Default IO_TLB_SEGSIZE (128) slabs may be not enough for some use cases.
This series adds support to customize io_tlb_segsize for each
restricted-dma-pool.
Example use case:
mtk-isp drivers[1] are controlled by mtk-scp[2] and allocate memory through
mtk-scp. In order to use the noncontiguous DMA API[3], we need to use
the swiotlb pool. mtk-scp needs to allocate memory with 2560 slabs.
mtk-isp drivers also needs to allocate memory with 200+ slabs. Both are
larger than the default IO_TLB_SEGSIZE (128) slabs.
[1] (not in upstream) https://patchwork.kernel.org/project/linux-media/cover/[email protected]/
[2] https://elixir.bootlin.com/linux/latest/source/drivers/remoteproc/mtk_scp.c
[3] https://patchwork.kernel.org/project/linux-media/cover/[email protected]/
Hsin-Yi Wang (3):
dma: swiotlb: Allow restricted-dma-pool to customize IO_TLB_SEGSIZE
dt-bindings: Add io-tlb-segsize property for restricted-dma-pool
arm64: dts: mt8183: use restricted swiotlb for scp mem
.../reserved-memory/shared-dma-pool.yaml | 8 +++++
.../arm64/boot/dts/mediatek/mt8183-kukui.dtsi | 4 +--
include/linux/swiotlb.h | 1 +
kernel/dma/swiotlb.c | 34 ++++++++++++++-----
4 files changed, 37 insertions(+), 10 deletions(-)
--
2.34.0.rc2.393.gf8c9666880-goog
Default IO_TLB_SEGSIZE is 128, but some use cases requires more slabs.
Otherwise swiotlb_find_slots() will fail.
This patch allows each mem pool to decide their own io-tlb-segsize
through dt property.
Signed-off-by: Hsin-Yi Wang <[email protected]>
---
include/linux/swiotlb.h | 1 +
kernel/dma/swiotlb.c | 34 ++++++++++++++++++++++++++--------
2 files changed, 27 insertions(+), 8 deletions(-)
diff --git a/include/linux/swiotlb.h b/include/linux/swiotlb.h
index 569272871375c4..73b3312f23e65b 100644
--- a/include/linux/swiotlb.h
+++ b/include/linux/swiotlb.h
@@ -95,6 +95,7 @@ struct io_tlb_mem {
unsigned long nslabs;
unsigned long used;
unsigned int index;
+ unsigned int io_tlb_segsize;
spinlock_t lock;
struct dentry *debugfs;
bool late_alloc;
diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
index 8e840fbbed7c7a..021eef1844ca4c 100644
--- a/kernel/dma/swiotlb.c
+++ b/kernel/dma/swiotlb.c
@@ -145,9 +145,10 @@ void swiotlb_print_info(void)
(mem->nslabs << IO_TLB_SHIFT) >> 20);
}
-static inline unsigned long io_tlb_offset(unsigned long val)
+static inline unsigned long io_tlb_offset(unsigned long val,
+ unsigned long io_tlb_segsize)
{
- return val & (IO_TLB_SEGSIZE - 1);
+ return val & (io_tlb_segsize - 1);
}
static inline unsigned long nr_slots(u64 val)
@@ -186,13 +187,16 @@ static void swiotlb_init_io_tlb_mem(struct io_tlb_mem *mem, phys_addr_t start,
mem->end = mem->start + bytes;
mem->index = 0;
mem->late_alloc = late_alloc;
+ if (!mem->io_tlb_segsize)
+ mem->io_tlb_segsize = IO_TLB_SEGSIZE;
if (swiotlb_force == SWIOTLB_FORCE)
mem->force_bounce = true;
spin_lock_init(&mem->lock);
for (i = 0; i < mem->nslabs; i++) {
- mem->slots[i].list = IO_TLB_SEGSIZE - io_tlb_offset(i);
+ mem->slots[i].list = mem->io_tlb_segsize -
+ io_tlb_offset(i, mem->io_tlb_segsize);
mem->slots[i].orig_addr = INVALID_PHYS_ADDR;
mem->slots[i].alloc_size = 0;
}
@@ -523,7 +527,7 @@ static int swiotlb_find_slots(struct device *dev, phys_addr_t orig_addr,
alloc_size - (offset + ((i - index) << IO_TLB_SHIFT));
}
for (i = index - 1;
- io_tlb_offset(i) != IO_TLB_SEGSIZE - 1 &&
+ io_tlb_offset(i, mem->io_tlb_segsize) != mem->io_tlb_segsize - 1 &&
mem->slots[i].list; i--)
mem->slots[i].list = ++count;
@@ -603,7 +607,7 @@ static void swiotlb_release_slots(struct device *dev, phys_addr_t tlb_addr)
* with slots below and above the pool being returned.
