From: Guanghui Feng <guanghuifeng@linux•alibaba.com>
To: adrian.larumbe@collabora•com, airlied@gmail•com,
alex@shazbot•org, baolu.lu@linux•intel.com,
boris.brezillon@collabora•com, dri-devel@lists•freedesktop.org,
dwmw2@infradead•org, iommu@lists•linux.dev, jgg@ziepe•ca,
joro@8bytes•org, kevin.tian@intel•com, kvm@vger•kernel.org,
linux-arm-kernel@lists•infradead.org,
linux-kernel@vger•kernel.org, liviu.dudau@arm•com,
maarten.lankhorst@linux•intel.com, mripard@kernel•org,
oliver.yang@linux•alibaba.com, robh@kernel•org,
robin.murphy@arm•com, shiyu.zsq@linux•alibaba.com,
steven.price@arm•com, suravee.suthikulpanit@amd•com,
tzimmermann@suse•de, wei.guo.simon@linux•alibaba.com,
will@kernel•org, xlpang@linux•alibaba.com
Cc: alikernel-developer@linux•alibaba.com
Subject: [PATCH v2 23/30] vfio/iommufd: use iova_to_phys_length for efficient unmap
Date: Tue, 2 Jun 2026 18:46:30 +0800 [thread overview]
Message-ID: <20260602104637.1219810-24-guanghuifeng@linux.alibaba.com> (raw)
In-Reply-To: <20260602104637.1219810-1-guanghuifeng@linux.alibaba.com>
Use iommu_iova_to_phys_length() to get PTE page size, allowing
traversal by actual mapping granularity instead of PAGE_SIZE steps.
Signed-off-by: Guanghui Feng <guanghuifeng@linux•alibaba.com>
Acked-by: Shiqiang Zhang <shiyu.zsq@linux•alibaba.com>
Acked-by: Simon Guo <wei.guo.simon@linux•alibaba.com>
---
drivers/iommu/iommufd/pages.c | 75 +++++++++++++++++++++++++++-----
drivers/iommu/iommufd/selftest.c | 2 +-
drivers/vfio/vfio_iommu_type1.c | 26 ++++++++---
3 files changed, 85 insertions(+), 18 deletions(-)
diff --git a/drivers/iommu/iommufd/pages.c b/drivers/iommu/iommufd/pages.c
index 9bdb2945afe1..aed05bd0b01c 100644
--- a/drivers/iommu/iommufd/pages.c
+++ b/drivers/iommu/iommufd/pages.c
@@ -417,17 +417,44 @@ static void batch_from_domain(struct pfn_batch *batch,
if (start_index == iopt_area_index(area))
page_offset = area->page_offset;
while (start_index <= last_index) {
+ size_t pgsize;
+ unsigned long npages;
+ unsigned long i;
+
/*
- * This is pretty slow, it would be nice to get the page size
- * back from the driver, or have the driver directly fill the
- * batch.
+ * Use iova_to_phys_length to get both the physical address
+ * and the PTE page size in a single page table walk, allowing
+ * us to skip ahead by the contiguous region size instead of
+ * walking the page tables for every PAGE_SIZE step.
*/
- phys = iommu_iova_to_phys(domain, iova) - page_offset;
- if (!batch_add_pfn(batch, PHYS_PFN(phys)))
- return;
- iova += PAGE_SIZE - page_offset;
+ phys = iommu_iova_to_phys_length(domain, iova, &pgsize);
+ if (WARN_ON(phys == PHYS_ADDR_MAX))
+ break;
+ phys -= page_offset;
+ if (WARN_ON(!pgsize || pgsize < PAGE_SIZE))
+ pgsize = PAGE_SIZE;
+
+ /*
+ * Calculate contiguous pages within this PTE from our
+ * position. phys points to the page-aligned start (backed
+ * up by page_offset), so pages available = bytes from phys
+ * to PTE end divided by PAGE_SIZE.
+ */
+ npages = (pgsize - (iova & (pgsize - 1)) + page_offset) /
+ PAGE_SIZE;
+ npages = min_t(unsigned long, npages,
+ last_index - start_index + 1);
+ if (!npages)
+ npages = 1;
+
+ for (i = 0; i < npages; i++) {
+ if (!batch_add_pfn(batch, PHYS_PFN(phys) + i))
+ return;
+ }
+
+ iova += npages * PAGE_SIZE - page_offset;
page_offset = 0;
- start_index++;
+ start_index += npages;
}
}
@@ -445,11 +472,35 @@ static struct page **raw_pages_from_domain(struct iommu_domain *domain,
if (start_index == iopt_area_index(area))
page_offset = area->page_offset;
while (start_index <= last_index) {
- phys = iommu_iova_to_phys(domain, iova) - page_offset;
- *(out_pages++) = pfn_to_page(PHYS_PFN(phys));
- iova += PAGE_SIZE - page_offset;
+ size_t pgsize;
+ unsigned long npages;
+ unsigned long i;
+
+ /*
+ * Resolve the PTE page size together with the physical
+ * address so we can fill multiple struct page pointers per
+ * page table walk when the IOMMU uses large pages.
