* [PATCH v2 1/9] PCI/P2PDMA: Add CONFIG_PCI_P2PDMA_CORE
2026-05-27 10:23 [PATCH v2 0/9] vfio/pci: Add mmap() for DMABUFs Matt Evans
@ 2026-05-27 10:23 ` Matt Evans
2026-05-27 16:07 ` Logan Gunthorpe
` (2 more replies)
2026-05-27 10:23 ` [PATCH v2 2/9] vfio/pci: Add a helper to look up PFNs for DMABUFs Matt Evans
` (8 subsequent siblings)
9 siblings, 3 replies; 25+ messages in thread
From: Matt Evans @ 2026-05-27 10:23 UTC (permalink / raw)
To: Alex Williamson, Leon Romanovsky, Jason Gunthorpe, Alex Mastro,
Christian König, Bjorn Helgaas, Logan Gunthorpe
Cc: Mahmoud Adam, David Matlack, Björn Töpel, Sumit Semwal,
Kevin Tian, Ankit Agrawal, Pranjal Shrivastava, Alistair Popple,
Vivek Kasireddy, linux-kernel, linux-media, dri-devel,
linaro-mm-sig, kvm, linux-pci
The P2PDMA code currently provides two features under the same
CONFIG_PCI_P2PDMA option:
1. Locate providers via pcim_p2pdma_provider()
2. Manage actual P2P DMA
Other code (such as vfio-pci) depends on 1, without having a hard
dependency on 2.
A future commit expands the use of DMABUF in vfio-pci for non-P2P
scenarios, relying on pcim_p2pdma_provider() always being present. If
that depended on CONFIG_PCI_P2PDMA, it would make vfio-pci only
available if CONFIG_ZONE_DEVICE is present (e.g. 64-bit systems), even
when P2P is not needed.
To resolve this, introduce CONFIG_PCI_P2PDMA_CORE which contains the
basic provider functionality to make it available even if the
CONFIG_PCI_P2PDMA feature is disabled or unavailable due to
!CONFIG_ZONE_DEVICE. Users such as vfio-pci can enable their own P2P
features based off the original CONFIG_PCI_P2PDMA (available when
CONFIG_ZONE_DEVICE is set).
Signed-off-by: Matt Evans <mattev@meta.com>
---
drivers/pci/Kconfig | 10 +++++-----
drivers/pci/Makefile | 2 +-
drivers/pci/p2pdma.c | 16 ++++++++++++++++
include/linux/pci-p2pdma.h | 24 ++++++++++++++----------
include/linux/pci.h | 2 +-
5 files changed, 37 insertions(+), 17 deletions(-)
diff --git a/drivers/pci/Kconfig b/drivers/pci/Kconfig
index 33c88432b728..59d70bc84cc9 100644
--- a/drivers/pci/Kconfig
+++ b/drivers/pci/Kconfig
@@ -206,11 +206,7 @@ config PCIE_TPH
config PCI_P2PDMA
bool "PCI peer-to-peer transfer support"
depends on ZONE_DEVICE
- #
- # The need for the scatterlist DMA bus address flag means PCI P2PDMA
- # requires 64bit
- #
- depends on 64BIT
+ select PCI_P2PDMA_CORE
select GENERIC_ALLOCATOR
select NEED_SG_DMA_FLAGS
help
@@ -226,6 +222,10 @@ config PCI_P2PDMA
If unsure, say N.
+config PCI_P2PDMA_CORE
+ default n
+ bool
+
config PCI_LABEL
def_bool y if (DMI || ACPI)
select NLS
diff --git a/drivers/pci/Makefile b/drivers/pci/Makefile
index 41ebc3b9a518..419b646a301d 100644
--- a/drivers/pci/Makefile
+++ b/drivers/pci/Makefile
@@ -30,7 +30,7 @@ obj-$(CONFIG_PCI_SYSCALL) += syscall.o
obj-$(CONFIG_PCI_STUB) += pci-stub.o
obj-$(CONFIG_PCI_PF_STUB) += pci-pf-stub.o
obj-$(CONFIG_PCI_ECAM) += ecam.o
-obj-$(CONFIG_PCI_P2PDMA) += p2pdma.o
+obj-$(CONFIG_PCI_P2PDMA_CORE) += p2pdma.o
obj-$(CONFIG_XEN_PCIDEV_FRONTEND) += xen-pcifront.o
obj-$(CONFIG_VGA_ARB) += vgaarb.o
obj-$(CONFIG_PCI_DOE) += doe.o
diff --git a/drivers/pci/p2pdma.c b/drivers/pci/p2pdma.c
index 7c898542af8d..619d46c652b8 100644
--- a/drivers/pci/p2pdma.c
+++ b/drivers/pci/p2pdma.c
@@ -28,6 +28,14 @@ struct pci_p2pdma {
struct p2pdma_provider mem[PCI_STD_NUM_BARS];
};
+/*
+ * CONFIG_PCI_P2PDMA_CORE provides just a bare-bones init and
+ * pcim_p2pdma_provider() interface (used by things like VFIO even if
+ * full P2PDMA isn't present). The full P2PDMA feature is under the
+ * CONFIG_PCI_P2PDMA option.
+ */
+#ifdef CONFIG_PCI_P2PDMA
+
struct pci_p2pdma_pagemap {
struct dev_pagemap pgmap;
struct p2pdma_provider *mem;
@@ -226,6 +234,8 @@ static const struct dev_pagemap_ops p2pdma_pgmap_ops = {
.folio_free = p2pdma_folio_free,
};
+#endif /* CONFIG_PCI_P2PDMA */
+
static void pci_p2pdma_release(void *data)
{
struct pci_dev *pdev = data;
@@ -241,11 +251,13 @@ static void pci_p2pdma_release(void *data)
synchronize_rcu();
xa_destroy(&p2pdma->map_types);
+#ifdef CONFIG_PCI_P2PDMA
if (!p2pdma->pool)
return;
gen_pool_destroy(p2pdma->pool);
sysfs_remove_group(&pdev->dev.kobj, &p2pmem_group);
+#endif
}
/**
@@ -330,6 +342,8 @@ struct p2pdma_provider *pcim_p2pdma_provider(struct pci_dev *pdev, int bar)
}
EXPORT_SYMBOL_GPL(pcim_p2pdma_provider);
+#ifdef CONFIG_PCI_P2PDMA
+
static int pci_p2pdma_setup_pool(struct pci_dev *pdev)
{
struct pci_p2pdma *p2pdma;
@@ -1207,3 +1221,5 @@ ssize_t pci_p2pdma_enable_show(char *page, struct pci_dev *p2p_dev,
return sprintf(page, "%s\n", pci_name(p2p_dev));
}
EXPORT_SYMBOL_GPL(pci_p2pdma_enable_show);
+
+#endif
diff --git a/include/linux/pci-p2pdma.h b/include/linux/pci-p2pdma.h
index 873de20a2247..4c42a7b2ee85 100644
--- a/include/linux/pci-p2pdma.h
+++ b/include/linux/pci-p2pdma.h
@@ -67,9 +67,22 @@ enum pci_p2pdma_map_type {
PCI_P2PDMA_MAP_THRU_HOST_BRIDGE,
};
-#ifdef CONFIG_PCI_P2PDMA
+#ifdef CONFIG_PCI_P2PDMA_CORE
int pcim_p2pdma_init(struct pci_dev *pdev);
struct p2pdma_provider *pcim_p2pdma_provider(struct pci_dev *pdev, int bar);
+#else
+static inline int pcim_p2pdma_init(struct pci_dev *pdev)
+{
+ return -EOPNOTSUPP;
+}
+static inline struct p2pdma_provider *pcim_p2pdma_provider(struct pci_dev *pdev,
+ int bar)
+{
+ return NULL;
+}
+#endif
+
+#ifdef CONFIG_PCI_P2PDMA
int pci_p2pdma_add_resource(struct pci_dev *pdev, int bar, size_t size,
u64 offset);
int pci_p2pdma_distance_many(struct pci_dev *provider, struct device **clients,
@@ -89,15 +102,6 @@ ssize_t pci_p2pdma_enable_show(char *page, struct pci_dev *p2p_dev,
enum pci_p2pdma_map_type pci_p2pdma_map_type(struct p2pdma_provider *provider,
struct device *dev);
#else /* CONFIG_PCI_P2PDMA */
-static inline int pcim_p2pdma_init(struct pci_dev *pdev)
-{
- return -EOPNOTSUPP;
-}
-static inline struct p2pdma_provider *pcim_p2pdma_provider(struct pci_dev *pdev,
- int bar)
-{
- return NULL;
-}
static inline int pci_p2pdma_add_resource(struct pci_dev *pdev, int bar,
size_t size, u64 offset)
{
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 2c4454583c11..531aec355686 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -557,7 +557,7 @@ struct pci_dev {
u16 pasid_cap; /* PASID Capability offset */
u16 pasid_features;
#endif
-#ifdef CONFIG_PCI_P2PDMA
+#ifdef CONFIG_PCI_P2PDMA_CORE
struct pci_p2pdma __rcu *p2pdma;
#endif
#ifdef CONFIG_PCI_DOE
--
2.47.3
^ permalink raw reply related [flat|nested] 25+ messages in thread* Re: [PATCH v2 1/9] PCI/P2PDMA: Add CONFIG_PCI_P2PDMA_CORE
2026-05-27 10:23 ` [PATCH v2 1/9] PCI/P2PDMA: Add CONFIG_PCI_P2PDMA_CORE Matt Evans
@ 2026-05-27 16:07 ` Logan Gunthorpe
2026-05-27 17:13 ` Matt Evans
2026-05-27 21:09 ` Alex Williamson
2026-05-28 2:28 ` Claude review: " Claude Code Review Bot
2 siblings, 1 reply; 25+ messages in thread
From: Logan Gunthorpe @ 2026-05-27 16:07 UTC (permalink / raw)
To: Matt Evans, Alex Williamson, Leon Romanovsky, Jason Gunthorpe,
Alex Mastro, Christian König, Bjorn Helgaas
Cc: Mahmoud Adam, David Matlack, Björn Töpel, Sumit Semwal,
Kevin Tian, Ankit Agrawal, Pranjal Shrivastava, Alistair Popple,
Vivek Kasireddy, linux-kernel, linux-media, dri-devel,
linaro-mm-sig, kvm, linux-pci
On 2026-05-27 04:23, Matt Evans wrote:
> The P2PDMA code currently provides two features under the same
> CONFIG_PCI_P2PDMA option:
>
> 1. Locate providers via pcim_p2pdma_provider()
> 2. Manage actual P2P DMA
>
> Other code (such as vfio-pci) depends on 1, without having a hard
> dependency on 2.
>
> A future commit expands the use of DMABUF in vfio-pci for non-P2P
> scenarios, relying on pcim_p2pdma_provider() always being present. If
> that depended on CONFIG_PCI_P2PDMA, it would make vfio-pci only
> available if CONFIG_ZONE_DEVICE is present (e.g. 64-bit systems), even
> when P2P is not needed.
>
> To resolve this, introduce CONFIG_PCI_P2PDMA_CORE which contains the
> basic provider functionality to make it available even if the
> CONFIG_PCI_P2PDMA feature is disabled or unavailable due to
> !CONFIG_ZONE_DEVICE. Users such as vfio-pci can enable their own P2P
> features based off the original CONFIG_PCI_P2PDMA (available when
> CONFIG_ZONE_DEVICE is set).
>
> Signed-off-by: Matt Evans <mattev@meta.com>
Largely this looks good to me. I have one minor nit below that you can
apply or not. Either way, feel free to add:
Reviewed-by: Logan Gunthorpe <logang@deltatee.com>
> static void pci_p2pdma_release(void *data)
> {
> struct pci_dev *pdev = data;
> @@ -241,11 +251,13 @@ static void pci_p2pdma_release(void *data)
> synchronize_rcu();
> xa_destroy(&p2pdma->map_types);
>
> +#ifdef CONFIG_PCI_P2PDMA
> if (!p2pdma->pool)
> return;
>
> gen_pool_destroy(p2pdma->pool);
> sysfs_remove_group(&pdev->dev.kobj, &p2pmem_group);
> +#endif
> }
I'm personally not a fan of adding #ifdefs inside functions like this.
This instance is small and easy to understand, but it can quickly become
a bit of a mess if we start adding more features. I probably would have
created a pci_p2pdma_release_pool() helper which does the inverse of
pci_p2pdma_setup_pool(), it would be called in pci_p2pdma_release() and
an empty implementation would be provided in the case where
CONFIG_PCI_P2PDMA is not set.
Logan
^ permalink raw reply [flat|nested] 25+ messages in thread* Re: [PATCH v2 1/9] PCI/P2PDMA: Add CONFIG_PCI_P2PDMA_CORE
2026-05-27 16:07 ` Logan Gunthorpe
@ 2026-05-27 17:13 ` Matt Evans
0 siblings, 0 replies; 25+ messages in thread
From: Matt Evans @ 2026-05-27 17:13 UTC (permalink / raw)
To: Logan Gunthorpe, Alex Williamson, Leon Romanovsky,
Jason Gunthorpe, Alex Mastro, Christian König, Bjorn Helgaas
Cc: Mahmoud Adam, David Matlack, Björn Töpel, Sumit Semwal,
Kevin Tian, Ankit Agrawal, Pranjal Shrivastava, Alistair Popple,
Vivek Kasireddy, linux-kernel, linux-media, dri-devel,
linaro-mm-sig, kvm, linux-pci
Hi Logan,
On 27/05/2026 17:07, Logan Gunthorpe wrote:
>
> On 2026-05-27 04:23, Matt Evans wrote:
>> The P2PDMA code currently provides two features under the same
>> CONFIG_PCI_P2PDMA option:
>>
>> 1. Locate providers via pcim_p2pdma_provider()
>> 2. Manage actual P2P DMA
>>
>> Other code (such as vfio-pci) depends on 1, without having a hard
>> dependency on 2.
>>
>> A future commit expands the use of DMABUF in vfio-pci for non-P2P
>> scenarios, relying on pcim_p2pdma_provider() always being present. If
>> that depended on CONFIG_PCI_P2PDMA, it would make vfio-pci only
>> available if CONFIG_ZONE_DEVICE is present (e.g. 64-bit systems), even
>> when P2P is not needed.
>>
>> To resolve this, introduce CONFIG_PCI_P2PDMA_CORE which contains the
>> basic provider functionality to make it available even if the
>> CONFIG_PCI_P2PDMA feature is disabled or unavailable due to
>> !CONFIG_ZONE_DEVICE. Users such as vfio-pci can enable their own P2P
>> features based off the original CONFIG_PCI_P2PDMA (available when
>> CONFIG_ZONE_DEVICE is set).
>>
>> Signed-off-by: Matt Evans <mattev@meta.com>
>
> Largely this looks good to me. I have one minor nit below that you can
> apply or not. Either way, feel free to add:
>
> Reviewed-by: Logan Gunthorpe <logang@deltatee.com>
>
>
>> static void pci_p2pdma_release(void *data)
>> {
>> struct pci_dev *pdev = data;
>> @@ -241,11 +251,13 @@ static void pci_p2pdma_release(void *data)
>> synchronize_rcu();
>> xa_destroy(&p2pdma->map_types);
>>
>> +#ifdef CONFIG_PCI_P2PDMA
>> if (!p2pdma->pool)
>> return;
>>
>> gen_pool_destroy(p2pdma->pool);
>> sysfs_remove_group(&pdev->dev.kobj, &p2pmem_group);
>> +#endif
>> }
>
> I'm personally not a fan of adding #ifdefs inside functions like this.
> This instance is small and easy to understand, but it can quickly become
> a bit of a mess if we start adding more features. I probably would have
> created a pci_p2pdma_release_pool() helper which does the inverse of
> pci_p2pdma_setup_pool(), it would be called in pci_p2pdma_release() and
> an empty implementation would be provided in the case where
> CONFIG_PCI_P2PDMA is not set.
That's cleaner, I'll do that. Thanks for the review.
Matt
^ permalink raw reply [flat|nested] 25+ messages in thread
* Re: [PATCH v2 1/9] PCI/P2PDMA: Add CONFIG_PCI_P2PDMA_CORE
2026-05-27 10:23 ` [PATCH v2 1/9] PCI/P2PDMA: Add CONFIG_PCI_P2PDMA_CORE Matt Evans
2026-05-27 16:07 ` Logan Gunthorpe
@ 2026-05-27 21:09 ` Alex Williamson
2026-05-28 2:28 ` Claude review: " Claude Code Review Bot
2 siblings, 0 replies; 25+ messages in thread
From: Alex Williamson @ 2026-05-27 21:09 UTC (permalink / raw)
To: Matt Evans
Cc: Leon Romanovsky, Jason Gunthorpe, Alex Mastro,
Christian König, Bjorn Helgaas, Logan Gunthorpe,
Mahmoud Adam, David Matlack, Björn Töpel, Sumit Semwal,
Kevin Tian, Ankit Agrawal, Pranjal Shrivastava, Alistair Popple,
Vivek Kasireddy, linux-kernel, linux-media, dri-devel,
linaro-mm-sig, kvm, linux-pci
On Wed, 27 May 2026 03:23:04 -0700
Matt Evans <mattev@meta.com> wrote:
> The P2PDMA code currently provides two features under the same
> CONFIG_PCI_P2PDMA option:
>
> 1. Locate providers via pcim_p2pdma_provider()
> 2. Manage actual P2P DMA
>
> Other code (such as vfio-pci) depends on 1, without having a hard
> dependency on 2.
>
> A future commit expands the use of DMABUF in vfio-pci for non-P2P
> scenarios, relying on pcim_p2pdma_provider() always being present. If
> that depended on CONFIG_PCI_P2PDMA, it would make vfio-pci only
> available if CONFIG_ZONE_DEVICE is present (e.g. 64-bit systems), even
> when P2P is not needed.
>
> To resolve this, introduce CONFIG_PCI_P2PDMA_CORE which contains the
> basic provider functionality to make it available even if the
> CONFIG_PCI_P2PDMA feature is disabled or unavailable due to
> !CONFIG_ZONE_DEVICE. Users such as vfio-pci can enable their own P2P
> features based off the original CONFIG_PCI_P2PDMA (available when
> CONFIG_ZONE_DEVICE is set).
>
> Signed-off-by: Matt Evans <mattev@meta.com>
> ---
> drivers/pci/Kconfig | 10 +++++-----
> drivers/pci/Makefile | 2 +-
> drivers/pci/p2pdma.c | 16 ++++++++++++++++
> include/linux/pci-p2pdma.h | 24 ++++++++++++++----------
> include/linux/pci.h | 2 +-
> 5 files changed, 37 insertions(+), 17 deletions(-)
>
> diff --git a/drivers/pci/Kconfig b/drivers/pci/Kconfig
> index 33c88432b728..59d70bc84cc9 100644
> --- a/drivers/pci/Kconfig
> +++ b/drivers/pci/Kconfig
> @@ -206,11 +206,7 @@ config PCIE_TPH
> config PCI_P2PDMA
> bool "PCI peer-to-peer transfer support"
> depends on ZONE_DEVICE
> - #
> - # The need for the scatterlist DMA bus address flag means PCI P2PDMA
> - # requires 64bit
> - #
> - depends on 64BIT
> + select PCI_P2PDMA_CORE
> select GENERIC_ALLOCATOR
> select NEED_SG_DMA_FLAGS
> help
> @@ -226,6 +222,10 @@ config PCI_P2PDMA
>
> If unsure, say N.
>
> +config PCI_P2PDMA_CORE
> + default n
> + bool
> +
> config PCI_LABEL
> def_bool y if (DMI || ACPI)
> select NLS
> diff --git a/drivers/pci/Makefile b/drivers/pci/Makefile
> index 41ebc3b9a518..419b646a301d 100644
> --- a/drivers/pci/Makefile
> +++ b/drivers/pci/Makefile
> @@ -30,7 +30,7 @@ obj-$(CONFIG_PCI_SYSCALL) += syscall.o
> obj-$(CONFIG_PCI_STUB) += pci-stub.o
> obj-$(CONFIG_PCI_PF_STUB) += pci-pf-stub.o
> obj-$(CONFIG_PCI_ECAM) += ecam.o
> -obj-$(CONFIG_PCI_P2PDMA) += p2pdma.o
> +obj-$(CONFIG_PCI_P2PDMA_CORE) += p2pdma.o
> obj-$(CONFIG_XEN_PCIDEV_FRONTEND) += xen-pcifront.o
> obj-$(CONFIG_VGA_ARB) += vgaarb.o
> obj-$(CONFIG_PCI_DOE) += doe.o
> diff --git a/drivers/pci/p2pdma.c b/drivers/pci/p2pdma.c
> index 7c898542af8d..619d46c652b8 100644
> --- a/drivers/pci/p2pdma.c
> +++ b/drivers/pci/p2pdma.c
> @@ -28,6 +28,14 @@ struct pci_p2pdma {
> struct p2pdma_provider mem[PCI_STD_NUM_BARS];
> };
>
> +/*
> + * CONFIG_PCI_P2PDMA_CORE provides just a bare-bones init and
> + * pcim_p2pdma_provider() interface (used by things like VFIO even if
> + * full P2PDMA isn't present). The full P2PDMA feature is under the
> + * CONFIG_PCI_P2PDMA option.
