Documentation/core-api/dma-attributes.rst | 38 ++++++++++++++++++++++++------- drivers/infiniband/core/umem.c | 5 ++-- drivers/iommu/dma-iommu.c | 21 +++++++++++++---- drivers/virtio/virtio_ring.c | 10 ++++---- include/linux/dma-mapping.h | 15 ++++++++---- include/trace/events/dma.h | 4 +++- kernel/dma/debug.c | 9 ++++---- kernel/dma/direct.h | 7 +++--- kernel/dma/mapping.c | 6 +++++ mm/hmm.c | 4 ++-- 10 files changed, 86 insertions(+), 33 deletions(-)
Add a new DMA_ATTR_REQUIRE_COHERENT attribute to the DMA API to mark
mappings that must run on a DMA‑coherent system. Such buffers cannot
use the SWIOTLB path, may overlap with CPU caches, and do not depend on
explicit cache flushing.
Mappings using this attribute are rejected on systems where cache
side‑effects could lead to data corruption, and therefore do not need
the cache‑overlap debugging logic. This series also includes fixes for
DMA_ATTR_CPU_CACHE_CLEAN handling.
Thanks.
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
Changes in v3:
- Enriched commit messages and documentation
- Added ROB tags
- Link to v2: https://patch.msgid.link/20260311-dma-debug-overlap-v2-0-e00bc2ca346d@nvidia.com
Changes in v2:
- Added DMA_ATTR_REQUIRE_COHERENT attribute
- Added HMM patch which needs this attribute as well
- Renamed DMA_ATTR_CPU_CACHE_CLEAN to be DMA_ATTR_DEBUGGING_IGNORE_CACHELINES
- Link to v1: https://patch.msgid.link/20260307-dma-debug-overlap-v1-0-c034c38872af@nvidia.com
---
Leon Romanovsky (8):
dma-debug: Allow multiple invocations of overlapping entries
dma-mapping: handle DMA_ATTR_CPU_CACHE_CLEAN in trace output
dma-mapping: Clarify valid conditions for CPU cache line overlap
dma-mapping: Introduce DMA require coherency attribute
dma-direct: prevent SWIOTLB path when DMA_ATTR_REQUIRE_COHERENT is set
iommu/dma: add support for DMA_ATTR_REQUIRE_COHERENT attribute
RDMA/umem: Tell DMA mapping that UMEM requires coherency
mm/hmm: Indicate that HMM requires DMA coherency
Documentation/core-api/dma-attributes.rst | 38 ++++++++++++++++++++++++-------
drivers/infiniband/core/umem.c | 5 ++--
drivers/iommu/dma-iommu.c | 21 +++++++++++++----
drivers/virtio/virtio_ring.c | 10 ++++----
include/linux/dma-mapping.h | 15 ++++++++----
include/trace/events/dma.h | 4 +++-
kernel/dma/debug.c | 9 ++++----
kernel/dma/direct.h | 7 +++---
kernel/dma/mapping.c | 6 +++++
mm/hmm.c | 4 ++--
10 files changed, 86 insertions(+), 33 deletions(-)
---
base-commit: 11439c4635edd669ae435eec308f4ab8a0804808
change-id: 20260305-dma-debug-overlap-21487c3fa02c
Best regards,
--
Leon Romanovsky <leonro@nvidia.com>
On Mon, Mar 16, 2026 at 09:06:44PM +0200, Leon Romanovsky wrote: > Add a new DMA_ATTR_REQUIRE_COHERENT attribute to the DMA API to mark > mappings that must run on a DMA‑coherent system. Such buffers cannot > use the SWIOTLB path, may overlap with CPU caches, and do not depend on > explicit cache flushing. > > Mappings using this attribute are rejected on systems where cache > side‑effects could lead to data corruption, and therefore do not need > the cache‑overlap debugging logic. This series also includes fixes for > DMA_ATTR_CPU_CACHE_CLEAN handling. > Thanks. <...> > --- > Leon Romanovsky (8): > dma-debug: Allow multiple invocations of overlapping entries > dma-mapping: handle DMA_ATTR_CPU_CACHE_CLEAN in trace output > dma-mapping: Clarify valid conditions for CPU cache line overlap > dma-mapping: Introduce DMA require coherency attribute > dma-direct: prevent SWIOTLB path when DMA_ATTR_REQUIRE_COHERENT is set > iommu/dma: add support for DMA_ATTR_REQUIRE_COHERENT attribute > RDMA/umem: Tell DMA mapping that UMEM requires coherency > mm/hmm: Indicate that HMM requires DMA coherency > > Documentation/core-api/dma-attributes.rst | 38 ++++++++++++++++++++++++------- > drivers/infiniband/core/umem.c | 5 ++-- > drivers/iommu/dma-iommu.c | 21 +++++++++++++---- > drivers/virtio/virtio_ring.c | 10 ++++---- > include/linux/dma-mapping.h | 15 ++++++++---- > include/trace/events/dma.h | 4 +++- > kernel/dma/debug.c | 9 ++++---- > kernel/dma/direct.h | 7 +++--- > kernel/dma/mapping.c | 6 +++++ > mm/hmm.c | 4 ++-- > 10 files changed, 86 insertions(+), 33 deletions(-) Marek, Despite the "RDMA ..." tag in the subject, the diffstat clearly shows that you are the appropriate person to take this patch. Thanks. > --- > base-commit: 11439c4635edd669ae435eec308f4ab8a0804808 > change-id: 20260305-dma-debug-overlap-21487c3fa02c > > Best regards, > -- > Leon Romanovsky <leonro@nvidia.com> >
Hi Leon, On 17.03.2026 20:05, Leon Romanovsky wrote: > On Mon, Mar 16, 2026 at 09:06:44PM +0200, Leon Romanovsky wrote: >> Add a new DMA_ATTR_REQUIRE_COHERENT attribute to the DMA API to mark >> mappings that must run on a DMA‑coherent system. Such buffers cannot >> use the SWIOTLB path, may overlap with CPU caches, and do not depend on >> explicit cache flushing. >> >> Mappings using this attribute are rejected on systems where cache >> side‑effects could lead to data corruption, and therefore do not need >> the cache‑overlap debugging logic. This series also includes fixes for >> DMA_ATTR_CPU_CACHE_CLEAN handling. >> Thanks. > <...> > >> --- >> Leon Romanovsky (8): >> dma-debug: Allow multiple invocations of overlapping entries >> dma-mapping: handle DMA_ATTR_CPU_CACHE_CLEAN in trace output >> dma-mapping: Clarify valid conditions for CPU cache line overlap >> dma-mapping: Introduce DMA require coherency attribute >> dma-direct: prevent SWIOTLB path when DMA_ATTR_REQUIRE_COHERENT is set >> iommu/dma: add support for DMA_ATTR_REQUIRE_COHERENT attribute >> RDMA/umem: Tell DMA mapping that UMEM requires coherency >> mm/hmm: Indicate that HMM requires DMA coherency >> >> Documentation/core-api/dma-attributes.rst | 38 ++++++++++++++++++++++++------- >> drivers/infiniband/core/umem.c | 5 ++-- >> drivers/iommu/dma-iommu.c | 21 +++++++++++++---- >> drivers/virtio/virtio_ring.c | 10 ++++---- >> include/linux/dma-mapping.h | 15 ++++++++---- >> include/trace/events/dma.h | 4 +++- >> kernel/dma/debug.c | 9 ++++---- >> kernel/dma/direct.h | 7 +++--- >> kernel/dma/mapping.c | 6 +++++ >> mm/hmm.c | 4 ++-- >> 10 files changed, 86 insertions(+), 33 deletions(-) > Marek, > > Despite the "RDMA ..." tag in the subject, the diffstat clearly shows that > you are the appropriate person to take this patch. I plan to take the first 2 patches to the dma-mapping-fixes branch (v7.0-rc) and the next to dma-mapping-for-next. Should I also take the RDMA and HMM patches, or do You want a stable branch for merging them via respective subsystem trees? Best regards -- Marek Szyprowski, PhD Samsung R&D Institute Poland
On Wed, Mar 18, 2026 at 09:03:00AM +0100, Marek Szyprowski wrote: > Hi Leon, > > On 17.03.2026 20:05, Leon Romanovsky wrote: > > On Mon, Mar 16, 2026 at 09:06:44PM +0200, Leon Romanovsky wrote: > >> Add a new DMA_ATTR_REQUIRE_COHERENT attribute to the DMA API to mark > >> mappings that must run on a DMA‑coherent system. Such buffers cannot > >> use the SWIOTLB path, may overlap with CPU caches, and do not depend on > >> explicit cache flushing. > >> > >> Mappings using this attribute are rejected on systems where cache > >> side‑effects could lead to data corruption, and therefore do not need > >> the cache‑overlap debugging logic. This series also includes fixes for > >> DMA_ATTR_CPU_CACHE_CLEAN handling. > >> Thanks. > > <...