2009-07-10 01:47:55

by FUJITA Tomonori

[permalink] [raw]
Subject: removing addr_needs_map in struct dma_mapping_ops

I'm trying to convert POWERPC to use asm-generic/dma-mapping-common.h.

POWERPC needs addr_needs_map() in struct dma_mapping_ops for SWIOTLB
support but I want to avoid add addr_needs_map() in struct
dma_map_ops. IIRC, you guys think it as a temporary solution and
talked about defining something like struct dma_data. Then we could

struct dev_archdata {
...

struct dma_data *ddata;
};

or

struct dev_archdata {
...

struct dma_data ddata;
};


struct dma_data needs dma_direct_offset, iommu_table, dma_base, and
dma_window_size, anything else?


Is it acceptable?


2009-07-13 21:51:06

by Becky Bruce

[permalink] [raw]
Subject: Re: removing addr_needs_map in struct dma_mapping_ops


On Jul 9, 2009, at 8:47 PM, FUJITA Tomonori wrote:

> I'm trying to convert POWERPC to use asm-generic/dma-mapping-common.h.
>
> POWERPC needs addr_needs_map() in struct dma_mapping_ops for SWIOTLB
> support but I want to avoid add addr_needs_map() in struct
> dma_map_ops. IIRC, you guys think it as a temporary solution and

That is correct - I was planning to update when the iotlb folks had
decided on a final scheme for this. We don't like the additional
indirection we're getting with the current implementation.

>
> talked about defining something like struct dma_data. Then we could
>
> struct dev_archdata {
> ...
>
> struct dma_data *ddata;
> };
>
> or
>
> struct dev_archdata {
> ...
>
> struct dma_data ddata;
> };
>
>
> struct dma_data needs dma_direct_offset, iommu_table, dma_base, and
> dma_window_size, anything else?

IIRC, what we had talked about was simpler - we talked about changing
the current dev_archdata from this:

struct dev_archdata {
struct device_node *of_node;
struct dma_mapping_ops *dma_ops;
void *dma_data;
};

to this:

struct dev_archdata {
struct device_node *of_node;
struct dma_mapping_ops *dma_ops;
unsigned long long dma_data;
#ifdef CONFIG_SWIOTLB
dma_addr_t max_direct_dma_addr;
#endif
};

Where max_direct_dma_addr is the address beyond which a specific
device must use swiotlb, and dma_data is the offset like it is now
(but wider on 32-bit systems than void *). I believe Ben had mentioned
wanting to make the max_direct_dma_addr part conditional so we don't
bloat archdata on platforms that don't ever bounce.

The change to the type of dma_data is actually in preparation for an
optimization I have planned for 64-bit PCI devices (and which probably
requires more discussion), so that doesn't need to happen now - just
leave it as a void *, and I can post a followup patch.

Let me know if I can help or do any testing - I've been meaning to
look into switching to dma_map_ops for a while now but it hasn't
managed to pop off my todo stack.

Cheers,
Becky

2009-07-14 00:50:11

by FUJITA Tomonori

[permalink] [raw]
Subject: Re: removing addr_needs_map in struct dma_mapping_ops

On Mon, 13 Jul 2009 16:50:43 -0500
Becky Bruce <[email protected]> wrote:

> > talked about defining something like struct dma_data. Then we could
> >
> > struct dev_archdata {
> > ...
> >
> > struct dma_data *ddata;
> > };
> >
> > or
> >
> > struct dev_archdata {
> > ...
> >
> > struct dma_data ddata;
> > };
> >
> >
> > struct dma_data needs dma_direct_offset, iommu_table, dma_base, and
> > dma_window_size, anything else?
>
> IIRC, what we had talked about was simpler - we talked about changing
> the current dev_archdata from this:
>
> struct dev_archdata {
> struct device_node *of_node;
> struct dma_mapping_ops *dma_ops;
> void *dma_data;
> };
>
> to this:
>
> struct dev_archdata {
> struct device_node *of_node;
> struct dma_mapping_ops *dma_ops;
> unsigned long long dma_data;
> #ifdef CONFIG_SWIOTLB
> dma_addr_t max_direct_dma_addr;
> #endif
> };
>
> Where max_direct_dma_addr is the address beyond which a specific
> device must use swiotlb, and dma_data is the offset like it is now
> (but wider on 32-bit systems than void *). I believe Ben had mentioned
> wanting to make the max_direct_dma_addr part conditional so we don't
> bloat archdata on platforms that don't ever bounce.

Only maximum address is enough? The minimum (dma_window_base_cur in
swiotlb_pci_addr_needs_map) is not necessary?


> The change to the type of dma_data is actually in preparation for an
> optimization I have planned for 64-bit PCI devices (and which probably
> requires more discussion), so that doesn't need to happen now - just
> leave it as a void *, and I can post a followup patch.
>
> Let me know if I can help or do any testing - I've been meaning to
> look into switching to dma_map_ops for a while now but it hasn't
> managed to pop off my todo stack.