> + */
> +#ifdef CONFIG_PCI_P2PDMA
> +
> struct pci_p2pdma_pagemap {
> struct dev_pagemap pgmap;
> struct p2pdma_provider *mem;
> @@ -226,6 +234,8 @@ static const struct dev_pagemap_ops p2pdma_pgmap_ops = {
> .folio_free = p2pdma_folio_free,
> };
>
> +#endif /* CONFIG_PCI_P2PDMA */
> +
> static void pci_p2pdma_release(void *data)
> {
> struct pci_dev *pdev = data;
> @@ -241,11 +251,13 @@ static void pci_p2pdma_release(void *data)
> synchronize_rcu();
> xa_destroy(&p2pdma->map_types);
>
> +#ifdef CONFIG_PCI_P2PDMA
> if (!p2pdma->pool)
> return;
>
> gen_pool_destroy(p2pdma->pool);
> sysfs_remove_group(&pdev->dev.kobj, &p2pmem_group);
> +#endif
> }
>
> /**
> @@ -330,6 +342,8 @@ struct p2pdma_provider *pcim_p2pdma_provider(struct pci_dev *pdev, int bar)
> }
> EXPORT_SYMBOL_GPL(pcim_p2pdma_provider);
>
> +#ifdef CONFIG_PCI_P2PDMA
> +
> static int pci_p2pdma_setup_pool(struct pci_dev *pdev)
> {
> struct pci_p2pdma *p2pdma;
> @@ -1207,3 +1221,5 @@ ssize_t pci_p2pdma_enable_show(char *page, struct pci_dev *p2p_dev,
> return sprintf(page, "%s\n", pci_name(p2p_dev));
> }
> EXPORT_SYMBOL_GPL(pci_p2pdma_enable_show);
> +
> +#endif
This deserves a /* CONFIG_PCI_P2PDMA */ reference like the previous
long span, though potentially a better solution would be to split the
core code out to a separate file with shared header. I asked opus to
see what this would look like and it generated the patch below
(unverified, compile tested only). Bigger diff, harder to follow
blame, but the shape of the core is much more apparent. You're welcome
to take or leave it. Thanks,
Alex
drivers/pci/Kconfig | 12 ++--
drivers/pci/Makefile | 1
drivers/pci/p2pdma.c | 115 ++++-----------------------------------
drivers/pci/p2pdma.h | 30 ++++++++++
drivers/pci/p2pdma_core.c | 119 +++++++++++++++++++++++++++++++++++++++++
include/linux/pci-p2pdma.h | 24 ++++----
include/linux/pci.h | 2
7 files changed, 184 insertions(+), 119 deletions(-)
diff --git a/drivers/pci/Kconfig b/drivers/pci/Kconfig
index 33c88432b728..2e5af8c576f5 100644
--- a/drivers/pci/Kconfig
+++ b/drivers/pci/Kconfig
@@ -206,11 +206,7 @@ config PCIE_TPH
config PCI_P2PDMA
bool "PCI peer-to-peer transfer support"
depends on ZONE_DEVICE
- #
- # The need for the scatterlist DMA bus address flag means PCI P2PDMA
- # requires 64bit
- #
- depends on 64BIT
+ select PCI_P2PDMA_CORE
select GENERIC_ALLOCATOR
select NEED_SG_DMA_FLAGS
help
@@ -226,6 +222,12 @@ config PCI_P2PDMA
If unsure, say N.
+# Selected by features that only need the pcim_p2pdma_provider() interface
+# (e.g. vfio-pci's DMABUF export of BAR ranges) without depending on the
+# full ZONE_DEVICE-backed allocator that PCI_P2PDMA pulls in.
+config PCI_P2PDMA_CORE
+ bool
+
config PCI_LABEL
def_bool y if (DMI || ACPI)
select NLS
diff --git a/drivers/pci/Makefile b/drivers/pci/Makefile
index 41ebc3b9a518..0b32572d57a1 100644
--- a/drivers/pci/Makefile
+++ b/drivers/pci/Makefile
@@ -30,6 +30,7 @@ obj-$(CONFIG_PCI_SYSCALL) += syscall.o
obj-$(CONFIG_PCI_STUB) += pci-stub.o
obj-$(CONFIG_PCI_PF_STUB) += pci-pf-stub.o
obj-$(CONFIG_PCI_ECAM) += ecam.o
+obj-$(CONFIG_PCI_P2PDMA_CORE) += p2pdma_core.o
obj-$(CONFIG_PCI_P2PDMA) += p2pdma.o
obj-$(CONFIG_XEN_PCIDEV_FRONTEND) += xen-pcifront.o
obj-$(CONFIG_VGA_ARB) += vgaarb.o
diff --git a/drivers/pci/p2pdma.c b/drivers/pci/p2pdma.c
index 7c898542af8d..158c274fda18 100644
--- a/drivers/pci/p2pdma.c
+++ b/drivers/pci/p2pdma.c
@@ -21,12 +21,7 @@
#include <linux/seq_buf.h>
#include <linux/xarray.h>
-struct pci_p2pdma {
- struct gen_pool *pool;
- bool p2pmem_published;
- struct xarray map_types;
- struct p2pdma_provider mem[PCI_STD_NUM_BARS];
-};
+#include "p2pdma.h"
struct pci_p2pdma_pagemap {
struct dev_pagemap pgmap;
@@ -226,110 +221,24 @@ static const struct dev_pagemap_ops p2pdma_pgmap_ops = {
.folio_free = p2pdma_folio_free,
};
-static void pci_p2pdma_release(void *data)
+/*
+ * Tear down the allocator pool side of pci_p2pdma. Called from the core
+ * release path (pci_p2pdma_release in p2pdma_core.c) while pdev->p2pdma
+ * has already been cleared but the structure itself has not yet been
+ * freed. Pairs with the rcu_dereference()s in pci_alloc_p2pmem() and
+ * friends, so that no allocator readers are in flight when we destroy
+ * the pool.
+ */
+void pci_p2pdma_release_pool(struct pci_p2pdma *p2pdma, struct pci_dev *pdev)
{
- struct pci_dev *pdev = data;
- struct pci_p2pdma *p2pdma;
-
- p2pdma = rcu_dereference_protected(pdev->p2pdma, 1);
- if (!p2pdma)
- return;
-
- /* Flush and disable pci_alloc_p2p_mem() */
- pdev->p2pdma = NULL;
- if (p2pdma->pool)
- synchronize_rcu();
- xa_destroy(&p2pdma->map_types);
-
if (!p2pdma->pool)
return;
+ synchronize_rcu();
gen_pool_destroy(p2pdma->pool);
sysfs_remove_group(&pdev->dev.kobj, &p2pmem_group);
}
-/**
- * pcim_p2pdma_init - Initialise peer-to-peer DMA providers
- * @pdev: The PCI device to enable P2PDMA for
- *
- * This function initializes the peer-to-peer DMA infrastructure
- * for a PCI device. It allocates and sets up the necessary data
- * structures to support P2PDMA operations, including mapping type
- * tracking.
- */
-int pcim_p2pdma_init(struct pci_dev *pdev)
-{
- struct pci_p2pdma *p2p;
- int i, ret;
-
- p2p = rcu_dereference_protected(pdev->p2pdma, 1);
- if (p2p)
- return 0;
-
- p2p = devm_kzalloc(&pdev->dev, sizeof(*p2p), GFP_KERNEL);
- if (!p2p)
- return -ENOMEM;
-
- xa_init(&p2p->map_types);
- /*
- * Iterate over all standard PCI BARs and record only those that
- * correspond to MMIO regions. Skip non-memory resources (e.g. I/O
- * port BARs) since they cannot be used for peer-to-peer (P2P)
- * transactions.
- */
- for (i = 0; i < PCI_STD_NUM_BARS; i++) {
- if (!(pci_resource_flags(pdev, i) & IORESOURCE_MEM))
- continue;
-
- p2p->mem[i].owner = &pdev->dev;
- p2p->mem[i].bus_offset =
- pci_bus_address(pdev, i) - pci_resource_start(pdev, i);
- }
-
- ret = devm_add_action_or_reset(&pdev->dev, pci_p2pdma_release, pdev);
- if (ret)
- goto out_p2p;
-
- rcu_assign_pointer(pdev->p2pdma, p2p);
- return 0;
-
-out_p2p:
- devm_kfree(&pdev->dev, p2p);
- return ret;
-}
-EXPORT_SYMBOL_GPL(pcim_p2pdma_init);
-
-/**
- * pcim_p2pdma_provider - Get peer-to-peer DMA provider
- * @pdev: The PCI device to enable P2PDMA for
- * @bar: BAR index to get provider
- *
- * This function gets peer-to-peer DMA provider for a PCI device. The lifetime
- * of the provider (and of course the MMIO) is bound to the lifetime of the
- * driver. A driver calling this function must ensure that all references to the
- * provider, and any DMA mappings created for any MMIO, are all cleaned up
- * before the driver remove() completes.
- *
- * Since P2P is almost always shared with a second driver this means some system
- * to notify, invalidate and revoke the MMIO's DMA must be in place to use this
- * function. For example a revoke can be built using DMABUF.
- */
-struct p2pdma_provider *pcim_p2pdma_provider(struct pci_dev *pdev, int bar)
-{
- struct pci_p2pdma *p2p;
-
- if (!(pci_resource_flags(pdev, bar) & IORESOURCE_MEM))
- return NULL;
-
- p2p = rcu_dereference_protected(pdev->p2pdma, 1);
- if (WARN_ON(!p2p))
- /* Someone forgot to call to pcim_p2pdma_init() before */
- return NULL;
-
- return &p2p->mem[bar];
-}
-EXPORT_SYMBOL_GPL(pcim_p2pdma_provider);
-
static int pci_p2pdma_setup_pool(struct pci_dev *pdev)
{
struct pci_p2pdma *p2pdma;
@@ -932,7 +841,7 @@ void *pci_alloc_p2pmem(struct pci_dev *pdev, size_t size)
struct pci_p2pdma *p2pdma;
/*
- * Pairs with synchronize_rcu() in pci_p2pdma_release() to
+ * Pairs with synchronize_rcu() in pci_p2pdma_release_pool() to
* ensure pdev->p2pdma is non-NULL for the duration of the
* read-lock.
*/
diff --git a/drivers/pci/p2pdma.h b/drivers/pci/p2pdma.h
new file mode 100644
index 000000000000..790e1df8b609
--- /dev/null
+++ b/drivers/pci/p2pdma.h
@@ -0,0 +1,30 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * PCI Peer-to-Peer DMA internal definitions, shared between p2pdma_core.c
+ * (always built when CONFIG_PCI_P2PDMA_CORE is set) and p2pdma.c (built
+ * only when the full CONFIG_PCI_P2PDMA feature is enabled).
+ */
+
+#ifndef DRIVERS_PCI_P2PDMA_H
+#define DRIVERS_PCI_P2PDMA_H
+
+#include <linux/genalloc.h>
+#include <linux/pci.h>
+#include <linux/pci-p2pdma.h>
+#include <linux/xarray.h>
+
+struct pci_p2pdma {
+ struct gen_pool *pool;
+ bool p2pmem_published;
+ struct xarray map_types;
+ struct p2pdma_provider mem[PCI_STD_NUM_BARS];
+};
+
+#ifdef CONFIG_PCI_P2PDMA
+void pci_p2pdma_release_pool(struct pci_p2pdma *p2pdma, struct pci_dev *pdev);
+#else
+static inline void pci_p2pdma_release_pool(struct pci_p2pdma *p2pdma,
+ struct pci_dev *pdev) {}
+#endif
+
+#endif /* DRIVERS_PCI_P2PDMA_H */
diff --git a/drivers/pci/p2pdma_core.c b/drivers/pci/p2pdma_core.c
new file mode 100644
index 000000000000..c7622613f03a
--- /dev/null
+++ b/drivers/pci/p2pdma_core.c
@@ -0,0 +1,119 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * PCI Peer-to-Peer DMA core: provider infrastructure shared by drivers
+ * that only need pcim_p2pdma_provider() (e.g. vfio-pci) without depending
+ * on the full CONFIG_PCI_P2PDMA allocator/pool feature.
+ *
+ * Copyright (c) 2016-2018, Logan Gunthorpe
+ * Copyright (c) 2016-2017, Microsemi Corporation
+ * Copyright (c) 2017, Christoph Hellwig
+ * Copyright (c) 2018, Eideticom Inc.
+ */
+
+#define pr_fmt(fmt) "pci-p2pdma: " fmt
+
+#include <linux/device.h>
+#include <linux/pci.h>
+#include <linux/pci-p2pdma.h>
+#include <linux/rcupdate.h>
+#include <linux/slab.h>
+#include <linux/xarray.h>
+
+#include "p2pdma.h"
+
+static void pci_p2pdma_release(void *data)
+{
+ struct pci_dev *pdev = data;
+ struct pci_p2pdma *p2pdma;
+
+ p2pdma = rcu_dereference_protected(pdev->p2pdma, 1);
+ if (!p2pdma)
+ return;
+
+ /* Flush and disable pci_alloc_p2p_mem() */
+ pdev->p2pdma = NULL;
+ pci_p2pdma_release_pool(p2pdma, pdev);
+ xa_destroy(&p2pdma->map_types);
+}
+
+/**
+ * pcim_p2pdma_init - Initialise peer-to-peer DMA providers
+ * @pdev: The PCI device to enable P2PDMA for
+ *
+ * This function initializes the peer-to-peer DMA infrastructure
+ * for a PCI device. It allocates and sets up the necessary data
+ * structures to support P2PDMA operations, including mapping type
+ * tracking.
+ */
+int pcim_p2pdma_init(struct pci_dev *pdev)
+{
+ struct pci_p2pdma *p2p;
+ int i, ret;
+
+ p2p = rcu_dereference_protected(pdev->p2pdma, 1);
+ if (p2p)
+ return 0;
+
+ p2p = devm_kzalloc(&pdev->dev, sizeof(*p2p), GFP_KERNEL);
+ if (!p2p)
+ return -ENOMEM;
+
+ xa_init(&p2p->map_types);
+ /*
+ * Iterate over all standard PCI BARs and record only those that
+ * correspond to MMIO regions. Skip non-memory resources (e.g. I/O
+ * port BARs) since they cannot be used for peer-to-peer (P2P)
+ * transactions.
+ */
+ for (i = 0; i < PCI_STD_NUM_BARS; i++) {
+ if (!(pci_resource_flags(pdev, i) & IORESOURCE_MEM))
+ continue;
+
+ p2p->mem[i].owner = &pdev->dev;
+ p2p->mem[i].bus_offset =
+ pci_bus_address(pdev, i) - pci_resource_start(pdev, i);
+ }
+
+ ret = devm_add_action_or_reset(&pdev->dev, pci_p2pdma_release, pdev);
+ if (ret)
+ goto out_p2p;
+
+ rcu_assign_pointer(pdev->p2pdma, p2p);
+ return 0;
+
+out_p2p:
+ devm_kfree(&pdev->dev, p2p);
+ return ret;
+}
+EXPORT_SYMBOL_GPL(pcim_p2pdma_init);
+
+/**
+ * pcim_p2pdma_provider - Get peer-to-peer DMA provider
+ * @pdev: The PCI device to enable P2PDMA for
+ * @bar: BAR index to get provider
+ *
+ * This function gets peer-to-peer DMA provider for a PCI device. The lifetime
+ * of the provider (and of course the MMIO) is bound to the lifetime of the
+ * driver. A driver calling this function must ensure that all references to the
+ * provider, and any DMA mappings created for any MMIO, are all cleaned up
+ * before the driver remove() completes.
+ *
+ * Since P2P is almost always shared with a second driver this means some system
+ * to notify, invalidate and revoke the MMIO's DMA must be in place to use this
+ * function. For example a revoke can be built using DMABUF.
+ */
+struct p2pdma_provider *pcim_p2pdma_provider(struct pci_dev *pdev, int bar)
+{
+ struct pci_p2pdma *p2p;
+
+ if (!(pci_resource_flags(pdev, bar) & IORESOURCE_MEM))
+ return NULL;
+
+ p2p = rcu_dereference_protected(pdev->p2pdma, 1);
+ if (WARN_ON(!p2p))
+ /* Someone forgot to call to pcim_p2pdma_init() before */
+ return NULL;
+
+ return &p2p->mem[bar];
+}
+EXPORT_SYMBOL_GPL(pcim_p2pdma_provider);
diff --git a/include/linux/pci-p2pdma.h b/include/linux/pci-p2pdma.h
index 873de20a2247..4c42a7b2ee85 100644
--- a/include/linux/pci-p2pdma.h
+++ b/include/linux/pci-p2pdma.h
@@ -67,9 +67,22 @@ enum pci_p2pdma_map_type {
PCI_P2PDMA_MAP_THRU_HOST_BRIDGE,
};
-#ifdef CONFIG_PCI_P2PDMA
+#ifdef CONFIG_PCI_P2PDMA_CORE
int pcim_p2pdma_init(struct pci_dev *pdev);
struct p2pdma_provider *pcim_p2pdma_provider(struct pci_dev *pdev, int bar);
+#else
+static inline int pcim_p2pdma_init(struct pci_dev *pdev)
+{
+ return -EOPNOTSUPP;
+}
+static inline struct p2pdma_provider *pcim_p2pdma_provider(struct pci_dev *pdev,
+ int bar)
+{
+ return NULL;
+}
+#endif
+
+#ifdef CONFIG_PCI_P2PDMA
int pci_p2pdma_add_resource(struct pci_dev *pdev, int bar, size_t size,
u64 offset);
int pci_p2pdma_distance_many(struct pci_dev *provider, struct device **clients,
@@ -89,15 +102,6 @@ ssize_t pci_p2pdma_enable_show(char *page, struct pci_dev *p2p_dev,
enum pci_p2pdma_map_type pci_p2pdma_map_type(struct p2pdma_provider *provider,
struct device *dev);
#else /* CONFIG_PCI_P2PDMA */
-static inline int pcim_p2pdma_init(struct pci_dev *pdev)
-{
- return -EOPNOTSUPP;
-}
-static inline struct p2pdma_provider *pcim_p2pdma_provider(struct pci_dev *pdev,
- int bar)
-{
- return NULL;
-}
static inline int pci_p2pdma_add_resource(struct pci_dev *pdev, int bar,
size_t size, u64 offset)
{
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 2c4454583c11..531aec355686 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -557,7 +557,7 @@ struct pci_dev {
u16 pasid_cap; /* PASID Capability offset */
u16 pasid_features;
#endif
-#ifdef CONFIG_PCI_P2PDMA
+#ifdef CONFIG_PCI_P2PDMA_CORE
struct pci_p2pdma __rcu *p2pdma;
#endif
#ifdef CONFIG_PCI_DOE
^ permalink raw reply related [flat|nested] 25+ messages in thread* Claude review: PCI/P2PDMA: Add CONFIG_PCI_P2PDMA_CORE
2026-05-27 10:23 ` [PATCH v2 1/9] PCI/P2PDMA: Add CONFIG_PCI_P2PDMA_CORE Matt Evans
2026-05-27 16:07 ` Logan Gunthorpe
2026-05-27 21:09 ` Alex Williamson
@ 2026-05-28 2:28 ` Claude Code Review Bot
2 siblings, 0 replies; 25+ messages in thread
From: Claude Code Review Bot @ 2026-05-28 2:28 UTC (permalink / raw)
To: dri-devel-reviews
Patch Review
Splits `CONFIG_PCI_P2PDMA` into a core part (provider lookup) and the full feature (requires `ZONE_DEVICE`). This unblocks vfio-pci on 32-bit platforms.
**Observation**: The patch removes the `depends on 64BIT` from `PCI_P2PDMA` but the comment it also removes says:
```
- #
- # The need for the scatterlist DMA bus address flag means PCI P2PDMA
- # requires 64bit
- #
- depends on 64BIT
```
This comment was explaining why `64BIT` was needed. Since `ZONE_DEVICE` itself already implies `64BIT` on most architectures (via `MEMORY_HOTPLUG`), the removal is likely fine in practice, but it would be worth confirming that `ZONE_DEVICE` is truly never selectable on 32-bit. If it were, the scatterlist DMA bus address flag concern would resurface.
**Minor**: `default n` is unnecessary for a bool Kconfig symbol -- `n` is already the default:
```
+config PCI_P2PDMA_CORE
+ default n
+ bool
```
The `#ifdef CONFIG_PCI_P2PDMA` guards in `p2pdma.c` look correct -- the file is now compiled under `_CORE`, and the full P2P functionality is conditionally compiled within it.
---
Generated by Claude Code Patch Reviewer
^ permalink raw reply [flat|nested] 25+ messages in thread
* [PATCH v2 2/9] vfio/pci: Add a helper to look up PFNs for DMABUFs
2026-05-27 10:23 [PATCH v2 0/9] vfio/pci: Add mmap() for DMABUFs Matt Evans
2026-05-27 10:23 ` [PATCH v2 1/9] PCI/P2PDMA: Add CONFIG_PCI_P2PDMA_CORE Matt Evans
@ 2026-05-27 10:23 ` Matt Evans
2026-05-27 22:38 ` Alex Williamson
2026-05-28 2:28 ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 3/9] vfio/pci: Add a helper to create a DMABUF for a BAR-map VMA Matt Evans
` (7 subsequent siblings)
9 siblings, 2 replies; 25+ messages in thread
From: Matt Evans @ 2026-05-27 10:23 UTC (permalink / raw)
To: Alex Williamson, Leon Romanovsky, Jason Gunthorpe, Alex Mastro,
Christian König, Bjorn Helgaas, Logan Gunthorpe
Cc: Mahmoud Adam, David Matlack, Björn Töpel, Sumit Semwal,
Kevin Tian, Ankit Agrawal, Pranjal Shrivastava, Alistair Popple,
Vivek Kasireddy, linux-kernel, linux-media, dri-devel,
linaro-mm-sig, kvm, linux-pci
Add vfio_pci_dma_buf_find_pfn(), which a VMA fault handler can use to
find a PFN.
This supports multi-range DMABUFs, which typically would be used to
represent scattered spans but might even represent overlapping or
aliasing spans of PFNs.
Because this is intended to be used in vfio_pci_core.c, we also need
to expose the struct vfio_pci_dma_buf in the vfio_pci_priv.h header.
