[PATCH] Fix bug in iommu_context_addr: Always get pointer to lower extended-context-table

[PATCH] Fix bug in iommu_context_addr: Always get pointer to lower extended-context-table

[Nan Xiao]

Hi David & Joerg,

In iommu_context_addr() function, we always use lower extended-context-table
even though upper-half of the extended root-entry is present.

Signed-off-by: Nan Xiao <nan-IAhD+5PxXTI3bvoXafbxaA@public.gmane.org>
---
 drivers/iommu/intel-iommu.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c
index 0649b94..4213598 100644
--- a/drivers/iommu/intel-iommu.c
+++ b/drivers/iommu/intel-iommu.c
@@ -759,10 +759,11 @@ static inline struct context_entry *iommu_context_addr(struct intel_iommu *iommu
 		if (devfn >= 0x80) {
 			devfn -= 0x80;
 			entry = &root->hi;
-		}
+		} else
+			entry = &root->lo;
 		devfn *= 2;
-	}
-	entry = &root->lo;
+	} else
+		entry = &root->lo;
 	if (*entry & 1)
 		context = phys_to_virt(*entry & VTD_PAGE_MASK);
 	else {
-- 
2.4.3

Best Regards
Nan Xiao

[PATCH] Fix bug in iommu_context_addr: Always get pointer to lower extended-context-table

[Xiao, Nan (Nan@HPS Performance, Beijing)]

Hi David & joerg,

In iommu_context_addr() function, we always use lower extended-context-table
even though upper-half of the extended root-entry is present.

Signed-off-by: Nan Xiao <nan.xiao@hp.com>
---
 drivers/iommu/intel-iommu.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c
index 0649b94..4213598 100644
--- a/drivers/iommu/intel-iommu.c
+++ b/drivers/iommu/intel-iommu.c
@@ -759,10 +759,11 @@ static inline struct context_entry *iommu_context_addr(struct intel_iommu *iommu
 		if (devfn >= 0x80) {
 			devfn -= 0x80;
 			entry = &root->hi;
-		}
+		} else
+			entry = &root->lo;
 		devfn *= 2;
-	}
-	entry = &root->lo;
+	} else
+		entry = &root->lo;
 	if (*entry & 1)
 		context = phys_to_virt(*entry & VTD_PAGE_MASK);
 	else {
-- 
2.4.3

Best Regards
Nan Xiao

Re: [PATCH] Fix bug in iommu_context_addr: Always get pointer to lower extended-context-table

[joro-zLv9SwRftAIdnm+yROfE0A]

Hi Nan,

On Mon, Aug 24, 2015 at 06:26:33AM +0000, Xiao, Nan (Nan@HPS Performance, Beijing) wrote:
>  drivers/iommu/intel-iommu.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c
> index 0649b94..4213598 100644
> --- a/drivers/iommu/intel-iommu.c
> +++ b/drivers/iommu/intel-iommu.c
> @@ -759,10 +759,11 @@ static inline struct context_entry *iommu_context_addr(struct intel_iommu *iommu
>  		if (devfn >= 0x80) {
>  			devfn -= 0x80;
>  			entry = &root->hi;
> -		}
> +		} else
> +			entry = &root->lo;
>  		devfn *= 2;
> -	}
> -	entry = &root->lo;
> +	} else
> +		entry = &root->lo;
>  	if (*entry & 1)
>  		context = phys_to_virt(*entry & VTD_PAGE_MASK);
>  	else {

Shouldn't this diff have the same effect?

diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c
index 0649b94..7553cb9 100644
--- a/drivers/iommu/intel-iommu.c
+++ b/drivers/iommu/intel-iommu.c
@@ -755,6 +755,7 @@ static inline struct context_entry *iommu_context_addr(struct intel_iommu *iommu
 	struct context_entry *context;
 	u64 *entry;
 
+	entry = &root->lo;
 	if (ecs_enabled(iommu)) {
 		if (devfn >= 0x80) {
 			devfn -= 0x80;
@@ -762,7 +763,6 @@ static inline struct context_entry *iommu_context_addr(struct intel_iommu *iommu
 		}
 		devfn *= 2;
 	}
-	entry = &root->lo;
 	if (*entry & 1)
 		context = phys_to_virt(*entry & VTD_PAGE_MASK);
 	else {

RE: [PATCH] Fix bug in iommu_context_addr: Always get pointer to lower extended-context-table

[Xiao, Nan (Nan@HPservers-Core-OE-PSC)]

Hi Joerg,

Yes, your patch is simple and better!

