When CONFIG_DMA_API_DEBUG_SG is enabled, importing a udmabuf into a DRM driver (e.g. amdgpu for video playback in GNOME Videos / Showtime) triggers a spurious warning:
DMA-API: amdgpu 0000:03:00.0: cacheline tracking EEXIST, \ overlapping mappings aren't supported WARNING: kernel/dma/debug.c:619 at add_dma_entry+0x473/0x5f0
The call chain is:
amdgpu_cs_ioctl -> amdgpu_ttm_backend_bind -> dma_buf_map_attachment -> [udmabuf] map_udmabuf -> get_sg_table -> dma_map_sgtable(dev, sg, direction, 0) // attrs=0 -> debug_dma_map_sg -> add_dma_entry -> EEXIST
This happens because udmabuf builds a per-page scatter-gather list via sg_set_folio(). When begin_cpu_udmabuf() has already created an sg table mapped for the misc device, and an importer such as amdgpu maps the same pages for its own device via map_udmabuf(), the DMA debug infrastructure sees two active mappings whose physical addresses share cacheline boundaries and warns about the overlap.
The DMA_ATTR_SKIP_CPU_SYNC flag suppresses this check in add_dma_entry() because it signals that no CPU cache maintenance is performed at map/unmap time, making the cacheline overlap harmless.
All other major dma-buf exporters already pass this flag: - drm_gem_map_dma_buf() passes DMA_ATTR_SKIP_CPU_SYNC - amdgpu_dma_buf_map() passes DMA_ATTR_SKIP_CPU_SYNC
The CPU sync at map/unmap time is also redundant for udmabuf: begin_cpu_udmabuf() and end_cpu_udmabuf() already perform explicit cache synchronization via dma_sync_sgtable_for_cpu/device() when CPU access is requested through the dma-buf interface.
Pass DMA_ATTR_SKIP_CPU_SYNC to dma_map_sgtable() and dma_unmap_sgtable() in udmabuf to suppress the spurious warning and skip the redundant sync.
Fixes: 284562e1f348 ("udmabuf: implement begin_cpu_access/end_cpu_access hooks") Cc: stable@vger.kernel.org Signed-off-by: Mikhail Gavrilov mikhail.v.gavrilov@gmail.com --- drivers/dma-buf/udmabuf.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/dma-buf/udmabuf.c b/drivers/dma-buf/udmabuf.c index 94b8ecb892bb..9c6f8785a28a 100644 --- a/drivers/dma-buf/udmabuf.c +++ b/drivers/dma-buf/udmabuf.c @@ -162,7 +162,7 @@ static struct sg_table *get_sg_table(struct device *dev, struct dma_buf *buf, sg_set_folio(sgl, ubuf->folios[i], PAGE_SIZE, ubuf->offsets[i]);
- ret = dma_map_sgtable(dev, sg, direction, 0); + ret = dma_map_sgtable(dev, sg, direction, DMA_ATTR_SKIP_CPU_SYNC); if (ret < 0) goto err_map; return sg; @@ -177,7 +177,7 @@ static struct sg_table *get_sg_table(struct device *dev, struct dma_buf *buf, static void put_sg_table(struct device *dev, struct sg_table *sg, enum dma_data_direction direction) { - dma_unmap_sgtable(dev, sg, direction, 0); + dma_unmap_sgtable(dev, sg, direction, DMA_ATTR_SKIP_CPU_SYNC); sg_free_table(sg); kfree(sg); }
Hi Mikhail,
Subject: [PATCH] dma-buf/udmabuf: skip redundant cpu sync to fix cacheline EEXIST warning
When CONFIG_DMA_API_DEBUG_SG is enabled, importing a udmabuf into a DRM driver (e.g. amdgpu for video playback in GNOME Videos / Showtime) triggers a spurious warning:
DMA-API: amdgpu 0000:03:00.0: cacheline tracking EEXIST, \ overlapping mappings aren't supported WARNING: kernel/dma/debug.c:619 at add_dma_entry+0x473/0x5f0
The call chain is:
amdgpu_cs_ioctl -> amdgpu_ttm_backend_bind -> dma_buf_map_attachment -> [udmabuf] map_udmabuf -> get_sg_table -> dma_map_sgtable(dev, sg, direction, 0) // attrs=0 -> debug_dma_map_sg -> add_dma_entry -> EEXIST
This happens because udmabuf builds a per-page scatter-gather list via sg_set_folio(). When begin_cpu_udmabuf() has already created an sg table mapped for the misc device, and an importer such as amdgpu maps the same pages for its own device via map_udmabuf(), the DMA debug infrastructure sees two active mappings whose physical addresses share cacheline boundaries and warns about the overlap.