> > > > >> --- > >> Leon Romanovsky (8): > >> dma-debug: Allow multiple invocations of overlapping entries > >> dma-mapping: handle DMA_ATTR_CPU_CACHE_CLEAN in trace output > >> dma-mapping: Clarify valid conditions for CPU cache line overlap > >> dma-mapping: Introduce DMA require coherency attribute > >> dma-direct: prevent SWIOTLB path when DMA_ATTR_REQUIRE_COHERENT is set > >> iommu/dma: add support for DMA_ATTR_REQUIRE_COHERENT attribute > >> RDMA/umem: Tell DMA mapping that UMEM requires coherency > >> mm/hmm: Indicate that HMM requires DMA coherency > >> > >> Documentation/core-api/dma-attributes.rst | 38 ++++++++++++++++++++++++------- > >> drivers/infiniband/core/umem.c | 5 ++-- > >> drivers/iommu/dma-iommu.c | 21 +++++++++++++---- > >> drivers/virtio/virtio_ring.c | 10 ++++---- > >> include/linux/dma-mapping.h | 15 ++++++++---- > >> include/trace/events/dma.h | 4 +++- > >> kernel/dma/debug.c | 9 ++++---- > >> kernel/dma/direct.h | 7 +++--- > >> kernel/dma/mapping.c | 6 +++++ > >> mm/hmm.c | 4 ++-- > >> 10 files changed, 86 insertions(+), 33 deletions(-) > > Marek, > > > > Despite the "RDMA ..." tag in the subject, the diffstat clearly shows that > > you are the appropriate person to take this patch. > > I plan to take the first 2 patches to the dma-mapping-fixes branch > (v7.0-rc) and the next to dma-mapping-for-next. Should I also take the > RDMA and HMM patches, or do You want a stable branch for merging them > via respective subsystem trees? I suggest taking all patches into the -fixes branch, as the "RDMA/..." patch also resolves the dmesg splat. With -fixes, there is no need to worry about a shared branch since we do not expect merge conflicts in that area. If you still prefer to split the series between -fixes and -next, it would be better to use a shared branch in that case. There are patches on the RDMA list targeted for -next that touch ib_umem_get(). Thanks > > Best regards > -- > Marek Szyprowski, PhD > Samsung R&D Institute Poland > >
Hi Leon, On 18.03.2026 09:18, Leon Romanovsky wrote: > On Wed, Mar 18, 2026 at 09:03:00AM +0100, Marek Szyprowski wrote: >> On 17.03.2026 20:05, Leon Romanovsky wrote: >>> On Mon, Mar 16, 2026 at 09:06:44PM +0200, Leon Romanovsky wrote: >>>> Add a new DMA_ATTR_REQUIRE_COHERENT attribute to the DMA API to mark >>>> mappings that must run on a DMA‑coherent system. Such buffers cannot >>>> use the SWIOTLB path, may overlap with CPU caches, and do not depend on >>>> explicit cache flushing. >>>> >>>> Mappings using this attribute are rejected on systems where cache >>>> side‑effects could lead to data corruption, and therefore do not need >>>> the cache‑overlap debugging logic. This series also includes fixes for >>>> DMA_ATTR_CPU_CACHE_CLEAN handling. >>>> Thanks. >>> <...> >>> >>>> --- >>>> Leon Romanovsky (8): >>>> dma-debug: Allow multiple invocations of overlapping entries >>>> dma-mapping: handle DMA_ATTR_CPU_CACHE_CLEAN in trace output >>>> dma-mapping: Clarify valid conditions for CPU cache line overlap >>>> dma-mapping: Introduce DMA require coherency attribute >>>> dma-direct: prevent SWIOTLB path when DMA_ATTR_REQUIRE_COHERENT is set >>>> iommu/dma: add support for DMA_ATTR_REQUIRE_COHERENT attribute >>>> RDMA/umem: Tell DMA mapping that UMEM requires coherency >>>> mm/hmm: Indicate that HMM requires DMA coherency >>>> >>>> Documentation/core-api/dma-attributes.rst | 38 ++++++++++++++++++++++++------- >>>> drivers/infiniband/core/umem.c | 5 ++-- >>>> drivers/iommu/dma-iommu.c | 21 +++++++++++++---- >>>> drivers/virtio/virtio_ring.c | 10 ++++---- >>>> include/linux/dma-mapping.h | 15 ++++++++---- >>>> include/trace/events/dma.h | 4 +++- >>>> kernel/dma/debug.c | 9 ++++---- >>>> kernel/dma/direct.h | 7 +++--- >>>> kernel/dma/mapping.c | 6 +++++ >>>> mm/hmm.c | 4 ++-- >>>> 10 files