[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Xen-devel] RE: [Xen-changelog] [xen-unstable]vt-d: Fixpanic in msi_msg_read_remap_rte with acpi=off



Hi Miroslav and Keir,
When acpi=off and we set iommu_enabled=0, the execution of xen can't reach the 
acpi_find_matched_drhd_unit Miroslav explicitly points out one by one.
I think I can give a cleaner fix. Please see the attached patch.

Thanks,
-- Dexuan



-----Original Message-----
From: Miroslav Rezanina [mailto:mrezanin@xxxxxxxxxx] 
Sent: 2009å10æ16æ 19:16
To: Cui, Dexuan
Cc: Keir Fraser; xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: Re: [Xen-changelog] [xen-unstable]vt-d: Fixpanic in 
msi_msg_read_remap_rte with acpi=off

Xen kernel panics when "acpi=off" is set. Problem is caused by dereferncing 
NULL pointer in drhd after calling acpi_find_matched_drhd_unit. As 
acpi_find_matched_drhd_unit can return NULL, checks has to be done before 
returned value is used. 

----- "Dexuan Cui" <dexuan.cui@xxxxxxxxx> wrote:
> But when I use c/s 20330 and use "iommu=1 acpi=off", I still get the
> the panic:
> 
> (XEN) Xen call trace:
> (XEN)    [<ffff82c4801401b5>] ats_device+0x53/0x10e
> (XEN)    [<ffff82c48013c8e5>] intel_iommu_domain_init+0xa8/0x200
> (XEN)    [<ffff82c4801379dd>] iommu_domain_init+0x74/0x76
> (XEN)    [<ffff82c48014c437>] arch_domain_create+0x544/0x94a
> (XEN)    [<ffff82c48010676a>] domain_create+0x26c/0x3fd
> (XEN)    [<ffff82c48024cdb5>] __start_xen+0x5264/0x557f
> 
> Actually when iommu=1 and acpi=off, in __start_xen() ->
> acpi_boot_init(), acpi_dmar_init() can't be invoked at all and hence
> parse_dmar_table() can't be invoked, as a result, we should not try to
> use VT-d at all, however, the global variable iommu_enabled is left
> set to 1.
> 
> So I don't think this changeset is the right fix. I think the correct
> one should be: in disable_acpi(), we force iommu_enabled to 0.
To prevent this problem, disable iommu if acpi is disabled.

Patch
---
diff -r 97684ba1303e xen/drivers/passthrough/iommu.c
--- a/xen/drivers/passthrough/iommu.c   Thu Oct 01 18:39:03 2009 +0100
+++ b/xen/drivers/passthrough/iommu.c   Fri Oct 16 13:08:36 2009 +0200
@@ -39,6 +39,7 @@
  *   no-intremap                Disable VT-d Interrupt Remapping
  */
 custom_param("iommu", parse_iommu_param);
+extern int acpi_disabled;
 int iommu_enabled = 0;
 int iommu_pv_enabled = 0;
 int force_iommu = 0;
@@ -266,6 +267,11 @@
     if ( !iommu_enabled )
         goto out;
 
+    if ( !acpi_disabled ) {
+       iommu_enabled = 0;
+       goto out;
+    }
+
     rc = iommu_hardware_setup();
 
     iommu_enabled = (rc == 0);
diff -r 97684ba1303e xen/drivers/passthrough/vtd/intremap.c
--- a/xen/drivers/passthrough/vtd/intremap.c    Thu Oct 01 18:39:03 2009 +0100
+++ b/xen/drivers/passthrough/vtd/intremap.c    Fri Oct 16 13:08:36 2009 +0200
@@ -461,6 +461,8 @@
     struct ir_ctrl *ir_ctrl;
 
     drhd = acpi_find_matched_drhd_unit(pdev);
+    if (!drhd)
+       return;
     iommu = drhd->iommu;
 
     ir_ctrl = iommu_ir_ctrl(iommu);
@@ -479,6 +481,8 @@
     struct ir_ctrl *ir_ctrl;
 
     drhd = acpi_find_matched_drhd_unit(pdev);
+    if (!drhd)
+       return;
     iommu = drhd->iommu;
 
     ir_ctrl = iommu_ir_ctrl(iommu);
diff -r 97684ba1303e xen/drivers/passthrough/vtd/iommu.c
--- a/xen/drivers/passthrough/vtd/iommu.c       Thu Oct 01 18:39:03 2009 +0100
+++ b/xen/drivers/passthrough/vtd/iommu.c       Fri Oct 16 13:08:36 2009 +0200
@@ -1412,6 +1412,8 @@
         return -ENODEV;
 
     drhd = acpi_find_matched_drhd_unit(pdev);
+    if (!drhd)
+       return -ENODEV;
     pdev_iommu = drhd->iommu;
     domain_context_unmap(source, bus, devfn);
 
@@ -1425,7 +1427,7 @@
     for_each_pdev ( source, pdev )
     {
         drhd = acpi_find_matched_drhd_unit(pdev);
-        if ( drhd->iommu == pdev_iommu )
+        if ( drhd && drhd->iommu == pdev_iommu )
         {
             found = 1;
             break;
---
Miroslav Rezanina
Software Engineer - Virtualization Team - XEN kernel

Attachment: disable_vtd_when_acpi_is_off.patch
Description: disable_vtd_when_acpi_is_off.patch

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.