The DMA_ATTR_SKIP_CPU_SYNC flag suppresses this check in add_dma_entry() because it signals that no CPU cache maintenance is performed at map/unmap time, making the cacheline overlap harmless.
All other major dma-buf exporters already pass this flag:
- drm_gem_map_dma_buf() passes DMA_ATTR_SKIP_CPU_SYNC
- amdgpu_dma_buf_map() passes DMA_ATTR_SKIP_CPU_SYNC
The CPU sync at map/unmap time is also redundant for udmabuf: begin_cpu_udmabuf() and end_cpu_udmabuf() already perform explicit cache synchronization via dma_sync_sgtable_for_cpu/device() when CPU access is requested through the dma-buf interface.
Pass DMA_ATTR_SKIP_CPU_SYNC to dma_map_sgtable() and dma_unmap_sgtable() in udmabuf to suppress the spurious warning and skip the redundant sync.
Fixes: 284562e1f348 ("udmabuf: implement begin_cpu_access/end_cpu_access hooks") Cc: stable@vger.kernel.org Signed-off-by: Mikhail Gavrilov mikhail.v.gavrilov@gmail.com
drivers/dma-buf/udmabuf.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/dma-buf/udmabuf.c b/drivers/dma-buf/udmabuf.c index 94b8ecb892bb..9c6f8785a28a 100644 --- a/drivers/dma-buf/udmabuf.c +++ b/drivers/dma-buf/udmabuf.c @@ -162,7 +162,7 @@ static struct sg_table *get_sg_table(struct device *dev, struct dma_buf *buf, sg_set_folio(sgl, ubuf->folios[i], PAGE_SIZE, ubuf->offsets[i]);
- ret = dma_map_sgtable(dev, sg, direction, 0);
- ret = dma_map_sgtable(dev, sg, direction,
DMA_ATTR_SKIP_CPU_SYNC); if (ret < 0) goto err_map; return sg; @@ -177,7 +177,7 @@ static struct sg_table *get_sg_table(struct device *dev, struct dma_buf *buf, static void put_sg_table(struct device *dev, struct sg_table *sg, enum dma_data_direction direction) {
- dma_unmap_sgtable(dev, sg, direction, 0);
- dma_unmap_sgtable(dev, sg, direction,
DMA_ATTR_SKIP_CPU_SYNC);
Looks OK to me but it would be nice if Christian or someone else can provide an Ack for this patch.
Thanks, Vivek
sg_free_table(sg); kfree(sg); } -- 2.53.0
On 3/18/26 06:40, Kasireddy, Vivek wrote:
Hi Mikhail,
Subject: [PATCH] dma-buf/udmabuf: skip redundant cpu sync to fix cacheline EEXIST warning
When CONFIG_DMA_API_DEBUG_SG is enabled, importing a udmabuf into a DRM driver (e.g. amdgpu for video playback in GNOME Videos / Showtime) triggers a spurious warning:
DMA-API: amdgpu 0000:03:00.0: cacheline tracking EEXIST, \ overlapping mappings aren't supported WARNING: kernel/dma/debug.c:619 at add_dma_entry+0x473/0x5f0
The call chain is:
amdgpu_cs_ioctl -> amdgpu_ttm_backend_bind -> dma_buf_map_attachment -> [udmabuf] map_udmabuf -> get_sg_table -> dma_map_sgtable(dev, sg, direction, 0) // attrs=0 -> debug_dma_map_sg -> add_dma_entry -> EEXIST
This happens because udmabuf builds a per-page scatter-gather list via sg_set_folio(). When begin_cpu_udmabuf() has already created an sg table mapped for the misc device, and an importer such as amdgpu maps the same pages for its own device via map_udmabuf(), the DMA debug infrastructure sees two active mappings whose physical addresses share cacheline boundaries and warns about the overlap.