Signed-off-by: Matt Evans <mattev@meta.com>
---
drivers/vfio/pci/vfio_pci_dmabuf.c | 142 ++++++++++++++++++++++++++---
drivers/vfio/pci/vfio_pci_priv.h | 20 ++++
2 files changed, 149 insertions(+), 13 deletions(-)
diff --git a/drivers/vfio/pci/vfio_pci_dmabuf.c b/drivers/vfio/pci/vfio_pci_dmabuf.c
index c16f460c01d6..0d132c4ca95f 100644
--- a/drivers/vfio/pci/vfio_pci_dmabuf.c
+++ b/drivers/vfio/pci/vfio_pci_dmabuf.c
@@ -9,19 +9,6 @@
MODULE_IMPORT_NS("DMA_BUF");
-struct vfio_pci_dma_buf {
- struct dma_buf *dmabuf;
- struct vfio_pci_core_device *vdev;
- struct list_head dmabufs_elm;
- size_t size;
- struct phys_vec *phys_vec;
- struct p2pdma_provider *provider;
- u32 nr_ranges;
- struct kref kref;
- struct completion comp;
- u8 revoked : 1;
-};
-
static int vfio_pci_dma_buf_attach(struct dma_buf *dmabuf,
struct dma_buf_attachment *attachment)
{
@@ -106,6 +93,135 @@ static const struct dma_buf_ops vfio_pci_dmabuf_ops = {
.release = vfio_pci_dma_buf_release,
};
+int vfio_pci_dma_buf_find_pfn(struct vfio_pci_dma_buf *vpdmabuf,
+ struct vm_area_struct *vma,
+ unsigned long address,
+ unsigned int order,
+ unsigned long *out_pfn)
+{
+ /*
+ * Given a VMA (start, end, pgoffs) and a fault address,
+ * search the corresponding DMABUF's phys_vec[] to find the
+ * range representing the address's offset into the VMA, and
+ * its PFN.
+ *
+ * The phys_vec[] ranges represent contiguous spans of VAs
+ * upwards from the buffer offset 0; the actual PFNs might be
+ * in any order, overlap/alias, etc. Calculate an offset of
+ * the desired page given VMA start/pgoff and address, then
+ * search upwards from 0 to find which span contains it.
+ *
+ * On success, a valid PFN for a page sized by 'order' is
+ * returned into out_pfn.
+ *
+ * Failure occurs if:
+ * - The page would cross the edge of the VMA
+ * - The page isn't entirely contained within a range
+ * - We find a range, but the final PFN isn't aligned to the
+ * requested order.
+ *
+ * (Upon failure, the caller is expected to try again with a
+ * smaller order; the tests above will always succeed for
+ * order=0 as the limit case.)
+ *
+ * It's suboptimal if DMABUFs are created with neigbouring
+ * ranges that are physically contiguous, since hugepages
+ * can't straddle range boundaries. (The construction of the
+ * ranges vector should merge such ranges.)
+ *
+ * Finally, vma_pgoff_adjust is used for a DMABUF representing
+ * a VFIO BAR mmap, which is created from the start of the
+ * offset region. It should be zero, or equal vm_pgoff.
+ */
+
+ const unsigned long pagesize = PAGE_SIZE << order;
+ unsigned long vma_off = ((vma->vm_pgoff - vpdmabuf->vma_pgoff_adjust) <<
+ PAGE_SHIFT) & VFIO_PCI_OFFSET_MASK;
+ unsigned long rounded_page_addr = ALIGN_DOWN(address, pagesize);
+ unsigned long rounded_page_end = rounded_page_addr + pagesize;
+ unsigned long page_buf_offset;
+ unsigned long range_buf_offset = 0;
+ unsigned int i;
+
+ if (rounded_page_addr < vma->vm_start || rounded_page_end > vma->vm_end) {
+ if (order > 0)
+ return -EAGAIN;
+
+ /* A fault address outside of the VMA is absurd. */
+ WARN(1, "Fault addr 0x%lx outside VMA 0x%lx-0x%lx\n",
+ address, vma->vm_start, vma->vm_end);
+ return -EFAULT;
+ }
+
+ if (vpdmabuf->vma_pgoff_adjust != 0 &&
+ vpdmabuf->vma_pgoff_adjust != (vma->vm_pgoff &
+ (VFIO_PCI_OFFSET_MASK >> PAGE_SHIFT))) {
+ WARN(1, "Unexpected vma_pgoff_adjust 0x%lx (vm_pgoff 0x%lx)\n",
+ vpdmabuf->vma_pgoff_adjust, vma->vm_pgoff);
+ return -EFAULT;
+ }
+
+ if (unlikely(check_add_overflow(rounded_page_addr - vma->vm_start,
+ vma_off, &page_buf_offset)))
+ return -EFAULT;
+
+ for (i = 0; i < vpdmabuf->nr_ranges; i++) {
+ unsigned long page_buf_offset_end;
+ size_t range_len = vpdmabuf->phys_vec[i].len;
+ phys_addr_t range_start = vpdmabuf->phys_vec[i].paddr;
+
+ if (unlikely(check_add_overflow(page_buf_offset, pagesize,
+ &page_buf_offset_end)))
+ return -EFAULT;
+ /*
+ * If the current range starts after the page's span,
+ * this and any future range won't match. Bail early.
+ */
+ if (page_buf_offset_end <= range_buf_offset)
+ break;
+
+ if (page_buf_offset >= range_buf_offset &&
+ page_buf_offset_end <= range_buf_offset + range_len) {
+ /*
+ * The faulting page is wholly contained
+ * within the span represented by the range.
+ * Validate PFN alignment for the order:
+ */
+ unsigned long pfn = (range_start + page_buf_offset -
+ range_buf_offset) / PAGE_SIZE;
+
+ if (IS_ALIGNED(pfn, 1 << order)) {
+ *out_pfn = pfn;
+ return 0;
+ }
+ /* Retry with smaller order */
+ return -EAGAIN;
+ }
+ range_buf_offset += range_len;
+ }
+
+ /*
+ * A hugepage straddling a range boundary will fail to match a
+ * range, but the address will (eventually) match when retried
+ * with a smaller page.
+ */
+ if (order > 0)
+ return -EAGAIN;
+
+ /*
+ * If we get here, the address fell outside of the span
+ * represented by the (concatenated) ranges. Setup of a
+ * mapping must ensure that the VMA is <= the total size of
+ * the ranges, so this should never happen. But, if it does,
+ * force SIGBUS for the access and warn.
+ */
+ WARN_ONCE(1, "No range for addr 0x%lx, order %d: VMA 0x%lx-0x%lx pgoff 0x%lx, %u ranges, size 0x%zx\n",
+ address, order, vma->vm_start, vma->vm_end, vma->vm_pgoff,
+ vpdmabuf->nr_ranges, vpdmabuf->size);
+
+ return -EFAULT;
+}
+
/*
* This is a temporary "private interconnect" between VFIO DMABUF and iommufd.
* It allows the two co-operating drivers to exchange the physical address of
diff --git a/drivers/vfio/pci/vfio_pci_priv.h b/drivers/vfio/pci/vfio_pci_priv.h
index fca9d0dfac90..c8f6f959056a 100644
--- a/drivers/vfio/pci/vfio_pci_priv.h
+++ b/drivers/vfio/pci/vfio_pci_priv.h
@@ -23,6 +23,20 @@ struct vfio_pci_ioeventfd {
bool test_mem;
};
+struct vfio_pci_dma_buf {
+ struct dma_buf *dmabuf;
+ struct vfio_pci_core_device *vdev;
+ struct list_head dmabufs_elm;
+ size_t size;
+ struct phys_vec *phys_vec;
+ struct p2pdma_provider *provider;
+ u32 nr_ranges;
+ struct kref kref;
+ struct completion comp;
+ unsigned long vma_pgoff_adjust;
+ u8 revoked : 1;
+};
+
bool vfio_pci_intx_mask(struct vfio_pci_core_device *vdev);
void vfio_pci_intx_unmask(struct vfio_pci_core_device *vdev);
@@ -114,6 +128,12 @@ static inline bool vfio_pci_is_vga(struct pci_dev *pdev)
return (pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA;
}
+int vfio_pci_dma_buf_find_pfn(struct vfio_pci_dma_buf *vpdmabuf,
+ struct vm_area_struct *vma,
+ unsigned long address,
+ unsigned int order,
+ unsigned long *out_pfn);
+
#ifdef CONFIG_VFIO_PCI_DMABUF
int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
struct vfio_device_feature_dma_buf __user *arg,
--
2.47.3
^ permalink raw reply related [flat|nested] 25+ messages in thread* Re: [PATCH v2 2/9] vfio/pci: Add a helper to look up PFNs for DMABUFs
2026-05-27 10:23 ` [PATCH v2 2/9] vfio/pci: Add a helper to look up PFNs for DMABUFs Matt Evans
@ 2026-05-27 22:38 ` Alex Williamson
2026-05-28 2:28 ` Claude review: " Claude Code Review Bot
1 sibling, 0 replies; 25+ messages in thread
From: Alex Williamson @ 2026-05-27 22:38 UTC (permalink / raw)
To: Matt Evans
Cc: Leon Romanovsky, Jason Gunthorpe, Alex Mastro,
Christian König, Bjorn Helgaas, Logan Gunthorpe,
Mahmoud Adam, David Matlack, Björn Töpel, Sumit Semwal,
Kevin Tian, Ankit Agrawal, Pranjal Shrivastava, Alistair Popple,
Vivek Kasireddy, linux-kernel, linux-media, dri-devel,
linaro-mm-sig, kvm, linux-pci, alex
On Wed, 27 May 2026 03:23:05 -0700
Matt Evans <mattev@meta.com> wrote:
> Add vfio_pci_dma_buf_find_pfn(), which a VMA fault handler can use to
> find a PFN.
>
> This supports multi-range DMABUFs, which typically would be used to
> represent scattered spans but might even represent overlapping or
> aliasing spans of PFNs.
>
> Because this is intended to be used in vfio_pci_core.c, we also need
> to expose the struct vfio_pci_dma_buf in the vfio_pci_priv.h header.
>
> Signed-off-by: Matt Evans <mattev@meta.com>
> ---
> drivers/vfio/pci/vfio_pci_dmabuf.c | 142 ++++++++++++++++++++++++++---
> drivers/vfio/pci/vfio_pci_priv.h | 20 ++++
> 2 files changed, 149 insertions(+), 13 deletions(-)
>
> diff --git a/drivers/vfio/pci/vfio_pci_dmabuf.c b/drivers/vfio/pci/vfio_pci_dmabuf.c
> index c16f460c01d6..0d132c4ca95f 100644
> --- a/drivers/vfio/pci/vfio_pci_dmabuf.c
> +++ b/drivers/vfio/pci/vfio_pci_dmabuf.c
> @@ -9,19 +9,6 @@
>
> MODULE_IMPORT_NS("DMA_BUF");
>
> -struct vfio_pci_dma_buf {
> - struct dma_buf *dmabuf;
> - struct vfio_pci_core_device *vdev;
> - struct list_head dmabufs_elm;
> - size_t size;
> - struct phys_vec *phys_vec;
> - struct p2pdma_provider *provider;
> - u32 nr_ranges;
> - struct kref kref;
> - struct completion comp;
> - u8 revoked : 1;
> -};
> -
> static int vfio_pci_dma_buf_attach(struct dma_buf *dmabuf,
> struct dma_buf_attachment *attachment)
> {
> @@ -106,6 +93,135 @@ static const struct dma_buf_ops vfio_pci_dmabuf_ops = {
> .release = vfio_pci_dma_buf_release,
> };
>
> +int vfio_pci_dma_buf_find_pfn(struct vfio_pci_dma_buf *vpdmabuf,
This private dmabuf object pointer is named priv pretty consistently
throughout the rest of the file.
> + struct vm_area_struct *vma,
> + unsigned long address,
> + unsigned int order,
> + unsigned long *out_pfn)
> +{
> + /*
> + * Given a VMA (start, end, pgoffs) and a fault address,
> + * search the corresponding DMABUF's phys_vec[] to find the
> + * range representing the address's offset into the VMA, and
> + * its PFN.
> + *
> + * The phys_vec[] ranges represent contiguous spans of VAs
> + * upwards from the buffer offset 0; the actual PFNs might be
> + * in any order, overlap/alias, etc. Calculate an offset of
> + * the desired page given VMA start/pgoff and address, then
> + * search upwards from 0 to find which span contains it.
> + *
> + * On success, a valid PFN for a page sized by 'order' is
> + * returned into out_pfn.
> + *
> + * Failure occurs if:
> + * - The page would cross the edge of the VMA
> + * - The page isn't entirely contained within a range
> + * - We find a range, but the final PFN isn't aligned to the
> + * requested order.
> + *
> + * (Upon failure, the caller is expected to try again with a
> + * smaller order; the tests above will always succeed for
> + * order=0 as the limit case.)
> + *
> + * It's suboptimal if DMABUFs are created with neigbouring
> + * ranges that are physically contiguous, since hugepages
> + * can't straddle range boundaries. (The construction of the
> + * ranges vector should merge such ranges.)
> + *
> + * Finally, vma_pgoff_adjust is used for a DMABUF representing
> + * a VFIO BAR mmap, which is created from the start of the
> + * offset region. It should be zero, or equal vm_pgoff.
> + */
> +
> + const unsigned long pagesize = PAGE_SIZE << order;
> + unsigned long vma_off = ((vma->vm_pgoff - vpdmabuf->vma_pgoff_adjust) <<
> + PAGE_SHIFT) & VFIO_PCI_OFFSET_MASK;
> + unsigned long rounded_page_addr = ALIGN_DOWN(address, pagesize);
> + unsigned long rounded_page_end = rounded_page_addr + pagesize;
> + unsigned long page_buf_offset;
> + unsigned long range_buf_offset = 0;
> + unsigned int i;
> +
> + if (rounded_page_addr < vma->vm_start || rounded_page_end > vma->vm_end) {
> + if (order > 0)
> + return -EAGAIN;
> +
> + /* A fault address outside of the VMA is absurd. */
> + WARN(1, "Fault addr 0x%lx outside VMA 0x%lx-0x%lx\n",
> + address, vma->vm_start, vma->vm_end);
> + return -EFAULT;
> + }
> +
> + if (vpdmabuf->vma_pgoff_adjust != 0 &&
> + vpdmabuf->vma_pgoff_adjust != (vma->vm_pgoff &
> + (VFIO_PCI_OFFSET_MASK >> PAGE_SHIFT))) {
> + WARN(1, "Unexpected vma_pgoff_adjust 0x%lx (vm_pgoff 0x%lx)\n",
> + vpdmabuf->vma_pgoff_adjust, vma->vm_pgoff);
> + return -EFAULT;
> + }
This appears to be user trigger'able, by causing a vma to be split, and
also unnecessary as it seems the math works out correctly for the split
case.
> +
> + if (unlikely(check_add_overflow(rounded_page_addr - vma->vm_start,
> + vma_off, &page_buf_offset)))
> + return -EFAULT;
> +
> + for (i = 0; i < vpdmabuf->nr_ranges; i++) {
> + unsigned long page_buf_offset_end;
> + size_t range_len = vpdmabuf->phys_vec[i].len;
> + phys_addr_t range_start = vpdmabuf->phys_vec[i].paddr;
> +
> + if (unlikely(check_add_overflow(page_buf_offset, pagesize,
> + &page_buf_offset_end)))
> + return -EFAULT;
Why is this inside the loop, the args are invariant of anything
modified in the loop. Thanks,
Alex
> + /*
> + * If the current range starts after the page's span,
> + * this and any future range won't match. Bail early.
> + */
> + if (page_buf_offset_end <= range_buf_offset)
> + break;
> +
> + if (page_buf_offset >= range_buf_offset &&
> + page_buf_offset_end <= range_buf_offset + range_len) {
> + /*
> + * The faulting page is wholly contained
> + * within the span represented by the range.
> + * Validate PFN alignment for the order:
> + */
> + unsigned long pfn = (range_start + page_buf_offset -
> + range_buf_offset) / PAGE_SIZE;
> +
> + if (IS_ALIGNED(pfn, 1 << order)) {
> + *out_pfn = pfn;
> + return 0;
> + }
> + /* Retry with smaller order */
> + return -EAGAIN;
> + }
> + range_buf_offset += range_len;
> + }
> +
> + /*
> + * A hugepage straddling a range boundary will fail to match a
> + * range, but the address will (eventually) match when retried
> + * with a smaller page.
> + */
> + if (order > 0)
> + return -EAGAIN;
> +
> + /*
> + * If we get here, the address fell outside of the span
> + * represented by the (concatenated) ranges. Setup of a
> + * mapping must ensure that the VMA is <= the total size of
> + * the ranges, so this should never happen. But, if it does,
> + * force SIGBUS for the access and warn.
> + */
> + WARN_ONCE(1, "No range for addr 0x%lx, order %d: VMA 0x%lx-0x%lx pgoff 0x%lx, %u ranges, size 0x%zx\n",
> + address, order, vma->vm_start, vma->vm_end, vma->vm_pgoff,
> + vpdmabuf->nr_ranges, vpdmabuf->size);
> +
> + return -EFAULT;
> +}
> +
> /*
> * This is a temporary "private interconnect" between VFIO DMABUF and iommufd.
> * It allows the two co-operating drivers to exchange the physical address of
> diff --git a/drivers/vfio/pci/vfio_pci_priv.h b/drivers/vfio/pci/vfio_pci_priv.h
> index fca9d0dfac90..c8f6f959056a 100644
> --- a/drivers/vfio/pci/vfio_pci_priv.h
> +++ b/drivers/vfio/pci/vfio_pci_priv.h
> @@ -23,6 +23,20 @@ struct vfio_pci_ioeventfd {
> bool test_mem;
> };
>
> +struct vfio_pci_dma_buf {
> + struct dma_buf *dmabuf;
> + struct vfio_pci_core_device *vdev;
> + struct list_head dmabufs_elm;
> + size_t size;
> + struct phys_vec *phys_vec;
> + struct p2pdma_provider *provider;
> + u32 nr_ranges;
> + struct kref kref;
> + struct completion comp;
> + unsigned long vma_pgoff_adjust;
> + u8 revoked : 1;
> +};
> +
> bool vfio_pci_intx_mask(struct vfio_pci_core_device *vdev);
> void vfio_pci_intx_unmask(struct vfio_pci_core_device *vdev);
>
> @@ -114,6 +128,12 @@ static inline bool vfio_pci_is_vga(struct pci_dev *pdev)
> return (pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA;
> }
>
> +int vfio_pci_dma_buf_find_pfn(struct vfio_pci_dma_buf *vpdmabuf,
> + struct vm_area_struct *vma,
> + unsigned long address,
> + unsigned int order,
> + unsigned long *out_pfn);
> +
> #ifdef CONFIG_VFIO_PCI_DMABUF
> int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
> struct vfio_device_feature_dma_buf __user *arg,
^ permalink raw reply [flat|nested] 25+ messages in thread* Claude review: vfio/pci: Add a helper to look up PFNs for DMABUFs
2026-05-27 10:23 ` [PATCH v2 2/9] vfio/pci: Add a helper to look up PFNs for DMABUFs Matt Evans
2026-05-27 22:38 ` Alex Williamson
@ 2026-05-28 2:28 ` Claude Code Review Bot
1 sibling, 0 replies; 25+ messages in thread
From: Claude Code Review Bot @ 2026-05-28 2:28 UTC (permalink / raw)
To: dri-devel-reviews
Patch Review
Adds `vfio_pci_dma_buf_find_pfn()` for translating fault addresses to PFNs via the DMABUF's `phys_vec[]` ranges.
**The comment block is quite long (lines 438-470)**. A 30+ line comment preceding the function duplicates what the code already communicates. Consider trimming to just the non-obvious parts (the `vma_pgoff_adjust` semantics and the hugepage/range-boundary limitation).
**The `vma_pgoff_adjust` field and its WARN**:
```c
+ if (vpdmabuf->vma_pgoff_adjust != 0 &&
+ vpdmabuf->vma_pgoff_adjust != (vma->vm_pgoff &
+ (VFIO_PCI_OFFSET_MASK >> PAGE_SHIFT))) {
+ WARN(1, "Unexpected vma_pgoff_adjust 0x%lx (vm_pgoff 0x%lx)\n",
```
This WARN fires if `vma_pgoff_adjust` is nonzero and doesn't match the VMA's pgoff. The check is a debugging assertion that should arguably be a `WARN_ON_ONCE` rather than `WARN(1, ...)` to avoid spamming dmesg.
**Struct movement**: Moving `struct vfio_pci_dma_buf` to the header and adding `vma_pgoff_adjust` is clean. Good that `vma_pgoff_adjust` is `unsigned long` to match `vm_pgoff`.
---
Generated by Claude Code Patch Reviewer
^ permalink raw reply [flat|nested] 25+ messages in thread
* [PATCH v2 3/9] vfio/pci: Add a helper to create a DMABUF for a BAR-map VMA
2026-05-27 10:23 [PATCH v2 0/9] vfio/pci: Add mmap() for DMABUFs Matt Evans
2026-05-27 10:23 ` [PATCH v2 1/9] PCI/P2PDMA: Add CONFIG_PCI_P2PDMA_CORE Matt Evans
2026-05-27 10:23 ` [PATCH v2 2/9] vfio/pci: Add a helper to look up PFNs for DMABUFs Matt Evans
@ 2026-05-27 10:23 ` Matt Evans
2026-05-27 22:59 ` Alex Williamson
2026-05-28 2:28 ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 4/9] vfio/pci: Convert BAR mmap() to use a DMABUF Matt Evans
` (6 subsequent siblings)
9 siblings, 2 replies; 25+ messages in thread
From: Matt Evans @ 2026-05-27 10:23 UTC (permalink / raw)
To: Alex Williamson, Leon Romanovsky, Jason Gunthorpe, Alex Mastro,
Christian König, Bjorn Helgaas, Logan Gunthorpe
Cc: Mahmoud Adam, David Matlack, Björn Töpel, Sumit Semwal,
Kevin Tian, Ankit Agrawal, Pranjal Shrivastava, Alistair Popple,
Vivek Kasireddy, linux-kernel, linux-media, dri-devel,
linaro-mm-sig, kvm, linux-pci
This helper, vfio_pci_core_mmap_prep_dmabuf(), creates a single-range
DMABUF for the purpose of mapping a PCI BAR. This is used in a future
commit by VFIO's ordinary mmap() path.