Ok, how about this? I'm not familiar with POWERPC so I might
misunderstand something.


diff --git a/arch/powerpc/include/asm/device.h b/arch/powerpc/include/asm/device.h
index 7d2277c..0086f8d 100644
--- a/arch/powerpc/include/asm/device.h
+++ b/arch/powerpc/include/asm/device.h
@@ -16,6 +16,9 @@ struct dev_archdata {
/* DMA operations on that device */
struct dma_mapping_ops *dma_ops;
void *dma_data;
+#ifdef CONFIG_SWIOTLB
+ dma_addr_t max_direct_dma_addr;
+#endif
};

static inline void dev_archdata_set_node(struct dev_archdata *ad,
diff --git a/arch/powerpc/include/asm/swiotlb.h b/arch/powerpc/include/asm/swiotlb.h
index 30891d6..b23a4f1 100644
--- a/arch/powerpc/include/asm/swiotlb.h
+++ b/arch/powerpc/include/asm/swiotlb.h
@@ -24,4 +24,6 @@ static inline void dma_mark_clean(void *addr, size_t size) {}
extern unsigned int ppc_swiotlb_enable;
int __init swiotlb_setup_bus_notifier(void);

+extern void pci_dma_dev_setup_swiotlb(struct pci_dev *pdev);
+
#endif /* __ASM_SWIOTLB_H */
diff --git a/arch/powerpc/kernel/dma-swiotlb.c b/arch/powerpc/kernel/dma-swiotlb.c
index 68ccf11..e21359e 100644
--- a/arch/powerpc/kernel/dma-swiotlb.c
+++ b/arch/powerpc/kernel/dma-swiotlb.c
@@ -56,39 +56,16 @@ swiotlb_arch_address_needs_mapping(struct device *hwdev, dma_addr_t addr,
size_t size)
{
struct dma_mapping_ops *dma_ops = get_dma_ops(hwdev);
+ struct dev_archdata *sd = &hwdev->archdata;

BUG_ON(!dma_ops);
- return dma_ops->addr_needs_map(hwdev, addr, size);
-}

-/*
- * Determine if an address is reachable by a pci device, or if we must bounce.
- */
-static int
-swiotlb_pci_addr_needs_map(struct device *hwdev, dma_addr_t addr, size_t size)
-{
- u64 mask = dma_get_mask(hwdev);
- dma_addr_t max;
- struct pci_controller *hose;
- struct pci_dev *pdev = to_pci_dev(hwdev);
-
- hose = pci_bus_to_host(pdev->bus);
- max = hose->dma_window_base_cur + hose->dma_window_size;
-
- /* check that we're within mapped pci window space */
- if ((addr + size > max) | (addr < hose->dma_window_base_cur))
+ if (sd->max_direct_dma_addr && addr + size > sd->max_direct_dma_addr)
return 1;

- return !is_buffer_dma_capable(mask, addr, size);
-}
-
-static int
-swiotlb_addr_needs_map(struct device *hwdev, dma_addr_t addr, size_t size)
-{
return !is_buffer_dma_capable(dma_get_mask(hwdev), addr, size);
}

-
/*
* At the moment, all platforms that use this code only require
* swiotlb to be used if we're operating on HIGHMEM. Since
@@ -104,7 +81,6 @@ struct dma_mapping_ops swiotlb_dma_ops = {
.dma_supported = swiotlb_dma_supported,
.map_page = swiotlb_map_page,
.unmap_page = swiotlb_unmap_page,
- .addr_needs_map = swiotlb_addr_needs_map,
.sync_single_range_for_cpu = swiotlb_sync_single_range_for_cpu,
.sync_single_range_for_device = swiotlb_sync_single_range_for_device,
.sync_sg_for_cpu = swiotlb_sync_sg_for_cpu,
@@ -119,13 +95,23 @@ struct dma_mapping_ops swiotlb_pci_dma_ops = {
.dma_supported = swiotlb_dma_supported,
.map_page = swiotlb_map_page,
.unmap_page = swiotlb_unmap_page,
- .addr_needs_map = swiotlb_pci_addr_needs_map,
.sync_single_range_for_cpu = swiotlb_sync_single_range_for_cpu,
.sync_single_range_for_device = swiotlb_sync_single_range_for_device,
.sync_sg_for_cpu = swiotlb_sync_sg_for_cpu,
.sync_sg_for_device = swiotlb_sync_sg_for_device
};

