mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* [PATCH] iommu/vt-d: Fix oops due to out of scope access
@ 2026-04-21  3:13 Zhenzhong Duan
  2026-04-21  8:04 ` Tian, Kevin
  2026-05-08  7:58 ` Baolu Lu
  0 siblings, 2 replies; 4+ messages in thread
From: Zhenzhong Duan @ 2026-04-21  3:13 UTC (permalink / raw)
  To: iommu, linux-kernel
  Cc: dwmw2, baolu.lu, joro, will, robin.murphy, kevin.tian,
	Zhenzhong Duan, Jacob Pan, Joerg Roedel

Below oops triggers when kill QEMU process:

  Oops: general protection fault, probably for non-canonical address 0x7fffffff844eaaa7: 0000 [#1] SMP NOPTI
  Call Trace:
   <TASK>
   do_raw_spin_lock+0xaa/0xc0
   _raw_spin_lock_irqsave+0x21/0x40
   domain_remove_dev_pasid+0x52/0x160
   intel_nested_set_dev_pasid+0x1b9/0x1e0
   __iommu_set_group_pasid+0x56/0x120
   pci_dev_reset_iommu_done+0xe3/0x180
   pcie_flr+0x65/0x160
   __pci_reset_function_locked+0x5b/0x120
   vfio_pci_core_close_device+0x63/0xe0 [vfio_pci_core]
   vfio_df_close+0x4f/0xa0
   vfio_df_unbind_iommufd+0x2d/0x60
   vfio_device_fops_release+0x3e/0x40
   __fput+0xe5/0x2c0
   task_work_run+0x58/0xa0
   do_exit+0x2c8/0x600
   do_group_exit+0x2f/0xa0
   get_signal+0x863/0x8c0
   arch_do_signal_or_restart+0x24/0x100
   exit_to_user_mode_loop+0x87/0x380
   do_syscall_64+0x2ff/0x11e0
   entry_SYSCALL_64_after_hwframe+0x76/0x7e

The global static blocked domain is a dummy domain without corresponding
dmar_domain structure, accessing beyond iommu_domain structure triggers
oops easily. Fix it by return early in domain_remove_dev_pasid() like
identity domain.

Fixes: 7d0c9da6c150 ("iommu/vt-d: Add set_dev_pasid callback for dma domain")
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
 drivers/iommu/intel/iommu.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
index c3d18cd77d2f..52aa12dbeea1 100644
--- a/drivers/iommu/intel/iommu.c
+++ b/drivers/iommu/intel/iommu.c
@@ -3530,8 +3530,8 @@ void domain_remove_dev_pasid(struct iommu_domain *domain,
 	if (!domain)
 		return;
 
-	/* Identity domain has no meta data for pasid. */
-	if (domain->type == IOMMU_DOMAIN_IDENTITY)
+	/* Identity domain and blocked domain have no meta data for pasid. */
+	if (domain->type == IOMMU_DOMAIN_IDENTITY || domain->type == IOMMU_DOMAIN_BLOCKED)
 		return;
 
 	dmar_domain = to_dmar_domain(domain);
-- 
2.47.3


^ permalink raw reply	[flat|nested] 4+ messages in thread

* RE: [PATCH] iommu/vt-d: Fix oops due to out of scope access
  2026-04-21  3:13 [PATCH] iommu/vt-d: Fix oops due to out of scope access Zhenzhong Duan
@ 2026-04-21  8:04 ` Tian, Kevin
  2026-04-21  9:44   ` Duan, Zhenzhong
  2026-05-08  7:58 ` Baolu Lu
  1 sibling, 1 reply; 4+ messages in thread
From: Tian, Kevin @ 2026-04-21  8:04 UTC (permalink / raw)
  To: Duan, Zhenzhong, iommu, linux-kernel
  Cc: dwmw2, baolu.lu, joro, will, robin.murphy, Jacob Pan, Joerg Roedel

> From: Duan, Zhenzhong <zhenzhong.duan@intel.com>
> Sent: Tuesday, April 21, 2026 11:14 AM
> 
> Below oops triggers when kill QEMU process:
> 
>   Oops: general protection fault, probably for non-canonical address
> 0x7fffffff844eaaa7: 0000 [#1] SMP NOPTI
>   Call Trace:
>    <TASK>
>    do_raw_spin_lock+0xaa/0xc0
>    _raw_spin_lock_irqsave+0x21/0x40
>    domain_remove_dev_pasid+0x52/0x160
>    intel_nested_set_dev_pasid+0x1b9/0x1e0
>    __iommu_set_group_pasid+0x56/0x120
>    pci_dev_reset_iommu_done+0xe3/0x180
>    pcie_flr+0x65/0x160
>    __pci_reset_function_locked+0x5b/0x120
>    vfio_pci_core_close_device+0x63/0xe0 [vfio_pci_core]
>    vfio_df_close+0x4f/0xa0
>    vfio_df_unbind_iommufd+0x2d/0x60
>    vfio_device_fops_release+0x3e/0x40
>    __fput+0xe5/0x2c0
>    task_work_run+0x58/0xa0
>    do_exit+0x2c8/0x600
>    do_group_exit+0x2f/0xa0
>    get_signal+0x863/0x8c0
>    arch_do_signal_or_restart+0x24/0x100
>    exit_to_user_mode_loop+0x87/0x380
>    do_syscall_64+0x2ff/0x11e0
>    entry_SYSCALL_64_after_hwframe+0x76/0x7e
> 
> The global static blocked domain is a dummy domain without corresponding
> dmar_domain structure, accessing beyond iommu_domain structure triggers
> oops easily. Fix it by return early in domain_remove_dev_pasid() like
> identity domain.