Best Regards
Nan Xiao

-----Original Message-----
From: joro-zLv9SwRftAIdnm+yROfE0A@public.gmane.org [mailto:joro-zLv9SwRftAIdnm+yROfE0A@public.gmane.org] 
Sent: Tuesday, August 25, 2015 4:42 PM
To: Xiao, Nan (Nan@HPservers-Core-OE-PSC)
Cc: dwmw2-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org; iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org; linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
Subject: Re: [PATCH] Fix bug in iommu_context_addr: Always get pointer to lower extended-context-table

Hi Nan,

On Mon, Aug 24, 2015 at 06:26:33AM +0000, Xiao, Nan (Nan@HPS Performance, Beijing) wrote:
>  drivers/iommu/intel-iommu.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c 
> index 0649b94..4213598 100644
> --- a/drivers/iommu/intel-iommu.c
> +++ b/drivers/iommu/intel-iommu.c
> @@ -759,10 +759,11 @@ static inline struct context_entry *iommu_context_addr(struct intel_iommu *iommu
>  		if (devfn >= 0x80) {
>  			devfn -= 0x80;
>  			entry = &root->hi;
> -		}
> +		} else
> +			entry = &root->lo;
>  		devfn *= 2;
> -	}
> -	entry = &root->lo;
> +	} else
> +		entry = &root->lo;
>  	if (*entry & 1)
>  		context = phys_to_virt(*entry & VTD_PAGE_MASK);
>  	else {

Shouldn't this diff have the same effect?

diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c index 0649b94..7553cb9 100644
--- a/drivers/iommu/intel-iommu.c
+++ b/drivers/iommu/intel-iommu.c
@@ -755,6 +755,7 @@ static inline struct context_entry *iommu_context_addr(struct intel_iommu *iommu
 	struct context_entry *context;
 	u64 *entry;
 
+	entry = &root->lo;
 	if (ecs_enabled(iommu)) {
 		if (devfn >= 0x80) {
 			devfn -= 0x80;
@@ -762,7 +763,6 @@ static inline struct context_entry *iommu_context_addr(struct intel_iommu *iommu
 		}
 		devfn *= 2;
 	}
-	entry = &root->lo;
 	if (*entry & 1)
 		context = phys_to_virt(*entry & VTD_PAGE_MASK);
 	else {

Re: [PATCH] Fix bug in iommu_context_addr: Always get pointer to lower extended-context-table

[joro-zLv9SwRftAIdnm+yROfE0A]

On Tue, Aug 25, 2015 at 08:46:27AM +0000, Xiao, Nan (Nan@HPservers-Core-OE-PSC) wrote:
> Yes, your patch is simple and better!

Okay, I queued this patch to my x86/vt-d branch:

>From 03932776424a799c3f065a69e98076a78530288b Mon Sep 17 00:00:00 2001
From: Joerg Roedel <jroedel-l3A5Bk7waGM@public.gmane.org>
Date: Tue, 25 Aug 2015 10:54:28 +0200
Subject: [PATCH] iommu/vt-d: Really use upper context table when necessary

There is a bug in iommu_context_addr() which will always use
the lower context table, event when the upper context table
needs to be used. Fix this issue.