+void pci_dma_dev_setup_swiotlb(struct pci_dev *pdev)
+{
+ struct pci_controller *hose;
+ struct dev_archdata *sd;
+
+ hose = pci_bus_to_host(pdev->bus);
+ sd = &pdev->dev.archdata;
+ sd->max_direct_dma_addr =
+ hose->dma_window_base_cur + hose->dma_window_size;
+}
+
static int ppc_swiotlb_bus_notify(struct notifier_block *nb,
unsigned long action, void *data)
{
diff --git a/arch/powerpc/platforms/85xx/mpc8536_ds.c b/arch/powerpc/platforms/85xx/mpc8536_ds.c
index 055ff41..401751b 100644
--- a/arch/powerpc/platforms/85xx/mpc8536_ds.c
+++ b/arch/powerpc/platforms/85xx/mpc8536_ds.c
@@ -136,6 +136,7 @@ define_machine(mpc8536_ds) {
.init_IRQ = mpc8536_ds_pic_init,
#ifdef CONFIG_PCI
.pcibios_fixup_bus = fsl_pcibios_fixup_bus,
+ .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
#endif
.get_irq = mpic_get_irq,
.restart = fsl_rstcr_restart,
diff --git a/arch/powerpc/platforms/85xx/mpc85xx_ds.c b/arch/powerpc/platforms/85xx/mpc85xx_ds.c
index 849c0ac..1ba8e38 100644
--- a/arch/powerpc/platforms/85xx/mpc85xx_ds.c
+++ b/arch/powerpc/platforms/85xx/mpc85xx_ds.c
@@ -277,6 +277,7 @@ define_machine(mpc8544_ds) {
.init_IRQ = mpc85xx_ds_pic_init,
#ifdef CONFIG_PCI
.pcibios_fixup_bus = fsl_pcibios_fixup_bus,
+ .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
#endif
.get_irq = mpic_get_irq,
.restart = fsl_rstcr_restart,
@@ -291,6 +292,7 @@ define_machine(mpc8572_ds) {
.init_IRQ = mpc85xx_ds_pic_init,
#ifdef CONFIG_PCI
.pcibios_fixup_bus = fsl_pcibios_fixup_bus,
+ .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
#endif
.get_irq = mpic_get_irq,
.restart = fsl_rstcr_restart,
@@ -305,6 +307,7 @@ define_machine(p2020_ds) {
.init_IRQ = mpc85xx_ds_pic_init,
#ifdef CONFIG_PCI
.pcibios_fixup_bus = fsl_pcibios_fixup_bus,
+ .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
#endif
.get_irq = mpic_get_irq,
.restart = fsl_rstcr_restart,
diff --git a/arch/powerpc/platforms/85xx/mpc85xx_mds.c b/arch/powerpc/platforms/85xx/mpc85xx_mds.c
index 60ed9c0..165a2de 100644
--- a/arch/powerpc/platforms/85xx/mpc85xx_mds.c
+++ b/arch/powerpc/platforms/85xx/mpc85xx_mds.c
@@ -356,6 +356,7 @@ define_machine(mpc8568_mds) {
.progress = udbg_progress,
#ifdef CONFIG_PCI
.pcibios_fixup_bus = fsl_pcibios_fixup_bus,
+ .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
#endif
};

@@ -377,5 +378,6 @@ define_machine(mpc8569_mds) {
.progress = udbg_progress,
#ifdef CONFIG_PCI
.pcibios_fixup_bus = fsl_pcibios_fixup_bus,
+ .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
#endif
};
diff --git a/arch/powerpc/platforms/86xx/mpc86xx_hpcn.c b/arch/powerpc/platforms/86xx/mpc86xx_hpcn.c
index 6632702..d1878f3 100644
--- a/arch/powerpc/platforms/86xx/mpc86xx_hpcn.c
+++ b/arch/powerpc/platforms/86xx/mpc86xx_hpcn.c
@@ -187,5 +187,6 @@ define_machine(mpc86xx_hpcn) {
.progress = udbg_progress,
#ifdef CONFIG_PCI
.pcibios_fixup_bus = fsl_pcibios_fixup_bus,
+ .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
#endif
};

2009-07-15 23:59:19

by Becky Bruce

[permalink] [raw]
Subject: Re: removing addr_needs_map in struct dma_mapping_ops


On Jul 13, 2009, at 7:49 PM, FUJITA Tomonori wrote:

> On Mon, 13 Jul 2009 16:50:43 -0500
> Becky Bruce <[email protected]> wrote:
>
>>> talked about defining something like struct dma_data. Then we could
>>>
>>> struct dev_archdata {
>>> ...
>>>
>>> struct dma_data *ddata;
>>> };
>>>
>>> or
>>>
>>> struct dev_archdata {
>>> ...
>>>
>>> struct dma_data ddata;
>>> };
>>>
>>>
>>> struct dma_data needs dma_direct_offset, iommu_table, dma_base, and
>>> dma_window_size, anything else?
>>
>> IIRC, what we had talked about was simpler - we talked about changing
>> the current dev_archdata from this:
>>
>> struct dev_archdata {
>> struct device_node *of_node;
>> struct dma_mapping_ops *dma_ops;
>> void *dma_data;
>> };
>>
>> to this:
>>
>> struct dev_archdata {
>> struct device_node *of_node;
>> struct dma_mapping_ops *dma_ops;
>> unsigned long long dma_data;
>> #ifdef CONFIG_SWIOTLB
>> dma_addr_t max_direct_dma_addr;
>> #endif
>> };
>>
>> Where max_direct_dma_addr is the address beyond which a specific
>> device must use swiotlb, and dma_data is the offset like it is now
>> (but wider on 32-bit systems than void *). I believe Ben had
>> mentioned
>> wanting to make the max_direct_dma_addr part conditional so we don't
>> bloat archdata on platforms that don't ever bounce.
>
> Only maximum address is enough? The minimum (dma_window_base_cur in
> swiotlb_pci_addr_needs_map) is not necessary?
>
>
>> The change to the type of dma_data is actually in preparation for an
>> optimization I have planned for 64-bit PCI devices (and which
>> probably
>> requires more discussion), so that doesn't need to happen now - just
>> leave it as a void *, and I can post a followup patch.
>>
>> Let me know if I can help or do any testing - I've been meaning to
>> look into switching to dma_map_ops for a while now but it hasn't
>> managed to pop off my todo stack.
>
> Ok, how about this? I'm not familiar with POWERPC so I might
> misunderstand something.

