[PATCH 1/1] mm: EXPORT_SYMBOL_GPL(find_vm_area);

[PATCH 1/1] mm: EXPORT_SYMBOL_GPL(find_vm_area);

[Sakari Ailus]

find_vm_area() is needed in implementing the DMA mapping API as a module.
Device specific IOMMUs with associated DMA mapping implementations should be
buildable as modules.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
---
 mm/vmalloc.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/mm/vmalloc.c b/mm/vmalloc.c
index 2faaa29..d06db45 100644
--- a/mm/vmalloc.c
+++ b/mm/vmalloc.c
@@ -1416,6 +1416,7 @@ struct vm_struct *find_vm_area(const void *addr)
 
 	return NULL;
 }
+EXPORT_SYMBOL_GPL(find_vm_area);
 
 /**
  *	remove_vm_area  -  find and remove a continuous kernel virtual area
-- 
2.1.0.231.g7484e3b

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH 1/1] mm: EXPORT_SYMBOL_GPL(find_vm_area);

[Andrew Morton]

On Wed, 11 Nov 2015 15:06:24 +0200 Sakari Ailus <sakari.ailus@linux.intel.com> wrote:

> find_vm_area() is needed in implementing the DMA mapping API as a module.
> Device specific IOMMUs with associated DMA mapping implementations should be
> buildable as modules.
> 
> ...
>
> --- a/mm/vmalloc.c
> +++ b/mm/vmalloc.c
> @@ -1416,6 +1416,7 @@ struct vm_struct *find_vm_area(const void *addr)
>  
>  	return NULL;
>  }
> +EXPORT_SYMBOL_GPL(find_vm_area);

Confused.  Who is setting CONFIG_HAS_DMA=m?

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH 1/1] mm: EXPORT_SYMBOL_GPL(find_vm_area);

[Sakari Ailus]

Hi Andrew,

Andrew Morton wrote:
> On Wed, 11 Nov 2015 15:06:24 +0200 Sakari Ailus <sakari.ailus@linux.intel.com> wrote:
> 
>> find_vm_area() is needed in implementing the DMA mapping API as a module.
>> Device specific IOMMUs with associated DMA mapping implementations should be
>> buildable as modules.
>>
>> ...
>>
>> --- a/mm/vmalloc.c
>> +++ b/mm/vmalloc.c
>> @@ -1416,6 +1416,7 @@ struct vm_struct *find_vm_area(const void *addr)
>>  
>>  	return NULL;
>>  }
>> +EXPORT_SYMBOL_GPL(find_vm_area);
> 
> Confused.  Who is setting CONFIG_HAS_DMA=m?
> 

Apologies for the late reply --- CONFIG_HAS_DMA isn't configured as a
module, but some devices are not DMA coherent even on x86. The existing
x86 DMA mapping implementation doesn't quite work for those at the
moment, and nothing prevents using another one (and as a module, in
which case this patch is required).

-- 
Kind regards,

Sakari Ailus
sakari.ailus@linux.intel.com

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH 1/1] mm: EXPORT_SYMBOL_GPL(find_vm_area);

[Andrew Morton]

On Mon, 11 Jan 2016 15:00:46 +0200 Sakari Ailus <sakari.ailus@linux.intel.com> wrote:

> Hi Andrew,
> 
> Andrew Morton wrote:
> > On Wed, 11 Nov 2015 15:06:24 +0200 Sakari Ailus <sakari.ailus@linux.intel.com> wrote:
> > 
> >> find_vm_area() is needed in implementing the DMA mapping API as a module.
> >> Device specific IOMMUs with associated DMA mapping implementations should be
> >> buildable as modules.
> >>
> >> ...
> >>
> >> --- a/mm/vmalloc.c
> >> +++ b/mm/vmalloc.c
> >> @@ -1416,6 +1416,7 @@ struct vm_struct *find_vm_area(const void *addr)
> >>  
> >>  	return NULL;
> >>  }
> >> +EXPORT_SYMBOL_GPL(find_vm_area);
> > 
> > Confused.  Who is setting CONFIG_HAS_DMA=m?
> > 
> 
> Apologies for the late reply --- CONFIG_HAS_DMA isn't configured as a
> module, but some devices are not DMA coherent even on x86. The existing
> x86 DMA mapping implementation doesn't quite work for those at the
> moment, and nothing prevents using another one (and as a module, in
> which case this patch is required).

hm, if you say so.

Please resend the patch sometime along with a much much more detailed
changelog?  If at all appropriate please cc the x86 maintainers so they
can understand why the current DMA mapping implementation is
inappropriate for whatever this requirement is.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH 1/1] mm: EXPORT_SYMBOL_GPL(find_vm_area);

[Dan Williams]

On Mon, Jan 11, 2016 at 5:00 AM, Sakari Ailus
<sakari.ailus@linux.intel.com> wrote:
> Hi Andrew,
>
> Andrew Morton wrote:
>> On Wed, 11 Nov 2015 15:06:24 +0200 Sakari Ailus <sakari.ailus@linux.intel.com> wrote:
>>
>>> find_vm_area() is needed in implementing the DMA mapping API as a module.
>>> Device specific IOMMUs with associated DMA mapping implementations should be
>>> buildable as modules.
>>>
>>> ...
>>>
>>> --- a/mm/vmalloc.c
>>> +++ b/mm/vmalloc.c
>>> @@ -1416,6 +1416,7 @@ struct vm_struct *find_vm_area(const void *addr)
>>>
>>>      return NULL;
>>>  }
>>> +EXPORT_SYMBOL_GPL(find_vm_area);
>>
>> Confused.  Who is setting CONFIG_HAS_DMA=m?
>>
>
> Apologies for the late reply --- CONFIG_HAS_DMA isn't configured as a
> module, but some devices are not DMA coherent even on x86. The existing
> x86 DMA mapping implementation doesn't quite work for those at the
> moment, and nothing prevents using another one (and as a module, in
> which case this patch is required).

Why not teach the DMA mapping api how to do cache management for such
devices?  Why would you need find_vm_area() exported?

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>