changed, 86 insertions(+), 33 deletions(-) >>> Marek, >>> >>> Despite the "RDMA ..." tag in the subject, the diffstat clearly shows that >>> you are the appropriate person to take this patch. >> I plan to take the first 2 patches to the dma-mapping-fixes branch >> (v7.0-rc) and the next to dma-mapping-for-next. Should I also take the >> RDMA and HMM patches, or do You want a stable branch for merging them >> via respective subsystem trees? > I suggest taking all patches into the -fixes branch, as the "RDMA/..." patch > also resolves the dmesg splat. With -fixes, there is no need to worry about > a shared branch since we do not expect merge conflicts in that area. > > If you still prefer to split the series between -fixes and -next, it would be > better to use a shared branch in that case. There are patches on the RDMA > list targeted for -next that touch ib_umem_get(). Okay, I will merge all patches to the -fixes branch then. Best regards -- Marek Szyprowski, PhD Samsung R&D Institute Poland
On 16.03.2026 20:06, Leon Romanovsky wrote: > Add a new DMA_ATTR_REQUIRE_COHERENT attribute to the DMA API to mark > mappings that must run on a DMA‑coherent system. Such buffers cannot > use the SWIOTLB path, may overlap with CPU caches, and do not depend on > explicit cache flushing. > > Mappings using this attribute are rejected on systems where cache > side‑effects could lead to data corruption, and therefore do not need > the cache‑overlap debugging logic. This series also includes fixes for > DMA_ATTR_CPU_CACHE_CLEAN handling. > Thanks. > > Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Applied to dma-mapping-fixes. Thanks! > --- > Changes in v3: > - Enriched commit messages and documentation > - Added ROB tags > - Link to v2: https://protect2.fireeye.com/v1/url?k=9c1ba148-fd90b40f-9c1a2a07-000babff99aa-86ebd022a97425b3&q=1&e=3c8e10cc-4c34-4bf6-aa9d-c339877d6a27&u=https%3A%2F%2Fpatch.msgid.link%2F20260311-dma-debug-overlap-v2-0-e00bc2ca346d%40nvidia.com > > Changes in v2: > - Added DMA_ATTR_REQUIRE_COHERENT attribute > - Added HMM patch which needs this attribute as well > - Renamed DMA_ATTR_CPU_CACHE_CLEAN to be DMA_ATTR_DEBUGGING_IGNORE_CACHELINES > - Link to v1: https://protect2.fireeye.com/v1/url?k=cc0590de-ad8e8599-cc041b91-000babff99aa-07e4da206b7e0d97&q=1&e=3c8e10cc-4c34-4bf6-aa9d-c339877d6a27&u=https%3A%2F%2Fpatch.msgid.link%2F20260307-dma-debug-overlap-v1-0-c034c38872af%40nvidia.com > > --- > Leon Romanovsky (8): > dma-debug: Allow multiple invocations of overlapping entries > dma-mapping: handle DMA_ATTR_CPU_CACHE_CLEAN in trace output > dma-mapping: Clarify valid conditions for CPU cache line overlap > dma-mapping: Introduce DMA require coherency attribute > dma-direct: prevent SWIOTLB path when DMA_ATTR_REQUIRE_COHERENT is set > iommu/dma: add support for DMA_ATTR_REQUIRE_COHERENT attribute > RDMA/umem: Tell DMA mapping that UMEM requires coherency > mm/hmm: Indicate that HMM requires DMA coherency > > Documentation/core-api/dma-attributes.rst | 38 ++++++++++++++++++++++++------- > drivers/infiniband/core/umem.c | 5 ++-- > drivers/iommu/dma-iommu.c | 21 +++++++++++++---- > drivers/virtio/virtio_ring.c | 10 ++++---- > include/linux/dma-mapping.h | 15 ++++++++---- > include/trace/events/dma.h | 4 +++- > kernel/dma/debug.c | 9 ++++---- > kernel/dma/direct.h | 7 +++--- > kernel/dma/mapping.c | 6 +++++ > mm/hmm.c | 4 ++-- > 10 files changed, 86 insertions(+), 33 deletions(-) > --- > base-commit: 11439c4635edd669ae435eec308f4ab8a0804808 > change-id: 20260305-dma-debug-overlap-21487c3fa02c > > Best regards, > -- > Leon Romanovsky <leonro@nvidia.com> > > Best regards -- Marek Szyprowski, PhD Samsung R&D Institute Poland
© 2016 - 2026 Red Hat, Inc.