This is close, but it misses the setup for non-pci devices. We have a
bus notifier that we use to set up archdata for those devices -
ppc_swiotlb_bus_notify() in arch/powerpc/kernel/dma-swiotlb.c. It
won't cause breakage to not have this set up, because those will fall
through to the dma_capable(), but I think we should initialize it
anyway (who knows what it will end up used for later....).

>
>
>
> diff --git a/arch/powerpc/include/asm/device.h b/arch/powerpc/
> include/asm/device.h
> index 7d2277c..0086f8d 100644
> --- a/arch/powerpc/include/asm/device.h
> +++ b/arch/powerpc/include/asm/device.h
> @@ -16,6 +16,9 @@ struct dev_archdata {
> /* DMA operations on that device */
> struct dma_mapping_ops *dma_ops;
> void *dma_data;
> +#ifdef CONFIG_SWIOTLB
> + dma_addr_t max_direct_dma_addr;
> +#endif
> };
>
> static inline void dev_archdata_set_node(struct dev_archdata *ad,
> diff --git a/arch/powerpc/include/asm/swiotlb.h b/arch/powerpc/
> include/asm/swiotlb.h
> index 30891d6..b23a4f1 100644
> --- a/arch/powerpc/include/asm/swiotlb.h
> +++ b/arch/powerpc/include/asm/swiotlb.h
> @@ -24,4 +24,6 @@ static inline void dma_mark_clean(void *addr,
> size_t size) {}
> extern unsigned int ppc_swiotlb_enable;
> int __init swiotlb_setup_bus_notifier(void);
>
> +extern void pci_dma_dev_setup_swiotlb(struct pci_dev *pdev);
> +
> #endif /* __ASM_SWIOTLB_H */
> diff --git a/arch/powerpc/kernel/dma-swiotlb.c b/arch/powerpc/kernel/
> dma-swiotlb.c
> index 68ccf11..e21359e 100644
> --- a/arch/powerpc/kernel/dma-swiotlb.c
> +++ b/arch/powerpc/kernel/dma-swiotlb.c
> @@ -56,39 +56,16 @@ swiotlb_arch_address_needs_mapping(struct device
> *hwdev, dma_addr_t addr,
> size_t size)
> {
> struct dma_mapping_ops *dma_ops = get_dma_ops(hwdev);
> + struct dev_archdata *sd = &hwdev->archdata;
>
> BUG_ON(!dma_ops);
> - return dma_ops->addr_needs_map(hwdev, addr, size);
> -}

You can get rid of the dma_ops stuff here.... it's no longer needed.

