Skip to contentSkip to conversation

iommu: avoid triggering ASSERT_UNREACHABLE() on ARM...

PATCH In review Paul Durrant 4 participants 6 messages Sep 26, 2019

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 iommu: avoid triggering ASSERT_UNREACHABLE() on ARM...

...when the IOMMU is not enabled.

80ff3d338dc9 "iommu: tidy up iommu_use_hap_pt() and need_iommu_pt_sync() macros" introduced CONFIG_IOMMU_FORCE_PT_SHARE, which causes the global 'iommu_hap_pt_share' option to be replaced with a #define-d value of true. In this configuration, calling clear_iommu_hap_pt_share() will result trigger the aforementioned ASSERT.

CONFIG_IOMMU_FORCE_PT_SHARE is always selected for ARM builds and, because clear_iommu_hap_pt_share() is called by the common iommu_setup() function if the IOMMU is not enabled, it is no longer safe to disable the IOMMU on ARM systems.

However, running with the IOMMU disabled is a valid configuration for ARM systems, so this patch rectifies the problem by removing the call to clear_iommu_hap_pt_share() from common code. As a side effect of this, however, it becomes possible on x86 systems for iommu_enabled to be false but iommu_hap_pt_share to be true. Thus the code in sysctl.c needs to be changed to make sure that XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share is not erroneously advertised when the IOMMU has been disabled.

---
Cc: Andrew Cooper <[email protected]>
Cc: George Dunlap <[email protected]>
Cc: Ian Jackson <[email protected]>
Cc: Jan Beulich <[email protected]>
Cc: Julien Grall <[email protected]>
Cc: Konrad Rzeszutek Wilk <[email protected]>
Cc: Stefano Stabellini <[email protected]>
Cc: Tim Deegan <[email protected]>
Cc: Wei Liu <[email protected]>

---
 xen/common/sysctl.c             | 6 ++++--
 xen/drivers/passthrough/iommu.c | 1 -
 2 files changed, 4 insertions(+), 3 deletions(-)
xen/common/sysctl.c +4 -2
index e8763c7fdf..f88a285e7f 100644
@@ -268,9 +268,11 @@ long do_sysctl(XEN_GUEST_HANDLE_PARAM(xen_sysctl_t) u_sysctl)
         pi->max_mfn = get_upper_mfn_bound();
         arch_do_physinfo(pi);
         if ( iommu_enabled )
+        {
             pi->capabilities |= XEN_SYSCTL_PHYSCAP_directio;
-        if ( iommu_hap_pt_share )
-            pi->capabilities |= XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share;
+            if ( iommu_hap_pt_share )
+                pi->capabilities |= XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share;
+        }
 
         if ( copy_to_guest(u_sysctl, op, 1) )
             ret = -EFAULT;
xen/drivers/passthrough/iommu.c +0 -1
index e8fddc2dc7..c33ca70ec9 100644
@@ -456,7 +456,6 @@ int __init iommu_setup(void)
     if ( !iommu_enabled )
     {
         iommu_intremap = 0;
-        clear_iommu_hap_pt_share();
     }
 
     if ( (force_iommu && !iommu_enabled) ||
Signature · 7 lines
2.20.1.2.gb21ebb671


_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Acked-by on v1 1/1
#1 lore raw

Hi Paul,

Quoted 21 lines · Paul Durrant

...when the IOMMU is not enabled.

80ff3d338dc9 "iommu: tidy up iommu_use_hap_pt() and need_iommu_pt_sync() macros" introduced CONFIG_IOMMU_FORCE_PT_SHARE, which causes the global 'iommu_hap_pt_share' option to be replaced with a #define-d value of true. In this configuration, calling clear_iommu_hap_pt_share() will result trigger the aforementioned ASSERT.

CONFIG_IOMMU_FORCE_PT_SHARE is always selected for ARM builds and, because clear_iommu_hap_pt_share() is called by the common iommu_setup() function if the IOMMU is not enabled, it is no longer safe to disable the IOMMU on ARM systems.

However, running with the IOMMU disabled is a valid configuration for ARM systems, so this patch rectifies the problem by removing the call to clear_iommu_hap_pt_share() from common code. As a side effect of this, however, it becomes possible on x86 systems for iommu_enabled to be false but iommu_hap_pt_share to be true. Thus the code in sysctl.c needs to be changed to make sure that XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share is not erroneously advertised when the IOMMU has been disabled.