This function transfers ownership of the VFIO device fd to the
DMABUF, which fput()s when it's released.
Refactor the existing vfio_pci_core_feature_dma_buf() to split out
export code common to the two paths, VFIO_DEVICE_FEATURE_DMA_BUF and
this new VFIO_BAR mmap().
Signed-off-by: Matt Evans <mattev@meta.com>
---
drivers/vfio/pci/vfio_pci_dmabuf.c | 140 ++++++++++++++++++++++-------
drivers/vfio/pci/vfio_pci_priv.h | 5 ++
2 files changed, 115 insertions(+), 30 deletions(-)
diff --git a/drivers/vfio/pci/vfio_pci_dmabuf.c b/drivers/vfio/pci/vfio_pci_dmabuf.c
index 0d132c4ca95f..782408c08a5e 100644
--- a/drivers/vfio/pci/vfio_pci_dmabuf.c
+++ b/drivers/vfio/pci/vfio_pci_dmabuf.c
@@ -82,6 +82,8 @@ static void vfio_pci_dma_buf_release(struct dma_buf *dmabuf)
up_write(&priv->vdev->memory_lock);
vfio_device_put_registration(&priv->vdev->vdev);
}
+ if (priv->vfile)
+ fput(priv->vfile);
kfree(priv->phys_vec);
kfree(priv);
}
@@ -222,6 +224,45 @@ int vfio_pci_dma_buf_find_pfn(struct vfio_pci_dma_buf *vpdmabuf,
return -EFAULT;
}
+/*
+ * Create a DMABUF corresponding to priv, add it to vdev->dmabufs list
+ * for tracking (meaning cleanup or revocation will zap it), and take
+ * a vfio_device registration.
+ */
+static int vfio_pci_dmabuf_export(struct vfio_pci_core_device *vdev,
+ struct vfio_pci_dma_buf *priv, uint32_t flags)
+{
+ DEFINE_DMA_BUF_EXPORT_INFO(exp_info);
+
+ if (!vfio_device_try_get_registration(&vdev->vdev))
+ return -ENODEV;
+
+ exp_info.ops = &vfio_pci_dmabuf_ops;
+ exp_info.size = priv->size;
+ exp_info.flags = flags;
+ exp_info.priv = priv;
+
+ priv->dmabuf = dma_buf_export(&exp_info);
+ if (IS_ERR(priv->dmabuf)) {
+ vfio_device_put_registration(&vdev->vdev);
+ return PTR_ERR(priv->dmabuf);
+ }
+
+ kref_init(&priv->kref);
+ init_completion(&priv->comp);
+
+ /* dma_buf_put() now frees priv */
+ INIT_LIST_HEAD(&priv->dmabufs_elm);
+ down_write(&vdev->memory_lock);
+ dma_resv_lock(priv->dmabuf->resv, NULL);
+ priv->revoked = !__vfio_pci_memory_enabled(vdev);
+ list_add_tail(&priv->dmabufs_elm, &vdev->dmabufs);
+ dma_resv_unlock(priv->dmabuf->resv);
+ up_write(&vdev->memory_lock);
+
+ return 0;
+}
+
/*
* This is a temporary "private interconnect" between VFIO DMABUF and iommufd.
* It allows the two co-operating drivers to exchange the physical address of
@@ -340,7 +381,6 @@ int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
{
struct vfio_device_feature_dma_buf get_dma_buf = {};
struct vfio_region_dma_range *dma_ranges;
- DEFINE_DMA_BUF_EXPORT_INFO(exp_info);
struct vfio_pci_dma_buf *priv;
size_t length;
int ret;
@@ -400,34 +440,9 @@ int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
kfree(dma_ranges);
dma_ranges = NULL;
- if (!vfio_device_try_get_registration(&vdev->vdev)) {
- ret = -ENODEV;
+ ret = vfio_pci_dmabuf_export(vdev, priv, get_dma_buf.open_flags);
+ if (ret)
goto err_free_phys;
- }
-
- exp_info.ops = &vfio_pci_dmabuf_ops;
- exp_info.size = priv->size;
- exp_info.flags = get_dma_buf.open_flags;
- exp_info.priv = priv;
-
- priv->dmabuf = dma_buf_export(&exp_info);
- if (IS_ERR(priv->dmabuf)) {
- ret = PTR_ERR(priv->dmabuf);
- goto err_dev_put;
- }
-
- kref_init(&priv->kref);
- init_completion(&priv->comp);
-
- /* dma_buf_put() now frees priv */
- INIT_LIST_HEAD(&priv->dmabufs_elm);
- down_write(&vdev->memory_lock);
- dma_resv_lock(priv->dmabuf->resv, NULL);
- priv->revoked = !__vfio_pci_memory_enabled(vdev);
- list_add_tail(&priv->dmabufs_elm, &vdev->dmabufs);
- dma_resv_unlock(priv->dmabuf->resv);
- up_write(&vdev->memory_lock);
-
/*
* dma_buf_fd() consumes the reference, when the file closes the dmabuf
* will be released.
@@ -438,8 +453,6 @@ int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
return ret;
-err_dev_put:
- vfio_device_put_registration(&vdev->vdev);
err_free_phys:
kfree(priv->phys_vec);
err_free_priv:
@@ -449,6 +462,73 @@ int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
return ret;
}
+int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
+ struct vm_area_struct *vma,
+ u64 phys_start, u64 req_len,
+ unsigned int res_index)
+{
+ struct vfio_pci_dma_buf *priv;
+ const unsigned int nr_ranges = 1;
+ unsigned long vma_pgoff = vma->vm_pgoff & (VFIO_PCI_OFFSET_MASK >> PAGE_SHIFT);
+ int ret;
+
+ priv = kzalloc_obj(*priv);
+ if (!priv)
+ return -ENOMEM;
+
+ priv->phys_vec = kzalloc_obj(*priv->phys_vec);
+ if (!priv->phys_vec) {
+ ret = -ENOMEM;
+ goto err_free_priv;
+ }
+
+ /*
+ * The DMABUF begins from the mmap()'s BAR offset, i.e. the
+ * start of the VMA corresponds to byte 0 of the DMABUF and
+ * byte (vma_pgoff << PAGE_SHIFT) of the BAR.
+ *
+ * vfio_pci_dma_buf_find_pfn() reverses this offset using
+ * vma_pgoff_adjust, so that ultimately a fault's offset from
+ * the start of the _VMA_ has a consistent usage whether the
+ * VMA originates from an mmap() of the VFIO device here or a
+ * direct DMABUF mmap().
+ */
+ priv->vdev = vdev;
+ priv->size = req_len;
+ priv->nr_ranges = nr_ranges;
+ priv->vma_pgoff_adjust = vma_pgoff;
+ priv->provider = pcim_p2pdma_provider(vdev->pdev, res_index);
+ if (!priv->provider) {
+ ret = -EINVAL;
+ goto err_free_phys;
+ }
+
+ priv->phys_vec[0].paddr = phys_start + ((u64)vma_pgoff << PAGE_SHIFT);
+ priv->phys_vec[0].len = priv->size;
+
+ ret = vfio_pci_dmabuf_export(vdev, priv, O_CLOEXEC | O_RDWR);
+ if (ret)
+ goto err_free_phys;
+
+ /*
+ * The VMA gets the DMABUF file so that other users can locate
+ * the DMABUF via a VA. Ownership of the original VFIO device
+ * file being mmap()ed transfers to priv, and is put when the
+ * DMABUF is released.
+ */
+ priv->vfile = vma->vm_file;
+ vma->vm_file = priv->dmabuf->file;
+ vma->vm_private_data = priv;
+
+ return 0;
+
+err_free_phys:
+ kfree(priv->phys_vec);
+err_free_priv:
+ kfree(priv);
+ return ret;
+}
+
void vfio_pci_dma_buf_move(struct vfio_pci_core_device *vdev, bool revoked)
{
struct vfio_pci_dma_buf *priv;
diff --git a/drivers/vfio/pci/vfio_pci_priv.h b/drivers/vfio/pci/vfio_pci_priv.h
index c8f6f959056a..06dc0fd3e230 100644
--- a/drivers/vfio/pci/vfio_pci_priv.h
+++ b/drivers/vfio/pci/vfio_pci_priv.h
@@ -30,6 +30,7 @@ struct vfio_pci_dma_buf {
size_t size;
struct phys_vec *phys_vec;
struct p2pdma_provider *provider;
+ struct file *vfile;
u32 nr_ranges;
struct kref kref;
struct completion comp;
@@ -133,6 +134,10 @@ int vfio_pci_dma_buf_find_pfn(struct vfio_pci_dma_buf *vpdmabuf,
unsigned long address,
unsigned int order,
unsigned long *out_pfn);
+int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
+ struct vm_area_struct *vma,
+ u64 phys_start, u64 req_len,
+ unsigned int res_index);
#ifdef CONFIG_VFIO_PCI_DMABUF
int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
--
2.47.3
^ permalink raw reply related [flat|nested] 25+ messages in thread* Re: [PATCH v2 3/9] vfio/pci: Add a helper to create a DMABUF for a BAR-map VMA
2026-05-27 10:23 ` [PATCH v2 3/9] vfio/pci: Add a helper to create a DMABUF for a BAR-map VMA Matt Evans
@ 2026-05-27 22:59 ` Alex Williamson
2026-05-28 2:28 ` Claude review: " Claude Code Review Bot
1 sibling, 0 replies; 25+ messages in thread
From: Alex Williamson @ 2026-05-27 22:59 UTC (permalink / raw)
To: Matt Evans
Cc: Leon Romanovsky, Jason Gunthorpe, Alex Mastro,
Christian König, Bjorn Helgaas, Logan Gunthorpe,
Mahmoud Adam, David Matlack, Björn Töpel, Sumit Semwal,
Kevin Tian, Ankit Agrawal, Pranjal Shrivastava, Alistair Popple,
Vivek Kasireddy, linux-kernel, linux-media, dri-devel,
linaro-mm-sig, kvm, linux-pci, alex
On Wed, 27 May 2026 03:23:06 -0700
Matt Evans <mattev@meta.com> wrote:
> This helper, vfio_pci_core_mmap_prep_dmabuf(), creates a single-range
> DMABUF for the purpose of mapping a PCI BAR. This is used in a future
> commit by VFIO's ordinary mmap() path.
>
> This function transfers ownership of the VFIO device fd to the
> DMABUF, which fput()s when it's released.
>
> Refactor the existing vfio_pci_core_feature_dma_buf() to split out
> export code common to the two paths, VFIO_DEVICE_FEATURE_DMA_BUF and
> this new VFIO_BAR mmap().
>
> Signed-off-by: Matt Evans <mattev@meta.com>
> ---
> drivers/vfio/pci/vfio_pci_dmabuf.c | 140 ++++++++++++++++++++++-------
> drivers/vfio/pci/vfio_pci_priv.h | 5 ++
> 2 files changed, 115 insertions(+), 30 deletions(-)
>
> diff --git a/drivers/vfio/pci/vfio_pci_dmabuf.c b/drivers/vfio/pci/vfio_pci_dmabuf.c
> index 0d132c4ca95f..782408c08a5e 100644
> --- a/drivers/vfio/pci/vfio_pci_dmabuf.c
> +++ b/drivers/vfio/pci/vfio_pci_dmabuf.c
> @@ -82,6 +82,8 @@ static void vfio_pci_dma_buf_release(struct dma_buf *dmabuf)
> up_write(&priv->vdev->memory_lock);
> vfio_device_put_registration(&priv->vdev->vdev);
> }
> + if (priv->vfile)
> + fput(priv->vfile);
> kfree(priv->phys_vec);
> kfree(priv);
> }
> @@ -222,6 +224,45 @@ int vfio_pci_dma_buf_find_pfn(struct vfio_pci_dma_buf *vpdmabuf,
> return -EFAULT;
> }
>
> +/*
> + * Create a DMABUF corresponding to priv, add it to vdev->dmabufs list
> + * for tracking (meaning cleanup or revocation will zap it), and take
> + * a vfio_device registration.
> + */
> +static int vfio_pci_dmabuf_export(struct vfio_pci_core_device *vdev,
> + struct vfio_pci_dma_buf *priv, uint32_t flags)
s/uint32_t/u32/?
> +{
> + DEFINE_DMA_BUF_EXPORT_INFO(exp_info);
> +
> + if (!vfio_device_try_get_registration(&vdev->vdev))
> + return -ENODEV;
> +
> + exp_info.ops = &vfio_pci_dmabuf_ops;
> + exp_info.size = priv->size;
> + exp_info.flags = flags;
> + exp_info.priv = priv;
> +
> + priv->dmabuf = dma_buf_export(&exp_info);
> + if (IS_ERR(priv->dmabuf)) {
> + vfio_device_put_registration(&vdev->vdev);
> + return PTR_ERR(priv->dmabuf);
> + }
> +
> + kref_init(&priv->kref);
> + init_completion(&priv->comp);
> +
> + /* dma_buf_put() now frees priv */
> + INIT_LIST_HEAD(&priv->dmabufs_elm);
> + down_write(&vdev->memory_lock);
> + dma_resv_lock(priv->dmabuf->resv, NULL);
> + priv->revoked = !__vfio_pci_memory_enabled(vdev);
> + list_add_tail(&priv->dmabufs_elm, &vdev->dmabufs);
> + dma_resv_unlock(priv->dmabuf->resv);
> + up_write(&vdev->memory_lock);
> +
> + return 0;
> +}
> +
> /*
> * This is a temporary "private interconnect" between VFIO DMABUF and iommufd.
> * It allows the two co-operating drivers to exchange the physical address of
> @@ -340,7 +381,6 @@ int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
> {
> struct vfio_device_feature_dma_buf get_dma_buf = {};
> struct vfio_region_dma_range *dma_ranges;
> - DEFINE_DMA_BUF_EXPORT_INFO(exp_info);
> struct vfio_pci_dma_buf *priv;
> size_t length;
> int ret;
> @@ -400,34 +440,9 @@ int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
> kfree(dma_ranges);
> dma_ranges = NULL;
>
> - if (!vfio_device_try_get_registration(&vdev->vdev)) {
> - ret = -ENODEV;
> + ret = vfio_pci_dmabuf_export(vdev, priv, get_dma_buf.open_flags);
> + if (ret)
> goto err_free_phys;
> - }
> -
> - exp_info.ops = &vfio_pci_dmabuf_ops;
> - exp_info.size = priv->size;
> - exp_info.flags = get_dma_buf.open_flags;
> - exp_info.priv = priv;
> -
> - priv->dmabuf = dma_buf_export(&exp_info);
> - if (IS_ERR(priv->dmabuf)) {
> - ret = PTR_ERR(priv->dmabuf);
> - goto err_dev_put;
> - }
> -
> - kref_init(&priv->kref);
> - init_completion(&priv->comp);
> -
> - /* dma_buf_put() now frees priv */
> - INIT_LIST_HEAD(&priv->dmabufs_elm);
> - down_write(&vdev->memory_lock);
> - dma_resv_lock(priv->dmabuf->resv, NULL);
> - priv->revoked = !__vfio_pci_memory_enabled(vdev);
> - list_add_tail(&priv->dmabufs_elm, &vdev->dmabufs);
> - dma_resv_unlock(priv->dmabuf->resv);
> - up_write(&vdev->memory_lock);
> -
> /*
> * dma_buf_fd() consumes the reference, when the file closes the dmabuf
> * will be released.
> @@ -438,8 +453,6 @@ int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
>
> return ret;
>
> -err_dev_put:
> - vfio_device_put_registration(&vdev->vdev);
> err_free_phys:
> kfree(priv->phys_vec);
> err_free_priv:
> @@ -449,6 +462,73 @@ int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
> return ret;
> }
>
> +int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
> + struct vm_area_struct *vma,
> + u64 phys_start, u64 req_len,
> + unsigned int res_index)
> +{
> + struct vfio_pci_dma_buf *priv;
> + const unsigned int nr_ranges = 1;
Why, versus priv->nr_ranges = 1; below? Thanks,
Alex
> + unsigned long vma_pgoff = vma->vm_pgoff & (VFIO_PCI_OFFSET_MASK >> PAGE_SHIFT);
> + int ret;
> +
> + priv = kzalloc_obj(*priv);
> + if (!priv)
> + return -ENOMEM;
> +
> + priv->phys_vec = kzalloc_obj(*priv->phys_vec);
> + if (!priv->phys_vec) {
> + ret = -ENOMEM;
> + goto err_free_priv;
> + }
> +
> + /*
> + * The DMABUF begins from the mmap()'s BAR offset, i.e. the
> + * start of the VMA corresponds to byte 0 of the DMABUF and
> + * byte (vma_pgoff << PAGE_SHIFT) of the BAR.
> + *
> + * vfio_pci_dma_buf_find_pfn() reverses this offset using
> + * vma_pgoff_adjust, so that ultimately a fault's offset from
> + * the start of the _VMA_ has a consistent usage whether the
> + * VMA originates from an mmap() of the VFIO device here or a
> + * direct DMABUF mmap().
> + */
> + priv->vdev = vdev;
> + priv->size = req_len;
> + priv->nr_ranges = nr_ranges;
> + priv->vma_pgoff_adjust = vma_pgoff;
> + priv->provider = pcim_p2pdma_provider(vdev->pdev, res_index);
> + if (!priv->provider) {
> + ret = -EINVAL;
> + goto err_free_phys;
> + }
> +
> + priv->phys_vec[0].paddr = phys_start + ((u64)vma_pgoff << PAGE_SHIFT);
> + priv->phys_vec[0].len = priv->size;
> +
> + ret = vfio_pci_dmabuf_export(vdev, priv, O_CLOEXEC | O_RDWR);
> + if (ret)
> + goto err_free_phys;
> +
> + /*
> + * The VMA gets the DMABUF file so that other users can locate
> + * the DMABUF via a VA. Ownership of the original VFIO device
> + * file being mmap()ed transfers to priv, and is put when the
> + * DMABUF is released.
> + */
> + priv->vfile = vma->vm_file;
> + vma->vm_file = priv->dmabuf->file;
> + vma->vm_private_data = priv;
> +
> + return 0;
> +
> +err_free_phys:
> + kfree(priv->phys_vec);
> +err_free_priv:
> + kfree(priv);
> + return ret;
> +}
> +
> void vfio_pci_dma_buf_move(struct vfio_pci_core_device *vdev, bool revoked)
> {
> struct vfio_pci_dma_buf *priv;
> diff --git a/drivers/vfio/pci/vfio_pci_priv.h b/drivers/vfio/pci/vfio_pci_priv.h
> index c8f6f959056a..06dc0fd3e230 100644
> --- a/drivers/vfio/pci/vfio_pci_priv.h
> +++ b/drivers/vfio/pci/vfio_pci_priv.h
> @@ -30,6 +30,7 @@ struct vfio_pci_dma_buf {
> size_t size;
> struct phys_vec *phys_vec;
> struct p2pdma_provider *provider;
> + struct file *vfile;
> u32 nr_ranges;
> struct kref kref;
> struct completion comp;
> @@ -133,6 +134,10 @@ int vfio_pci_dma_buf_find_pfn(struct vfio_pci_dma_buf *vpdmabuf,
> unsigned long address,
> unsigned int order,
> unsigned long *out_pfn);
> +int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
> + struct vm_area_struct *vma,
> + u64 phys_start, u64 req_len,
> + unsigned int res_index);
>
> #ifdef CONFIG_VFIO_PCI_DMABUF
> int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
^ permalink raw reply [flat|nested] 25+ messages in thread* Claude review: vfio/pci: Add a helper to create a DMABUF for a BAR-map VMA
2026-05-27 10:23 ` [PATCH v2 3/9] vfio/pci: Add a helper to create a DMABUF for a BAR-map VMA Matt Evans
2026-05-27 22:59 ` Alex Williamson
@ 2026-05-28 2:28 ` Claude Code Review Bot
1 sibling, 0 replies; 25+ messages in thread
From: Claude Code Review Bot @ 2026-05-28 2:28 UTC (permalink / raw)
To: dri-devel-reviews
Patch Review
Refactors DMABUF export into `vfio_pci_dmabuf_export()` and adds `vfio_pci_core_mmap_prep_dmabuf()`.
**File ownership transfer**: The pattern of swapping `vma->vm_file`:
```c
+ priv->vfile = vma->vm_file;
+ vma->vm_file = priv->dmabuf->file;
```
and adding `fput(priv->vfile)` in release is correct but subtle. The `vfile` field is added to the struct but only used in the BAR mmap path (not the explicit export path). Consider adding a brief comment at the `fput` in `vfio_pci_dma_buf_release`:
```c
+ if (priv->vfile)
+ fput(priv->vfile);
```
to note this is the original VFIO device fd that was replaced in `vm_file`.
**`kzalloc_obj`**: The use of `kzalloc_obj(*priv)` and `kzalloc_obj(*priv->phys_vec)` -- I'm not sure this macro exists in mainline. If this is a Meta-internal helper, it won't compile upstream. This should be `kzalloc(sizeof(*priv), GFP_KERNEL)`.
**`get_file_rcu` / file refcounting**: The `vma->vm_file` swap happens without holding any particular lock beyond the mmap_lock that the kernel holds during `mmap()`. This should be safe since the VMA hasn't been installed yet at this point.
---
Generated by Claude Code Patch Reviewer
^ permalink raw reply [flat|nested] 25+ messages in thread
* [PATCH v2 4/9] vfio/pci: Convert BAR mmap() to use a DMABUF
2026-05-27 10:23 [PATCH v2 0/9] vfio/pci: Add mmap() for DMABUFs Matt Evans
` (2 preceding siblings ...)