>
>
> -/*
> - * Determine if an address is reachable by a pci device, or if we
> must bounce.
> - */
> -static int
> -swiotlb_pci_addr_needs_map(struct device *hwdev, dma_addr_t addr,
> size_t size)
> -{
> - u64 mask = dma_get_mask(hwdev);
> - dma_addr_t max;
> - struct pci_controller *hose;
> - struct pci_dev *pdev = to_pci_dev(hwdev);
> -
> - hose = pci_bus_to_host(pdev->bus);
> - max = hose->dma_window_base_cur + hose->dma_window_size;
> -
> - /* check that we're within mapped pci window space */
> - if ((addr + size > max) | (addr < hose->dma_window_base_cur))
> + if (sd->max_direct_dma_addr && addr + size > sd-
> >max_direct_dma_addr)
> return 1;
>
> - return !is_buffer_dma_capable(mask, addr, size);
> -}
> -
> -static int
> -swiotlb_addr_needs_map(struct device *hwdev, dma_addr_t addr,
> size_t size)
> -{
> return !is_buffer_dma_capable(dma_get_mask(hwdev), addr, size);
> }
>
> -
> /*
> * At the moment, all platforms that use this code only require
> * swiotlb to be used if we're operating on HIGHMEM. Since
> @@ -104,7 +81,6 @@ struct dma_mapping_ops swiotlb_dma_ops = {
> .dma_supported = swiotlb_dma_supported,
> .map_page = swiotlb_map_page,
> .unmap_page = swiotlb_unmap_page,
> - .addr_needs_map = swiotlb_addr_needs_map,
> .sync_single_range_for_cpu = swiotlb_sync_single_range_for_cpu,
> .sync_single_range_for_device = swiotlb_sync_single_range_for_device,
> .sync_sg_for_cpu = swiotlb_sync_sg_for_cpu,
> @@ -119,13 +95,23 @@ struct dma_mapping_ops swiotlb_pci_dma_ops = {
> .dma_supported = swiotlb_dma_supported,
> .map_page = swiotlb_map_page,
> .unmap_page = swiotlb_unmap_page,
> - .addr_needs_map = swiotlb_pci_addr_needs_map,
> .sync_single_range_for_cpu = swiotlb_sync_single_range_for_cpu,
> .sync_single_range_for_device = swiotlb_sync_single_range_for_device,
> .sync_sg_for_cpu = swiotlb_sync_sg_for_cpu,
> .sync_sg_for_device = swiotlb_sync_sg_for_device
> };
>
> +void pci_dma_dev_setup_swiotlb(struct pci_dev *pdev)
> +{
> + struct pci_controller *hose;
> + struct dev_archdata *sd;
> +
> + hose = pci_bus_to_host(pdev->bus);
> + sd = &pdev->dev.archdata;
> + sd->max_direct_dma_addr =
> + hose->dma_window_base_cur + hose->dma_window_size;
> +}
> +
> static int ppc_swiotlb_bus_notify(struct notifier_block *nb,
> unsigned long action, void *data)
> {
> diff --git a/arch/powerpc/platforms/85xx/mpc8536_ds.c b/arch/powerpc/
> platforms/85xx/mpc8536_ds.c
> index 055ff41..401751b 100644
> --- a/arch/powerpc/platforms/85xx/mpc8536_ds.c
> +++ b/arch/powerpc/platforms/85xx/mpc8536_ds.c
> @@ -136,6 +136,7 @@ define_machine(mpc8536_ds) {
> .init_IRQ = mpc8536_ds_pic_init,
> #ifdef CONFIG_PCI
> .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
> + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
> #endif
> .get_irq = mpic_get_irq,
> .restart = fsl_rstcr_restart,
> diff --git a/arch/powerpc/platforms/85xx/mpc85xx_ds.c b/arch/powerpc/
> platforms/85xx/mpc85xx_ds.c
> index 849c0ac..1ba8e38 100644
> --- a/arch/powerpc/platforms/85xx/mpc85xx_ds.c
> +++ b/arch/powerpc/platforms/85xx/mpc85xx_ds.c
> @@ -277,6 +277,7 @@ define_machine(mpc8544_ds) {
> .init_IRQ = mpc85xx_ds_pic_init,
> #ifdef CONFIG_PCI
> .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
> + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
> #endif
> .get_irq = mpic_get_irq,
> .restart = fsl_rstcr_restart,
> @@ -291,6 +292,7 @@ define_machine(mpc8572_ds) {
> .init_IRQ = mpc85xx_ds_pic_init,
> #ifdef CONFIG_PCI
> .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
> + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
> #endif
> .get_irq = mpic_get_irq,
> .restart = fsl_rstcr_restart,
> @@ -305,6 +307,7 @@ define_machine(p2020_ds) {
> .init_IRQ = mpc85xx_ds_pic_init,
> #ifdef CONFIG_PCI
> .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
> + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
> #endif
> .get_irq = mpic_get_irq,
> .restart = fsl_rstcr_restart,
> diff --git a/arch/powerpc/platforms/85xx/mpc85xx_mds.c b/arch/
> powerpc/platforms/85xx/mpc85xx_mds.c
> index 60ed9c0..165a2de 100644
> --- a/arch/powerpc/platforms/85xx/mpc85xx_mds.c
> +++ b/arch/powerpc/platforms/85xx/mpc85xx_mds.c
> @@ -356,6 +356,7 @@ define_machine(mpc8568_mds) {
> .progress = udbg_progress,
> #ifdef CONFIG_PCI
> .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
> + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
> #endif
> };
>
> @@ -377,5 +378,6 @@ define_machine(mpc8569_mds) {
> .progress = udbg_progress,
> #ifdef CONFIG_PCI
> .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
> + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
> #endif
> };
> diff --git a/arch/powerpc/platforms/86xx/mpc86xx_hpcn.c b/arch/
> powerpc/platforms/86xx/mpc86xx_hpcn.c
> index 6632702..d1878f3 100644
> --- a/arch/powerpc/platforms/86xx/mpc86xx_hpcn.c
> +++ b/arch/powerpc/platforms/86xx/mpc86xx_hpcn.c
> @@ -187,5 +187,6 @@ define_machine(mpc86xx_hpcn) {
> .progress = udbg_progress,
> #ifdef CONFIG_PCI
> .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
> + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
> #endif
> };

Instead of initializing this here (which has problems if !
CONFIG_SWIOTLB), place this in the xxxxx_xxxx_setup_arch function in
the same files, which already have an #ifdef CONFIG_SWIOTLB in which
this can be embedded.

