Skip to contentSkip to conversation

iommu: fix PVH dom0 settings

PATCH In review Paul Durrant 4 participants 6 messages Sep 27, 2019

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 iommu: fix PVH dom0 settings

PVH dom0 must operate with the iommu settings in 'strict' mode i.e. only the domain's own pages will be mapped in the IOMMU. The check_hwdom_reqs() is supposed to ensure this. Unfortunately the test for a PVH dom0 is made using paging_mode_translate() and, when commit f89f5558 "remove late (on-demand) construction of IOMMU page tables" moved the call of check_hwdom_reqs() from iommu_hwdom_init() to iommu_domain_init(), that test became ineffective (because iommu_domain_init() is called before paging_enable()).

This patch replaces the test of paging_mode_translate() with a test of hap_enabled(), and also verifies 'strict' mode is turned on in arch_iommu_check_autotranslated_hwdom().

---
Cc: Jan Beulich <[email protected]>

---
 xen/drivers/passthrough/iommu.c     | 6 +++---
 xen/drivers/passthrough/x86/iommu.c | 3 +++
 2 files changed, 6 insertions(+), 3 deletions(-)
xen/drivers/passthrough/iommu.c +3 -3
index 2733b320ec..8b550f909b 100644
@@ -154,13 +154,13 @@ custom_param("dom0-iommu", parse_dom0_iommu_param);
 
 static void __hwdom_init check_hwdom_reqs(struct domain *d)
 {
-    if ( iommu_hwdom_none || !paging_mode_translate(d) )
+    if ( iommu_hwdom_none || !hap_enabled(d) )
         return;
 
-    arch_iommu_check_autotranslated_hwdom(d);
-
     iommu_hwdom_passthrough = false;
     iommu_hwdom_strict = true;
+
+    arch_iommu_check_autotranslated_hwdom(d);
 }
 
 int iommu_domain_init(struct domain *d, unsigned int opts)
xen/drivers/passthrough/x86/iommu.c +3 -0
index 47a3e55213..f54805babd 100644
@@ -85,6 +85,9 @@ void __hwdom_init arch_iommu_check_autotranslated_hwdom(struct domain *d)
 {
     if ( !is_iommu_enabled(d) )
         panic("Presently, iommu must be enabled for PVH hardware domain\n");
+
+    if ( !iommu_hwdom_strict )
+        panic("PVH hardware domain iommu must be set in 'strict' mode\n");
 }
 
 int arch_iommu_domain_init(struct domain *d)
Signature · 7 lines
2.20.1.2.gb21ebb671


_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Reviewed-by on v1 1/1
#1 lore raw
Quoted 9 lines · Paul Durrant

PVH dom0 must operate with the iommu settings in 'strict' mode i.e. only the domain's own pages will be mapped in the IOMMU. The check_hwdom_reqs() is supposed to ensure this. Unfortunately the test for a PVH dom0 is made using paging_mode_translate() and, when commit f89f5558 "remove late (on-demand) construction of IOMMU page tables" moved the call of check_hwdom_reqs() from iommu_hwdom_init() to iommu_domain_init(), that test became ineffective (because iommu_domain_init() is called before paging_enable()).

Quoted 6 lines

This patch replaces the test of paging_mode_translate() with a test of hap_enabled(), and also verifies 'strict' mode is turned on in arch_iommu_check_autotranslated_hwdom().

Nit: Conventionally your two tags should be the other way around, to represent events chronologically.

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#2 lore raw
In reply to Paul Durrant
Quoted 9 lines · Paul Durrant

PVH dom0 must operate with the iommu settings in 'strict' mode i.e. only the domain's own pages will be mapped in the IOMMU. The check_hwdom_reqs() is supposed to ensure this. Unfortunately the test for a PVH dom0 is made using paging_mode_translate() and, when commit f89f5558 "remove late (on-demand) construction of IOMMU page tables" moved the call of check_hwdom_reqs() from iommu_hwdom_init() to iommu_domain_init(), that test became ineffective (because iommu_domain_init() is called before paging_enable()).

Quoted 6 lines

This patch replaces the test of paging_mode_translate() with a test of hap_enabled(), and also verifies 'strict' mode is turned on in arch_iommu_check_autotranslated_hwdom().