+ */
+ phys = iommu_iova_to_phys_length(domain, iova, &pgsize);
+ if (WARN_ON(phys == PHYS_ADDR_MAX))
+ break;
+ phys -= page_offset;
+ if (WARN_ON(!pgsize || pgsize < PAGE_SIZE))
+ pgsize = PAGE_SIZE;
+
+ npages = (pgsize - (iova & (pgsize - 1)) + page_offset) /
+ PAGE_SIZE;
+ npages = min_t(unsigned long, npages,
+ last_index - start_index + 1);
+ if (!npages)
+ npages = 1;
+
+ for (i = 0; i < npages; i++)
+ *(out_pages++) = pfn_to_page(PHYS_PFN(phys) + i);
+
+ iova += npages * PAGE_SIZE - page_offset;
page_offset = 0;
- start_index++;
+ start_index += npages;
}
return out_pages;
}
diff --git a/drivers/iommu/iommufd/selftest.c b/drivers/iommu/iommufd/selftest.c
index af07c642a526..4b9c3ffc9523 100644
--- a/drivers/iommu/iommufd/selftest.c
+++ b/drivers/iommu/iommufd/selftest.c
@@ -1214,7 +1214,7 @@ static int iommufd_test_md_check_pa(struct iommufd_ucmd *ucmd,
pfn = page_to_pfn(pages[0]);
put_page(pages[0]);
- io_phys = mock->domain.ops->iova_to_phys(&mock->domain, iova);
+ io_phys = iommu_iova_to_phys(&mock->domain, iova);
if (io_phys !=
pfn * PAGE_SIZE + ((uintptr_t)uptr % PAGE_SIZE)) {
rc = -EINVAL;
diff --git a/drivers/vfio/vfio_iommu_type1.c b/drivers/vfio/vfio_iommu_type1.c
index c8151ba54de3..c86315b1fcda 100644
--- a/drivers/vfio/vfio_iommu_type1.c
+++ b/drivers/vfio/vfio_iommu_type1.c
@@ -1177,25 +1177,41 @@ static long vfio_unmap_unpin(struct vfio_iommu *iommu, struct vfio_dma *dma,
iommu_iotlb_gather_init(&iotlb_gather);
while (pos < dma->size) {
- size_t unmapped, len;
+ size_t unmapped, len, pgsize;
phys_addr_t phys, next;
dma_addr_t iova = dma->iova + pos;
- phys = iommu_iova_to_phys(domain->domain, iova);
- if (WARN_ON(!phys)) {
+ /* Single page table walk returns both phys and PTE size */
+ phys = iommu_iova_to_phys_length(domain->domain, iova,
+ &pgsize);
+ if (WARN_ON(phys == PHYS_ADDR_MAX)) {
pos += PAGE_SIZE;
continue;
}
+ if (WARN_ON(!pgsize || pgsize < PAGE_SIZE))
+ pgsize = PAGE_SIZE;
/*
* To optimize for fewer iommu_unmap() calls, each of which
* may require hardware cache flushing, try to find the
* largest contiguous physical memory chunk to unmap.
+ *
+ * Calculate remaining contiguous bytes within this PTE from
+ * our position, then try to join following physically
+ * contiguous PTEs.