I'm about to be off-list for a few days but will be happy to help when
I'm back next week.

Thanks!
Becky


>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-
> kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/

2009-07-23 07:10:11

by FUJITA Tomonori

[permalink] [raw]
Subject: Re: removing addr_needs_map in struct dma_mapping_ops

On Wed, 15 Jul 2009 18:59:03 -0500
Becky Bruce <[email protected]> wrote:

>
> On Jul 13, 2009, at 7:49 PM, FUJITA Tomonori wrote:
>
> > On Mon, 13 Jul 2009 16:50:43 -0500
> > Becky Bruce <[email protected]> wrote:
> >
> >>> talked about defining something like struct dma_data. Then we could
> >>>
> >>> struct dev_archdata {
> >>> ...
> >>>
> >>> struct dma_data *ddata;
> >>> };
> >>>
> >>> or
> >>>
> >>> struct dev_archdata {
> >>> ...
> >>>
> >>> struct dma_data ddata;
> >>> };
> >>>
> >>>
> >>> struct dma_data needs dma_direct_offset, iommu_table, dma_base, and
> >>> dma_window_size, anything else?
> >>
> >> IIRC, what we had talked about was simpler - we talked about changing
> >> the current dev_archdata from this:
> >>
> >> struct dev_archdata {
> >> struct device_node *of_node;
> >> struct dma_mapping_ops *dma_ops;
> >> void *dma_data;
> >> };
> >>
> >> to this:
> >>
> >> struct dev_archdata {
> >> struct device_node *of_node;
> >> struct dma_mapping_ops *dma_ops;
> >> unsigned long long dma_data;
> >> #ifdef CONFIG_SWIOTLB
> >> dma_addr_t max_direct_dma_addr;
> >> #endif
> >> };
> >>
> >> Where max_direct_dma_addr is the address beyond which a specific
> >> device must use swiotlb, and dma_data is the offset like it is now
> >> (but wider on 32-bit systems than void *). I believe Ben had
> >> mentioned
> >> wanting to make the max_direct_dma_addr part conditional so we don't
> >> bloat archdata on platforms that don't ever bounce.
> >
> > Only maximum address is enough? The minimum (dma_window_base_cur in
> > swiotlb_pci_addr_needs_map) is not necessary?
> >
> >
> >> The change to the type of dma_data is actually in preparation for an
> >> optimization I have planned for 64-bit PCI devices (and which
> >> probably
> >> requires more discussion), so that doesn't need to happen now - just
> >> leave it as a void *, and I can post a followup patch.
> >>
> >> Let me know if I can help or do any testing - I've been meaning to
> >> look into switching to dma_map_ops for a while now but it hasn't
> >> managed to pop off my todo stack.
> >
> > Ok, how about this? I'm not familiar with POWERPC so I might
> > misunderstand something.
>
> This is close, but it misses the setup for non-pci devices. We have a
> bus notifier that we use to set up archdata for those devices -
> ppc_swiotlb_bus_notify() in arch/powerpc/kernel/dma-swiotlb.c. It
> won't cause breakage to not have this set up, because those will fall
> through to the dma_capable(), but I think we should initialize it
> anyway (who knows what it will end up used for later....).

You mean that you like to initialize max_direct_dma_addr to zero
explicitly in ppc_swiotlb_bus_notify()?


> > diff --git a/arch/powerpc/include/asm/device.h b/arch/powerpc/
> > include/asm/device.h
> > index 7d2277c..0086f8d 100644
> > --- a/arch/powerpc/include/asm/device.h
> > +++ b/arch/powerpc/include/asm/device.h
> > @@ -16,6 +16,9 @@ struct dev_archdata {
> > /* DMA operations on that device */
> > struct dma_mapping_ops *dma_ops;
> > void *dma_data;
> > +#ifdef CONFIG_SWIOTLB
> > + dma_addr_t max_direct_dma_addr;
> > +#endif
> > };
> >
> > static inline void dev_archdata_set_node(struct dev_archdata *ad,
> > diff --git a/arch/powerpc/include/asm/swiotlb.h b/arch/powerpc/
> > include/asm/swiotlb.h
> > index 30891d6..b23a4f1 100644
> > --- a/arch/powerpc/include/asm/swiotlb.h
> > +++ b/arch/powerpc/include/asm/swiotlb.h
> > @@ -24,4 +24,6 @@ static inline void dma_mark_clean(void *addr,
> > size_t size) {}
> > extern unsigned int ppc_swiotlb_enable;
> > int __init swiotlb_setup_bus_notifier(void);
> >
> > +extern void pci_dma_dev_setup_swiotlb(struct pci_dev *pdev);
> > +
> > #endif /* __ASM_SWIOTLB_H */
> > diff --git a/arch/powerpc/kernel/dma-swiotlb.c b/arch/powerpc/kernel/
> > dma-swiotlb.c
> > index 68ccf11..e21359e 100644
> > --- a/arch/powerpc/kernel/dma-swiotlb.c
> > +++ b/arch/powerpc/kernel/dma-swiotlb.c
> > @@ -56,39 +56,16 @@ swiotlb_arch_address_needs_mapping(struct device
> > *hwdev, dma_addr_t addr,
> > size_t size)
> > {
> > struct dma_mapping_ops *dma_ops = get_dma_ops(hwdev);
> > + struct dev_archdata *sd = &hwdev->archdata;
> >
> > BUG_ON(!dma_ops);
> > - return dma_ops->addr_needs_map(hwdev, addr, size);
> > -}
>
> You can get rid of the dma_ops stuff here.... it's no longer needed.

