dma_ops->map_page == NULL

dma_ops->map_page == NULL

[Kári Davíðsson]

I am doing a driver that uses dma_map_single().

After changing to to linux 2.6.29.3 I am getting
segfaults in dma_map_single() because dma_ops->map_page is NULL.
Actually dma_ops looks funky too.

The driver is an of_platform_driver which is declared as an child of
the lbp (fsl,lpb) node of the device tree.

This is on powerpc 5200b platform.

rg
kd

Re: dma_ops->map_page == NULL

[Mark Nelson]

On Tuesday 07 July 2009 03:51:00 K=E1ri Dav=ED=F0sson wrote:
> I am doing a driver that uses dma_map_single().
>=20
> After changing to to linux 2.6.29.3 I am getting
> segfaults in dma_map_single() because dma_ops->map_page is NULL.
> Actually dma_ops looks funky too.

When the 32 and 64bit DMA code was merged in .28 , map_/unmap_page() was
added in favour of map_/unmap_single() (which was later removed in .29)
so you'll have to replace your calls to dma_map_single() with
dma_map_page(). Just pass it the page and offset rather than the address.

Hope that helps!
Mark.

>=20
> The driver is an of_platform_driver which is declared as an child of
> the lbp (fsl,lpb) node of the device tree.
>=20
> This is on powerpc 5200b platform.
>=20
> rg
> kd

Re: dma_ops->map_page == NULL

[Benjamin Herrenschmidt]

On Tue, 2009-07-07 at 10:15 +1000, Mark Nelson wrote:
> 
> When the 32 and 64bit DMA code was merged in .28 , map_/unmap_page() was
> added in favour of map_/unmap_single() (which was later removed in .29)
> so you'll have to replace your calls to dma_map_single() with
> dma_map_page(). Just pass it the page and offset rather than the address.

Wait a minute ... dma_map_single() should still work, it will just call
dma_map_page() underneath. All dma_ops should have a ->map page
callback.

Do you have any dma_ops set for your device at all ? I wonder how we
set the dma_ops for platform devices nowadays ... We use to have this
fallback to direct ops when the dma_ops are NULL but that is gone and
I see no suitable replacement to set them on platform devices for
embedded archs ... oops...

Kumar, Becky, what's the situation there ?

Cheers,
Ben.

Re: dma_ops->map_page == NULL

[Kumar Gala]

On Jul 7, 2009, at 6:08 AM, Benjamin Herrenschmidt wrote:

> On Tue, 2009-07-07 at 10:15 +1000, Mark Nelson wrote:
>>
>> When the 32 and 64bit DMA code was merged in .28 , map_/ 
>> unmap_page() was
>> added in favour of map_/unmap_single() (which was later removed in . 
>> 29)
>> so you'll have to replace your calls to dma_map_single() with
>> dma_map_page(). Just pass it the page and offset rather than the  
>> address.
>
> Wait a minute ... dma_map_single() should still work, it will just  
> call
> dma_map_page() underneath. All dma_ops should have a ->map page
> callback.
>
> Do you have any dma_ops set for your device at all ? I wonder how we
> set the dma_ops for platform devices nowadays ... We use to have this
> fallback to direct ops when the dma_ops are NULL but that is gone and
> I see no suitable replacement to set them on platform devices for
> embedded archs ... oops...
>
> Kumar, Becky, what's the situation there ?
>
> Cheers,
> Ben.

Is it possible the dev pointer is not valid?  I can't remember if that  
was a .29 or .30 change that requires us to start passing a valid dev  
pointer to get the proper dma_ops.

- k

Re: dma_ops->map_page == NULL

[Kári Davíðsson]

Yes the device pointer was invalid.

I was passing the of_device pointer instead of
the address of of_device->dev.

But I am sure this was working (passing of_device pointer) with
earlier kernels.

Thanks for the help.

rg
kd

Kumar Gala wrote:
> On Jul 7, 2009, at 6:08 AM, Benjamin Herrenschmidt wrote:
> 
>> On Tue, 2009-07-07 at 10:15 +1000, Mark Nelson wrote:
>>> When the 32 and 64bit DMA code was merged in .28 , map_/ 
>>> unmap_page() was
>>> added in favour of map_/unmap_single() (which was later removed in . 
>>> 29)
>>> so you'll have to replace your calls to dma_map_single() with
>>> dma_map_page(). Just pass it the page and offset rather than the  
>>> address.
>> Wait a minute ... dma_map_single() should still work, it will just  
>> call
>> dma_map_page() underneath. All dma_ops should have a ->map page
>> callback.
>>
>> Do you have any dma_ops set for your device at all ? I wonder how we
>> set the dma_ops for platform devices nowadays ... We use to have this
>> fallback to direct ops when the dma_ops are NULL but that is gone and
>> I see no suitable replacement to set them on platform devices for
>> embedded archs ... oops...
>>
>> Kumar, Becky, what's the situation there ?
>>
>> Cheers,
>> Ben.
> 
> Is it possible the dev pointer is not valid?  I can't remember if that  
> was a .29 or .30 change that requires us to start passing a valid dev  
> pointer to get the proper dma_ops.
> 
> - k