*/
- for (len = PAGE_SIZE; pos + len < dma->size; len += PAGE_SIZE) {
- next = iommu_iova_to_phys(domain->domain, iova + len);
+ len = pgsize - (iova & (pgsize - 1));
+ for (; pos + len < dma->size; ) {
+ size_t next_pgsize;
+
+ next = iommu_iova_to_phys_length(domain->domain,
+ iova + len,
+ &next_pgsize);
if (next != phys + len)
break;
+ if (WARN_ON(!next_pgsize || next_pgsize < PAGE_SIZE))
+ next_pgsize = PAGE_SIZE;
+ len += next_pgsize;
}
/*
--
2.43.7
next prev parent reply other threads:[~2026-06-02 10:47 UTC|newest]
Thread overview: 95+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <20260529115116.GR2487554@ziepe.ca>
2026-05-31 9:36 ` [PATCH 0/9] iommu: introduce iova_to_phys_length for efficient IOVA-to-physical translation Guanghui Feng
2026-05-31 9:36 ` [PATCH 1/9] iommu: introduce iova_to_phys_length in iommu_domain_ops Guanghui Feng
2026-05-31 23:51 ` Jason Gunthorpe
2026-06-01 8:41 ` guanghuifeng
2026-06-01 13:43 ` Jason Gunthorpe
2026-06-01 14:14 ` guanghuifeng
2026-06-01 14:31 ` Jason Gunthorpe
2026-05-31 9:36 ` [PATCH 2/9] iommu/io-pgtable: introduce iova_to_phys_length in io_pgtable_ops Guanghui Feng
2026-05-31 9:36 ` [PATCH 3/9] iommu/generic_pt: implement iova_to_phys_length Guanghui Feng
2026-05-31 23:54 ` Jason Gunthorpe
2026-06-01 9:23 ` guanghuifeng
[not found] ` <fa924b86-1ca9-4819-8330-0d5f6ede8923@linux.alibaba.com>
2026-06-01 14:32 ` Jason Gunthorpe
2026-06-02 7:20 ` guanghuifeng
2026-06-02 12:32 ` Jason Gunthorpe
2026-05-31 9:36 ` [PATCH 4/9] iommu/arm-smmu: " Guanghui Feng
2026-05-31 9:36 ` [PATCH 5/9] iommu: apple-dart/ipmmu/mtk_iommu " Guanghui Feng
2026-05-31 9:36 ` [PATCH 6/9] iommu: direct page-table drivers " Guanghui Feng
2026-05-31 9:36 ` [PATCH 7/9] vfio/iommufd: use iova_to_phys_length for efficient unmap Guanghui Feng
2026-05-31 23:58 ` Jason Gunthorpe
2026-05-31 9:36 ` [PATCH 8/9] drm/gpu, iommu/io-pgtable: switch to iova_to_phys_length Guanghui Feng
2026-05-31 9:36 ` [PATCH 9/9] iommu: remove deprecated iova_to_phys from domain_ops and io_pgtable_ops Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 00/30] iommu: introduce iova_to_phys_length for efficient IOVA-to-physical translation Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 01/30] iommu: introduce iova_to_phys_length in iommu_domain_ops Guanghui Feng
2026-06-03 1:08 ` Jason Gunthorpe
2026-06-02 10:46 ` [PATCH v2 02/30] iommu/io-pgtable-arm: introduce iova_to_phys_length in io_pgtable_ops Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 03/30] iommu/io-pgtable-arm-v7s: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 04/30] iommu/io-pgtable-dart: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 05/30] iommu/generic_pt: implement iova_to_phys_length Guanghui Feng
2026-06-03 1:11 ` Jason Gunthorpe
2026-06-02 10:46 ` [PATCH v2 06/30] iommu/arm-smmu-v3: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 07/30] iommu/arm-smmu: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 08/30] iommu/qcom_iommu: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 09/30] iommu/apple-dart: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 10/30] iommu/ipmmu-vmsa: " Guanghui Feng
2026-06-03 1:13 ` Jason Gunthorpe
2026-06-02 10:46 ` [PATCH v2 11/30] iommu/mtk_iommu: " Guanghui Feng
2026-06-03 1:17 ` Jason Gunthorpe
2026-06-02 10:46 ` [PATCH v2 12/30] iommu/exynos: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 13/30] iommu/fsl_pamu: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 14/30] iommu/msm: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 15/30] iommu/mtk_v1: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 16/30] iommu/omap: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 17/30] iommu/rockchip: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 18/30] iommu/s390: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 19/30] iommu/sprd: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 20/30] iommu/sun50i: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 21/30] iommu/tegra-smmu: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 22/30] iommu/virtio: " Guanghui Feng
2026-06-02 10:46 ` Guanghui Feng [this message]
2026-06-02 10:46 ` [PATCH v2 24/30] drm/panfrost: switch to iova_to_phys_length Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 25/30] drm/panthor: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 26/30] iommu/io-pgtable: selftests " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 27/30] iommu/io-pgtable-arm: remove deprecated iova_to_phys wrapper Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 28/30] iommu/io-pgtable-arm-v7s: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 29/30] iommu/io-pgtable-dart: " Guanghui Feng