Quoted 2 lines

With one NIT below:

Quoted 38 lines

---
Cc: Andrew Cooper <[email protected]>
Cc: George Dunlap <[email protected]>
Cc: Ian Jackson <[email protected]>
Cc: Jan Beulich <[email protected]>
Cc: Julien Grall <[email protected]>
Cc: Konrad Rzeszutek Wilk <[email protected]>
Cc: Stefano Stabellini <[email protected]>
Cc: Tim Deegan <[email protected]>
Cc: Wei Liu <[email protected]>

---
  xen/common/sysctl.c             | 6 ++++--
  xen/drivers/passthrough/iommu.c | 1 -
  2 files changed, 4 insertions(+), 3 deletions(-)
xen/common/sysctl.c +4 -2
index e8763c7fdf..f88a285e7f 100644
@@ -268,9 +268,11 @@ long do_sysctl(XEN_GUEST_HANDLE_PARAM(xen_sysctl_t) u_sysctl)
          pi->max_mfn = get_upper_mfn_bound();
          arch_do_physinfo(pi);
          if ( iommu_enabled )
+        {
              pi->capabilities |= XEN_SYSCTL_PHYSCAP_directio;
-        if ( iommu_hap_pt_share )
-            pi->capabilities |= XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share;
+            if ( iommu_hap_pt_share )
+                pi->capabilities |= XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share;
+        }
 
          if ( copy_to_guest(u_sysctl, op, 1) )
              ret = -EFAULT;
xen/drivers/passthrough/iommu.c +0 -0
index e8fddc2dc7..c33ca70ec9 100644

xen/drivers/passthrough/iommu.c int __init iommu_setup(void) · line 459

Quoted 5 lines
xen/drivers/passthrough/iommu.c +0 -1
@@ -456,7 +456,6 @@ int __init iommu_setup(void)
      if ( !iommu_enabled )
      {
          iommu_intremap = 0;
-        clear_iommu_hap_pt_share();
      }

NIT: The {} can now be removed.

I can fix it on commit.

Cheers,

Signature · 6 lines
Julien Grall

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Tested-by Oleksandr Tyshchenko
#2 lore raw

Hi Paul, Julien

Quoted 24 lines · Julien Grall

Hi Paul,

Quoted 21 lines · Paul Durrant

...when the IOMMU is not enabled.

80ff3d338dc9 "iommu: tidy up iommu_use_hap_pt() and need_iommu_pt_sync() macros" introduced CONFIG_IOMMU_FORCE_PT_SHARE, which causes the global 'iommu_hap_pt_share' option to be replaced with a #define-d value of true. In this configuration, calling clear_iommu_hap_pt_share() will result trigger the aforementioned ASSERT.

CONFIG_IOMMU_FORCE_PT_SHARE is always selected for ARM builds and, because clear_iommu_hap_pt_share() is called by the common iommu_setup() function if the IOMMU is not enabled, it is no longer safe to disable the IOMMU on ARM systems.

However, running with the IOMMU disabled is a valid configuration for ARM systems, so this patch rectifies the problem by removing the call to clear_iommu_hap_pt_share() from common code. As a side effect of this, however, it becomes possible on x86 systems for iommu_enabled to be false but iommu_hap_pt_share to be true. Thus the code in sysctl.c needs to be changed to make sure that XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share is not erroneously advertised when the IOMMU has been disabled.

Quoted 6 lines
Quoted 2 lines

With one NIT below:

Could you, please, change to:

You can also add if really needed:

[with IOMMU disabled on Arm]

Signature · 9 lines
Regards,

Oleksandr Tyshchenko


_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
#3 lore raw
In reply to Julien Grall
Quoted 87 lines

-----Original Message-----
From: Julien Grall <[email protected]>
Sent: 26 September 2019 12:24
To: Paul Durrant <[email protected]>; [email protected]
Cc: Oleksandr <[email protected]>; Andrew Cooper <[email protected]>; George Dunlap <[email protected]>; Ian Jackson <[email protected]>; Jan Beulich <[email protected]>; Konrad Rzeszutek Wilk <[email protected]>; Stefano Stabellini <[email protected]>; Tim (Xen.org) <[email protected]>; Wei Liu <[email protected]>
Subject: Re: [PATCH] iommu: avoid triggering ASSERT_UNREACHABLE() on ARM...

Hi Paul,

Quoted 21 lines · Paul Durrant

...when the IOMMU is not enabled.

80ff3d338dc9 "iommu: tidy up iommu_use_hap_pt() and need_iommu_pt_sync() macros" introduced CONFIG_IOMMU_FORCE_PT_SHARE, which causes the global 'iommu_hap_pt_share' option to be replaced with a #define-d value of true. In this configuration, calling clear_iommu_hap_pt_share() will result trigger the aforementioned ASSERT.

CONFIG_IOMMU_FORCE_PT_SHARE is always selected for ARM builds and, because clear_iommu_hap_pt_share() is called by the common iommu_setup() function if the IOMMU is not enabled, it is no longer safe to disable the IOMMU on ARM systems.

However, running with the IOMMU disabled is a valid configuration for ARM systems, so this patch rectifies the problem by removing the call to clear_iommu_hap_pt_share() from common code. As a side effect of this, however, it becomes possible on x86 systems for iommu_enabled to be false but iommu_hap_pt_share to be true. Thus the code in sysctl.c needs to be changed to make sure that XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share is not erroneously advertised when the IOMMU has been disabled.

