Re: [PATCH v2 2/7] mpi3mr: add support for driver commands

[Bart Van Assche <bvanassche@acm.org>]

On 3/29/22 11:06, Sumit Saxena wrote:
> +/**
> + * struct mpi3mr_nvme_pt_sge -  Structure to store SGEs for NVMe
> + * Encapsulated commands.
> + *
> + * @base_addr: Physical address
> + * @length: SGE length
> + * @rsvd: Reserved
> + * @rsvd1: Reserved
> + * @sgl_type: sgl type
> + */
> +struct mpi3mr_nvme_pt_sge {
> +	u64 base_addr;
> +	u32 length;
> +	u16 rsvd;
> +	u8 rsvd1;
> +	u8 sgl_type;
> +};

Does the above data structure force user space software to pass a 
physical address to the kernel? A kernel driver should not do this 
unless there is a very good reason to do so. Passing a physical address 
from user space to the kernel requires freezing the virtual-to-physical 
mapping. Some user space developers erroneously use mlock() for this 
purpose. Is there any other way than the VFIO_IOMMU_MAP_DMA ioctl to 
prevent that the kernel modifies the virtual-to-physical mapping?

Thanks,

Bart.