*/
spin_lock_irqsave(&mem->lock, flags);
- if (index + nslots < ALIGN(index + 1, IO_TLB_SEGSIZE))
+ if (index + nslots < ALIGN(index + 1, mem->io_tlb_segsize))
count = mem->slots[index + nslots].list;
else
count = 0;
@@ -623,8 +627,8 @@ static void swiotlb_release_slots(struct device *dev, phys_addr_t tlb_addr)
* available (non zero)
*/
for (i = index - 1;
- io_tlb_offset(i) != IO_TLB_SEGSIZE - 1 && mem->slots[i].list;
- i--)
+ io_tlb_offset(i, mem->io_tlb_segsize) != mem->io_tlb_segsize - 1 &&
+ mem->slots[i].list; i--)
mem->slots[i].list = ++count;
mem->used -= nslots;
spin_unlock_irqrestore(&mem->lock, flags);
@@ -701,7 +705,9 @@ dma_addr_t swiotlb_map(struct device *dev, phys_addr_t paddr, size_t size,
size_t swiotlb_max_mapping_size(struct device *dev)
{
- return ((size_t)IO_TLB_SIZE) * IO_TLB_SEGSIZE;
+ struct io_tlb_mem *mem = dev->dma_io_tlb_mem;
+
+ return ((size_t)IO_TLB_SIZE) * mem->io_tlb_segsize;
}
bool is_swiotlb_active(struct device *dev)
@@ -788,6 +794,7 @@ static int rmem_swiotlb_device_init(struct reserved_mem *rmem,
{
struct io_tlb_mem *mem = rmem->priv;
unsigned long nslabs = rmem->size >> IO_TLB_SHIFT;
+ struct device_node *np;
/*
* Since multiple devices can share the same pool, the private data,
@@ -808,6 +815,17 @@ static int rmem_swiotlb_device_init(struct reserved_mem *rmem,
set_memory_decrypted((unsigned long)phys_to_virt(rmem->base),
rmem->size >> PAGE_SHIFT);
+
+ np = of_find_node_by_phandle(rmem->phandle);
+ if (np) {
+ if (!of_property_read_u32(np, "io-tlb-segsize",
+ &mem->io_tlb_segsize)) {
+ if (hweight32(mem->io_tlb_segsize) != 1)
+ mem->io_tlb_segsize = IO_TLB_SEGSIZE;
+ }
+ of_node_put(np);
+ }
+
swiotlb_init_io_tlb_mem(mem, rmem->base, nslabs, false);
mem->force_bounce = true;
mem->for_alloc = true;
--
2.34.0.rc2.393.gf8c9666880-goog
Add a io-tlb-segsize property that each restricted-dma-pool can set its
own io_tlb_segsize since some use cases require slabs larger than default
value (128).