Yeah, I'll do later in this patchset that converts POWERPC to use
dma_map_ops structure; this is the first patch of the patchset.


> > - * Determine if an address is reachable by a pci device, or if we
> > must bounce.
> > - */
> > -static int
> > -swiotlb_pci_addr_needs_map(struct device *hwdev, dma_addr_t addr,
> > size_t size)
> > -{
> > - u64 mask = dma_get_mask(hwdev);
> > - dma_addr_t max;
> > - struct pci_controller *hose;
> > - struct pci_dev *pdev = to_pci_dev(hwdev);
> > -
> > - hose = pci_bus_to_host(pdev->bus);
> > - max = hose->dma_window_base_cur + hose->dma_window_size;
> > -
> > - /* check that we're within mapped pci window space */
> > - if ((addr + size > max) | (addr < hose->dma_window_base_cur))
> > + if (sd->max_direct_dma_addr && addr + size > sd-
> > >max_direct_dma_addr)
> > return 1;
> >
> > - return !is_buffer_dma_capable(mask, addr, size);
> > -}
> > -
> > -static int
> > -swiotlb_addr_needs_map(struct device *hwdev, dma_addr_t addr,
> > size_t size)
> > -{
> > return !is_buffer_dma_capable(dma_get_mask(hwdev), addr, size);
> > }
> >
> > -
> > /*
> > * At the moment, all platforms that use this code only require
> > * swiotlb to be used if we're operating on HIGHMEM. Since
> > @@ -104,7 +81,6 @@ struct dma_mapping_ops swiotlb_dma_ops = {
> > .dma_supported = swiotlb_dma_supported,
> > .map_page = swiotlb_map_page,
> > .unmap_page = swiotlb_unmap_page,
> > - .addr_needs_map = swiotlb_addr_needs_map,
> > .sync_single_range_for_cpu = swiotlb_sync_single_range_for_cpu,
> > .sync_single_range_for_device = swiotlb_sync_single_range_for_device,
> > .sync_sg_for_cpu = swiotlb_sync_sg_for_cpu,
> > @@ -119,13 +95,23 @@ struct dma_mapping_ops swiotlb_pci_dma_ops = {
> > .dma_supported = swiotlb_dma_supported,
> > .map_page = swiotlb_map_page,
> > .unmap_page = swiotlb_unmap_page,
> > - .addr_needs_map = swiotlb_pci_addr_needs_map,
> > .sync_single_range_for_cpu = swiotlb_sync_single_range_for_cpu,
> > .sync_single_range_for_device = swiotlb_sync_single_range_for_device,
> > .sync_sg_for_cpu = swiotlb_sync_sg_for_cpu,
> > .sync_sg_for_device = swiotlb_sync_sg_for_device
> > };
> >
> > +void pci_dma_dev_setup_swiotlb(struct pci_dev *pdev)
> > +{
> > + struct pci_controller *hose;
> > + struct dev_archdata *sd;
> > +
> > + hose = pci_bus_to_host(pdev->bus);
> > + sd = &pdev->dev.archdata;
> > + sd->max_direct_dma_addr =
> > + hose->dma_window_base_cur + hose->dma_window_size;
> > +}
> > +
> > static int ppc_swiotlb_bus_notify(struct notifier_block *nb,
> > unsigned long action, void *data)
> > {
> > diff --git a/arch/powerpc/platforms/85xx/mpc8536_ds.c b/arch/powerpc/
> > platforms/85xx/mpc8536_ds.c
> > index 055ff41..401751b 100644
> > --- a/arch/powerpc/platforms/85xx/mpc8536_ds.c
> > +++ b/arch/powerpc/platforms/85xx/mpc8536_ds.c
> > @@ -136,6 +136,7 @@ define_machine(mpc8536_ds) {
> > .init_IRQ = mpc8536_ds_pic_init,
> > #ifdef CONFIG_PCI
> > .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
> > + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
> > #endif
> > .get_irq = mpic_get_irq,
> > .restart = fsl_rstcr_restart,
> > diff --git a/arch/powerpc/platforms/85xx/mpc85xx_ds.c b/arch/powerpc/
> > platforms/85xx/mpc85xx_ds.c
> > index 849c0ac..1ba8e38 100644
> > --- a/arch/powerpc/platforms/85xx/mpc85xx_ds.c
> > +++ b/arch/powerpc/platforms/85xx/mpc85xx_ds.c
> > @@ -277,6 +277,7 @@ define_machine(mpc8544_ds) {
> > .init_IRQ = mpc85xx_ds_pic_init,
> > #ifdef CONFIG_PCI
> > .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
> > + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
> > #endif
> > .get_irq = mpic_get_irq,
> > .restart = fsl_rstcr_restart,
> > @@ -291,6 +292,7 @@ define_machine(mpc8572_ds) {
> > .init_IRQ = mpc85xx_ds_pic_init,
> > #ifdef CONFIG_PCI
> > .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
> > + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
> > #endif
> > .get_irq = mpic_get_irq,
> > .restart = fsl_rstcr_restart,
> > @@ -305,6 +307,7 @@ define_machine(p2020_ds) {
> > .init_IRQ = mpc85xx_ds_pic_init,
> > #ifdef CONFIG_PCI
> > .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
> > + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
> > #endif
> > .get_irq = mpic_get_irq,
> > .restart = fsl_rstcr_restart,
> > diff --git a/arch/powerpc/platforms/85xx/mpc85xx_mds.c b/arch/
> > powerpc/platforms/85xx/mpc85xx_mds.c
> > index 60ed9c0..165a2de 100644
> > --- a/arch/powerpc/platforms/85xx/mpc85xx_mds.c
> > +++ b/arch/powerpc/platforms/85xx/mpc85xx_mds.c
> > @@ -356,6 +356,7 @@ define_machine(mpc8568_mds) {
> > .progress = udbg_progress,
> > #ifdef CONFIG_PCI
> > .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
> > + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
> > #endif
> > };
> >
> > @@ -377,5 +378,6 @@ define_machine(mpc8569_mds) {
> > .progress = udbg_progress,
> > #ifdef CONFIG_PCI
> > .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
> > + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
> > #endif
> > };
> > diff --git a/arch/powerpc/platforms/86xx/mpc86xx_hpcn.c b/arch/
> > powerpc/platforms/86xx/mpc86xx_hpcn.c
> > index 6632702..d1878f3 100644
> > --- a/arch/powerpc/platforms/86xx/mpc86xx_hpcn.c
> > +++ b/arch/powerpc/platforms/86xx/mpc86xx_hpcn.c
> > @@ -187,5 +187,6 @@ define_machine(mpc86xx_hpcn) {
> > .progress = udbg_progress,
> > #ifdef CONFIG_PCI
> > .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
> > + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
> > #endif
> > };
>
> Instead of initializing this here (which has problems if !
> CONFIG_SWIOTLB),

