From: Nuno Das Neves <nunodasneves@linux.microsoft.com>
To: Stanislav Kinsburskii <skinsburskii@linux.microsoft.com>,
kys@microsoft.com, haiyangz@microsoft.com, wei.liu@kernel.org,
decui@microsoft.com
Cc: linux-hyperv@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v4 3/5] Drivers: hv: Batch GPA unmap operations to improve large region performance
Date: Fri, 10 Oct 2025 16:06:43 -0700 [thread overview]
Message-ID: <5be44d00-2a34-49cf-828f-105c2e07874b@linux.microsoft.com> (raw)
In-Reply-To: <175976318688.16834.16198650808431263017.stgit@skinsburskii-cloud-desktop.internal.cloudapp.net>
On 10/6/2025 8:06 AM, Stanislav Kinsburskii wrote:
> Reduce overhead when unmapping large memory regions by batching GPA unmap
> operations in 2MB-aligned chunks.
>
> Use a dedicated constant for batch size to improve code clarity and
> maintainability.
>
> Signed-off-by: Stanislav Kinsburskii <skinsburskii@linux.microsoft.com>
> ---
> drivers/hv/mshv_root.h | 2 ++
> drivers/hv/mshv_root_hv_call.c | 2 +-
> drivers/hv/mshv_root_main.c | 28 +++++++++++++++++++++++++---
> 3 files changed, 28 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/hv/mshv_root.h b/drivers/hv/mshv_root.h
> index e3931b0f12693..97e64d5341b6e 100644
> --- a/drivers/hv/mshv_root.h
> +++ b/drivers/hv/mshv_root.h
> @@ -32,6 +32,8 @@ static_assert(HV_HYP_PAGE_SIZE == MSHV_HV_PAGE_SIZE);
>
> #define MSHV_PIN_PAGES_BATCH_SIZE (0x10000000ULL / HV_HYP_PAGE_SIZE)
>
> +#define MSHV_MAX_UNMAP_GPA_PAGES 512
> +
> struct mshv_vp {
> u32 vp_index;
> struct mshv_partition *vp_partition;
> diff --git a/drivers/hv/mshv_root_hv_call.c b/drivers/hv/mshv_root_hv_call.c
> index c9c274f29c3c6..0696024ccfe31 100644
> --- a/drivers/hv/mshv_root_hv_call.c
> +++ b/drivers/hv/mshv_root_hv_call.c
> @@ -17,7 +17,7 @@
> /* Determined empirically */
> #define HV_INIT_PARTITION_DEPOSIT_PAGES 208
> #define HV_MAP_GPA_DEPOSIT_PAGES 256
> -#define HV_UMAP_GPA_PAGES 512
> +#define HV_UMAP_GPA_PAGES MSHV_MAX_UNMAP_GPA_PAGES
>
> #define HV_PAGE_COUNT_2M_ALIGNED(pg_count) (!((pg_count) & (0x200 - 1)))
>
> diff --git a/drivers/hv/mshv_root_main.c b/drivers/hv/mshv_root_main.c
> index 97e322f3c6b5e..b61bef6b9c132 100644
> --- a/drivers/hv/mshv_root_main.c
> +++ b/drivers/hv/mshv_root_main.c
> @@ -1378,6 +1378,7 @@ mshv_map_user_memory(struct mshv_partition *partition,
> static void mshv_partition_destroy_region(struct mshv_mem_region *region)
> {
> struct mshv_partition *partition = region->partition;
> + u64 gfn, gfn_count, start_gfn, end_gfn;
> u32 unmap_flags = 0;
> int ret;
>
> @@ -1396,9 +1397,30 @@ static void mshv_partition_destroy_region(struct mshv_mem_region *region)
> if (region->flags.large_pages)
> unmap_flags |= HV_UNMAP_GPA_LARGE_PAGE;
>
> - /* ignore unmap failures and continue as process may be exiting */
> - hv_call_unmap_gpa_pages(partition->pt_id, region->start_gfn,
> - region->nr_pages, unmap_flags);
> + start_gfn = region->start_gfn;
> + end_gfn = region->start_gfn + region->nr_pages;
> +
> + for (gfn = start_gfn; gfn < end_gfn; gfn += gfn_count) {
> + if (gfn % MSHV_MAX_UNMAP_GPA_PAGES)
> + gfn_count = ALIGN(gfn, MSHV_MAX_UNMAP_GPA_PAGES) - gfn;
> + else
> + gfn_count = MSHV_MAX_UNMAP_GPA_PAGES;
> +
> + if (gfn + gfn_count > end_gfn)
> + gfn_count = end_gfn - gfn;
> +
> + /* Skip if all pages in this range if none is mapped */
nit: comment grammar. Suggest "Skip all pages in this range if none are mapped"
Looks good otherwise.
Reviewed-by: Nuno Das Neves <nunodasneves@linux.microsoft.com>
> + if (!memchr_inv(region->pages + (gfn - start_gfn), 0,
> + gfn_count * sizeof(struct page *)))
> + continue;
> +
> + ret = hv_call_unmap_gpa_pages(partition->pt_id, gfn,
> + gfn_count, unmap_flags);
> + if (ret)
> + pt_err(partition,
> + "Failed to unmap GPA pages %#llx-%#llx: %d\n",
> + gfn, gfn + gfn_count - 1, ret);
> + }
>
> mshv_region_invalidate(region);
>
>
>
next prev parent reply other threads:[~2025-10-10 23:06 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-10-06 15:06 [PATCH v4 0/5] Introduce movable pages for Hyper-V guests Stanislav Kinsburskii
2025-10-06 15:06 ` [PATCH v4 1/5] Drivers: hv: Refactor and rename memory region handling functions Stanislav Kinsburskii
2025-10-10 22:49 ` Nuno Das Neves
2025-10-06 15:06 ` [PATCH v4 2/5] Drivers: hv: Centralize guest memory region destruction Stanislav Kinsburskii
2025-10-10 22:54 ` Nuno Das Neves
2025-10-06 15:06 ` [PATCH v4 3/5] Drivers: hv: Batch GPA unmap operations to improve large region performance Stanislav Kinsburskii
2025-10-06 17:09 ` Michael Kelley
2025-10-06 23:32 ` Stanislav Kinsburskii
2025-10-07 0:02 ` Michael Kelley
2025-10-10 23:06 ` Nuno Das Neves [this message]
2025-10-06 15:06 ` [PATCH v4 4/5] Drivers: hv: Ensure large page GPA mapping is PMD-aligned Stanislav Kinsburskii
2025-10-10 23:07 ` Nuno Das Neves
2025-10-06 15:06 ` [PATCH v4 5/5] Drivers: hv: Add support for movable memory regions Stanislav Kinsburskii
2025-10-10 17:48 ` kernel test robot
2025-10-11 1:12 ` kernel test robot
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5be44d00-2a34-49cf-828f-105c2e07874b@linux.microsoft.com \
--to=nunodasneves@linux.microsoft.com \
--cc=decui@microsoft.com \
--cc=haiyangz@microsoft.com \
--cc=kys@microsoft.com \
--cc=linux-hyperv@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=skinsburskii@linux.microsoft.com \
--cc=wei.liu@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®