The DMA_ATTR_SKIP_CPU_SYNC flag suppresses this check in add_dma_entry() because it signals that no CPU cache maintenance is performed at map/unmap time, making the cacheline overlap harmless.
All other major dma-buf exporters already pass this flag:
- drm_gem_map_dma_buf() passes DMA_ATTR_SKIP_CPU_SYNC
- amdgpu_dma_buf_map() passes DMA_ATTR_SKIP_CPU_SYNC
The CPU sync at map/unmap time is also redundant for udmabuf: begin_cpu_udmabuf() and end_cpu_udmabuf() already perform explicit cache synchronization via dma_sync_sgtable_for_cpu/device() when CPU access is requested through the dma-buf interface.
Pass DMA_ATTR_SKIP_CPU_SYNC to dma_map_sgtable() and dma_unmap_sgtable() in udmabuf to suppress the spurious warning and skip the redundant sync.
Fixes: 284562e1f348 ("udmabuf: implement begin_cpu_access/end_cpu_access hooks") Cc: stable@vger.kernel.org Signed-off-by: Mikhail Gavrilov mikhail.v.gavrilov@gmail.com
drivers/dma-buf/udmabuf.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/dma-buf/udmabuf.c b/drivers/dma-buf/udmabuf.c index 94b8ecb892bb..9c6f8785a28a 100644 --- a/drivers/dma-buf/udmabuf.c +++ b/drivers/dma-buf/udmabuf.c @@ -162,7 +162,7 @@ static struct sg_table *get_sg_table(struct device *dev, struct dma_buf *buf, sg_set_folio(sgl, ubuf->folios[i], PAGE_SIZE, ubuf->offsets[i]);
- ret = dma_map_sgtable(dev, sg, direction, 0);
- ret = dma_map_sgtable(dev, sg, direction,
DMA_ATTR_SKIP_CPU_SYNC); if (ret < 0) goto err_map; return sg; @@ -177,7 +177,7 @@ static struct sg_table *get_sg_table(struct device *dev, struct dma_buf *buf, static void put_sg_table(struct device *dev, struct sg_table *sg, enum dma_data_direction direction) {
- dma_unmap_sgtable(dev, sg, direction, 0);
- dma_unmap_sgtable(dev, sg, direction,
DMA_ATTR_SKIP_CPU_SYNC);
Looks OK to me but it would be nice if Christian or someone else can provide an Ack for this patch.
The details of the udmabuf handling is absolutely not my field of expertise.
Feel free to add my Acked-by since it obviously seems to fix a bug, but it would be nice if somebody could do an in deep review as well.
Regards, Christian.
Thanks, Vivek
sg_free_table(sg); kfree(sg); } -- 2.53.0
On Tue, Mar 17, 2026 at 10:37 AM Mikhail Gavrilov mikhail.v.gavrilov@gmail.com wrote:
When CONFIG_DMA_API_DEBUG_SG is enabled, importing a udmabuf into a DRM driver (e.g. amdgpu for video playback in GNOME Videos / Showtime) triggers a spurious warning:
DMA-API: amdgpu 0000:03:00.0: cacheline tracking EEXIST, \ overlapping mappings aren't supported WARNING: kernel/dma/debug.c:619 at add_dma_entry+0x473/0x5f0
The call chain is:
amdgpu_cs_ioctl -> amdgpu_ttm_backend_bind -> dma_buf_map_attachment -> [udmabuf] map_udmabuf -> get_sg_table -> dma_map_sgtable(dev, sg, direction, 0) // attrs=0 -> debug_dma_map_sg -> add_dma_entry -> EEXIST
This happens because udmabuf builds a per-page scatter-gather list via sg_set_folio(). When begin_cpu_udmabuf() has already created an sg table mapped for the misc device, and an importer such as amdgpu maps the same pages for its own device via map_udmabuf(), the DMA debug infrastructure sees two active mappings whose physical addresses share cacheline boundaries and warns about the overlap.