Signed-off-by: Hsin-Yi Wang <[email protected]>
---
.../bindings/reserved-memory/shared-dma-pool.yaml | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/Documentation/devicetree/bindings/reserved-memory/shared-dma-pool.yaml b/Documentation/devicetree/bindings/reserved-memory/shared-dma-pool.yaml
index a4bf757d6881de..6198bf6b76f0b2 100644
--- a/Documentation/devicetree/bindings/reserved-memory/shared-dma-pool.yaml
+++ b/Documentation/devicetree/bindings/reserved-memory/shared-dma-pool.yaml
@@ -56,6 +56,14 @@ properties:
If this property is present, then Linux will use the region for
the default pool of the consistent DMA allocator.
+ io-tlb-segsize:
+ type: u32
+ description: >
+ Each restricted-dma-pool can use this property to set its own
+ io_tlb_segsize. If not set, it will use the default value
+ IO_TLB_SEGSIZE defined in include/linux/swiotlb.h. The value has
+ to be a power of 2, otherwise it will fall back to IO_TLB_SEGSIZE.
+
unevaluatedProperties: false
examples:
--
2.34.0.rc2.393.gf8c9666880-goog
Use restricted-dma-pool for mtk_scp's reserved memory. And set the
io-tlb-segsize to 4096 since the driver needs at least 2560 slabs to
allocate memory.
Signed-off-by: Hsin-Yi Wang <[email protected]>
---
arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi b/arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi
index 94c13c45919445..de94b2fd7f33e7 100644
--- a/arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi
+++ b/arch/arm64/boot/dts/mediatek/mt8183-kukui.dtsi
@@ -109,9 +109,9 @@ reserved_memory: reserved-memory {
ranges;
scp_mem_reserved: scp_mem_region {
- compatible = "shared-dma-pool";
+ compatible = "restricted-dma-pool";
reg = <0 0x50000000 0 0x2900000>;
- no-map;
+ io-tlb-segsize = <4096>;
};
};
--
2.34.0.rc2.393.gf8c9666880-goog
On 2021-11-23 11:21, Hsin-Yi Wang wrote:
> Default IO_TLB_SEGSIZE (128) slabs may be not enough for some use cases.
> This series adds support to customize io_tlb_segsize for each
> restricted-dma-pool.
>
> Example use case:
>
> mtk-isp drivers[1] are controlled by mtk-scp[2] and allocate memory through
> mtk-scp. In order to use the noncontiguous DMA API[3], we need to use
> the swiotlb pool. mtk-scp needs to allocate memory with 2560 slabs.
> mtk-isp drivers also needs to allocate memory with 200+ slabs. Both are
> larger than the default IO_TLB_SEGSIZE (128) slabs.
Are drivers really doing streaming DMA mappings that large? If so, that
seems like it might be worth trying to address in its own right for the
sake of efficiency - allocating ~5MB of memory twice and copying it back
and forth doesn't sound like the ideal thing to do.
If it's really about coherent DMA buffer allocation, I thought the plan
was that devices which expect to use a significant amount and/or size of
coherent buffers would continue to use a shared-dma-pool for that? It's
still what the binding implies. My understanding was that
swiotlb_alloc() is mostly just a fallback for the sake of drivers which
mostly do streaming DMA but may allocate a handful of pages worth of
coherent buffers here and there. Certainly looking at the mtk_scp
driver, that seems like it shouldn't be going anywhere near SWIOTLB at all.
Robin.
> [1] (not in upstream) https://patchwork.kernel.org/project/linux-media/cover/[email protected]/
> [2] https://elixir.bootlin.com/linux/latest/source/drivers/remoteproc/mtk_scp.c
> [3] https://patchwork.kernel.org/project/linux-media/cover/[email protected]/
>
> Hsin-Yi Wang (3):
> dma: swiotlb: Allow restricted-dma-pool to customize IO_TLB_SEGSIZE
> dt-bindings: Add io-tlb-segsize property for restricted-dma-pool
> arm64: dts: mt8183: use restricted swiotlb for scp mem
>
> .../reserved-memory/shared-dma-pool.yaml | 8 +++++
> .../arm64/boot/dts/mediatek/mt8183-kukui.dtsi | 4 +--
> include/linux/swiotlb.h | 1 +
> kernel/dma/swiotlb.c | 34 ++++++++++++++-----
> 4 files changed, 37 insertions(+), 10 deletions(-)
>
On Tue, 23 Nov 2021 19:21:03 +0800, Hsin-Yi Wang wrote:
> Add a io-tlb-segsize property that each restricted-dma-pool can set its
> own io_tlb_segsize since some use cases require slabs larger than default
> value (128).