Oh, and I guess you've also meant to Cc Jürgen for a release ack (now done)?

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#3 lore raw
Quoted 14 lines · Jan Beulich
Quoted 13 lines · Paul Durrant

PVH dom0 must operate with the iommu settings in 'strict' mode i.e. only the domain's own pages will be mapped in the IOMMU. The check_hwdom_reqs() is supposed to ensure this. Unfortunately the test for a PVH dom0 is made using paging_mode_translate() and, when commit f89f5558 "remove late (on-demand) construction of IOMMU page tables" moved the call of check_hwdom_reqs() from iommu_hwdom_init() to iommu_domain_init(), that test became ineffective (because iommu_domain_init() is called before paging_enable()).

This patch replaces the test of paging_mode_translate() with a test of hap_enabled(), and also verifies 'strict' mode is turned on in arch_iommu_check_autotranslated_hwdom().

Quoted 5 lines
Quoted 2 lines

Oh, and I guess you've also meant to Cc Jürgen for a release ack (now done)?

Thanks for the Cc!

Juergen

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#4 lore raw
In reply to Paul Durrant
Quoted 27 lines · Paul Durrant

PVH dom0 must operate with the iommu settings in 'strict' mode i.e. only the domain's own pages will be mapped in the IOMMU. The check_hwdom_reqs() is supposed to ensure this. Unfortunately the test for a PVH dom0 is made using paging_mode_translate() and, when commit f89f5558 "remove late (on-demand) construction of IOMMU page tables" moved the call of check_hwdom_reqs() from iommu_hwdom_init() to iommu_domain_init(), that test became ineffective (because iommu_domain_init() is called before paging_enable()).

This patch replaces the test of paging_mode_translate() with a test of hap_enabled(), and also verifies 'strict' mode is turned on in arch_iommu_check_autotranslated_hwdom().

---
Cc: Jan Beulich <[email protected]>

---
 xen/drivers/passthrough/iommu.c     | 6 +++---
 xen/drivers/passthrough/x86/iommu.c | 3 +++
 2 files changed, 6 insertions(+), 3 deletions(-)
xen/drivers/passthrough/iommu.c +0 -0
index 2733b320ec..8b550f909b 100644

xen/drivers/passthrough/iommu.c custom_param("dom0-iommu", parse_dom0_iommu_param); · line 157

Quoted 5 lines
xen/drivers/passthrough/iommu.c +1 -1
@@ -154,13 +154,13 @@ custom_param("dom0-iommu", parse_dom0_iommu_param);
 
 static void __hwdom_init check_hwdom_reqs(struct domain *d)
 {
-    if ( iommu_hwdom_none || !paging_mode_translate(d) )
+    if ( iommu_hwdom_none || !hap_enabled(d) )

Since dom0 PVH can also be used with shadow paging (not sure how useful that is TBH), I'm not sure explicitly checking for hap is fine. What about using is_hvm_domain instead?

That ought to cover both shadow and hap, and a classic PV dom0 won't be affected by it (which is the intention).

Thanks, Roger.

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#5 lore raw
Quoted 14 lines · Roger Pau Monné

xen/drivers/passthrough/iommu.c custom_param("dom0-iommu", parse_dom0_iommu_param); · line 157

Quoted 8 lines · Paul Durrant
xen/drivers/passthrough/iommu.c +1 -1
@@ -154,13 +154,13 @@ custom_param("dom0-iommu", parse_dom0_iommu_param);
 
 static void __hwdom_init check_hwdom_reqs(struct domain *d)
 {
-    if ( iommu_hwdom_none || !paging_mode_translate(d) )
+    if ( iommu_hwdom_none || !hap_enabled(d) )

Since dom0 PVH can also be used with shadow paging (not sure how useful that is TBH), I'm not sure explicitly checking for hap is fine. What about using is_hvm_domain instead?

Quoted 2 lines

That ought to cover both shadow and hap, and a classic PV dom0 won't be affected by it (which is the intention).

Oh, indeed - we shouldn't prevent shadow mode use. That'll need to be an incremental change though, as I've already committed the one here.

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel