9C=8827=E6=97=A5=E5=91=A8=
=E4=B8=80 14:48=E5=86=99=E9=81=93=EF=BC=9A
>
> From: Xinghui Li <korantli@tencent.com>
>
> In the legacy, the vmd msi-mode can only be adjusted by configing
> vmd_ids table.This patch adds another way to adjust msi mode by
> adjusting module param, which allow users easier to adjust the vmd
> according to the I/O scenario without rebuilding driver.There are two
> params could be recognized: remapping, bypass.
>
> Signed-off-by: Xinghui Li <korantli@tencent.com>
> ---
>  drivers/pci/controller/vmd.c | 24 ++++++++++++++++++++++++
>  1 file changed, 24 insertions(+)
>
> diff --git a/drivers/pci/controller/vmd.c b/drivers/pci/controller/vmd.c
> index 769eedeb8802..6d3272f5598d 100644
> --- a/drivers/pci/controller/vmd.c
> +++ b/drivers/pci/controller/vmd.c
> @@ -34,6 +34,18 @@
>  #define MB2_SHADOW_OFFSET      0x2000
>  #define MB2_SHADOW_SIZE                16
>
> +/*
> + * The VMD msi_mode module parameter provides the alternative way
> + * to adjust msi_mode when load vmd.ko other than vmd_ids table.
> + * There are two params could be recognized:
> + *
> + * 1-remapping
> + * 2-bypass
> + *
> + */
> +char *msi_mode =3D "default";
> +module_param(msi_mode, charp, 0444);
> +
>  enum vmd_features {
>         /*
>          * Device may contain registers which hint the physical location =
of the
> @@ -820,6 +832,7 @@ static int vmd_enable_domain(struct vmd_dev *vmd, uns=
igned long features)
>                         return ret;
>
>                 vmd_set_msi_remapping(vmd, true);
> +               dev_info(&vmd->dev->dev, "init vmd with remapping msi-x\n=
");
>
>                 ret =3D vmd_create_irq_domain(vmd);
>                 if (ret)
> @@ -832,6 +845,7 @@ static int vmd_enable_domain(struct vmd_dev *vmd, uns=
igned long features)
>                 irq_domain_update_bus_token(vmd->irq_domain, DOMAIN_BUS_V=
MD_MSI);
>         } else {
>                 vmd_set_msi_remapping(vmd, false);
> +               dev_info(&vmd->dev->dev, "init vmd with bypass msi-x\n");
>         }
>
>         pci_add_resource(&resources, &vmd->resources[0]);
> @@ -898,6 +912,14 @@ static int vmd_enable_domain(struct vmd_dev *vmd, un=
signed long features)
>         return 0;
>  }
>
> +static void vmd_config_msi_mode_param(unsigned long *features)
> +{
> +       if (strcmp(msi_mode, "remapping") =3D=3D 0)
> +               *features &=3D ~(VMD_FEAT_CAN_BYPASS_MSI_REMAP);
> +       else if (strcmp(msi_mode, "bypass") =3D=3D 0)
> +               *features |=3D VMD_FEAT_CAN_BYPASS_MSI_REMAP;
> +}
> +
>  static int vmd_probe(struct pci_dev *dev, const struct pci_device_id *id=
)
>  {
>         unsigned long features =3D (unsigned long) id->driver_data;
> @@ -927,6 +949,8 @@ static int vmd_probe(struct pci_dev *dev, const struc=
t pci_device_id *id)
>         if (err < 0)
>                 goto out_release_instance;
>
> +       vmd_config_msi_mode_param(&features);
> +
>         vmd->cfgbar =3D pcim_iomap(dev, VMD_CFGBAR, 0);
>         if (!vmd->cfgbar) {
>                 err =3D -ENOMEM;
> --
> 2.34.1
>

Hi all:
As we discussed in last patch (PCI: vmd: Do not disable MSI-X
remapping in VMD 28C0 controller).
I add the module param in vmd driver to adjust msi-x working mode. I
try not to modify the default mode.
And as for the `programmatic way`, I design to add the `automatic`
parameter in the next patch, which needs
more testing of different IO scenarios.