>
> Signed-off-by: Hsin-Yi Wang <[email protected]>
> ---
> .../bindings/reserved-memory/shared-dma-pool.yaml | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
My bot found errors running 'make DT_CHECKER_FLAGS=-m dt_binding_check'
on your patch (DT_CHECKER_FLAGS is new in v5.13):
yamllint warnings/errors:
dtschema/dtc warnings/errors:
/builds/robherring/linux-dt-review/Documentation/devicetree/bindings/reserved-memory/shared-dma-pool.yaml: properties:io-tlb-segsize:type: 'anyOf' conditional failed, one must be fixed:
'u32' is not one of ['array', 'boolean', 'integer', 'null', 'number', 'object', 'string']
'u32' is not of type 'array'
from schema $id: http://json-schema.org/draft-07/schema#
/builds/robherring/linux-dt-review/Documentation/devicetree/bindings/reserved-memory/shared-dma-pool.yaml: properties:io-tlb-segsize:type: 'u32' is not one of ['boolean', 'object']
from schema $id: http://devicetree.org/meta-schemas/core.yaml#
/builds/robherring/linux-dt-review/Documentation/devicetree/bindings/reserved-memory/shared-dma-pool.yaml: ignoring, error in schema: properties: io-tlb-segsize: type
warning: no schema found in file: ./Documentation/devicetree/bindings/reserved-memory/shared-dma-pool.yaml
Documentation/devicetree/bindings/display/msm/gpu.example.dt.yaml:0:0: /example-1/reserved-memory/gpu@8f200000: failed to match any schema with compatible: ['shared-dma-pool']
Documentation/devicetree/bindings/reserved-memory/shared-dma-pool.example.dt.yaml:0:0: /example-0/reserved-memory/linux,cma: failed to match any schema with compatible: ['shared-dma-pool']
Documentation/devicetree/bindings/reserved-memory/shared-dma-pool.example.dt.yaml:0:0: /example-0/reserved-memory/restricted-dma-pool@50000000: failed to match any schema with compatible: ['restricted-dma-pool']
Documentation/devicetree/bindings/dsp/fsl,dsp.example.dt.yaml:0:0: /example-1/vdev0buffer@94300000: failed to match any schema with compatible: ['shared-dma-pool']
Documentation/devicetree/bindings/remoteproc/ti,omap-remoteproc.example.dt.yaml:0:0: /example-0/reserved-memory/dsp-memory@98000000: failed to match any schema with compatible: ['shared-dma-pool']
Documentation/devicetree/bindings/remoteproc/ti,omap-remoteproc.example.dt.yaml:0:0: /example-1/reserved-memory/ipu-memory@95800000: failed to match any schema with compatible: ['shared-dma-pool']
Documentation/devicetree/bindings/remoteproc/ti,omap-remoteproc.example.dt.yaml:0:0: /example-2/reserved-memory/dsp1-memory@99000000: failed to match any schema with compatible: ['shared-dma-pool']
Documentation/devicetree/bindings/sound/google,cros-ec-codec.example.dt.yaml:0:0: /example-0/reserved-mem@52800000: failed to match any schema with compatible: ['shared-dma-pool']
doc reference errors (make refcheckdocs):
See https://patchwork.ozlabs.org/patch/1558503
This check can fail if there are any dependencies. The base for a patch
series is generally the most recent rc1.