2026-05-27 10:23 ` [PATCH v2 3/9] vfio/pci: Add a helper to create a DMABUF for a BAR-map VMA Matt Evans
@ 2026-05-27 10:23 ` Matt Evans
2026-05-28 2:28 ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 5/9] vfio/pci: Provide a user-facing name for BAR mappings Matt Evans
` (5 subsequent siblings)
9 siblings, 1 reply; 25+ messages in thread
From: Matt Evans @ 2026-05-27 10:23 UTC (permalink / raw)
To: Alex Williamson, Leon Romanovsky, Jason Gunthorpe, Alex Mastro,
Christian König, Bjorn Helgaas, Logan Gunthorpe
Cc: Mahmoud Adam, David Matlack, Björn Töpel, Sumit Semwal,
Kevin Tian, Ankit Agrawal, Pranjal Shrivastava, Alistair Popple,
Vivek Kasireddy, linux-kernel, linux-media, dri-devel,
linaro-mm-sig, kvm, linux-pci
Convert the VFIO device fd fops->mmap to create a DMABUF representing
the BAR mapping, and make the VMA fault handler look up PFNs from the
corresponding DMABUF. This supports future code mmap()ing BAR
DMABUFs, and iommufd work to support Type1 P2P.
First, vfio_pci_core_mmap() uses the new
vfio_pci_core_mmap_prep_dmabuf() helper to export a DMABUF
representing a single BAR range. Then, the vfio_pci_mmap_huge_fault()
callback is updated to understand revoked buffers, and uses the new
vfio_pci_dma_buf_find_pfn() helper to determine the PFN for a given
fault address.
Now that the VFIO DMABUFs can be mmap()ed, vfio_pci_dma_buf_move()
zaps PTEs (used on the revocation and cleanup paths).
CONFIG_VFIO_PCI_CORE now unconditionally depends on
CONFIG_DMA_SHARED_BUFFER and CONFIG_PCI_P2PDMA_CORE. The
CONFIG_VFIO_PCI_DMABUF feature conditionally includes support for
VFIO_DEVICE_FEATURE_DMA_BUF, depending on the availability of
CONFIG_PCI_P2PDMA.
Signed-off-by: Matt Evans <mattev@meta.com>
---
drivers/vfio/pci/Kconfig | 4 +-
drivers/vfio/pci/Makefile | 3 +-
drivers/vfio/pci/vfio_pci_core.c | 79 +++++++++++++++++++-----------
drivers/vfio/pci/vfio_pci_dmabuf.c | 12 +++++
drivers/vfio/pci/vfio_pci_priv.h | 11 +----
5 files changed, 68 insertions(+), 41 deletions(-)
diff --git a/drivers/vfio/pci/Kconfig b/drivers/vfio/pci/Kconfig
index 296bf01e185e..9197343a7301 100644
--- a/drivers/vfio/pci/Kconfig
+++ b/drivers/vfio/pci/Kconfig
@@ -6,6 +6,8 @@ config VFIO_PCI_CORE
tristate
select VFIO_VIRQFD
select IRQ_BYPASS_MANAGER
+ select PCI_P2PDMA_CORE
+ select DMA_SHARED_BUFFER
config VFIO_PCI_INTX
def_bool y if !S390
@@ -56,7 +58,7 @@ config VFIO_PCI_ZDEV_KVM
To enable s390x KVM vfio-pci extensions, say Y.
config VFIO_PCI_DMABUF
- def_bool y if VFIO_PCI_CORE && PCI_P2PDMA && DMA_SHARED_BUFFER
+ def_bool y if PCI_P2PDMA
source "drivers/vfio/pci/mlx5/Kconfig"
diff --git a/drivers/vfio/pci/Makefile b/drivers/vfio/pci/Makefile
index 6138f1bf241d..881452ea89be 100644
--- a/drivers/vfio/pci/Makefile
+++ b/drivers/vfio/pci/Makefile
@@ -1,8 +1,7 @@
# SPDX-License-Identifier: GPL-2.0-only
-vfio-pci-core-y := vfio_pci_core.o vfio_pci_intrs.o vfio_pci_rdwr.o vfio_pci_config.o
+vfio-pci-core-y := vfio_pci_core.o vfio_pci_intrs.o vfio_pci_rdwr.o vfio_pci_config.o vfio_pci_dmabuf.o
vfio-pci-core-$(CONFIG_VFIO_PCI_ZDEV_KVM) += vfio_pci_zdev.o
-vfio-pci-core-$(CONFIG_VFIO_PCI_DMABUF) += vfio_pci_dmabuf.o
obj-$(CONFIG_VFIO_PCI_CORE) += vfio-pci-core.o
vfio-pci-y := vfio_pci.o
diff --git a/drivers/vfio/pci/vfio_pci_core.c b/drivers/vfio/pci/vfio_pci_core.c
index 041243a84d81..c5f934905ce0 100644
--- a/drivers/vfio/pci/vfio_pci_core.c
+++ b/drivers/vfio/pci/vfio_pci_core.c
@@ -1683,18 +1683,6 @@ void vfio_pci_memory_unlock_and_restore(struct vfio_pci_core_device *vdev, u16 c
up_write(&vdev->memory_lock);
}
-static unsigned long vma_to_pfn(struct vm_area_struct *vma)
-{
- struct vfio_pci_core_device *vdev = vma->vm_private_data;
- int index = vma->vm_pgoff >> (VFIO_PCI_OFFSET_SHIFT - PAGE_SHIFT);
- u64 pgoff;
-
- pgoff = vma->vm_pgoff &
- ((1U << (VFIO_PCI_OFFSET_SHIFT - PAGE_SHIFT)) - 1);
-
- return (pci_resource_start(vdev->pdev, index) >> PAGE_SHIFT) + pgoff;
-}
-
vm_fault_t vfio_pci_vmf_insert_pfn(struct vfio_pci_core_device *vdev,
struct vm_fault *vmf,
unsigned long pfn,
@@ -1722,23 +1710,42 @@ static vm_fault_t vfio_pci_mmap_huge_fault(struct vm_fault *vmf,
unsigned int order)
{
struct vm_area_struct *vma = vmf->vma;
- struct vfio_pci_core_device *vdev = vma->vm_private_data;
- unsigned long addr = vmf->address & ~((PAGE_SIZE << order) - 1);
- unsigned long pgoff = (addr - vma->vm_start) >> PAGE_SHIFT;
- unsigned long pfn = vma_to_pfn(vma) + pgoff;
- vm_fault_t ret = VM_FAULT_FALLBACK;
-
- if (is_aligned_for_order(vma, addr, pfn, order)) {
- scoped_guard(rwsem_read, &vdev->memory_lock)
- ret = vfio_pci_vmf_insert_pfn(vdev, vmf, pfn, order);
- }
+ struct vfio_pci_dma_buf *priv = vma->vm_private_data;
+ struct vfio_pci_core_device *vdev;
+ unsigned long pfn = 0;
+ vm_fault_t ret = VM_FAULT_SIGBUS;
- dev_dbg_ratelimited(&vdev->pdev->dev,
- "%s(,order = %d) BAR %ld page offset 0x%lx: 0x%x\n",
- __func__, order,
- vma->vm_pgoff >>
- (VFIO_PCI_OFFSET_SHIFT - PAGE_SHIFT),
- pgoff, (unsigned int)ret);
+ /*
+ * We can rely on the existence of both a DMABUF (priv) and
+ * the VFIO device it was exported from (vdev). This fault's
+ * VMA was established using vfio_pci_core_mmap_prep_dmabuf()
+ * which transfers ownership of the VFIO device fd to the
+ * DMABUF, and so the VFIO device is held open because the
+ * VMA's vm_file (DMABUF) is open.
+ *
+ * Since vfio_pci_dma_buf_cleanup() cannot have happened,
+ * vdev must be valid; we can take memory_lock.
+ */
+ vdev = READ_ONCE(priv->vdev);
+
+ scoped_guard(rwsem_read, &vdev->memory_lock) {
+ if (!priv->revoked) {
+ int pres = vfio_pci_dma_buf_find_pfn(priv, vma,
+ vmf->address,
+ order, &pfn);
+
+ if (pres == 0)
+ ret = vfio_pci_vmf_insert_pfn(vdev, vmf,
+ pfn, order);
+ else if (pres == -EAGAIN)
+ ret = VM_FAULT_FALLBACK;
+ }
+
+ dev_dbg_ratelimited(&vdev->pdev->dev,
+ "%s(order = %d) PFN 0x%lx, VA 0x%lx, pgoff 0x%lx: 0x%x\n",
+ __func__, order, pfn, vmf->address,
+ vma->vm_pgoff, (unsigned int)ret);
+ }
return ret;
}
@@ -1763,6 +1770,7 @@ int vfio_pci_core_mmap(struct vfio_device *core_vdev, struct vm_area_struct *vma
unsigned int index;
u64 phys_len, req_len, pgoff, req_start;
void __iomem *bar_io;
+ int ret;
index = vma->vm_pgoff >> (VFIO_PCI_OFFSET_SHIFT - PAGE_SHIFT);
@@ -1802,7 +1810,20 @@ int vfio_pci_core_mmap(struct vfio_device *core_vdev, struct vm_area_struct *vma
if (IS_ERR(bar_io))
return PTR_ERR(bar_io);
- vma->vm_private_data = vdev;
+ /*
+ * Create a DMABUF with a single range corresponding to this
+ * mapping, and wire it into vma->vm_private_data. The VMA's
+ * vm_file becomes that of the DMABUF, and the DMABUF takes
+ * ownership of the VFIO device file (put upon DMABUF
+ * release). This maintains the behaviour of a live VMA
+ * mapping holding the VFIO device file open.
+ */
+ ret = vfio_pci_core_mmap_prep_dmabuf(vdev, vma,
+ pci_resource_start(pdev, index),
+ req_len, index);
+ if (ret)
+ return ret;
+
vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
vma->vm_page_prot = pgprot_decrypted(vma->vm_page_prot);
diff --git a/drivers/vfio/pci/vfio_pci_dmabuf.c b/drivers/vfio/pci/vfio_pci_dmabuf.c
index 782408c08a5e..f7797f58d44b 100644
--- a/drivers/vfio/pci/vfio_pci_dmabuf.c
+++ b/drivers/vfio/pci/vfio_pci_dmabuf.c
@@ -9,6 +9,7 @@
MODULE_IMPORT_NS("DMA_BUF");
+#ifdef CONFIG_VFIO_PCI_DMABUF
static int vfio_pci_dma_buf_attach(struct dma_buf *dmabuf,
struct dma_buf_attachment *attachment)
{
@@ -25,6 +26,7 @@ static int vfio_pci_dma_buf_attach(struct dma_buf *dmabuf,
return 0;
}
+#endif /* CONFIG_VFIO_PCI_DMABUF */
static void vfio_pci_dma_buf_done(struct kref *kref)
{
@@ -89,7 +91,9 @@ static void vfio_pci_dma_buf_release(struct dma_buf *dmabuf)
}
static const struct dma_buf_ops vfio_pci_dmabuf_ops = {
+#ifdef CONFIG_VFIO_PCI_DMABUF
.attach = vfio_pci_dma_buf_attach,
+#endif
.map_dma_buf = vfio_pci_dma_buf_map,
.unmap_dma_buf = vfio_pci_dma_buf_unmap,
.release = vfio_pci_dma_buf_release,
@@ -263,6 +267,7 @@ static int vfio_pci_dmabuf_export(struct vfio_pci_core_device *vdev,
return 0;
}
+#ifdef CONFIG_VFIO_PCI_DMABUF
/*
* This is a temporary "private interconnect" between VFIO DMABUF and iommufd.
* It allows the two co-operating drivers to exchange the physical address of
@@ -461,6 +466,7 @@ int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
kfree(dma_ranges);
return ret;
}
+#endif /* CONFIG_VFIO_PCI_DMABUF */
int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
struct vm_area_struct *vma,
@@ -535,6 +541,10 @@ void vfio_pci_dma_buf_move(struct vfio_pci_core_device *vdev, bool revoked)
struct vfio_pci_dma_buf *tmp;
lockdep_assert_held_write(&vdev->memory_lock);
+ /*
+ * Holding memory_lock ensures a racing VMA fault observes
+ * priv->revoked properly.
+ */
list_for_each_entry_safe(priv, tmp, &vdev->dmabufs, dmabufs_elm) {
if (!get_file_active(&priv->dmabuf->file))
@@ -552,6 +562,8 @@ void vfio_pci_dma_buf_move(struct vfio_pci_core_device *vdev, bool revoked)
if (revoked) {
kref_put(&priv->kref, vfio_pci_dma_buf_done);
wait_for_completion(&priv->comp);
+ unmap_mapping_range(priv->dmabuf->file->f_mapping,
+ 0, priv->size, 1);
/*
* Re-arm the registered kref reference and the
* completion so the post-revoke state matches the
diff --git a/drivers/vfio/pci/vfio_pci_priv.h b/drivers/vfio/pci/vfio_pci_priv.h
index 06dc0fd3e230..d38e1b98b2e9 100644
--- a/drivers/vfio/pci/vfio_pci_priv.h
+++ b/drivers/vfio/pci/vfio_pci_priv.h
@@ -138,13 +138,13 @@ int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
struct vm_area_struct *vma,
u64 phys_start, u64 req_len,
unsigned int res_index);
+void vfio_pci_dma_buf_cleanup(struct vfio_pci_core_device *vdev);
+void vfio_pci_dma_buf_move(struct vfio_pci_core_device *vdev, bool revoked);
#ifdef CONFIG_VFIO_PCI_DMABUF
int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
struct vfio_device_feature_dma_buf __user *arg,
size_t argsz);
-void vfio_pci_dma_buf_cleanup(struct vfio_pci_core_device *vdev);
-void vfio_pci_dma_buf_move(struct vfio_pci_core_device *vdev, bool revoked);
#else
static inline int
vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
@@ -153,13 +153,6 @@ vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
{
return -ENOTTY;
}
-static inline void vfio_pci_dma_buf_cleanup(struct vfio_pci_core_device *vdev)
-{
-}
-static inline void vfio_pci_dma_buf_move(struct vfio_pci_core_device *vdev,
- bool revoked)
-{
-}
#endif
#endif
--
2.47.3
^ permalink raw reply related [flat|nested] 25+ messages in thread* Claude review: vfio/pci: Convert BAR mmap() to use a DMABUF
2026-05-27 10:23 ` [PATCH v2 4/9] vfio/pci: Convert BAR mmap() to use a DMABUF Matt Evans
@ 2026-05-28 2:28 ` Claude Code Review Bot
0 siblings, 0 replies; 25+ messages in thread
From: Claude Code Review Bot @ 2026-05-28 2:28 UTC (permalink / raw)
To: dri-devel-reviews
Patch Review
The main conversion patch. Removes `vma_to_pfn()`, updates the fault handler, and makes `VFIO_PCI_CORE` depend on `DMA_SHARED_BUFFER` and `PCI_P2PDMA_CORE`.
**Kconfig change**:
```c
+ select PCI_P2PDMA_CORE
+ select DMA_SHARED_BUFFER
```
Making these unconditional `select`s for `VFIO_PCI_CORE` is a significant dependency change. `DMA_SHARED_BUFFER` is generally available, but this should be called out in the commit message more prominently since it affects all VFIO PCI users.
**Conditional `.attach` in dma_buf_ops**:
```c
static const struct dma_buf_ops vfio_pci_dmabuf_ops = {
+#ifdef CONFIG_VFIO_PCI_DMABUF
.attach = vfio_pci_dma_buf_attach,
+#endif
```
Setting `.attach = NULL` when `CONFIG_VFIO_PCI_DMABUF` is disabled means DMA-BUF attach will silently succeed (the core code skips the callback if NULL). This is intentional -- BAR mmap DMABUFs don't need P2P attach semantics -- but it's worth noting that this means the DMABUFs created on the mmap path can't be used for DMA by importers when P2P is disabled.
**Default fault return changed**: The old code defaulted to `VM_FAULT_FALLBACK`; the new code defaults to `VM_FAULT_SIGBUS`. This is correct since a revoked buffer should SIGBUS, and the fallback case is now explicit in the `-EAGAIN` return from `find_pfn`.
---
Generated by Claude Code Patch Reviewer
^ permalink raw reply [flat|nested] 25+ messages in thread
* [PATCH v2 5/9] vfio/pci: Provide a user-facing name for BAR mappings
2026-05-27 10:23 [PATCH v2 0/9] vfio/pci: Add mmap() for DMABUFs Matt Evans
` (3 preceding siblings ...)
2026-05-27 10:23 ` [PATCH v2 4/9] vfio/pci: Convert BAR mmap() to use a DMABUF Matt Evans
@ 2026-05-27 10:23 ` Matt Evans
2026-05-28 2:28 ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 6/9] vfio/pci: Clean up BAR zap and revocation Matt Evans
` (4 subsequent siblings)
9 siblings, 1 reply; 25+ messages in thread
From: Matt Evans @ 2026-05-27 10:23 UTC (permalink / raw)
To: Alex Williamson, Leon Romanovsky, Jason Gunthorpe, Alex Mastro,
Christian König, Bjorn Helgaas, Logan Gunthorpe
Cc: Mahmoud Adam, David Matlack, Björn Töpel, Sumit Semwal,
Kevin Tian, Ankit Agrawal, Pranjal Shrivastava, Alistair Popple,
Vivek Kasireddy, linux-kernel, linux-media, dri-devel,
linaro-mm-sig, kvm, linux-pci
Since converting BAR mmap()s to using DMABUFs, we lose the original
device path in /proc/<pid>/maps, lsof, etc. Generate a debug-oriented
synthetic 'filename' based on the cdev, plus BDF, plus resource index.
This applies only to BAR mappings via the VFIO device fd, as
explicitly-exported DMABUFs are named by userspace via the
DMA_BUF_SET_NAME ioctl.
Signed-off-by: Matt Evans <mattev@meta.com>
---
drivers/vfio/pci/vfio_pci_dmabuf.c | 27 +++++++++++++++++++++++++--
1 file changed, 25 insertions(+), 2 deletions(-)
diff --git a/drivers/vfio/pci/vfio_pci_dmabuf.c b/drivers/vfio/pci/vfio_pci_dmabuf.c
index f7797f58d44b..733607371082 100644
--- a/drivers/vfio/pci/vfio_pci_dmabuf.c
+++ b/drivers/vfio/pci/vfio_pci_dmabuf.c
@@ -4,6 +4,7 @@
#include <linux/dma-buf-mapping.h>
#include <linux/pci-p2pdma.h>
#include <linux/dma-resv.h>
+#include <uapi/linux/dma-buf.h>
#include "vfio_pci_priv.h"
@@ -476,6 +477,7 @@ int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
struct vfio_pci_dma_buf *priv;
const unsigned int nr_ranges = 1;
unsigned long vma_pgoff = vma->vm_pgoff & (VFIO_PCI_OFFSET_MASK >> PAGE_SHIFT);
+ char *bufname;
int ret;
priv = kzalloc_obj(*priv);
@@ -488,6 +490,20 @@ int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
goto err_free_priv;
}
+ bufname = kzalloc(DMA_BUF_NAME_LEN, GFP_KERNEL);
+ if (!bufname) {
+ ret = -ENOMEM;
+ goto err_free_phys;
+ }
+
+ /*
+ * Maximum size of the friendly debug name is
+ * vfio1234567890:ffff:ff:3f.7/5 = 30, which fits within
+ * DMA_BUF_NAME_LEN.
+ */
+ snprintf(bufname, DMA_BUF_NAME_LEN, "%s:%s/%x",
+ dev_name(&vdev->vdev.device), pci_name(vdev->pdev), res_index);
+
/*
* The DMABUF begins from the mmap()'s BAR offset, i.e. the
* start of the VMA corresponds to byte 0 of the DMABUF and
@@ -506,7 +522,7 @@ int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
priv->provider = pcim_p2pdma_provider(vdev->pdev, res_index);
if (!priv->provider) {
ret = -EINVAL;
- goto err_free_phys;
+ goto err_free_name;
}
priv->phys_vec[0].paddr = phys_start + ((u64)vma_pgoff << PAGE_SHIFT);
@@ -514,7 +530,7 @@ int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
ret = vfio_pci_dmabuf_export(vdev, priv, O_CLOEXEC | O_RDWR);
if (ret)
- goto err_free_phys;
+ goto err_free_name;
/*
* The VMA gets the DMABUF file so that other users can locate
@@ -526,8 +542,15 @@ int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
vma->vm_file = priv->dmabuf->file;
vma->vm_private_data = priv;
+ spin_lock(&priv->dmabuf->name_lock);
+ kfree(priv->dmabuf->name);
+ priv->dmabuf->name = bufname;
+ spin_unlock(&priv->dmabuf->name_lock);
+
return 0;
+err_free_name:
+ kfree(bufname);
err_free_phys:
kfree(priv->phys_vec);
err_free_priv:
--
2.47.3
^ permalink raw reply related [flat|nested] 25+ messages in thread* Claude review: vfio/pci: Provide a user-facing name for BAR mappings
2026-05-27 10:23 ` [PATCH v2 5/9] vfio/pci: Provide a user-facing name for BAR mappings Matt Evans
@ 2026-05-28 2:28 ` Claude Code Review Bot
0 siblings, 0 replies; 25+ messages in thread
From: Claude Code Review Bot @ 2026-05-28 2:28 UTC (permalink / raw)
To: dri-devel-reviews
Patch Review
Nice quality-of-life patch for debugging. Generates names like `vfio1234567890:ffff:ff:3f.7/5`.
**Direct access to `priv->dmabuf->name_lock`**:
```c
+ spin_lock(&priv->dmabuf->name_lock);
+ kfree(priv->dmabuf->name);
+ priv->dmabuf->name = bufname;
+ spin_unlock(&priv->dmabuf->name_lock);
```
This directly reaches into `struct dma_buf` internals rather than using the `DMA_BUF_SET_NAME` path. While functionally fine (and avoids an unnecessary `copy_from_user`), it's fragile if the dma-buf subsystem changes its internal locking. Consider using `dma_buf_set_name()` if one exists, or at minimum add a comment explaining why direct access is used.
**`#include <uapi/linux/dma-buf.h>`**: Kernel code should include `<linux/dma-buf.h>` rather than the UAPI header. The UAPI header is pulled for `DMA_BUF_NAME_LEN`, which should be available through the normal header.
---
Generated by Claude Code Patch Reviewer
^ permalink raw reply [flat|nested] 25+ messages in thread
* [PATCH v2 6/9] vfio/pci: Clean up BAR zap and revocation
2026-05-27 10:23 [PATCH v2 0/9] vfio/pci: Add mmap() for DMABUFs Matt Evans
` (4 preceding siblings ...)
2026-05-27 10:23 ` [PATCH v2 5/9] vfio/pci: Provide a user-facing name for BAR mappings Matt Evans
@ 2026-05-27 10:23 ` Matt Evans
2026-05-28 2:28 ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 7/9] vfio/pci: Support mmap() of a VFIO DMABUF Matt Evans
` (3 subsequent siblings)
9 siblings, 1 reply; 25+ messages in thread
From: Matt Evans @ 2026-05-27 10:23 UTC (permalink / raw)
To: Alex Williamson, Leon Romanovsky, Jason Gunthorpe, Alex Mastro,
Christian König, Bjorn Helgaas, Logan Gunthorpe
Cc: Mahmoud Adam, David Matlack, Björn Töpel, Sumit Semwal,
Kevin Tian, Ankit Agrawal, Pranjal Shrivastava, Alistair Popple,
Vivek Kasireddy, linux-kernel, linux-media, dri-devel,
linaro-mm-sig, kvm, linux-pci
Previously, vfio_pci_zap_bars() (and the wrapper
vfio_pci_zap_and_down_write_memory_lock()) calls were paired with
calls of vfio_pci_dma_buf_move().
This commit replaces them a unified new function,
vfio_pci_zap_revoke_bars() containing both the vfio_pci_dma_buf_move()
and the unmap_mapping_range(), making it harder for callers to omit
one. It adds a wrapper, vfio_pci_lock_zap_revoke_bars(), which takes
the write memory_lock before zapping, and adds a new
vfio_pci_unrevoke_bars() for the re-enable path.
However, as of "vfio/pci: Convert BAR mmap() to use a DMABUF" the
unmap_mapping_range() to zap is entirely redundant for plain vfio-pci,
since the DMABUFs used for BAR mappings already zap PTEs when the
vfio_pci_dma_buf_move() occurs.
One exception remains as a FIXME: in nvgrace-gpu, some BAR VMAs
conditionally use custom vm_ops, which have not moved to be backed by
DMABUFs. If these BARs are mmap()ed, the vdev enables the existing
behaviour of unmap_mapping_range() for the device fd address space.
Signed-off-by: Matt Evans <mattev@meta.com>
---
drivers/vfio/pci/nvgrace-gpu/main.c | 5 +++
drivers/vfio/pci/vfio_pci_config.c | 30 ++++++-------
drivers/vfio/pci/vfio_pci_core.c | 66 ++++++++++++++++++++---------
drivers/vfio/pci/vfio_pci_priv.h | 3 +-
include/linux/vfio_pci_core.h | 1 +
5 files changed, 66 insertions(+), 39 deletions(-)
diff --git a/drivers/vfio/pci/nvgrace-gpu/main.c b/drivers/vfio/pci/nvgrace-gpu/main.c
index 15e2f03c6cd4..cfa649200a7f 100644
--- a/drivers/vfio/pci/nvgrace-gpu/main.c
+++ b/drivers/vfio/pci/nvgrace-gpu/main.c
@@ -364,6 +364,8 @@ static int nvgrace_gpu_mmap(struct vfio_device *core_vdev,
struct nvgrace_gpu_pci_core_device *nvdev =
container_of(core_vdev, struct nvgrace_gpu_pci_core_device,
core_device.vdev);
+ struct vfio_pci_core_device *vdev =
+ container_of(core_vdev, struct vfio_pci_core_device, vdev);
struct mem_region *memregion;
u64 req_len, pgoff, end;
unsigned int index;
@@ -374,6 +376,9 @@ static int nvgrace_gpu_mmap(struct vfio_device *core_vdev,
if (!memregion)
return vfio_pci_core_mmap(core_vdev, vma);
+ /* Non-DMABUF BAR mappings need an extra zap */
+ vdev->bar_needs_zap = true;
+
/*
* Request to mmap the BAR. Map to the CPU accessible memory on the
* GPU using the memory information gathered from the system ACPI
diff --git a/drivers/vfio/pci/vfio_pci_config.c b/drivers/vfio/pci/vfio_pci_config.c
index a10ed733f0e3..8bfab0da481c 100644
--- a/drivers/vfio/pci/vfio_pci_config.c
+++ b/drivers/vfio/pci/vfio_pci_config.c
@@ -590,12 +590,10 @@ static int vfio_basic_config_write(struct vfio_pci_core_device *vdev, int pos,
virt_mem = !!(le16_to_cpu(*virt_cmd) & PCI_COMMAND_MEMORY);
new_mem = !!(new_cmd & PCI_COMMAND_MEMORY);
- if (!new_mem) {
- vfio_pci_zap_and_down_write_memory_lock(vdev);
- vfio_pci_dma_buf_move(vdev, true);
- } else {
+ if (!new_mem)
+ vfio_pci_lock_zap_revoke_bars(vdev);
+ else
down_write(&vdev->memory_lock);
- }
/*
* If the user is writing mem/io enable (new_mem/io) and we
@@ -631,7 +629,7 @@ static int vfio_basic_config_write(struct vfio_pci_core_device *vdev, int pos,
*virt_cmd |= cpu_to_le16(new_cmd & mask);
if (__vfio_pci_memory_enabled(vdev))
- vfio_pci_dma_buf_move(vdev, false);
+ vfio_pci_unrevoke_bars(vdev);
up_write(&vdev->memory_lock);
}
@@ -712,16 +710,14 @@ static int __init init_pci_cap_basic_perm(struct perm_bits *perm)
static void vfio_lock_and_set_power_state(struct vfio_pci_core_device *vdev,
pci_power_t state)
{
- if (state >= PCI_D3hot) {
- vfio_pci_zap_and_down_write_memory_lock(vdev);
- vfio_pci_dma_buf_move(vdev, true);
- } else {
+ if (state >= PCI_D3hot)
+ vfio_pci_lock_zap_revoke_bars(vdev);
+ else
down_write(&vdev->memory_lock);
- }
vfio_pci_set_power_state(vdev, state);
if (__vfio_pci_memory_enabled(vdev))
- vfio_pci_dma_buf_move(vdev, false);
+ vfio_pci_unrevoke_bars(vdev);
up_write(&vdev->memory_lock);
}
@@ -908,11 +904,10 @@ static int vfio_exp_config_write(struct vfio_pci_core_device *vdev, int pos,
&cap);
if (!ret && (cap & PCI_EXP_DEVCAP_FLR)) {
- vfio_pci_zap_and_down_write_memory_lock(vdev);
- vfio_pci_dma_buf_move(vdev, true);
+ vfio_pci_lock_zap_revoke_bars(vdev);
pci_try_reset_function(vdev->pdev);
if (__vfio_pci_memory_enabled(vdev))
- vfio_pci_dma_buf_move(vdev, false);
+ vfio_pci_unrevoke_bars(vdev);
up_write(&vdev->memory_lock);
}
}
@@ -993,11 +988,10 @@ static int vfio_af_config_write(struct vfio_pci_core_device *vdev, int pos,
&cap);
if (!ret && (cap & PCI_AF_CAP_FLR) && (cap & PCI_AF_CAP_TP)) {
- vfio_pci_zap_and_down_write_memory_lock(vdev);
- vfio_pci_dma_buf_move(vdev, true);
+ vfio_pci_lock_zap_revoke_bars(vdev);
pci_try_reset_function(vdev->pdev);
if (__vfio_pci_memory_enabled(vdev))
- vfio_pci_dma_buf_move(vdev, false);
+ vfio_pci_unrevoke_bars(vdev);
up_write(&vdev->memory_lock);
}
}
diff --git a/drivers/vfio/pci/vfio_pci_core.c b/drivers/vfio/pci/vfio_pci_core.c
index c5f934905ce0..cfea59806a4f 100644
--- a/drivers/vfio/pci/vfio_pci_core.c
+++ b/drivers/vfio/pci/vfio_pci_core.c
@@ -319,8 +319,7 @@ static int vfio_pci_runtime_pm_entry(struct vfio_pci_core_device *vdev,
* The vdev power related flags are protected with 'memory_lock'
* semaphore.
*/
- vfio_pci_zap_and_down_write_memory_lock(vdev);
- vfio_pci_dma_buf_move(vdev, true);
+ vfio_pci_lock_zap_revoke_bars(vdev);
if (vdev->pm_runtime_engaged) {
up_write(&vdev->memory_lock);
@@ -406,7 +405,7 @@ static void vfio_pci_runtime_pm_exit(struct vfio_pci_core_device *vdev)
down_write(&vdev->memory_lock);
__vfio_pci_runtime_pm_exit(vdev);
if (__vfio_pci_memory_enabled(vdev))
- vfio_pci_dma_buf_move(vdev, false);
+ vfio_pci_unrevoke_bars(vdev);
up_write(&vdev->memory_lock);
}
@@ -1256,6 +1255,8 @@ static int vfio_pci_ioctl_set_irqs(struct vfio_pci_core_device *vdev,
return ret;
}
+static void vfio_pci_zap_revoke_bars(struct vfio_pci_core_device *vdev);
+
static int vfio_pci_ioctl_reset(struct vfio_pci_core_device *vdev,
void __user *arg)
{
@@ -1264,7 +1265,7 @@ static int vfio_pci_ioctl_reset(struct vfio_pci_core_device *vdev,
if (!vdev->reset_works)
return -EINVAL;
- vfio_pci_zap_and_down_write_memory_lock(vdev);
+ down_write(&vdev->memory_lock);
/*
* This function can be invoked while the power state is non-D0. If
@@ -1277,10 +1278,11 @@ static int vfio_pci_ioctl_reset(struct vfio_pci_core_device *vdev,
*/
vfio_pci_set_power_state(vdev, PCI_D0);
- vfio_pci_dma_buf_move(vdev, true);
+ vfio_pci_zap_revoke_bars(vdev);
+
ret = pci_try_reset_function(vdev->pdev);
if (__vfio_pci_memory_enabled(vdev))
- vfio_pci_dma_buf_move(vdev, false);
+ vfio_pci_unrevoke_bars(vdev);
up_write(&vdev->memory_lock);
return ret;
@@ -1648,20 +1650,44 @@ ssize_t vfio_pci_core_write(struct vfio_device *core_vdev, const char __user *bu
}
EXPORT_SYMBOL_GPL(vfio_pci_core_write);
-static void vfio_pci_zap_bars(struct vfio_pci_core_device *vdev)
+static void vfio_pci_zap_revoke_bars(struct vfio_pci_core_device *vdev)
{
- struct vfio_device *core_vdev = &vdev->vdev;
- loff_t start = VFIO_PCI_INDEX_TO_OFFSET(VFIO_PCI_BAR0_REGION_INDEX);
- loff_t end = VFIO_PCI_INDEX_TO_OFFSET(VFIO_PCI_ROM_REGION_INDEX);
- loff_t len = end - start;
+ lockdep_assert_held_write(&vdev->memory_lock);
+ vfio_pci_dma_buf_move(vdev, true);
- unmap_mapping_range(core_vdev->inode->i_mapping, start, len, true);
+ /*
+ * All VFIO PCI BARs are backed by DMABUFs, with the current
+ * exception of the nvgrace-gpu device which uses its own
+ * vm_ops for a subset of BARs. For this, BAR mappings are
+ * still made in the vdev's address_space, and a zap is
+ * required. The tracking is crude, and will (harmlessly)
+ * continue to zap if the special BAR is unmapped, but that
+ * behaviour isn't the common case.
+ *
+ * FIXME: This can go away if the special nvgrace-gpu mapping
+ * is converted to use DMABUF.
+ */
+ if (vdev->bar_needs_zap) {
+ struct vfio_device *core_vdev = &vdev->vdev;
+ loff_t start = VFIO_PCI_INDEX_TO_OFFSET(VFIO_PCI_BAR0_REGION_INDEX);
+ loff_t end = VFIO_PCI_INDEX_TO_OFFSET(VFIO_PCI_ROM_REGION_INDEX);
+ loff_t len = end - start;
+
+ unmap_mapping_range(core_vdev->inode->i_mapping,
+ start, len, true);
+ }
}
-void vfio_pci_zap_and_down_write_memory_lock(struct vfio_pci_core_device *vdev)
+void vfio_pci_lock_zap_revoke_bars(struct vfio_pci_core_device *vdev)
{
down_write(&vdev->memory_lock);
- vfio_pci_zap_bars(vdev);
+ vfio_pci_zap_revoke_bars(vdev);
+}
+
+void vfio_pci_unrevoke_bars(struct vfio_pci_core_device *vdev)
+{
+ lockdep_assert_held_write(&vdev->memory_lock);
+ vfio_pci_dma_buf_move(vdev, false);
}
u16 vfio_pci_memory_lock_and_enable(struct vfio_pci_core_device *vdev)
@@ -2517,9 +2543,10 @@ static int vfio_pci_dev_set_hot_reset(struct vfio_device_set *dev_set,
}
/*
- * Take the memory write lock for each device and zap BAR
- * mappings to prevent the user accessing the device while in
- * reset. Locking multiple devices is prone to deadlock,
+ * Take the memory write lock for each device and
+ * zap/revoke BAR mappings to prevent the user (or
+ * peers) accessing the device while in reset.
+ * Locking multiple devices is prone to deadlock,
* runaway and unwind if we hit contention.
*/
if (!down_write_trylock(&vdev->memory_lock)) {
@@ -2527,8 +2554,7 @@ static int vfio_pci_dev_set_hot_reset(struct vfio_device_set *dev_set,
break;
}
- vfio_pci_dma_buf_move(vdev, true);
- vfio_pci_zap_bars(vdev);
+ vfio_pci_zap_revoke_bars(vdev);
}
if (!list_entry_is_head(vdev,
@@ -2558,7 +2584,7 @@ static int vfio_pci_dev_set_hot_reset(struct vfio_device_set *dev_set,
list_for_each_entry_from_reverse(vdev, &dev_set->device_list,
vdev.dev_set_list) {
if (vdev->vdev.open_count && __vfio_pci_memory_enabled(vdev))
- vfio_pci_dma_buf_move(vdev, false);
+ vfio_pci_unrevoke_bars(vdev);
up_write(&vdev->memory_lock);
}
diff --git a/drivers/vfio/pci/vfio_pci_priv.h b/drivers/vfio/pci/vfio_pci_priv.h
index d38e1b98b2e9..10833aabd7fb 100644
--- a/drivers/vfio/pci/vfio_pci_priv.h
+++ b/drivers/vfio/pci/vfio_pci_priv.h
@@ -83,7 +83,8 @@ void vfio_config_free(struct vfio_pci_core_device *vdev);
int vfio_pci_set_power_state(struct vfio_pci_core_device *vdev,
pci_power_t state);
-void vfio_pci_zap_and_down_write_memory_lock(struct vfio_pci_core_device *vdev);
+void vfio_pci_lock_zap_revoke_bars(struct vfio_pci_core_device *vdev);
+void vfio_pci_unrevoke_bars(struct vfio_pci_core_device *vdev);
u16 vfio_pci_memory_lock_and_enable(struct vfio_pci_core_device *vdev);
void vfio_pci_memory_unlock_and_restore(struct vfio_pci_core_device *vdev,
u16 cmd);
diff --git a/include/linux/vfio_pci_core.h b/include/linux/vfio_pci_core.h
index 7accd0eac457..e35e82c24c8c 100644
--- a/include/linux/vfio_pci_core.h
+++ b/include/linux/vfio_pci_core.h
@@ -127,6 +127,7 @@ struct vfio_pci_core_device {
bool needs_pm_restore:1;
bool pm_intx_masked:1;
bool pm_runtime_engaged:1;
+ bool bar_needs_zap:1;
struct pci_saved_state *pci_saved_state;
struct pci_saved_state *pm_save;
int ioeventfds_nr;
--
2.47.3
^ permalink raw reply related [flat|nested] 25+ messages in thread* Claude review: vfio/pci: Clean up BAR zap and revocation
2026-05-27 10:23 ` [PATCH v2 6/9] vfio/pci: Clean up BAR zap and revocation Matt Evans
@ 2026-05-28 2:28 ` Claude Code Review Bot
0 siblings, 0 replies; 25+ messages in thread
From: Claude Code Review Bot @ 2026-05-28 2:28 UTC (permalink / raw)
To: dri-devel-reviews
Patch Review
Consolidates `vfio_pci_zap_bars()` + `vfio_pci_dma_buf_move()` into `vfio_pci_zap_revoke_bars()`. Clean refactoring.
**`vfio_pci_ioctl_reset` change**: The lock acquisition changes from `vfio_pci_zap_and_down_write_memory_lock` (which did lock+zap atomically) to separate `down_write` + `vfio_pci_zap_revoke_bars`:
```c
- vfio_pci_zap_and_down_write_memory_lock(vdev);
+ down_write(&vdev->memory_lock);
...
- vfio_pci_dma_buf_move(vdev, true);
+ vfio_pci_zap_revoke_bars(vdev);
```
This moves the zap/revoke to *after* the D0 power transition, which the changelog notes. This is intentional per the v2 notes but is a behavioral change worth PCI maintainer attention.
**`bar_needs_zap` flag**: Set in nvgrace-gpu but never cleared. This means once any nvgrace-gpu BAR is mmap'd, the zap path will always fire even after that BAR is unmapped. The FIXME acknowledges this, and the overhead is minimal (an extra `unmap_mapping_range` on an empty range), so this is acceptable as an interim solution.
---
Generated by Claude Code Patch Reviewer
^ permalink raw reply [flat|nested] 25+ messages in thread
* [PATCH v2 7/9] vfio/pci: Support mmap() of a VFIO DMABUF
2026-05-27 10:23 [PATCH v2 0/9] vfio/pci: Add mmap() for DMABUFs Matt Evans
` (5 preceding siblings ...)
2026-05-27 10:23 ` [PATCH v2 6/9] vfio/pci: Clean up BAR zap and revocation Matt Evans
@ 2026-05-27 10:23 ` Matt Evans
2026-05-28 2:28 ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 8/9] vfio/pci: Permanently revoke a DMABUF on request Matt Evans
` (2 subsequent siblings)
9 siblings, 1 reply; 25+ messages in thread
From: Matt Evans @ 2026-05-27 10:23 UTC (permalink / raw)
To: Alex Williamson, Leon Romanovsky, Jason Gunthorpe, Alex Mastro,
Christian König, Bjorn Helgaas, Logan Gunthorpe
Cc: Mahmoud Adam, David Matlack, Björn Töpel, Sumit Semwal,
Kevin Tian, Ankit Agrawal, Pranjal Shrivastava, Alistair Popple,
Vivek Kasireddy, linux-kernel, linux-media, dri-devel,
linaro-mm-sig, kvm, linux-pci
A VFIO DMABUF can export a subset of a BAR to userspace by fd; add
support for mmap() of this fd. This provides another route for a
process to map BARs, except one where the process can only map a specific
subset of a BAR represented by the exported DMABUF.
mmap() support enables userspace driver designs that safely delegate
access to BAR sub-ranges to other client processes by sharing a DMABUF
fd, without having to share the (omnipotent) VFIO device fd with them.
Since the main VFIO BAR mmap() is now DMABUF-aware, this path reuses
the existing vm_ops. But, since the lifecycle of an exported DMABUF
is still decoupled from that of the device fd it came from, the device
fd might now be closed concurrent with a VMA fault.
Extra synchronisation is added to deal with the possibility of a fault
racing with the DMABUF cleanup path. (Note that this differs to a
DMABUF implicitly created on the mmap() path, which holds ownership of
the device fd and so prevents close-during-fault scenarios in order to
maintain the same user-facing behaviour on close.) It does this by
temporarily taking a VFIO device registration to ensure vdev remains
valid, then vdev->memory_lock can be taken.
Signed-off-by: Matt Evans <mattev@meta.com>
---
drivers/vfio/pci/vfio_pci_core.c | 79 ++++++++++++++++++++++++++----
drivers/vfio/pci/vfio_pci_dmabuf.c | 27 ++++++++++
drivers/vfio/pci/vfio_pci_priv.h | 2 +
3 files changed, 99 insertions(+), 9 deletions(-)
diff --git a/drivers/vfio/pci/vfio_pci_core.c b/drivers/vfio/pci/vfio_pci_core.c
index cfea59806a4f..41e049fa9a8a 100644
--- a/drivers/vfio/pci/vfio_pci_core.c
+++ b/drivers/vfio/pci/vfio_pci_core.c
@@ -12,6 +12,8 @@
#include <linux/aperture.h>
#include <linux/device.h>
+#include <linux/dma-buf.h>
+#include <linux/dma-resv.h>
#include <linux/eventfd.h>
#include <linux/file.h>
#include <linux/interrupt.h>
@@ -1742,19 +1744,77 @@ static vm_fault_t vfio_pci_mmap_huge_fault(struct vm_fault *vmf,
vm_fault_t ret = VM_FAULT_SIGBUS;
/*
- * We can rely on the existence of both a DMABUF (priv) and
- * the VFIO device it was exported from (vdev). This fault's
- * VMA was established using vfio_pci_core_mmap_prep_dmabuf()
- * which transfers ownership of the VFIO device fd to the
- * DMABUF, and so the VFIO device is held open because the
- * VMA's vm_file (DMABUF) is open.
+ * The only thing this can rely on is that the DMABUF relating
+ * to the VMA's vm_file exists (priv).