Fixes: 03ecc32c5274 ("iommu/vt-d: support extended root and context entries")
Reported-by: Xiao, Nan <nan.xiao-VXdhtT5mjnY@public.gmane.org>
Signed-off-by: Joerg Roedel <jroedel-l3A5Bk7waGM@public.gmane.org>
---
 drivers/iommu/intel-iommu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c
index a85077d..63daf1b 100644
--- a/drivers/iommu/intel-iommu.c
+++ b/drivers/iommu/intel-iommu.c
@@ -803,6 +803,7 @@ static inline struct context_entry *iommu_context_addr(struct intel_iommu *iommu
 	struct context_entry *context;
 	u64 *entry;
 
+	entry = &root->lo;
 	if (ecs_enabled(iommu)) {
 		if (devfn >= 0x80) {
 			devfn -= 0x80;
@@ -810,7 +811,6 @@ static inline struct context_entry *iommu_context_addr(struct intel_iommu *iommu
 		}
 		devfn *= 2;
 	}
-	entry = &root->lo;
 	if (*entry & 1)
 		context = phys_to_virt(*entry & VTD_PAGE_MASK);
 	else {
-- 
1.8.4.5

RE: [PATCH] Fix bug in iommu_context_addr: Always get pointer to lower extended-context-table

[Xiao, Nan (Nan@HPservers-Core-OE-PSC)]

Hi Joerg,

In commit message:

> There is a bug in iommu_context_addr() which will always use the lower context table, event when the upper context table needs to be used. Fix this issue.

I think it should be " even when the upper context table ", not "event".

Thanks!

Best Regards
Nan Xiao

-----Original Message-----
From: joro-zLv9SwRftAIdnm+yROfE0A@public.gmane.org [mailto:joro-zLv9SwRftAIdnm+yROfE0A@public.gmane.org] 
Sent: Tuesday, August 25, 2015 5:00 PM
To: Xiao, Nan (Nan@HPservers-Core-OE-PSC)
Cc: dwmw2-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org; iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org; linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
Subject: Re: [PATCH] Fix bug in iommu_context_addr: Always get pointer to lower extended-context-table

On Tue, Aug 25, 2015 at 08:46:27AM +0000, Xiao, Nan (Nan@HPservers-Core-OE-PSC) wrote:
> Yes, your patch is simple and better!

Okay, I queued this patch to my x86/vt-d branch:

>From 03932776424a799c3f065a69e98076a78530288b Mon Sep 17 00:00:00 2001
From: Joerg Roedel <jroedel-l3A5Bk7waGM@public.gmane.org>
Date: Tue, 25 Aug 2015 10:54:28 +0200
Subject: [PATCH] iommu/vt-d: Really use upper context table when necessary

There is a bug in iommu_context_addr() which will always use the lower context table, event when the upper context table needs to be used. Fix this issue.

Fixes: 03ecc32c5274 ("iommu/vt-d: support extended root and context entries")
Reported-by: Xiao, Nan <nan.xiao-VXdhtT5mjnY@public.gmane.org>
Signed-off-by: Joerg Roedel <jroedel-l3A5Bk7waGM@public.gmane.org>
---
 drivers/iommu/intel-iommu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c index a85077d..63daf1b 100644
--- a/drivers/iommu/intel-iommu.c
+++ b/drivers/iommu/intel-iommu.c
@@ -803,6 +803,7 @@ static inline struct context_entry *iommu_context_addr(struct intel_iommu *iommu
 	struct context_entry *context;
 	u64 *entry;
 
+	entry = &root->lo;
 	if (ecs_enabled(iommu)) {
 		if (devfn >= 0x80) {
 			devfn -= 0x80;
@@ -810,7 +811,6 @@ static inline struct context_entry *iommu_context_addr(struct intel_iommu *iommu
 		}
 		devfn *= 2;
 	}
-	entry = &root->lo;
 	if (*entry & 1)
 		context = phys_to_virt(*entry & VTD_PAGE_MASK);
 	else {
--
1.8.4.5

Re: [PATCH] Fix bug in iommu_context_addr: Always get pointer to lower extended-context-table

[joro]

On Tue, Aug 25, 2015 at 09:15:39AM +0000, Xiao, Nan (Nan@HPservers-Core-OE-PSC) wrote:
> In commit message:
> 
> > There is a bug in iommu_context_addr() which will always use the lower context table, event when the upper context table needs to be used. Fix this issue.
> 
> I think it should be " even when the upper context table ", not "event".

Right, fixed that. Thanks for noticing.


	Joerg