If you already ran 'make dt_binding_check' and didn't see the above
error(s), then make sure 'yamllint' is installed and dt-schema is up to
date:
pip3 install dtschema --upgrade
Please check and re-submit.
On Tue, Nov 23, 2021 at 7:58 PM Robin Murphy <[email protected]> wrote:
>
> On 2021-11-23 11:21, Hsin-Yi Wang wrote:
> > Default IO_TLB_SEGSIZE (128) slabs may be not enough for some use cases.
> > This series adds support to customize io_tlb_segsize for each
> > restricted-dma-pool.
> >
> > Example use case:
> >
> > mtk-isp drivers[1] are controlled by mtk-scp[2] and allocate memory through
> > mtk-scp. In order to use the noncontiguous DMA API[3], we need to use
> > the swiotlb pool. mtk-scp needs to allocate memory with 2560 slabs.
> > mtk-isp drivers also needs to allocate memory with 200+ slabs. Both are
> > larger than the default IO_TLB_SEGSIZE (128) slabs.
>
> Are drivers really doing streaming DMA mappings that large? If so, that
> seems like it might be worth trying to address in its own right for the
> sake of efficiency - allocating ~5MB of memory twice and copying it back
> and forth doesn't sound like the ideal thing to do.
>
> If it's really about coherent DMA buffer allocation, I thought the plan
> was that devices which expect to use a significant amount and/or size of
> coherent buffers would continue to use a shared-dma-pool for that? It's
> still what the binding implies. My understanding was that
> swiotlb_alloc() is mostly just a fallback for the sake of drivers which
> mostly do streaming DMA but may allocate a handful of pages worth of
> coherent buffers here and there. Certainly looking at the mtk_scp
> driver, that seems like it shouldn't be going anywhere near SWIOTLB at all.
>
mtk_scp on its own can use the shared-dma-pool, which it currently uses.
The reason we switched to restricted-dma-pool is that we want to use
the noncontiguous DMA API for mtk-isp. The noncontiguous DMA API is
designed for devices with iommu, and if a device doesn't have an
iommu, it will fallback using swiotlb. But currently noncontiguous DMA
API doesn't work with the shared-dma-pool.
vb2_dc_alloc() -> dma_alloc_noncontiguous() -> alloc_single_sgt() ->
__dma_alloc_pages() -> dma_direct_alloc_pages() ->
__dma_direct_alloc_pages() -> swiotlb_alloc().
> Robin.
>
> > [1] (not in upstream) https://patchwork.kernel.org/project/linux-media/cover/[email protected]/
> > [2] https://elixir.bootlin.com/linux/latest/source/drivers/remoteproc/mtk_scp.c
> > [3] https://patchwork.kernel.org/project/linux-media/cover/[email protected]/
> >
> > Hsin-Yi Wang (3):
> > dma: swiotlb: Allow restricted-dma-pool to customize IO_TLB_SEGSIZE
> > dt-bindings: Add io-tlb-segsize property for restricted-dma-pool
> > arm64: dts: mt8183: use restricted swiotlb for scp mem
> >
> > .../reserved-memory/shared-dma-pool.yaml | 8 +++++
> > .../arm64/boot/dts/mediatek/mt8183-kukui.dtsi | 4 +--
> > include/linux/swiotlb.h | 1 +
> > kernel/dma/swiotlb.c | 34 ++++++++++++++-----
> > 4 files changed, 37 insertions(+), 10 deletions(-)
> >
On 2021-11-24 03:55, Hsin-Yi Wang wrote:
> On Tue, Nov 23, 2021 at 7:58 PM Robin Murphy <[email protected]> wrote:
>>
>> On 2021-11-23 11:21, Hsin-Yi Wang wrote:
>>> Default IO_TLB_SEGSIZE (128) slabs may be not enough for some use cases.
>>> This series adds support to customize io_tlb_segsize for each
>>> restricted-dma-pool.
