netdev - Re: [PATCH V4 mlx5-next 11/13] vfio/mlx5: Implement vfio

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]

Date:   Tue, 26 Oct 2021 16:42:28 -0600
From:   Alex Williamson <alex.williamson@...hat.com>
To:     Yishai Hadas <yishaih@...dia.com>
Cc:     <bhelgaas@...gle.com>, <jgg@...dia.com>, <saeedm@...dia.com>,
        <linux-pci@...r.kernel.org>, <kvm@...r.kernel.org>,
        <netdev@...r.kernel.org>, <kuba@...nel.org>, <leonro@...dia.com>,
        <kwankhede@...dia.com>, <mgurtovoy@...dia.com>, <maorg@...dia.com>
Subject: Re: [PATCH V4 mlx5-next 11/13] vfio/mlx5: Implement vfio_pci driver
 for mlx5 devices

On Tue, 26 Oct 2021 12:06:03 +0300
Yishai Hadas <yishaih@...dia.com> wrote:
> +static int mlx5vf_pci_set_device_state(struct mlx5vf_pci_core_device *mvdev,
> +				       u32 state)
> +{
> +	struct mlx5vf_pci_migration_info *vmig = &mvdev->vmig;
> +	u32 old_state = vmig->vfio_dev_state;
> +	u32 flipped_bits = old_state ^ state;
> +	int ret = 0;
> +
> +	if (old_state == VFIO_DEVICE_STATE_ERROR ||
> +	    !VFIO_DEVICE_STATE_VALID(state) ||
> +	    (state & ~MLX5VF_SUPPORTED_DEVICE_STATES))
> +		return -EINVAL;
> +
> +	/* Running switches off */
> +	if ((flipped_bits & VFIO_DEVICE_STATE_RUNNING) &&
> +	    !(state & VFIO_DEVICE_STATE_RUNNING)) {
> +		ret = mlx5vf_pci_quiesce_device(mvdev);
> +		if (ret)
> +			return ret;
> +		ret = mlx5vf_pci_freeze_device(mvdev);
> +		if (ret) {
> +			vmig->vfio_dev_state = VFIO_DEVICE_STATE_ERROR;
> +			return ret;

Is it not possible for this to unwind, only entering the error state if
unquiesce also fails?

> +		}
> +	}
> +
> +	/* Saving switches on and not running */
> +	if ((flipped_bits &
> +	     (VFIO_DEVICE_STATE_RUNNING | VFIO_DEVICE_STATE_SAVING)) &&
> +	    ((state & (VFIO_DEVICE_STATE_RUNNING |
> +	      VFIO_DEVICE_STATE_SAVING)) == VFIO_DEVICE_STATE_SAVING)) {

Can't this be reduced to:

 if ((flipped_bits & ~VFIO_DEVICE_STATE_RESUMING) &&
     (state == VFIO_DEVICE_STATE_SAVING)) {

Maybe there's an argument for the original to be more invariant of TBD
device_state bits?  Thanks,

Alex

> +		ret = mlx5vf_pci_save_device_data(mvdev);
> +		if (ret)
> +			return ret;
> +	}
> +
> +	/* Resuming switches on */
> +	if ((flipped_bits & VFIO_DEVICE_STATE_RESUMING) &&
> +	    (state & VFIO_DEVICE_STATE_RESUMING)) {
> +		mlx5vf_reset_mig_state(mvdev);
> +		ret = mlx5vf_pci_new_write_window(mvdev);
> +		if (ret)
> +			return ret;
> +	}
> +
> +	/* Resuming switches off */
> +	if ((flipped_bits & VFIO_DEVICE_STATE_RESUMING) &&
> +	    !(state & VFIO_DEVICE_STATE_RESUMING)) {
> +		/* deserialize state into the device */
> +		ret = mlx5vf_load_state(mvdev);
> +		if (ret) {
> +			vmig->vfio_dev_state = VFIO_DEVICE_STATE_ERROR;
> +			return ret;
> +		}
> +	}
> +
> +	/* Running switches on */
> +	if ((flipped_bits & VFIO_DEVICE_STATE_RUNNING) &&
> +	    (state & VFIO_DEVICE_STATE_RUNNING)) {
> +		ret = mlx5vf_pci_unfreeze_device(mvdev);
> +		if (ret)
> +			return ret;
> +		ret = mlx5vf_pci_unquiesce_device(mvdev);
> +		if (ret) {
> +			vmig->vfio_dev_state = VFIO_DEVICE_STATE_ERROR;
> +			return ret;
> +		}
> +	}
> +
> +	vmig->vfio_dev_state = state;
> +	return 0;
> +}