Re: dma_ops->map_page == NULL

[Kumar Gala]

On Jul 7, 2009, at 10:24 AM, K=E1ri Dav=ED=F0sson wrote:

> Yes the device pointer was invalid.
>
> I was passing the of_device pointer instead of
> the address of of_device->dev.
>
> But I am sure this was working (passing of_device pointer) with
> earlier kernels.
>
> Thanks for the help.
>
> rg
> kd

earlier kernels didn't care if the dev pointer was valid and thus let =20=

it pass if you didn't give the proper thing.

- k=

Re: dma_ops->map_page == NULL

[Becky Bruce]

On Jul 7, 2009, at 9:37 AM, Kumar Gala wrote:

>
> On Jul 7, 2009, at 6:08 AM, Benjamin Herrenschmidt wrote:
>
>> On Tue, 2009-07-07 at 10:15 +1000, Mark Nelson wrote:
>>>
>>> When the 32 and 64bit DMA code was merged in .28 , map_/ 
>>> unmap_page() was
>>> added in favour of map_/unmap_single() (which was later removed  
>>> in .29)
>>> so you'll have to replace your calls to dma_map_single() with
>>> dma_map_page(). Just pass it the page and offset rather than the  
>>> address.
>>
>> Wait a minute ... dma_map_single() should still work, it will just  
>> call
>> dma_map_page() underneath. All dma_ops should have a ->map page
>> callback.

>>
>>
>> Do you have any dma_ops set for your device at all ? I wonder how we
>> set the dma_ops for platform devices nowadays ... We use to have this
>> fallback to direct ops when the dma_ops are NULL but that is gone and
>> I see no suitable replacement to set them on platform devices for
>> embedded archs ... oops...

There is a platform/of bus notifier that sets the dma_ops to a default  
value in arch/powerpc/kernel/setup-common.c.

>>
>>
>> Kumar, Becky, what's the situation there ?
>>
>> Cheers,
>> Ben.
>
> Is it possible the dev pointer is not valid?  I can't remember if  
> that was a .29 or .30 change that requires us to start passing a  
> valid dev pointer to get the proper dma_ops.

I'm pretty sure that went into .29.  And invalid dev pointer is the  
most likely culprit.  IIRC, the usual cause of this is that you're  
passing in the *wrong* dev pointer.  There are often struct  
hierarchies with, confusingly, multiple struct device pointers.  You  
need the one which has archdata dma_ops setup properly.  For an of  
device, you want the dev pointer that is part of the of_device struct.

If this isn't your problem, then please post some code so we can look  
at this further, and post a log of what's happening.

Cheers,
Becky

Re: dma_ops->map_page == NULL

[Mark Nelson]

On Tuesday 07 July 2009 21:08:25 Benjamin Herrenschmidt wrote:
> On Tue, 2009-07-07 at 10:15 +1000, Mark Nelson wrote:
> > 
> > When the 32 and 64bit DMA code was merged in .28 , map_/unmap_page() was
> > added in favour of map_/unmap_single() (which was later removed in .29)
> > so you'll have to replace your calls to dma_map_single() with
> > dma_map_page(). Just pass it the page and offset rather than the address.
> 
> Wait a minute ... dma_map_single() should still work, it will just call
> dma_map_page() underneath. All dma_ops should have a ->map page
> callback.

Sorry my mistake - I was thinking of when we removed the map/unmap_single()
from the dma_mapping_ops.

Mark.

Re: dma_ops->map_page == NULL

[Jon Smirl]

On Mon, Jul 6, 2009 at 1:51 PM, K=E1ri Dav=ED=F0sson<kari.davidsson@marel.c=
om> wrote:
> I am doing a driver that uses dma_map_single().
>
> After changing to to linux 2.6.29.3 I am getting
> segfaults in dma_map_single() because dma_ops->map_page is NULL.
> Actually dma_ops looks funky too.
>
> The driver is an of_platform_driver which is declared as an child of
> the lbp (fsl,lpb) node of the device tree.
>
> This is on powerpc 5200b platform.

I have hit this before, in my case the driver pointer was null. It
faulted on dma_ops->map_page which had nothing to do with the real
problem which was the null driver pointer. I spent a long time looking
at this until I figured out that the reported error was bogus compared
to the actual problem.


>
> rg
> kd
>
> _______________________________________________
> Linuxppc-dev mailing list
> Linuxppc-dev@lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/linuxppc-dev
>



--=20
Jon Smirl
jonsmirl@gmail.com