>>>
>>> Example use case:
>>>
>>> mtk-isp drivers[1] are controlled by mtk-scp[2] and allocate memory through
>>> mtk-scp. In order to use the noncontiguous DMA API[3], we need to use
>>> the swiotlb pool. mtk-scp needs to allocate memory with 2560 slabs.
>>> mtk-isp drivers also needs to allocate memory with 200+ slabs. Both are
>>> larger than the default IO_TLB_SEGSIZE (128) slabs.
>>
>> Are drivers really doing streaming DMA mappings that large? If so, that
>> seems like it might be worth trying to address in its own right for the
>> sake of efficiency - allocating ~5MB of memory twice and copying it back
>> and forth doesn't sound like the ideal thing to do.
>>
>> If it's really about coherent DMA buffer allocation, I thought the plan
>> was that devices which expect to use a significant amount and/or size of
>> coherent buffers would continue to use a shared-dma-pool for that? It's
>> still what the binding implies. My understanding was that
>> swiotlb_alloc() is mostly just a fallback for the sake of drivers which
>> mostly do streaming DMA but may allocate a handful of pages worth of
>> coherent buffers here and there. Certainly looking at the mtk_scp
>> driver, that seems like it shouldn't be going anywhere near SWIOTLB at all.
>>
> mtk_scp on its own can use the shared-dma-pool, which it currently uses.
> The reason we switched to restricted-dma-pool is that we want to use
> the noncontiguous DMA API for mtk-isp. The noncontiguous DMA API is
> designed for devices with iommu, and if a device doesn't have an
> iommu, it will fallback using swiotlb. But currently noncontiguous DMA
> API doesn't work with the shared-dma-pool.
>
> vb2_dc_alloc() -> dma_alloc_noncontiguous() -> alloc_single_sgt() ->
> __dma_alloc_pages() -> dma_direct_alloc_pages() ->
> __dma_direct_alloc_pages() -> swiotlb_alloc().
OK, thanks for clarifying. My gut feeling is that drivers should
probably only be calling the noncontiguous API when they *know* that
they have a scatter-gather-capable device or IOMMU that can cope with
it, but either way I'm still not convinced that it makes sense to hack
up SWIOTLB with DT ABI baggage for an obscure fallback case. It would
seem a lot more sensible to fix alloc_single_sgt() to not ignore
per-device pools once it has effectively fallen back to the normal
dma_alloc_attrs() flow, but I guess that's not technically guaranteed to
uphold the assumption that we can allocate struct-page-backed memory.
Still, if we've got to the point of needing to use a SWIOTLB pool as
nothing more than a bad reinvention of CMA, rather than an actual bounce
buffer, that reeks of a fundamental design issue and adding more hacks
on top to bodge around it is not the right way to go - we need to take a
step back and properly reconsider how dma_alloc_noncontiguous() is
supposed to interact with DMA protection schemes.
Thanks,
Robin.
>>> [1] (not in upstream) https://patchwork.kernel.org/project/linux-media/cover/[email protected]/
>>> [2] https://elixir.bootlin.com/linux/latest/source/drivers/remoteproc/mtk_scp.c
>>> [3] https://patchwork.kernel.org/project/linux-media/cover/[email protected]/
>>>
>>> Hsin-Yi Wang (3):
>>> dma: swiotlb: Allow restricted-dma-pool to customize IO_TLB_SEGSIZE
>>> dt-bindings: Add io-tlb-segsize property for restricted-dma-pool
>>> arm64: dts: mt8183: use restricted swiotlb for scp mem
>>>
>>> .../reserved-memory/shared-dma-pool.yaml | 8 +++++
>>> .../arm64/boot/dts/mediatek/mt8183-kukui.dtsi | 4 +--
>>> include/linux/swiotlb.h | 1 +
>>> kernel/dma/swiotlb.c | 34 ++++++++++++++-----
>>> 4 files changed, 37 insertions(+), 10 deletions(-)
>>>
> _______________________________________________
> iommu mailing list
> [email protected]
> https://lists.linuxfoundation.org/mailman/listinfo/iommu
>
Hi Robin,
On Tue, Nov 23, 2021 at 8:59 PM Robin Murphy <[email protected]> wrote:
>
> On 2021-11-23 11:21, Hsin-Yi Wang wrote:
> > Default IO_TLB_SEGSIZE (128) slabs may be not enough for some use cases.