2026-06-02 10:46 ` [PATCH v2 30/30] iommu: remove iova_to_phys from domain_ops and io_pgtable_ops Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 00/32] iommu: introduce iova_to_phys_length and remove iova_to_phys Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 01/32] iommu: introduce iova_to_phys_length in iommu_domain_ops Guanghui Feng
2026-06-04 2:44 ` Baolu Lu
2026-06-04 14:16 ` Jason Gunthorpe
2026-06-03 15:17 ` [PATCH v3 02/32] iommu/io-pgtable-arm: introduce iova_to_phys_length in io_pgtable_ops Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 03/32] iommu/io-pgtable-arm-v7s: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 04/32] iommu/io-pgtable-dart: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 05/32] iommu/generic_pt: implement iova_to_phys_length Guanghui Feng
2026-06-04 3:30 ` Baolu Lu
2026-06-04 14:12 ` Jason Gunthorpe
2026-06-03 15:17 ` [PATCH v3 06/32] iommu/arm-smmu-v3: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 07/32] iommu/arm-smmu: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 08/32] iommu/qcom_iommu: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 09/32] iommu/apple-dart: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 10/32] iommu/ipmmu-vmsa: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 11/32] iommu/mtk_iommu: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 12/32] iommu/exynos: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 13/32] iommu/fsl_pamu: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 14/32] iommu/msm: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 15/32] iommu/mtk_v1: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 16/32] iommu/omap: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 17/32] iommu/rockchip: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 18/32] iommu/s390: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 19/32] iommu/sprd: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 20/32] iommu/sun50i: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 21/32] iommu/tegra-smmu: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 22/32] iommu/virtio: " Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 23/32] vfio: use iova_to_phys_length for efficient unmap Guanghui Feng
2026-06-04 14:27 ` Jason Gunthorpe
2026-06-03 15:17 ` [PATCH v3 24/32] iommufd: " Guanghui Feng
2026-06-04 14:26 ` Jason Gunthorpe
2026-06-03 15:17 ` [PATCH v3 25/32] iommufd/selftest: switch to iommu_iova_to_phys_length Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 26/32] drm/panfrost: switch to iova_to_phys_length Guanghui Feng
2026-06-03 15:17 ` [PATCH v3 27/32] drm/panthor: " Guanghui Feng
2026-06-03 15:18 ` [PATCH v3 28/32] iommu/io-pgtable: selftests " Guanghui Feng
2026-06-03 15:18 ` [PATCH v3 29/32] iommu/io-pgtable-arm: remove deprecated iova_to_phys wrapper Guanghui Feng
2026-06-03 15:18 ` [PATCH v3 30/32] iommu/io-pgtable-arm-v7s: " Guanghui Feng
2026-06-03 15:18 ` [PATCH v3 31/32] iommu/io-pgtable-dart: " Guanghui Feng
2026-06-03 15:18 ` [PATCH v3 32/32] iommu: remove iova_to_phys from domain_ops and io_pgtable_ops Guanghui Feng
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=20260602104637.1219810-24-guanghuifeng@linux.alibaba.com \
--to=guanghuifeng@linux$(echo .)alibaba.com \
--cc=adrian.larumbe@collabora$(echo .)com \
--cc=airlied@gmail$(echo .)com \
--cc=alex@shazbot$(echo .)org \
--cc=alikernel-developer@linux$(echo .)alibaba.com \
--cc=baolu.lu@linux$(echo .)intel.com \
--cc=boris.brezillon@collabora$(echo .)com \
--cc=dri-devel@lists$(echo .)freedesktop.org \
--cc=dwmw2@infradead$(echo .)org \
--cc=iommu@lists$(echo .)linux.dev \
--cc=jgg@ziepe$(echo .)ca \
--cc=joro@8bytes$(echo .)org \
--cc=kevin.tian@intel$(echo .)com \
--cc=kvm@vger$(echo .)kernel.org \
--cc=linux-arm-kernel@lists$(echo .)infradead.org \
--cc=linux-kernel@vger$(echo .)kernel.org \
--cc=liviu.dudau@arm$(echo .)com \
--cc=maarten.lankhorst@linux$(echo .)intel.com \
--cc=mripard@kernel$(echo .)org \
--cc=oliver.yang@linux$(echo .)alibaba.com \
--cc=robh@kernel$(echo .)org \
--cc=robin.murphy@arm$(echo .)com \
--cc=shiyu.zsq@linux$(echo .)alibaba.com \
--cc=steven.price@arm$(echo .)com \
--cc=suravee.suthikulpanit@amd$(echo .)com \
--cc=tzimmermann@suse$(echo .)de \
--cc=wei.guo.simon@linux$(echo .)alibaba.com \
--cc=will@kernel$(echo .)org \
--cc=xlpang@linux$(echo .)alibaba.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
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