Re: [PATCH for-next v3 4/7] RDMA/rxe: Add page invalidation support

From: Jason Gunthorpe <jgg@nvidia.com>
To: Daisuke Matsuda <matsuda-daisuke@fujitsu.com>
Cc: linux-rdma@vger.kernel.org, leonro@nvidia.com,
	zyjzyj2000@gmail.com, nvdimm@lists.linux.dev,
	linux-kernel@vger.kernel.org, rpearsonhpe@gmail.com,
	yangx.jy@fujitsu.com, lizhijian@fujitsu.com, y-goto@fujitsu.com
Subject: Re: [PATCH for-next v3 4/7] RDMA/rxe: Add page invalidation support
Date: Mon, 16 Jan 2023 14:23:56 -0400	[thread overview]
Message-ID: <Y8WWPAHvZsXpwc0M@nvidia.com> (raw)
In-Reply-To: <97b0362e256dd3eb022d81c30941edd8fb0caba9.1671772917.git.matsuda-daisuke@fujitsu.com>

On Fri, Dec 23, 2022 at 03:51:55PM +0900, Daisuke Matsuda wrote:

> +static bool rxe_ib_invalidate_range(struct mmu_interval_notifier *mni,
> +				    const struct mmu_notifier_range *range,
> +				    unsigned long cur_seq)
> +{
> +	struct ib_umem_odp *umem_odp =
> +		container_of(mni, struct ib_umem_odp, notifier);
> +	unsigned long start;
> +	unsigned long end;
> +
> +	if (!mmu_notifier_range_blockable(range))
> +		return false;
> +
> +	mutex_lock(&umem_odp->umem_mutex);
> +	mmu_interval_set_seq(mni, cur_seq);
> +
> +	start = max_t(u64, ib_umem_start(umem_odp), range->start);
> +	end = min_t(u64, ib_umem_end(umem_odp), range->end);
> +
> +	ib_umem_odp_unmap_dma_pages(umem_odp, start, end);

After bob's xarray conversion this can be done alot faster, it just an
xa_for_each_range and make the xarray items non-present

non-present is probably just a null struct page in the xarray.

Jason