> > This series adds support to customize io_tlb_segsize for each
> > restricted-dma-pool.
> >
> > Example use case:
> >
> > mtk-isp drivers[1] are controlled by mtk-scp[2] and allocate memory through
> > mtk-scp. In order to use the noncontiguous DMA API[3], we need to use
> > the swiotlb pool. mtk-scp needs to allocate memory with 2560 slabs.
> > mtk-isp drivers also needs to allocate memory with 200+ slabs. Both are
> > larger than the default IO_TLB_SEGSIZE (128) slabs.
>
> Are drivers really doing streaming DMA mappings that large? If so, that
> seems like it might be worth trying to address in its own right for the
> sake of efficiency - allocating ~5MB of memory twice and copying it back
> and forth doesn't sound like the ideal thing to do.
>
> If it's really about coherent DMA buffer allocation, I thought the plan
> was that devices which expect to use a significant amount and/or size of
> coherent buffers would continue to use a shared-dma-pool for that? It's
> still what the binding implies. My understanding was that
> swiotlb_alloc() is mostly just a fallback for the sake of drivers which
> mostly do streaming DMA but may allocate a handful of pages worth of
> coherent buffers here and there. Certainly looking at the mtk_scp
> driver, that seems like it shouldn't be going anywhere near SWIOTLB at all.
First, thanks a lot for taking a look at this patch series.
The drivers would do streaming DMA within a reserved region that is
the only memory accessible to them for security reasons. This seems to
exactly match the definition of the restricted pool as merged
recently.
The new dma_alloc_noncontiguous() API would allow allocating suitable
memory directly from the pool, which would eliminate the need to copy.
However, for a restricted pool, this would exercise the SWIOTLB
allocator, which currently suffers from the limitation as described by
Hsin-Yi. Since the allocator in general is quite general purpose and
already used for coherent allocations as per the current restricted
pool implementation, I think it indeed makes sense to lift the
limitation, rather than trying to come up with yet another thing.
Best regards,
Tomasz
>
> Robin.
>
> > [1] (not in upstream) https://patchwork.kernel.org/project/linux-media/cover/[email protected]/
> > [2] https://elixir.bootlin.com/linux/latest/source/drivers/remoteproc/mtk_scp.c
> > [3] https://patchwork.kernel.org/project/linux-media/cover/[email protected]/
> >
> > Hsin-Yi Wang (3):
> > dma: swiotlb: Allow restricted-dma-pool to customize IO_TLB_SEGSIZE
> > dt-bindings: Add io-tlb-segsize property for restricted-dma-pool
> > arm64: dts: mt8183: use restricted swiotlb for scp mem
> >
> > .../reserved-memory/shared-dma-pool.yaml | 8 +++++
> > .../arm64/boot/dts/mediatek/mt8183-kukui.dtsi | 4 +--
> > include/linux/swiotlb.h | 1 +
> > kernel/dma/swiotlb.c | 34 ++++++++++++++-----
> > 4 files changed, 37 insertions(+), 10 deletions(-)
> >
On 2021-11-25 07:35, Tomasz Figa wrote:
> Hi Robin,
>
> On Tue, Nov 23, 2021 at 8:59 PM Robin Murphy <[email protected]> wrote:
>>
>> On 2021-11-23 11:21, Hsin-Yi Wang wrote:
>>> Default IO_TLB_SEGSIZE (128) slabs may be not enough for some use cases.
>>> This series adds support to customize io_tlb_segsize for each
>>> restricted-dma-pool.
