Re: [PATCH v2 03/11] coco: guest: arm64: Add support for guest initiated TDI bind/unbind

[Jonathan Cameron]

On Mon, 17 Nov 2025 19:29:59 +0530
"Aneesh Kumar K.V (Arm)" <aneesh.kumar@xxxxxxxxxx> wrote:

> Add RHI for VDEV_SET_TDI_STATE
> 
> Note: This is not part of RHI spec. This is a POC implementation
> and will be later converted to correct interface defined by RHI.
> 
> Signed-off-by: Aneesh Kumar K.V (Arm) <aneesh.kumar@xxxxxxxxxx>

> diff --git a/drivers/virt/coco/arm-cca-guest/rsi-da.c b/drivers/virt/coco/arm-cca-guest/rsi-da.c
> new file mode 100644
> index 000000000000..6770861629f2
> --- /dev/null
> +++ b/drivers/virt/coco/arm-cca-guest/rsi-da.c
> @@ -0,0 +1,36 @@
> +// SPDX-License-Identifier: GPL-2.0-only
> +/*
> + * Copyright (C) 2025 ARM Ltd.
> + */
> +
> +#include <linux/pci.h>
> +#include <asm/rsi_cmds.h>
> +
> +#include "rsi-da.h"
> +#include "rhi-da.h"
> +
> +#define PCI_TDISP_MESSAGE_VERSION_10	0x10

Not sure why this define is here.  It sounds generic
and looks ot be the TDISPVersion field content for first
byte of a TDISP message.  If so should be in a PCI header
not here.

> +
> +int cca_device_lock(struct pci_dev *pdev)
> +{
> +	int ret;
> +
> +	ret = rhi_vdev_set_tdi_state(pdev, RHI_DA_TDI_CONFIG_LOCKED);
> +	if (ret) {
> +		pci_err(pdev, "failed to lock the device (%u)\n", ret);
> +		return -EIO;
Why eat ret?  It might have a useful error value to the caller.
If there is a reason -EIO is special then add a comment here to explain
that.
> +	}
> +	return 0;
> +}
> +
> +int cca_device_unlock(struct pci_dev *pdev)
> +{
> +	int ret;
> +
> +	ret = rhi_vdev_set_tdi_state(pdev, RHI_DA_TDI_CONFIG_UNLOCKED);
> +	if (ret) {
> +		pci_err(pdev, "failed to unlock the device (%u)\n", ret);
> +		return -EIO;
Same as above.
> +	}
> +	return 0;
> +}