[RFC patch] clean up swiotlb check in intel_iommu_init

[RFC patch] clean up swiotlb check in intel_iommu_init

[Luming Yu]

[-- Attachment #1: Type: text/plain, Size: 955 bytes --]

Hi there,

Since swiotlb is an fall back option that might have
initialized. We don't need to stop intel_iommu_init
because of swiotlb..

Please review. If make sense, please apply.

**The patch is enclosed in text attachment*
**Using web client to send the patch* *
**below is for review, please apply attached  patch*/

Thanks,
Luming

Signed-off-by: Yu Luming <luming.yu@intel.com>

 intel-iommu.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)


diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c
index 2314ad7..246acb9 100644
--- a/drivers/pci/intel-iommu.c
+++ b/drivers/pci/intel-iommu.c
@@ -3194,7 +3194,7 @@ int __init intel_iommu_init(void)
 	 * Check the need for DMA-remapping initialization now.
 	 * Above initialization will also be used by Interrupt-remapping.
 	 */
-	if (no_iommu || (swiotlb && !iommu_pass_through) || dmar_disabled)
+	if (no_iommu || dmar_disabled)
 		return -ENODEV;

 	iommu_init_mempool();

[-- Attachment #2: 6 --]
[-- Type: application/octet-stream, Size: 496 bytes --]

diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c
index 2314ad7..246acb9 100644
--- a/drivers/pci/intel-iommu.c
+++ b/drivers/pci/intel-iommu.c
@@ -3194,7 +3194,7 @@ int __init intel_iommu_init(void)
 	 * Check the need for DMA-remapping initialization now.
 	 * Above initialization will also be used by Interrupt-remapping.
 	 */
-	if (no_iommu || (swiotlb && !iommu_pass_through) || dmar_disabled)
+	if (no_iommu || dmar_disabled)
 		return -ENODEV;
 
 	iommu_init_mempool();

Re: [RFC patch] clean up swiotlb check in intel_iommu_init

[FUJITA Tomonori]

On Thu, 13 Aug 2009 11:11:27 +0800
Luming Yu <luming.yu@gmail.com> wrote:

> Hi there,
> 
> Since swiotlb is an fall back option that might have
> initialized. We don't need to stop intel_iommu_init
> because of swiotlb..

I think that this doesn't work on X86; breaks "soft" boot option.


> Please review. If make sense, please apply.
> 
> **The patch is enclosed in text attachment*
> **Using web client to send the patch* *
> **below is for review, please apply attached  patch*/
> 
> Thanks,
> Luming
> 
> Signed-off-by: Yu Luming <luming.yu@intel.com>
> 
>  intel-iommu.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> 
> diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c
> index 2314ad7..246acb9 100644
> --- a/drivers/pci/intel-iommu.c
> +++ b/drivers/pci/intel-iommu.c
> @@ -3194,7 +3194,7 @@ int __init intel_iommu_init(void)
>  	 * Check the need for DMA-remapping initialization now.
>  	 * Above initialization will also be used by Interrupt-remapping.
>  	 */
> -	if (no_iommu || (swiotlb && !iommu_pass_through) || dmar_disabled)
> +	if (no_iommu || dmar_disabled)
>  		return -ENODEV;
> 
>  	iommu_init_mempool();

Re: [RFC patch] clean up swiotlb check in intel_iommu_init

[Luming Yu]

>>
>> Since swiotlb is an fall back option that might have
>> initialized. We don't need to stop intel_iommu_init
>> because of swiotlb..
>
> I think that this doesn't work on X86; breaks "soft" boot option.

Assumption is that swiotlb is default dma_ops.
I got this patch when working on ia64, will check x86 later.
But it is only needed when most of us agree that swiotlb should be
the default.

>
>
>> Please review. If make sense, please apply.
>>
>> **The patch is enclosed in text attachment*
>> **Using web client to send the patch* *
>> **below is for review, please apply attached  patch*/
>>
>> Thanks,
>> Luming
>>
>> Signed-off-by: Yu Luming <luming.yu@intel.com>
>>
>>  intel-iommu.c |    2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>>
>> diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c
>> index 2314ad7..246acb9 100644
>> --- a/drivers/pci/intel-iommu.c
>> +++ b/drivers/pci/intel-iommu.c
>> @@ -3194,7 +3194,7 @@ int __init intel_iommu_init(void)
>>        * Check the need for DMA-remapping initialization now.
>>        * Above initialization will also be used by Interrupt-remapping.
>>        */
>> -     if (no_iommu || (swiotlb && !iommu_pass_through) || dmar_disabled)
>> +     if (no_iommu || dmar_disabled)
>>               return -ENODEV;
>>
>>       iommu_init_mempool();
>

Re: [RFC patch] clean up swiotlb check in intel_iommu_init

[David Woodhouse]

On Thu, 2009-08-13 at 11:11 +0800, Luming Yu wrote:
> 
> Since swiotlb is an fall back option that might have
> initialized. We don't need to stop intel_iommu_init
> because of swiotlb..
> 
> Please review. If make sense, please apply.

It doesn't apply to the iommu-2.6.git tree.

-- 
David Woodhouse                            Open Source Technology Centre
David.Woodhouse@intel.com                              Intel Corporation

Re: [RFC patch] clean up swiotlb check in intel_iommu_init

[Luming Yu]

On Thu, Aug 13, 2009 at 4:23 PM, David Woodhouse<dwmw2@infradead.org> wrote:
> On Thu, 2009-08-13 at 11:11 +0800, Luming Yu wrote:
>>
>> Since swiotlb is an fall back option that might have
>> initialized. We don't need to stop intel_iommu_init
>> because of swiotlb..
>>
>> Please review. If make sense, please apply.
>
> It doesn't apply to the iommu-2.6.git tree.

The patch is against 31-rc5.

If you were not using the attached patch, then I need to use another
mail client to re-send the patch to you. But it is not as convenient
as web gmail
that I'm current using...

If I must generate patch against iommu-2.6.git tree, please point out
the url of that tree,
I can refresh this patch later.

Please let me know what I should do to push this patch in..