*
- * Since vfio_pci_dma_buf_cleanup() cannot have happened,
- * vdev must be valid; we can take memory_lock.
+ * A DMABUF for a VFIO device fd mmap() holds a reference to
+ * the original VFIO device fd, but an explicitly-exported
+ * DMABUF does not. The original fd might have closed,
+ * meaning this fault can race with
+ * vfio_pci_dma_buf_cleanup(), meaning priv->vdev might be
+ * NULL, and the VFIO device registration might have been
+ * dropped.
+ *
+ * With the goal of taking vdev->memory_lock in a world where
+ * vdev might not still exist:
+ *
+ * 1. Take the resv lock on the DMABUF:
+ * - If racing cleanup got in first, the buffer is revoked;
+ * stop/exit if so.
+ * - If we got in first, the buffer is not revoked so vdev is
+ * non-NULL, accessible, and cleanup _has not yet put the
+ * VFIO device registration_. So, the device refcount must
+ * be >0.
+ *
+ * 2. Take vfio_device registration (refcount guaranteed >0
+ * hereafter).
+ *
+ * 3. Unlock the DMABUF's resv lock:
+ * - A racing cleanup can now complete.
+ * - But, the device refcount >0, meaning the vfio_device
+ * (and vfio_pcie_core device vdev) have not yet been
+ * freed. vdev is accessible, even if the DMABUF has been
+ * revoked or cleanup has happened, because
+ * vfio_unregister_group_dev() can't complete.
+ *
+ * 4. Take the vdev->memory_lock
+ * - Either the DMABUF is usable, or has been cleaned up.
+ * Whichever, it can no longer change under us.
+ * - Test the DMABUF revocation status again: if it was
+ * revoked between 1 and 4 return a SIGBUS. Otherwise,
+ * return a PFN.
+ * - It's not necessary to also take the resv lock, because
+ * the status/vdev can't change while memory_lock is held.
+ *
+ * 5. Unlock, done.
*/
+
+ dma_resv_lock(priv->dmabuf->resv, NULL);
vdev = READ_ONCE(priv->vdev);
+ if (priv->revoked || !vdev) {
+ pr_debug_ratelimited("%s VA 0x%lx, pgoff 0x%lx: DMABUF revoked/cleaned up\n",
+ __func__, vmf->address, vma->vm_pgoff);
+ dma_resv_unlock(priv->dmabuf->resv);
+ return VM_FAULT_SIGBUS;
+ }
+ /* vdev is usable */
+
+ if (!vfio_device_try_get_registration(&vdev->vdev)) {
+ /*
+ * If vdev != NULL (above), the registration should
+ * already be >0 and so this try_get should never
+ * fail.
+ */
+ dev_warn(&vdev->pdev->dev, "%s: Unexpected registration failure\n",
+ __func__);
+ dma_resv_unlock(priv->dmabuf->resv);
+ return VM_FAULT_SIGBUS;
+ }
+ dma_resv_unlock(priv->dmabuf->resv);
+
scoped_guard(rwsem_read, &vdev->memory_lock) {
+ /* Revocation status must be re-read, under memory_lock */
if (!priv->revoked) {
int pres = vfio_pci_dma_buf_find_pfn(priv, vma,
vmf->address,
@@ -1773,6 +1833,7 @@ static vm_fault_t vfio_pci_mmap_huge_fault(struct vm_fault *vmf,
vma->vm_pgoff, (unsigned int)ret);
}
+ vfio_device_put_registration(&vdev->vdev);
return ret;
}
@@ -1781,7 +1842,7 @@ static vm_fault_t vfio_pci_mmap_page_fault(struct vm_fault *vmf)
return vfio_pci_mmap_huge_fault(vmf, 0);
}
-static const struct vm_operations_struct vfio_pci_mmap_ops = {
+const struct vm_operations_struct vfio_pci_mmap_ops = {
.fault = vfio_pci_mmap_page_fault,
#ifdef CONFIG_ARCH_SUPPORTS_HUGE_PFNMAP
.huge_fault = vfio_pci_mmap_huge_fault,
diff --git a/drivers/vfio/pci/vfio_pci_dmabuf.c b/drivers/vfio/pci/vfio_pci_dmabuf.c
index 733607371082..4b3b15655f1d 100644
--- a/drivers/vfio/pci/vfio_pci_dmabuf.c
+++ b/drivers/vfio/pci/vfio_pci_dmabuf.c
@@ -27,6 +27,32 @@ static int vfio_pci_dma_buf_attach(struct dma_buf *dmabuf,
return 0;
}
+
+static int vfio_pci_dma_buf_mmap(struct dma_buf *dmabuf, struct vm_area_struct *vma)
+{
+ struct vfio_pci_dma_buf *priv = dmabuf->priv;
+
+ if (priv->revoked)
+ return -ENODEV;
+ if ((vma->vm_flags & VM_SHARED) == 0)
+ return -EINVAL;
+
+ /*
+ * dma_buf_mmap_internal() has asserted that the VMA is
+ * contained within the DMABUF size before calling this.
+ */
+
+ vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
+ vma->vm_page_prot = pgprot_decrypted(vma->vm_page_prot);
+
+ /* See comments in vfio_pci_core_mmap() re VM_ALLOW_ANY_UNCACHED. */
+ vm_flags_set(vma, VM_ALLOW_ANY_UNCACHED | VM_IO | VM_PFNMAP |
+ VM_DONTEXPAND | VM_DONTDUMP);
+ vma->vm_private_data = priv;
+ vma->vm_ops = &vfio_pci_mmap_ops;
+
+ return 0;
+}
#endif /* CONFIG_VFIO_PCI_DMABUF */
static void vfio_pci_dma_buf_done(struct kref *kref)
@@ -94,6 +120,7 @@ static void vfio_pci_dma_buf_release(struct dma_buf *dmabuf)
static const struct dma_buf_ops vfio_pci_dmabuf_ops = {
#ifdef CONFIG_VFIO_PCI_DMABUF
.attach = vfio_pci_dma_buf_attach,
+ .mmap = vfio_pci_dma_buf_mmap,
#endif
.map_dma_buf = vfio_pci_dma_buf_map,
.unmap_dma_buf = vfio_pci_dma_buf_unmap,
diff --git a/drivers/vfio/pci/vfio_pci_priv.h b/drivers/vfio/pci/vfio_pci_priv.h
index 10833aabd7fb..db2e2aeae88f 100644
--- a/drivers/vfio/pci/vfio_pci_priv.h
+++ b/drivers/vfio/pci/vfio_pci_priv.h
@@ -38,6 +38,8 @@ struct vfio_pci_dma_buf {
u8 revoked : 1;
};
+extern const struct vm_operations_struct vfio_pci_mmap_ops;
+
bool vfio_pci_intx_mask(struct vfio_pci_core_device *vdev);
void vfio_pci_intx_unmask(struct vfio_pci_core_device *vdev);
--
2.47.3
^ permalink raw reply related [flat|nested] 25+ messages in thread* Claude review: vfio/pci: Support mmap() of a VFIO DMABUF
2026-05-27 10:23 ` [PATCH v2 7/9] vfio/pci: Support mmap() of a VFIO DMABUF Matt Evans
@ 2026-05-28 2:28 ` Claude Code Review Bot
0 siblings, 0 replies; 25+ messages in thread
From: Claude Code Review Bot @ 2026-05-28 2:28 UTC (permalink / raw)
To: dri-devel-reviews
Patch Review
This is the most complex patch. Adds `fops->mmap` for DMABUF fds and handles the lifecycle where the device fd might close while a DMABUF VMA fault is in-flight.
**The 5-step synchronization protocol** in `vfio_pci_mmap_huge_fault` is the core contribution. The logic is:
1. Lock DMABUF resv -> check revoked/vdev
2. `vfio_device_try_get_registration` (guaranteed >0 here)
3. Unlock resv (cleanup can now proceed)
4. Take `memory_lock`, re-check revocation
5. Service fault or SIGBUS
This is correct but very subtle. The key invariant is that cleanup holds `resv` when setting `revoked`, so if the fault sees `!revoked` under `resv`, the device registration must still be positive.
**`vfio_pci_dma_buf_mmap` hardcodes UC**:
```c
+ vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
```
This is overridden in patch 9 to support WC. Fine for the series ordering.
**Missing `.mmap` for non-DMABUF case**: The `vfio_pci_dmabuf_ops` struct conditionally includes `.mmap` only under `CONFIG_VFIO_PCI_DMABUF`. But the BAR mmap path (which always creates DMABUFs) never goes through `dma_buf_mmap()` -- it installs `vfio_pci_mmap_ops` directly. So this is correct: `.mmap` is only needed for explicitly-exported DMABUFs.
---
Generated by Claude Code Patch Reviewer
^ permalink raw reply [flat|nested] 25+ messages in thread
* [PATCH v2 8/9] vfio/pci: Permanently revoke a DMABUF on request
2026-05-27 10:23 [PATCH v2 0/9] vfio/pci: Add mmap() for DMABUFs Matt Evans
` (6 preceding siblings ...)
2026-05-27 10:23 ` [PATCH v2 7/9] vfio/pci: Support mmap() of a VFIO DMABUF Matt Evans
@ 2026-05-27 10:23 ` Matt Evans
2026-05-28 2:28 ` Claude review: " Claude Code Review Bot
2026-05-27 10:23 ` [PATCH v2 9/9] vfio/pci: Add mmap() attributes to DMABUF feature Matt Evans
2026-05-28 2:28 ` Claude review: vfio/pci: Add mmap() for DMABUFs Claude Code Review Bot
9 siblings, 1 reply; 25+ messages in thread
From: Matt Evans @ 2026-05-27 10:23 UTC (permalink / raw)
To: Alex Williamson, Leon Romanovsky, Jason Gunthorpe, Alex Mastro,
Christian König, Bjorn Helgaas, Logan Gunthorpe
Cc: Mahmoud Adam, David Matlack, Björn Töpel, Sumit Semwal,
Kevin Tian, Ankit Agrawal, Pranjal Shrivastava, Alistair Popple,
Vivek Kasireddy, linux-kernel, linux-media, dri-devel,
linaro-mm-sig, kvm, linux-pci
Expand the VFIO DMABUF revocation state to three states:
Not revoked, temporarily revoked, and permanently revoked.
The first two are for existing transient revocation, e.g. across a
function reset, and the DMABUF is put into the last in response to a
new ioctl(VFIO_DEVICE_PCI_DMABUF_REVOKE) request.
This VFIO device fd ioctl() passes a DMABUF by fd and requests that
the DMABUF is permanently revoked. On success, it's guaranteed that
the buffer can never be imported/attached/mmap()ed in future, that
dynamic imports have been cleanly detached, and that all mappings have
been made inaccessible/PTEs zapped.
This is useful for lifecycle management, to reclaim VFIO PCI BAR
ranges previously delegated to a subordinate client process: The
driver process can ensure that the loaned resources are revoked when
the client is deemed "done", and exported ranges can be safely re-used
elsewhere.
Refactor the revocation code out of vfio_pci_dma_buf_move() to a
function common to move and the new ioctl path.
Signed-off-by: Matt Evans <mattev@meta.com>
---
drivers/vfio/pci/vfio_pci_core.c | 21 ++++-
drivers/vfio/pci/vfio_pci_dmabuf.c | 146 +++++++++++++++++++++--------
drivers/vfio/pci/vfio_pci_priv.h | 14 ++-
include/uapi/linux/vfio.h | 30 ++++++
4 files changed, 170 insertions(+), 41 deletions(-)
diff --git a/drivers/vfio/pci/vfio_pci_core.c b/drivers/vfio/pci/vfio_pci_core.c
index 41e049fa9a8a..5184b3cac160 100644
--- a/drivers/vfio/pci/vfio_pci_core.c
+++ b/drivers/vfio/pci/vfio_pci_core.c
@@ -1500,6 +1500,21 @@ static int vfio_pci_ioctl_ioeventfd(struct vfio_pci_core_device *vdev,
ioeventfd.fd);
}
+static int vfio_pci_ioctl_dmabuf_revoke(struct vfio_pci_core_device *vdev,
+ struct vfio_pci_dmabuf_revoke __user *arg)
+{
+ unsigned long minsz = offsetofend(struct vfio_pci_dmabuf_revoke, dmabuf_fd);
+ struct vfio_pci_dmabuf_revoke revoke;
+
+ if (copy_from_user(&revoke, arg, minsz))
+ return -EFAULT;
+
+ if (revoke.argsz < minsz)
+ return -EINVAL;
+
+ return vfio_pci_dma_buf_revoke(vdev, revoke.dmabuf_fd);
+}
+
long vfio_pci_core_ioctl(struct vfio_device *core_vdev, unsigned int cmd,
unsigned long arg)
{
@@ -1522,6 +1537,8 @@ long vfio_pci_core_ioctl(struct vfio_device *core_vdev, unsigned int cmd,
return vfio_pci_ioctl_reset(vdev, uarg);
case VFIO_DEVICE_SET_IRQS:
return vfio_pci_ioctl_set_irqs(vdev, uarg);
+ case VFIO_DEVICE_PCI_DMABUF_REVOKE:
+ return vfio_pci_ioctl_dmabuf_revoke(vdev, uarg);
default:
return -ENOTTY;
}
@@ -1792,7 +1809,7 @@ static vm_fault_t vfio_pci_mmap_huge_fault(struct vm_fault *vmf,
dma_resv_lock(priv->dmabuf->resv, NULL);
vdev = READ_ONCE(priv->vdev);
- if (priv->revoked || !vdev) {
+ if (priv->status != VFIO_PCI_DMABUF_OK || !vdev) {
pr_debug_ratelimited("%s VA 0x%lx, pgoff 0x%lx: DMABUF revoked/cleaned up\n",
__func__, vmf->address, vma->vm_pgoff);
dma_resv_unlock(priv->dmabuf->resv);
@@ -1815,7 +1832,7 @@ static vm_fault_t vfio_pci_mmap_huge_fault(struct vm_fault *vmf,
scoped_guard(rwsem_read, &vdev->memory_lock) {
/* Revocation status must be re-read, under memory_lock */
- if (!priv->revoked) {
+ if (priv->status == VFIO_PCI_DMABUF_OK) {
int pres = vfio_pci_dma_buf_find_pfn(priv, vma,
vmf->address,
order, &pfn);
diff --git a/drivers/vfio/pci/vfio_pci_dmabuf.c b/drivers/vfio/pci/vfio_pci_dmabuf.c
index 4b3b15655f1d..3fa14760898f 100644
--- a/drivers/vfio/pci/vfio_pci_dmabuf.c
+++ b/drivers/vfio/pci/vfio_pci_dmabuf.c
@@ -19,7 +19,7 @@ static int vfio_pci_dma_buf_attach(struct dma_buf *dmabuf,
if (!attachment->peer2peer)
return -EOPNOTSUPP;
- if (priv->revoked)
+ if (priv->status != VFIO_PCI_DMABUF_OK)
return -ENODEV;
if (!dma_buf_attach_revocable(attachment))
@@ -32,7 +32,7 @@ static int vfio_pci_dma_buf_mmap(struct dma_buf *dmabuf, struct vm_area_struct *
{
struct vfio_pci_dma_buf *priv = dmabuf->priv;
- if (priv->revoked)
+ if (priv->status != VFIO_PCI_DMABUF_OK)
return -ENODEV;
if ((vma->vm_flags & VM_SHARED) == 0)
return -EINVAL;
@@ -72,7 +72,7 @@ vfio_pci_dma_buf_map(struct dma_buf_attachment *attachment,
dma_resv_assert_held(priv->dmabuf->resv);
- if (priv->revoked)
+ if (priv->status != VFIO_PCI_DMABUF_OK)
return ERR_PTR(-ENODEV);
ret = dma_buf_phys_vec_to_sgt(attachment, priv->provider,
@@ -287,7 +287,8 @@ static int vfio_pci_dmabuf_export(struct vfio_pci_core_device *vdev,
INIT_LIST_HEAD(&priv->dmabufs_elm);
down_write(&vdev->memory_lock);
dma_resv_lock(priv->dmabuf->resv, NULL);
- priv->revoked = !__vfio_pci_memory_enabled(vdev);
+ priv->status = __vfio_pci_memory_enabled(vdev) ? VFIO_PCI_DMABUF_OK :
+ VFIO_PCI_DMABUF_TEMP_REVOKED;
list_add_tail(&priv->dmabufs_elm, &vdev->dmabufs);
dma_resv_unlock(priv->dmabuf->resv);
up_write(&vdev->memory_lock);
@@ -318,7 +319,7 @@ int vfio_pci_dma_buf_iommufd_map(struct dma_buf_attachment *attachment,
return -EOPNOTSUPP;
priv = attachment->dmabuf->priv;
- if (priv->revoked)
+ if (priv->status != VFIO_PCI_DMABUF_OK)
return -ENODEV;
/* More than one range to iommufd will require proper DMABUF support */
@@ -585,6 +586,63 @@ int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
return ret;
}
+static void __vfio_pci_dma_buf_revoke(struct vfio_pci_dma_buf *priv, bool revoked,
+ bool permanently)
+{
+ bool was_revoked;
+
+ lockdep_assert_held_write(&priv->vdev->memory_lock);
+
+ if ((priv->status == VFIO_PCI_DMABUF_PERM_REVOKED) ||
+ (priv->status == VFIO_PCI_DMABUF_OK && !revoked) ||
+ (priv->status == VFIO_PCI_DMABUF_TEMP_REVOKED && revoked && !permanently)) {
+ return;
+ }
+
+ dma_resv_lock(priv->dmabuf->resv, NULL);
+ was_revoked = priv->status != VFIO_PCI_DMABUF_OK;
+
+ if (revoked)
+ priv->status = permanently ? VFIO_PCI_DMABUF_PERM_REVOKED :
+ VFIO_PCI_DMABUF_TEMP_REVOKED;
+
+ /*
+ * If TEMP_REVOKED is being upgraded to PERM_REVOKED, the
+ * buffer is already gone. Don't wait on it again.
+ */
+ if (was_revoked && revoked) {
+ dma_resv_unlock(priv->dmabuf->resv);
+ return;
+ }
+
+ dma_buf_invalidate_mappings(priv->dmabuf);
+ dma_resv_wait_timeout(priv->dmabuf->resv,
+ DMA_RESV_USAGE_BOOKKEEP, false,
+ MAX_SCHEDULE_TIMEOUT);
+ dma_resv_unlock(priv->dmabuf->resv);
+ if (revoked) {
+ kref_put(&priv->kref, vfio_pci_dma_buf_done);
+ wait_for_completion(&priv->comp);
+ unmap_mapping_range(priv->dmabuf->file->f_mapping,
+ 0, priv->size, 1);
+ /*
+ * Re-arm the registered kref reference and the
+ * completion so the post-revoke state matches the
+ * post-creation state. An un-revoke followed by a
+ * new mapping needs the kref to be non-zero before
+ * kref_get(), and vfio_pci_dma_buf_cleanup()
+ * delegates its drain back through this revoke
+ * path on a possibly-already-revoked dma-buf.
+ */
+ kref_init(&priv->kref);
+ reinit_completion(&priv->comp);
+ } else {
+ dma_resv_lock(priv->dmabuf->resv, NULL);
+ priv->status = VFIO_PCI_DMABUF_OK;
+ dma_resv_unlock(priv->dmabuf->resv);
+ }
+}
+
void vfio_pci_dma_buf_move(struct vfio_pci_core_device *vdev, bool revoked)
{
struct vfio_pci_dma_buf *priv;
@@ -593,44 +651,13 @@ void vfio_pci_dma_buf_move(struct vfio_pci_core_device *vdev, bool revoked)
lockdep_assert_held_write(&vdev->memory_lock);
/*
* Holding memory_lock ensures a racing VMA fault observes
- * priv->revoked properly.
+ * priv->status properly.
*/
list_for_each_entry_safe(priv, tmp, &vdev->dmabufs, dmabufs_elm) {
if (!get_file_active(&priv->dmabuf->file))
continue;
-
- if (priv->revoked != revoked) {
- dma_resv_lock(priv->dmabuf->resv, NULL);
- if (revoked)
- priv->revoked = true;
- dma_buf_invalidate_mappings(priv->dmabuf);
- dma_resv_wait_timeout(priv->dmabuf->resv,
- DMA_RESV_USAGE_BOOKKEEP, false,
- MAX_SCHEDULE_TIMEOUT);
- dma_resv_unlock(priv->dmabuf->resv);
- if (revoked) {
- kref_put(&priv->kref, vfio_pci_dma_buf_done);
- wait_for_completion(&priv->comp);
- unmap_mapping_range(priv->dmabuf->file->f_mapping,
- 0, priv->size, 1);
- /*
- * Re-arm the registered kref reference and the
- * completion so the post-revoke state matches the
- * post-creation state. An un-revoke followed by a
- * new mapping needs the kref to be non-zero before
- * kref_get(), and vfio_pci_dma_buf_cleanup()
- * delegates its drain back through this revoke
- * path on a possibly-already-revoked dma-buf.
- */
- kref_init(&priv->kref);
- reinit_completion(&priv->comp);
- } else {
- dma_resv_lock(priv->dmabuf->resv, NULL);
- priv->revoked = false;
- dma_resv_unlock(priv->dmabuf->resv);
- }
- }
+ __vfio_pci_dma_buf_revoke(priv, revoked, false);
fput(priv->dmabuf->file);
}
}
@@ -662,3 +689,46 @@ void vfio_pci_dma_buf_cleanup(struct vfio_pci_core_device *vdev)
}
up_write(&vdev->memory_lock);
}
+
+#ifdef CONFIG_VFIO_PCI_DMABUF
+int vfio_pci_dma_buf_revoke(struct vfio_pci_core_device *vdev, int dmabuf_fd)
+{
+ struct vfio_pci_dma_buf *priv;
+ struct dma_buf *dmabuf;
+ int ret = 0;
+
+ dmabuf = dma_buf_get(dmabuf_fd);
+ if (IS_ERR(dmabuf))
+ return PTR_ERR(dmabuf);
+
+ priv = dmabuf->priv;
+ /*
+ * Sanity-check the DMABUF is really a vfio_pci_dma_buf _and_
+ * relates to the VFIO device it was provided with.
