On Fri, Oct 14 2022 at 10:10, Nipun Gupta wrote:
> Since CDX devices are not linked to of node they need a
> separate MSI domain for handling device ID to be provided to
> the GIC ITS domain.
>
> This also introduces APIs to alloc and free IRQs for CDX domain.
This is not really much of an explanation.
> +#include "cdx.h"
> +
> +#define REQ_ID_SHIFT 10
> +
> +/*
> + * Convert an msi_desc to a globaly unique identifier (per-device
> + * reqid + msi_desc position in the msi_list).
> + */
> +static irq_hw_number_t cdx_domain_calc_hwirq(struct cdx_device *dev,
> + struct msi_desc *desc)
> +{
> + return (dev->req_id << REQ_ID_SHIFT) | desc->msi_index;
> +}
> +
> +static void cdx_msi_set_desc(msi_alloc_info_t *arg,
> + struct msi_desc *desc)
> +{
> + arg->desc = desc;
> + arg->hwirq = cdx_domain_calc_hwirq(to_cdx_device(desc->dev), desc);
> +}
> +
> +static void cdx_msi_write_msg(struct irq_data *irq_data,
> + struct msi_msg *msg)
> +{
> + struct msi_desc *msi_desc = irq_data_get_msi_desc(irq_data);
> + struct cdx_device *cdx_dev = to_cdx_device(msi_desc->dev);
> + struct cdx_controller_t *cdx = cdx_dev->cdx;
> + uint64_t addr;
> + int ret;
> +
> + addr = ((uint64_t)(msi_desc->msg.address_hi) << 32) |
> + msi_desc->msg.address_lo;
> +
> + ret = cdx->ops.write_msi(cdx, cdx_dev->bus_num, cdx_dev->dev_num,
> + msi_desc->msi_index, msi_desc->msg.data,
> + addr);
> + if (ret)
> + dev_err(&cdx_dev->dev, "Write MSI failed to CDX controller\n");
> +}
So this is yet another variant of platform-MSI which is about to be
kicked from the planet.
Please check:
https://lore.kernel.org/r/20221111131813.914374272@xxxxxxxxxxxxx
https://lore.kernel.org/r/20221111133158.196269823@xxxxxxxxxxxxx
and the ARM/GIC conversion in:
https://git.kernel.org/pub/scm/linux/kernel/git/tglx/devel.git/log/?h=devmsi-arm
The latter is WIP, but functional.
I'm sure you replace your yet another magic MSI domain which only
provides a device specific MSI write function with the new
infrastructure trivially.
Thanks,
tglx