The DMA_ATTR_SKIP_CPU_SYNC flag suppresses this check in add_dma_entry() because it signals that no CPU cache maintenance is performed at map/unmap time, making the cacheline overlap harmless.
All other major dma-buf exporters already pass this flag:
- drm_gem_map_dma_buf() passes DMA_ATTR_SKIP_CPU_SYNC
- amdgpu_dma_buf_map() passes DMA_ATTR_SKIP_CPU_SYNC
The CPU sync at map/unmap time is also redundant for udmabuf: begin_cpu_udmabuf() and end_cpu_udmabuf() already perform explicit cache synchronization via dma_sync_sgtable_for_cpu/device() when CPU access is requested through the dma-buf interface.
Pass DMA_ATTR_SKIP_CPU_SYNC to dma_map_sgtable() and dma_unmap_sgtable() in udmabuf to suppress the spurious warning and skip the redundant sync.
Fixes: 284562e1f348 ("udmabuf: implement begin_cpu_access/end_cpu_access hooks") Cc: stable@vger.kernel.org Signed-off-by: Mikhail Gavrilov mikhail.v.gavrilov@gmail.com
drivers/dma-buf/udmabuf.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/dma-buf/udmabuf.c b/drivers/dma-buf/udmabuf.c index 94b8ecb892bb..9c6f8785a28a 100644 --- a/drivers/dma-buf/udmabuf.c +++ b/drivers/dma-buf/udmabuf.c @@ -162,7 +162,7 @@ static struct sg_table *get_sg_table(struct device *dev, struct dma_buf *buf, sg_set_folio(sgl, ubuf->folios[i], PAGE_SIZE, ubuf->offsets[i]);
ret = dma_map_sgtable(dev, sg, direction, 0);
ret = dma_map_sgtable(dev, sg, direction, DMA_ATTR_SKIP_CPU_SYNC); if (ret < 0) goto err_map; return sg;@@ -177,7 +177,7 @@ static struct sg_table *get_sg_table(struct device *dev, struct dma_buf *buf, static void put_sg_table(struct device *dev, struct sg_table *sg, enum dma_data_direction direction) {
dma_unmap_sgtable(dev, sg, direction, 0);
dma_unmap_sgtable(dev, sg, direction, DMA_ATTR_SKIP_CPU_SYNC); sg_free_table(sg); kfree(sg);}
2.53.0
Gentle ping on this patch.
To summarize the review so far: Vivek: "Looks OK to me" Christian: Acked-by
Note: while my separate dma-debug patch [1] would also suppress this specific warning on x86, the udmabuf fix here is still valuable as it removes the redundant CPU sync at map/unmap time and aligns udmabuf with other dma-buf exporters that already pass DMA_ATTR_SKIP_CPU_SYNC.
[1] https://lore.kernel.org/all/20260327124156.24820-1-mikhail.v.gavrilov@gmail....
Gerd, could you take a look when you have a chance?
Hi Mikhail,
Subject: Re: [PATCH] dma-buf/udmabuf: skip redundant cpu sync to fix cacheline EEXIST warning
On Tue, Mar 17, 2026 at 10:37 AM Mikhail Gavrilov mikhail.v.gavrilov@gmail.com wrote:
When CONFIG_DMA_API_DEBUG_SG is enabled, importing a udmabuf
into a DRM
driver (e.g. amdgpu for video playback in GNOME Videos / Showtime) triggers a spurious warning:
DMA-API: amdgpu 0000:03:00.0: cacheline tracking EEXIST, \ overlapping mappings aren't supported WARNING: kernel/dma/debug.c:619 at add_dma_entry+0x473/0x5f0
The call chain is:
amdgpu_cs_ioctl -> amdgpu_ttm_backend_bind -> dma_buf_map_attachment -> [udmabuf] map_udmabuf -> get_sg_table -> dma_map_sgtable(dev, sg, direction, 0) // attrs=0 -> debug_dma_map_sg -> add_dma_entry -> EEXIST
This happens because udmabuf builds a per-page scatter-gather list
via
sg_set_folio(). When begin_cpu_udmabuf() has already created an sg table mapped for the misc device, and an importer such as amdgpu
maps
the same pages for its own device via map_udmabuf(), the DMA
debug
infrastructure sees two active mappings whose physical addresses
share
cacheline boundaries and warns about the overlap.
The DMA_ATTR_SKIP_CPU_SYNC flag suppresses this check in add_dma_entry() because it signals that no CPU cache maintenance is performed at map/unmap time, making the cacheline overlap
harmless.
All other major dma-buf exporters already pass this flag:
- drm_gem_map_dma_buf() passes DMA_ATTR_SKIP_CPU_SYNC
- amdgpu_dma_buf_map() passes DMA_ATTR_SKIP_CPU_SYNC
The CPU sync at map/unmap time is also redundant for udmabuf: begin_cpu_udmabuf() and end_cpu_udmabuf() already perform
explicit
cache synchronization via dma_sync_sgtable_for_cpu/device() when
CPU
access is requested through the dma-buf interface.
Pass DMA_ATTR_SKIP_CPU_SYNC to dma_map_sgtable() and dma_unmap_sgtable() in udmabuf to suppress the spurious warning
and
skip the redundant sync.
Fixes: 284562e1f348 ("udmabuf: implement
begin_cpu_access/end_cpu_access hooks")
Cc: stable@vger.kernel.org Signed-off-by: Mikhail Gavrilov mikhail.v.gavrilov@gmail.com
drivers/dma-buf/udmabuf.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/dma-buf/udmabuf.c b/drivers/dma-buf/udmabuf.c index 94b8ecb892bb..9c6f8785a28a 100644 --- a/drivers/dma-buf/udmabuf.c +++ b/drivers/dma-buf/udmabuf.c @@ -162,7 +162,7 @@ static struct sg_table *get_sg_table(struct
device *dev, struct dma_buf *buf,
sg_set_folio(sgl, ubuf->folios[i], PAGE_SIZE, ubuf->offsets[i]);
ret = dma_map_sgtable(dev, sg, direction, 0);
ret = dma_map_sgtable(dev, sg, direction,DMA_ATTR_SKIP_CPU_SYNC);
if (ret < 0) goto err_map; return sg;@@ -177,7 +177,7 @@ static struct sg_table *get_sg_table(struct
device *dev, struct dma_buf *buf,
static void put_sg_table(struct device *dev, struct sg_table *sg, enum dma_data_direction direction) {
dma_unmap_sgtable(dev, sg, direction, 0);
dma_unmap_sgtable(dev, sg, direction,DMA_ATTR_SKIP_CPU_SYNC);
sg_free_table(sg); kfree(sg);}
2.53.0
Gentle ping on this patch.
To summarize the review so far: Vivek: "Looks OK to me" Christian: Acked-by
Note: while my separate dma-debug patch [1] would also suppress this specific warning on x86, the udmabuf fix here is still valuable as it removes the redundant CPU sync at map/unmap time and aligns udmabuf with other dma-buf exporters that already pass DMA_ATTR_SKIP_CPU_SYNC.
[1] https://lore.kernel.org/all/20260327124156.24820-1- mikhail.v.gavrilov@gmail.com/
Gerd, could you take a look when you have a chance?
I am convinced that this patch is useful in some cases, so I'll add Christian's Ack and push it (into drm-misc-next) in a day or two.
Thanks, Vivek
-- Best Regards, Mike Gavrilov.
Hi Mikhail,
Subject: RE: [PATCH] dma-buf/udmabuf: skip redundant cpu sync to fix cacheline EEXIST warning
When CONFIG_DMA_API_DEBUG_SG is enabled, importing a udmabuf
into a DRM
driver (e.g. amdgpu for video playback in GNOME Videos / Showtime) triggers a spurious warning:
DMA-API: amdgpu 0000:03:00.0: cacheline tracking EEXIST, \ overlapping mappings aren't supported WARNING: kernel/dma/debug.c:619 at add_dma_entry+0x473/0x5f0
The call chain is:
amdgpu_cs_ioctl -> amdgpu_ttm_backend_bind -> dma_buf_map_attachment -> [udmabuf] map_udmabuf -> get_sg_table -> dma_map_sgtable(dev, sg, direction, 0) // attrs=0 -> debug_dma_map_sg -> add_dma_entry -> EEXIST
This happens because udmabuf builds a per-page scatter-gather list
via
sg_set_folio(). When begin_cpu_udmabuf() has already created an sg table mapped for the misc device, and an importer such as amdgpu
maps
the same pages for its own device via map_udmabuf(), the DMA
debug
infrastructure sees two active mappings whose physical addresses
share
cacheline boundaries and warns about the overlap.