+ *
+ * If the DMABUF relates to this vdev then priv->vdev is
+ * stable because this open fd prevents cleanup.
+ *
+ * If it relates to a different vdev, reading priv->vdev might
+ * race with a concurrent cleanup on that device. But if so,
+ * it points to a non-matching vdev or NULL and is unusable
+ * either way.
+ */
+ if (dmabuf->ops != &vfio_pci_dmabuf_ops || priv->vdev != vdev) {
+ ret = -ENODEV;
+ goto out_put_buf;
+ }
+
+ scoped_guard(rwsem_write, &vdev->memory_lock) {
+ if (priv->status == VFIO_PCI_DMABUF_PERM_REVOKED)
+ ret = -EBADFD;
+ else
+ __vfio_pci_dma_buf_revoke(priv, true, true);
+ }
+
+ out_put_buf:
+ dma_buf_put(dmabuf);
+
+ return ret;
+}
+#endif /* CONFIG_VFIO_PCI_DMABUF */
diff --git a/drivers/vfio/pci/vfio_pci_priv.h b/drivers/vfio/pci/vfio_pci_priv.h
index db2e2aeae88f..a1e0f4fcb1dc 100644
--- a/drivers/vfio/pci/vfio_pci_priv.h
+++ b/drivers/vfio/pci/vfio_pci_priv.h
@@ -23,6 +23,12 @@ struct vfio_pci_ioeventfd {
bool test_mem;
};
+enum vfio_pci_dma_buf_status {
+ VFIO_PCI_DMABUF_OK = 0,
+ VFIO_PCI_DMABUF_TEMP_REVOKED = 1,
+ VFIO_PCI_DMABUF_PERM_REVOKED = 2,
+};
+
struct vfio_pci_dma_buf {
struct dma_buf *dmabuf;
struct vfio_pci_core_device *vdev;
@@ -35,7 +41,7 @@ struct vfio_pci_dma_buf {
struct kref kref;
struct completion comp;
unsigned long vma_pgoff_adjust;
- u8 revoked : 1;
+ enum vfio_pci_dma_buf_status status;
};
extern const struct vm_operations_struct vfio_pci_mmap_ops;
@@ -148,6 +154,7 @@ void vfio_pci_dma_buf_move(struct vfio_pci_core_device *vdev, bool revoked);
int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
struct vfio_device_feature_dma_buf __user *arg,
size_t argsz);
+int vfio_pci_dma_buf_revoke(struct vfio_pci_core_device *vdev, int dmabuf_fd);
#else
static inline int
vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
@@ -156,6 +163,11 @@ vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
{
return -ENOTTY;
}
+static inline int vfio_pci_dma_buf_revoke(struct vfio_pci_core_device *vdev,
+ int dmabuf_fd)
+{
+ return -ENODEV;
+}
#endif
#endif
diff --git a/include/uapi/linux/vfio.h b/include/uapi/linux/vfio.h
index 5de618a3a5ee..02366e9f8e16 100644
--- a/include/uapi/linux/vfio.h
+++ b/include/uapi/linux/vfio.h
@@ -1321,6 +1321,36 @@ struct vfio_precopy_info {
#define VFIO_MIG_GET_PRECOPY_INFO _IO(VFIO_TYPE, VFIO_BASE + 21)
+/**
+ * VFIO_DEVICE_PCI_DMABUF_REVOKE - _IO(VFIO_TYPE, VFIO_BASE + 22)
+ *
+ * This ioctl is used on the device FD, and requests that access to
+ * the buffer corresponding to the DMABUF FD parameter is immediately
+ * and permanently revoked. On successful return, the buffer is not
+ * accessible through any mmap() or dma-buf import. The request fails
+ * if the buffer is pinned; otherwise, the exporter marks the buffer
+ * as inaccessible and uses the move_notify callback to inform
+ * importers of the change. The buffer is permanently disabled, and
+ * VFIO refuses all map, mmap, attach, etc. requests.
+ *
+ * Returns:
+ *
+ * Return: 0 on success, -1 and errno set on failure:
+ *
+ * ENODEV if the associated dmabuf FD no longer exists/is closed,
+ * or is not a DMABUF created for this device.
+ * EINVAL if the dmabuf_fd parameter isn't a DMABUF.
+ * EBADF if the dmabuf_fd parameter isn't a valid file number.
+ * EBADFD if the buffer has already been revoked.
+ *
+ */
+struct vfio_pci_dmabuf_revoke {
+ __u32 argsz;
+ __s32 dmabuf_fd;
+};
+
+#define VFIO_DEVICE_PCI_DMABUF_REVOKE _IO(VFIO_TYPE, VFIO_BASE + 22)
+
/*
* Upon VFIO_DEVICE_FEATURE_SET, allow the device to be moved into a low power
* state with the platform-based power management. Device use of lower power
--
2.47.3
^ permalink raw reply related [flat|nested] 25+ messages in thread* Claude review: vfio/pci: Permanently revoke a DMABUF on request
2026-05-27 10:23 ` [PATCH v2 8/9] vfio/pci: Permanently revoke a DMABUF on request Matt Evans
@ 2026-05-28 2:28 ` Claude Code Review Bot
0 siblings, 0 replies; 25+ messages in thread
From: Claude Code Review Bot @ 2026-05-28 2:28 UTC (permalink / raw)
To: dri-devel-reviews
Patch Review
Adds three-state revocation (OK/TEMP_REVOKED/PERM_REVOKED) and a new `VFIO_DEVICE_PCI_DMABUF_REVOKE` ioctl.
**Ioctl number**: Uses `VFIO_BASE + 22` which is confirmed available.
**Security check in `vfio_pci_dma_buf_revoke`**:
```c
+ if (dmabuf->ops != &vfio_pci_dmabuf_ops || priv->vdev != vdev) {
```
Checking `dmabuf->ops` prevents accepting arbitrary dmabufs. Good. The `priv->vdev != vdev` check prevents cross-device revocation. The comment about racing cleanup on a different vdev is well-reasoned.
**The `__vfio_pci_dma_buf_revoke` function**: The early-return logic for state transitions is clear:
```c
+ if ((priv->status == VFIO_PCI_DMABUF_PERM_REVOKED) ||
+ (priv->status == VFIO_PCI_DMABUF_OK && !revoked) ||
+ (priv->status == VFIO_PCI_DMABUF_TEMP_REVOKED && revoked && !permanently)) {
+ return;
+ }
```
The third condition (temp-revoked and requesting non-permanent revoke) correctly avoids redundant work. TEMP->PERM upgrade skips the invalidation path since the buffer is already gone. Sound.
**Missing `flags` field question**: The author asks whether a `flags` field should be added to `struct vfio_pci_dmabuf_revoke`. Given the `argsz` pattern already allows extension, I'd say no -- add it when needed. UAPI extensibility is already handled.
---
Generated by Claude Code Patch Reviewer
^ permalink raw reply [flat|nested] 25+ messages in thread
* [PATCH v2 9/9] vfio/pci: Add mmap() attributes to DMABUF feature
2026-05-27 10:23 [PATCH v2 0/9] vfio/pci: Add mmap() for DMABUFs Matt Evans
` (7 preceding siblings ...)
2026-05-27 10:23 ` [PATCH v2 8/9] vfio/pci: Permanently revoke a DMABUF on request Matt Evans
@ 2026-05-27 10:23 ` Matt Evans
2026-05-28 2:28 ` Claude review: " Claude Code Review Bot
2026-05-28 2:28 ` Claude review: vfio/pci: Add mmap() for DMABUFs Claude Code Review Bot
9 siblings, 1 reply; 25+ messages in thread
From: Matt Evans @ 2026-05-27 10:23 UTC (permalink / raw)
To: Alex Williamson, Leon Romanovsky, Jason Gunthorpe, Alex Mastro,
Christian König, Bjorn Helgaas, Logan Gunthorpe
Cc: Mahmoud Adam, David Matlack, Björn Töpel, Sumit Semwal,
Kevin Tian, Ankit Agrawal, Pranjal Shrivastava, Alistair Popple,
Vivek Kasireddy, linux-kernel, linux-media, dri-devel,
linaro-mm-sig, kvm, linux-pci
A new VFIO feature, VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR, is added to
set (and get) CPU-facing memory type attributes for a DMABUF exported
from vfio-pci. These are used for subsequent mmap()s of the buffer.
There are two attributes supported:
- The default, VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR_UC
- VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR_WC, which results in WC
PTEs for the DMABUF's BAR region.
Signed-off-by: Matt Evans <mattev@meta.com>
---
drivers/vfio/pci/vfio_pci_core.c | 2 +
drivers/vfio/pci/vfio_pci_dmabuf.c | 70 +++++++++++++++++++++++++++++-
drivers/vfio/pci/vfio_pci_priv.h | 12 +++++
include/uapi/linux/vfio.h | 27 ++++++++++++
4 files changed, 110 insertions(+), 1 deletion(-)
diff --git a/drivers/vfio/pci/vfio_pci_core.c b/drivers/vfio/pci/vfio_pci_core.c
index 5184b3cac160..e256a925e7ce 100644
--- a/drivers/vfio/pci/vfio_pci_core.c
+++ b/drivers/vfio/pci/vfio_pci_core.c
@@ -1590,6 +1590,8 @@ int vfio_pci_core_ioctl_feature(struct vfio_device *device, u32 flags,
return vfio_pci_core_feature_token(vdev, flags, arg, argsz);
case VFIO_DEVICE_FEATURE_DMA_BUF:
return vfio_pci_core_feature_dma_buf(vdev, flags, arg, argsz);
+ case VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR:
+ return vfio_pci_core_feature_dma_buf_memattr(vdev, flags, arg, argsz);
default:
return -ENOTTY;
}
diff --git a/drivers/vfio/pci/vfio_pci_dmabuf.c b/drivers/vfio/pci/vfio_pci_dmabuf.c
index 3fa14760898f..db8b95ddbe18 100644
--- a/drivers/vfio/pci/vfio_pci_dmabuf.c
+++ b/drivers/vfio/pci/vfio_pci_dmabuf.c
@@ -42,7 +42,10 @@ static int vfio_pci_dma_buf_mmap(struct dma_buf *dmabuf, struct vm_area_struct *
* contained within the DMABUF size before calling this.
*/
- vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
+ if (READ_ONCE(priv->memattr) == VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR_WC)
+ vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot);
+ else
+ vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
vma->vm_page_prot = pgprot_decrypted(vma->vm_page_prot);
/* See comments in vfio_pci_core_mmap() re VM_ALLOW_ANY_UNCACHED. */
@@ -464,6 +467,7 @@ int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
priv->vdev = vdev;
priv->nr_ranges = get_dma_buf.nr_ranges;
priv->size = length;
+ priv->memattr = VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR_NC;
ret = vdev->pci_ops->get_dmabuf_phys(vdev, &priv->provider,
get_dma_buf.region_index,
priv->phys_vec, dma_ranges,
@@ -731,4 +735,68 @@ int vfio_pci_dma_buf_revoke(struct vfio_pci_core_device *vdev, int dmabuf_fd)
return ret;
}
+
+int vfio_pci_core_feature_dma_buf_memattr(
+ struct vfio_pci_core_device *vdev, u32 flags,
+ struct vfio_device_feature_dma_buf_memattr __user *arg,
+ size_t argsz)
+{
+ struct vfio_device_feature_dma_buf_memattr db_attr;
+ struct vfio_pci_dma_buf *priv;
+ struct dma_buf *dmabuf;
+ int ret;
+
+ if (!vdev->pci_ops || !vdev->pci_ops->get_dmabuf_phys)
+ return -EOPNOTSUPP;
+
+ ret = vfio_check_feature(flags, argsz,
+ VFIO_DEVICE_FEATURE_GET |
+ VFIO_DEVICE_FEATURE_SET,
+ sizeof(db_attr));
+ if (ret != 1)
+ return ret;
+
+ if (copy_from_user(&db_attr, arg, sizeof(db_attr)))
+ return -EFAULT;
+
+ dmabuf = dma_buf_get(db_attr.dmabuf_fd);
+ if (IS_ERR(dmabuf))
+ return PTR_ERR(dmabuf);
+
+ /* Verify DMABUF: see comments in vfio_pci_dma_buf_revoke() */
+ priv = dmabuf->priv;
+ if (dmabuf->ops != &vfio_pci_dmabuf_ops || priv->vdev != vdev) {
+ ret = -ENODEV;
+ goto out_put_buf;
+ }
+
+ ret = 0;
+ scoped_guard(rwsem_write, &vdev->memory_lock) {
+ uint32_t old_attr = priv->memattr;
+
+ if (flags & VFIO_DEVICE_FEATURE_SET) {
+ switch(db_attr.memattr) {
+ case VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR_NC:
+ case VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR_WC:
+ priv->memattr = db_attr.memattr;
+ break;
+
+ default:
+ ret = -ENOTSUPP;
+ }
+ }
+ db_attr.memattr = old_attr;
+ }
+
+ if (!ret && (flags & VFIO_DEVICE_FEATURE_GET)) {
+ if (copy_to_user(arg, &db_attr, sizeof(db_attr)))
+ ret = -EFAULT;
+ }
+
+ out_put_buf:
+ dma_buf_put(dmabuf);
+
+ return ret;
+
+}
#endif /* CONFIG_VFIO_PCI_DMABUF */
diff --git a/drivers/vfio/pci/vfio_pci_priv.h b/drivers/vfio/pci/vfio_pci_priv.h
index a1e0f4fcb1dc..8067be45beb0 100644
--- a/drivers/vfio/pci/vfio_pci_priv.h
+++ b/drivers/vfio/pci/vfio_pci_priv.h
@@ -41,6 +41,7 @@ struct vfio_pci_dma_buf {
struct kref kref;
struct completion comp;
unsigned long vma_pgoff_adjust;
+ u32 memattr;
enum vfio_pci_dma_buf_status status;
};
@@ -154,6 +155,10 @@ void vfio_pci_dma_buf_move(struct vfio_pci_core_device *vdev, bool revoked);
int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
struct vfio_device_feature_dma_buf __user *arg,
size_t argsz);
+int vfio_pci_core_feature_dma_buf_memattr(
+ struct vfio_pci_core_device *vdev, u32 flags,
+ struct vfio_device_feature_dma_buf_memattr __user *arg,
+ size_t argsz);
int vfio_pci_dma_buf_revoke(struct vfio_pci_core_device *vdev, int dmabuf_fd);
#else
static inline int
@@ -163,6 +168,13 @@ vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 flags,
{
return -ENOTTY;
}
+static inline int vfio_pci_core_feature_dma_buf_memattr(
+ struct vfio_pci_core_device *vdev, u32 flags,
+ struct vfio_device_feature_dma_buf_memattr __user *arg,
+ size_t argsz)
+{
+ return -ENODEV;
+}
static inline int vfio_pci_dma_buf_revoke(struct vfio_pci_core_device *vdev,
int dmabuf_fd)
{
diff --git a/include/uapi/linux/vfio.h b/include/uapi/linux/vfio.h
index 02366e9f8e16..9b0b68f8a1ef 100644
--- a/include/uapi/linux/vfio.h
+++ b/include/uapi/linux/vfio.h
@@ -1564,6 +1564,33 @@ struct vfio_device_feature_dma_buf {
*/
#define VFIO_DEVICE_FEATURE_MIG_PRECOPY_INFOv2 12
+/**
+ * Given a dma_buf fd previously created by
+ * VFIO_DEVICE_FEATURE_DMA_BUF, GET or SET the memory attribute that
+ * will be used by future mmap()s of that fd. SETting a new attribute
+ * does not affect existing VMAs.
+ *
+ * The default, if no previous SET has been performed, is NC.
+ *
+ * Return: 0 on success, -1 and errno is set on failure:
+ *
+ * ENOTSUPP: The given memattr is not supported.
+ * EBADF, EINVAL: dmabuf_fd is not a DMABUF fd.
+ * ENODEV: The dmabuf_fd does not match this VFIO device.
+ */
+#define VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR 13
+
+/* Valid memory attributes for the memattr field */
+enum vfio_device_dma_buf_memattr {
+ VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR_NC = 0, /* pgprot_noncached */
+ VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR_WC = 1, /* pgprot_writecombine */
+};
+
+struct vfio_device_feature_dma_buf_memattr {
+ __s32 dmabuf_fd;
+ __u32 memattr;
+};
+
/* -------- API for Type1 VFIO IOMMU -------- */
/**
--
2.47.3
^ permalink raw reply related [flat|nested] 25+ messages in thread* Claude review: vfio/pci: Add mmap() attributes to DMABUF feature
2026-05-27 10:23 ` [PATCH v2 9/9] vfio/pci: Add mmap() attributes to DMABUF feature Matt Evans
@ 2026-05-28 2:28 ` Claude Code Review Bot
0 siblings, 0 replies; 25+ messages in thread
From: Claude Code Review Bot @ 2026-05-28 2:28 UTC (permalink / raw)
To: dri-devel-reviews
Patch Review
Adds `VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR` for setting UC/WC memory attributes on exported DMABUFs.
**Bug -- wrong errno**:
```c
+ default:
+ ret = -ENOTSUPP;
```
`-ENOTSUPP` is `524` (kernel-internal, `EOPNOTSUPP` equivalent for in-kernel use only). For a UAPI-facing path that returns to userspace, this should be `-EOPNOTSUPP` (`95`). The docstring even says `ENOTSUPP` but userspace will see the raw `524` value which is non-standard. Fix:
```c
ret = -EOPNOTSUPP;
```
**GET+SET combined semantics**: When both `VFIO_DEVICE_FEATURE_GET` and `VFIO_DEVICE_FEATURE_SET` are specified, the code applies SET first, then returns the *old* value via GET:
```c
+ uint32_t old_attr = priv->memattr;
+ if (flags & VFIO_DEVICE_FEATURE_SET) { ... }
+ db_attr.memattr = old_attr;
+ }
+ if (!ret && (flags & VFIO_DEVICE_FEATURE_GET)) {
+ if (copy_to_user(arg, &db_attr, sizeof(db_attr)))
```
This "swap" semantic (set new, return old) is unusual and should be documented in the UAPI header. Users would typically expect GET to return the current (post-SET) value. Consider whether this is the intended behavior or a bug.
**`memattr` initialization for BAR mmap DMABUFs**: In `vfio_pci_core_mmap_prep_dmabuf()`, `priv->memattr` is not explicitly set, relying on `kzalloc` zero-initialization to give `VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR_NC = 0`. This is fine but fragile -- if the enum values ever change, this would break. The explicit `VFIO_DEVICE_FEATURE_DMA_BUF_MEMATTR_NC` enum name is used in patch 9 for the explicit export path, so it would be consistent to also set it explicitly in the mmap path. (Low priority.)
**Naming inconsistency**: The enum uses `_NC` (noncached) but the UAPI docstring says "The default ... is NC." Meanwhile patch 7 uses `pgprot_noncached`. The cover letter and commit message say "UC" (uncached). These are the same thing in practice, but the UAPI should pick one term and stick with it. `UC` is more conventional for MMIO.
---
Generated by Claude Code Patch Reviewer
^ permalink raw reply [flat|nested] 25+ messages in thread
* Claude review: vfio/pci: Add mmap() for DMABUFs
2026-05-27 10:23 [PATCH v2 0/9] vfio/pci: Add mmap() for DMABUFs Matt Evans
` (8 preceding siblings ...)
2026-05-27 10:23 ` [PATCH v2 9/9] vfio/pci: Add mmap() attributes to DMABUF feature Matt Evans
@ 2026-05-28 2:28 ` Claude Code Review Bot
9 siblings, 0 replies; 25+ messages in thread
From: Claude Code Review Bot @ 2026-05-28 2:28 UTC (permalink / raw)
To: dri-devel-reviews
Overall Series Review
Subject: vfio/pci: Add mmap() for DMABUFs
Author: Matt Evans <mattev@meta.com>
Patches: 15
Reviewed: 2026-05-28T12:28:51.795261
---
This is a well-structured v2 series from Matt Evans (Meta) that converts VFIO PCI BAR mmap() to be backed by DMABUFs, then adds mmap() support for explicitly-exported DMABUFs, plus ioctl-based permanent revocation and memory attribute control. The goal is to allow userspace drivers to safely delegate BAR sub-ranges to subordinate processes via DMABUF fds.
**Architecture**: The approach of unifying the BAR mmap path and explicit DMABUF mmap path under common fault-handling code is sound and follows prior review feedback. The lifecycle management -- with the BAR mmap DMABUF taking ownership of the device fd -- is a clever way to preserve existing behavior.
**Concerns**:
1. The `#ifdef CONFIG_PCI_P2PDMA` / `#ifdef CONFIG_VFIO_PCI_DMABUF` guards sprinkled through `vfio_pci_dmabuf.c` (patch 4) create a file that compiles in multiple configurations with interleaved `#ifdef` blocks, which is fragile.
2. Patch 7's fault handler synchronization protocol (resv lock -> try_get_registration -> unlock resv -> memory_lock) is complex with 5 numbered steps. While thoroughly commented, this is a subtle ordering that future maintainers need to preserve.
3. The `bar_needs_zap` flag (patch 6) for nvgrace-gpu is a reasonable interim hack but deserves an actual timeline for removal.
4. Patch 9's use of `-ENOTSUPP` (kernel-internal) instead of `-EOPNOTSUPP` (POSIX) in a UAPI-facing path is a bug.
Overall the series is in good shape for a v2. The code is well-commented, the cover letter is thorough, and the incremental approach is easy to follow.
---
---
Generated by Claude Code Patch Reviewer
^ permalink raw reply [flat|nested] 25+ messages in thread