Re: [patch 0/4] Intel IOMMU Supspend/Resume Support

Re: [patch 0/4] Intel IOMMU Supspend/Resume Support

[fenghua.yu]

>> Current Intel IOMMU does not support suspend and resume. In S3 event, kernel
> crashes when DMAR or interrupt remapping is running.
>>
>> The attached patch set implements the suspend and resume feature for Intel
>> IOMMU. It hooks to kernel suspend and resume interface. When suspend happens, itsaves necessary hardware registers. When resume happens, it restores the
>> registers and restarts IOMMU.
>>
>> This patch set can be applied on the tip tree. 
>  
> Looks good to me, apart from the fact that your #4 patch actually needs
>to come before #1. Shall I apply it to my iommu-2.6.git tree or let Ingo
>take it? If the latter, 
>
>Acked-by: David Woodhouse <David.Woodhouse@intel.com>
>Acked-by: David Woodhouse <David.Woodhouse@intel.com>
>Acked-by: David Woodhouse <David.Woodhouse@intel.com>
>Acked-by: David Woodhouse <David.Woodhouse@intel.com>
>
>(if you look carefully, you'll see those are in order 4, 1, 2, 3)
>
>Have we tested this on IA64?

IA64 platforms don't support suspend/resume. The original 4 patches can be
compiled and run on IA64. Only problem is the sysdev interface iommu is
installed but not used.

On the top of the 4 patches, you can apply the following small patch for IA64.
With this small patch, IA64 kernel won't build IOMMU suspend/resume code at all.


Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>

---

 intel-iommu.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c
index 5066668..2d9be1d 100644
--- a/drivers/pci/intel-iommu.c
+++ b/drivers/pci/intel-iommu.c
@@ -2564,7 +2564,7 @@ static void __init init_no_remapping_devices(void)
 	}
 }
 
-#ifdef CONFIG_PM
+#ifdef CONFIG_SUSPEND
 static int init_iommu_hw(void)
 {
 	struct dmar_drhd_unit *drhd;
@@ -2702,7 +2702,7 @@ static int __init init_iommu_sysfs(void)
 }
 
 #else
-static init __init init_iommu_sysfs(void)
+static int __init init_iommu_sysfs(void)
 {
 	return 0;
 }
-- 

Re: [patch 0/4] Intel IOMMU Supspend/Resume Support

[fenghua.yu]

>>
>> The attached patch set implements the suspend and resume feature for
>Intel
>> IOMMU. It hooks to kernel suspend and resume interface. When suspend
>happens, itsaves necessary hardware registers. When resume happens, it
>restores the
>> registers and restarts IOMMU.
>>
>> This patch set can be applied on the tip tree.
>
>Looks good to me, apart from the fact that your #4 patch actually needs
>to come before #1. Shall I apply it to my iommu-2.6.git tree or let Ingo
>take it? If the latter,
>
>Acked-by: David Woodhouse <David.Woodhouse@intel.com>
>Acked-by: David Woodhouse <David.Woodhouse@intel.com>
>Acked-by: David Woodhouse <David.Woodhouse@intel.com>
>Acked-by: David Woodhouse <David.Woodhouse@intel.com>
>
>(if you look carefully, you'll see those are in order 4, 1, 2, 3)
>
>Have we tested this on IA64?

Since ia64 platforms don't support suspend/resume, the code won't take action onia64. But it still installs unused sys device interface.

I'll clean up the code for ia64 and re-send the patch set.

Thanks.

-Fenghua

-- 

Re: [patch 0/4] Intel IOMMU Supspend/Resume Support

[Andrew Lutomirski]

Works for me (applied to current -tip).  I suspended and resumed just
fine, with no errors logged and all hardware apparently still working.

Tested-by: Andy Lutomirski <amluto@gmail.com>

Thanks for the patch!
Andy

On Fri, Mar 27, 2009 at 5:22 PM, Fenghua Yu <fenghua.yu@intel.com> wrote:
> Current Intel IOMMU does not support suspend and resume. In S3 event, kernel
> crashes when DMAR or interrupt remapping is running.
>
> The attached patch set implements the suspend and resume feature for Intel
> IOMMU. It hooks to kernel suspend and resume interface. When suspend happens, itsaves necessary hardware registers. When resume happens, it restores the
> registers and restarts IOMMU.
>
> This patch set can be applied on the tip tree.
>
> --
>
>

Re: [patch 0/4] Intel IOMMU Supspend/Resume Support

[David Woodhouse]

On Fri, 2009-03-27 at 14:22 -0700, Fenghua Yu wrote:
> Current Intel IOMMU does not support suspend and resume. In S3 event, kernel
> crashes when DMAR or interrupt remapping is running.
> 
> The attached patch set implements the suspend and resume feature for Intel
> IOMMU. It hooks to kernel suspend and resume interface. When suspend happens, itsaves necessary hardware registers. When resume happens, it restores the
> registers and restarts IOMMU.
> 
> This patch set can be applied on the tip tree. 

Looks good to me, apart from the fact that your #4 patch actually needs
to come before #1. Shall I apply it to my iommu-2.6.git tree or let Ingo
take it? If the latter, 

Acked-by: David Woodhouse <David.Woodhouse@intel.com>
Acked-by: David Woodhouse <David.Woodhouse@intel.com>
Acked-by: David Woodhouse <David.Woodhouse@intel.com>
Acked-by: David Woodhouse <David.Woodhouse@intel.com>

(if you look carefully, you'll see those are in order 4, 1, 2, 3)

Have we tested this on IA64?

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