Quoted 2 lines

With one NIT below:

Quoted 38 lines

---
Cc: Andrew Cooper <[email protected]>
Cc: George Dunlap <[email protected]>
Cc: Ian Jackson <[email protected]>
Cc: Jan Beulich <[email protected]>
Cc: Julien Grall <[email protected]>
Cc: Konrad Rzeszutek Wilk <[email protected]>
Cc: Stefano Stabellini <[email protected]>
Cc: Tim Deegan <[email protected]>
Cc: Wei Liu <[email protected]>

---
  xen/common/sysctl.c             | 6 ++++--
  xen/drivers/passthrough/iommu.c | 1 -
  2 files changed, 4 insertions(+), 3 deletions(-)
xen/common/sysctl.c +4 -2
index e8763c7fdf..f88a285e7f 100644
@@ -268,9 +268,11 @@ long do_sysctl(XEN_GUEST_HANDLE_PARAM(xen_sysctl_t) u_sysctl)
          pi->max_mfn = get_upper_mfn_bound();
          arch_do_physinfo(pi);
          if ( iommu_enabled )
+        {
              pi->capabilities |= XEN_SYSCTL_PHYSCAP_directio;
-        if ( iommu_hap_pt_share )
-            pi->capabilities |= XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share;
+            if ( iommu_hap_pt_share )
+                pi->capabilities |= XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share;
+        }
 
          if ( copy_to_guest(u_sysctl, op, 1) )
              ret = -EFAULT;
xen/drivers/passthrough/iommu.c +0 -0
index e8fddc2dc7..c33ca70ec9 100644

xen/drivers/passthrough/iommu.c int __init iommu_setup(void) · line 459

Quoted 5 lines
xen/drivers/passthrough/iommu.c +0 -1
@@ -456,7 +456,6 @@ int __init iommu_setup(void)
      if ( !iommu_enabled )
      {
          iommu_intremap = 0;
-        clear_iommu_hap_pt_share();
      }

NIT: The {} can now be removed.

Quoted 1 line

I can fix it on commit.

Good point. Thanks,

Paul

Quoted 5 lines

Cheers,

--
Julien Grall

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

Reviewed-by on v1 1/1
#4 lore raw
In reply to Paul Durrant
Quoted 21 lines · Paul Durrant

...when the IOMMU is not enabled.

80ff3d338dc9 "iommu: tidy up iommu_use_hap_pt() and need_iommu_pt_sync() macros" introduced CONFIG_IOMMU_FORCE_PT_SHARE, which causes the global 'iommu_hap_pt_share' option to be replaced with a #define-d value of true. In this configuration, calling clear_iommu_hap_pt_share() will result trigger the aforementioned ASSERT.

CONFIG_IOMMU_FORCE_PT_SHARE is always selected for ARM builds and, because clear_iommu_hap_pt_share() is called by the common iommu_setup() function if the IOMMU is not enabled, it is no longer safe to disable the IOMMU on ARM systems.

However, running with the IOMMU disabled is a valid configuration for ARM systems, so this patch rectifies the problem by removing the call to clear_iommu_hap_pt_share() from common code. As a side effect of this, however, it becomes possible on x86 systems for iommu_enabled to be false but iommu_hap_pt_share to be true. Thus the code in sysctl.c needs to be changed to make sure that XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share is not erroneously advertised when the IOMMU has been disabled.

Quoted 2 lines

Preferably with the adjustments mantioned elsewhere

Jan

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

#5 lore raw

Hi Jan,

Quoted 22 lines · Jan Beulich
Quoted 21 lines · Paul Durrant

...when the IOMMU is not enabled.

80ff3d338dc9 "iommu: tidy up iommu_use_hap_pt() and need_iommu_pt_sync() macros" introduced CONFIG_IOMMU_FORCE_PT_SHARE, which causes the global 'iommu_hap_pt_share' option to be replaced with a #define-d value of true. In this configuration, calling clear_iommu_hap_pt_share() will result trigger the aforementioned ASSERT.

CONFIG_IOMMU_FORCE_PT_SHARE is always selected for ARM builds and, because clear_iommu_hap_pt_share() is called by the common iommu_setup() function if the IOMMU is not enabled, it is no longer safe to disable the IOMMU on ARM systems.

However, running with the IOMMU disabled is a valid configuration for ARM systems, so this patch rectifies the problem by removing the call to clear_iommu_hap_pt_share() from common code. As a side effect of this, however, it becomes possible on x86 systems for iommu_enabled to be false but iommu_hap_pt_share to be true. Thus the code in sysctl.c needs to be changed to make sure that XEN_SYSCTL_PHYSCAP_iommu_hap_pt_share is not erroneously advertised when the IOMMU has been disabled.

Quoted 5 lines
Quoted 2 lines

Preferably with the adjustments mantioned elsewhere

I have done it while committing the patch.

Cheers,

Signature · 6 lines
Julien Grall

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