Oops, I overlooked it.


> place this in the xxxxx_xxxx_setup_arch function in
> the same files, which already have an #ifdef CONFIG_SWIOTLB in which
> this can be embedded.

But the xxxxx_xxxx_setup_arch function doesn't access to each device
so we need to add something like for_each_pci_dev()? You prefer that?


> I'm about to be off-list for a few days but will be happy to help when
> I'm back next week.

Thanks!

2009-07-23 15:44:51

by Kumar Gala

[permalink] [raw]
Subject: Re: removing addr_needs_map in struct dma_mapping_ops


On Jul 23, 2009, at 2:09 AM, FUJITA Tomonori wrote:

>>> diff --git a/arch/powerpc/platforms/86xx/mpc86xx_hpcn.c b/arch/
>>> powerpc/platforms/86xx/mpc86xx_hpcn.c
>>> index 6632702..d1878f3 100644
>>> --- a/arch/powerpc/platforms/86xx/mpc86xx_hpcn.c
>>> +++ b/arch/powerpc/platforms/86xx/mpc86xx_hpcn.c
>>> @@ -187,5 +187,6 @@ define_machine(mpc86xx_hpcn) {
>>> .progress = udbg_progress,
>>> #ifdef CONFIG_PCI
>>> .pcibios_fixup_bus = fsl_pcibios_fixup_bus,
>>> + .pci_dma_dev_setup = pci_dma_dev_setup_swiotlb,
>>> #endif
>>> };
>>
>> Instead of initializing this here (which has problems if !
>> CONFIG_SWIOTLB),
>
> Oops, I overlooked it.
>
>
>> place this in the xxxxx_xxxx_setup_arch function in
>> the same files, which already have an #ifdef CONFIG_SWIOTLB in which
>> this can be embedded.
>
> But the xxxxx_xxxx_setup_arch function doesn't access to each device
> so we need to add something like for_each_pci_dev()? You prefer that?

No.. I think what we want is:

#ifdef CONFIG_SWIOTLB
if (lmb_end_of_DRAM() > max) {
ppc_swiotlb_enable = 1;
set_pci_dma_ops(&swiotlb_pci_dma_ops);
ppc_md.pci_dma_dev_setup = pci_dma_dev_setup_swiotlb;
}
#endif

Buts its been a few days since Becky & I chatted about this and I feel
like I'm forgetting something.

- k