>>>
>>> Example use case:
>>>
>>> mtk-isp drivers[1] are controlled by mtk-scp[2] and allocate memory through
>>> mtk-scp. In order to use the noncontiguous DMA API[3], we need to use
>>> the swiotlb pool. mtk-scp needs to allocate memory with 2560 slabs.
>>> mtk-isp drivers also needs to allocate memory with 200+ slabs. Both are
>>> larger than the default IO_TLB_SEGSIZE (128) slabs.
>>
>> Are drivers really doing streaming DMA mappings that large? If so, that
>> seems like it might be worth trying to address in its own right for the
>> sake of efficiency - allocating ~5MB of memory twice and copying it back
>> and forth doesn't sound like the ideal thing to do.
>>
>> If it's really about coherent DMA buffer allocation, I thought the plan
>> was that devices which expect to use a significant amount and/or size of
>> coherent buffers would continue to use a shared-dma-pool for that? It's
>> still what the binding implies. My understanding was that
>> swiotlb_alloc() is mostly just a fallback for the sake of drivers which
>> mostly do streaming DMA but may allocate a handful of pages worth of
>> coherent buffers here and there. Certainly looking at the mtk_scp
>> driver, that seems like it shouldn't be going anywhere near SWIOTLB at all.
>
> First, thanks a lot for taking a look at this patch series.
>
> The drivers would do streaming DMA within a reserved region that is
> the only memory accessible to them for security reasons. This seems to
> exactly match the definition of the restricted pool as merged
> recently.
Huh? Of the drivers indicated, the SCP driver is doing nothing but
coherent allocations, and I'm not entirely sure what those ISP driver
patches are supposed to be doing but I suspect it's probably just buffer
allocation too. I don't see any actual streaming DMA anywhere :/
> The new dma_alloc_noncontiguous() API would allow allocating suitable
> memory directly from the pool, which would eliminate the need to copy.
Can you clarify what's being copied, and where? I'm not all that
familiar with the media APIs, but I thought it was all based around
preallocated DMA buffers (the whole dedicated "videobuf" thing)? The few
instances of actual streaming DMA I can see in drivers/media/ look to be
mostly PCI drivers mapping private descriptors, whereas the MTK ISP
appears to be entirely register-based.
> However, for a restricted pool, this would exercise the SWIOTLB
> allocator, which currently suffers from the limitation as described by
> Hsin-Yi. Since the allocator in general is quite general purpose and
> already used for coherent allocations as per the current restricted
> pool implementation, I think it indeed makes sense to lift the
> limitation, rather than trying to come up with yet another thing.
No, just fix the dma_alloc_noncontiguous() fallback case to split the
allocation into dma_max_mapping_size() chunks. *That* makes sense.
Thanks,
Robin.
>
> Best regards,
> Tomasz
>
>>
>> Robin.
>>
>>> [1] (not in upstream) https://patchwork.kernel.org/project/linux-media/cover/[email protected]/
>>> [2] https://elixir.bootlin.com/linux/latest/source/drivers/remoteproc/mtk_scp.c
>>> [3] https://patchwork.kernel.org/project/linux-media/cover/[email protected]/
>>>
>>> Hsin-Yi Wang (3):
>>> dma: swiotlb: Allow restricted-dma-pool to customize IO_TLB_SEGSIZE
>>> dt-bindings: Add io-tlb-segsize property for restricted-dma-pool
>>> arm64: dts: mt8183: use restricted swiotlb for scp mem
>>>
>>> .../reserved-memory/shared-dma-pool.yaml | 8 +++++
>>> .../arm64/boot/dts/mediatek/mt8183-kukui.dtsi | 4 +--
>>> include/linux/swiotlb.h | 1 +
>>> kernel/dma/swiotlb.c | 34 ++++++++++++++-----
>>> 4 files changed, 37 insertions(+), 10 deletions(-)
>>>