The DMA_ATTR_SKIP_CPU_SYNC flag suppresses this check in add_dma_entry() because it signals that no CPU cache maintenance is performed at map/unmap time, making the cacheline overlap
harmless.
All other major dma-buf exporters already pass this flag:
- drm_gem_map_dma_buf() passes DMA_ATTR_SKIP_CPU_SYNC
- amdgpu_dma_buf_map() passes DMA_ATTR_SKIP_CPU_SYNC
The CPU sync at map/unmap time is also redundant for udmabuf: begin_cpu_udmabuf() and end_cpu_udmabuf() already perform
explicit
cache synchronization via dma_sync_sgtable_for_cpu/device() when
CPU
access is requested through the dma-buf interface.
Pass DMA_ATTR_SKIP_CPU_SYNC to dma_map_sgtable() and dma_unmap_sgtable() in udmabuf to suppress the spurious warning
and
skip the redundant sync.
Fixes: 284562e1f348 ("udmabuf: implement
begin_cpu_access/end_cpu_access hooks")
Cc: stable@vger.kernel.org Signed-off-by: Mikhail Gavrilov mikhail.v.gavrilov@gmail.com
drivers/dma-buf/udmabuf.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/dma-buf/udmabuf.c b/drivers/dma-buf/udmabuf.c index 94b8ecb892bb..9c6f8785a28a 100644 --- a/drivers/dma-buf/udmabuf.c +++ b/drivers/dma-buf/udmabuf.c @@ -162,7 +162,7 @@ static struct sg_table *get_sg_table(struct
device *dev, struct dma_buf *buf,
sg_set_folio(sgl, ubuf->folios[i], PAGE_SIZE, ubuf->offsets[i]);
ret = dma_map_sgtable(dev, sg, direction, 0);
ret = dma_map_sgtable(dev, sg, direction,DMA_ATTR_SKIP_CPU_SYNC);
if (ret < 0) goto err_map; return sg;@@ -177,7 +177,7 @@ static struct sg_table *get_sg_table(struct
device *dev, struct dma_buf *buf,
static void put_sg_table(struct device *dev, struct sg_table *sg, enum dma_data_direction direction) {
dma_unmap_sgtable(dev, sg, direction, 0);
dma_unmap_sgtable(dev, sg, direction,DMA_ATTR_SKIP_CPU_SYNC);
sg_free_table(sg); kfree(sg);}
2.53.0
Gentle ping on this patch.
To summarize the review so far: Vivek: "Looks OK to me" Christian: Acked-by
Note: while my separate dma-debug patch [1] would also suppress this specific warning on x86, the udmabuf fix here is still valuable as it removes the redundant CPU sync at map/unmap time and aligns udmabuf with other dma-buf exporters that already pass DMA_ATTR_SKIP_CPU_SYNC.
[1] https://lore.kernel.org/all/20260327124156.24820-1- mikhail.v.gavrilov@gmail.com/
Gerd, could you take a look when you have a chance?
I am convinced that this patch is useful in some cases, so I'll add Christian's Ack and push it (into drm-misc-next) in a day or two.
I ran into a merge conflict while applying this patch on top of other udmabuf patches. So, please resend this patch after rebasing on top of drm-tip: https://gitlab.freedesktop.org/drm/tip/-/commits/drm-tip?ref_type=HEADS
Thanks, Vivek
Thanks, Vivek
-- Best Regards, Mike Gavrilov.
linaro-mm-sig@lists.linaro.org
-
Christian König -
Kasireddy, Vivek -
Mikhail Gavrilov