It's interesting why this doesn't hurt us in normal attach/detach paths...

> 
> Fixes: 7d0c9da6c150 ("iommu/vt-d: Add set_dev_pasid callback for dma
> domain")
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>

Reviewed-by: Kevin Tian <kevin.tian@intel.com>

btw Sashiko captured another issue in existing code. Could you help fix it too?

https://sashiko.dev/#/patchset/20260421031347.1408890-1-zhenzhong.duan%40intel.com

^ permalink raw reply	[flat|nested] 4+ messages in thread

* RE: [PATCH] iommu/vt-d: Fix oops due to out of scope access
  2026-04-21  8:04 ` Tian, Kevin
@ 2026-04-21  9:44   ` Duan, Zhenzhong
  0 siblings, 0 replies; 4+ messages in thread
From: Duan, Zhenzhong @ 2026-04-21  9:44 UTC (permalink / raw)
  To: Tian, Kevin, iommu, linux-kernel
  Cc: dwmw2, baolu.lu, joro, will, robin.murphy, Jacob Pan, Joerg Roedel



>-----Original Message-----
>From: Tian, Kevin <kevin.tian@intel.com>
>Subject: RE: [PATCH] iommu/vt-d: Fix oops due to out of scope access
>
>> From: Duan, Zhenzhong <zhenzhong.duan@intel.com>
>> Sent: Tuesday, April 21, 2026 11:14 AM
>>
>> Below oops triggers when kill QEMU process:
>>
>>   Oops: general protection fault, probably for non-canonical address
>> 0x7fffffff844eaaa7: 0000 [#1] SMP NOPTI
>>   Call Trace:
>>    <TASK>
>>    do_raw_spin_lock+0xaa/0xc0
>>    _raw_spin_lock_irqsave+0x21/0x40
>>    domain_remove_dev_pasid+0x52/0x160
>>    intel_nested_set_dev_pasid+0x1b9/0x1e0
>>    __iommu_set_group_pasid+0x56/0x120
>>    pci_dev_reset_iommu_done+0xe3/0x180
>>    pcie_flr+0x65/0x160
>>    __pci_reset_function_locked+0x5b/0x120
>>    vfio_pci_core_close_device+0x63/0xe0 [vfio_pci_core]
>>    vfio_df_close+0x4f/0xa0
>>    vfio_df_unbind_iommufd+0x2d/0x60
>>    vfio_device_fops_release+0x3e/0x40
>>    __fput+0xe5/0x2c0
>>    task_work_run+0x58/0xa0
>>    do_exit+0x2c8/0x600
>>    do_group_exit+0x2f/0xa0
>>    get_signal+0x863/0x8c0
>>    arch_do_signal_or_restart+0x24/0x100
>>    exit_to_user_mode_loop+0x87/0x380
>>    do_syscall_64+0x2ff/0x11e0
>>    entry_SYSCALL_64_after_hwframe+0x76/0x7e
>>
>> The global static blocked domain is a dummy domain without corresponding
>> dmar_domain structure, accessing beyond iommu_domain structure triggers
>> oops easily. Fix it by return early in domain_remove_dev_pasid() like
>> identity domain.
>
>It's interesting why this doesn't hurt us in normal attach/detach paths...

IIUC, other call path never pass in blocked domain as old domain.

>
>>
>> Fixes: 7d0c9da6c150 ("iommu/vt-d: Add set_dev_pasid callback for dma
>> domain")
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>
>Reviewed-by: Kevin Tian <kevin.tian@intel.com>
>
>btw Sashiko captured another issue in existing code. Could you help fix it too?

Will do.

Thanks
Zhenzhong

>
>https://sashiko.dev/#/patchset/20260421031347.1408890-1-
>zhenzhong.duan%40intel.com

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] iommu/vt-d: Fix oops due to out of scope access
  2026-04-21  3:13 [PATCH] iommu/vt-d: Fix oops due to out of scope access Zhenzhong Duan
  2026-04-21  8:04 ` Tian, Kevin
@ 2026-05-08  7:58 ` Baolu Lu
  1 sibling, 0 replies; 4+ messages in thread
From: Baolu Lu @ 2026-05-08  7:58 UTC (permalink / raw)
  To: Zhenzhong Duan, iommu, linux-kernel
  Cc: dwmw2, joro, will, robin.murphy, kevin.tian, Jacob Pan, Joerg Roedel

On 4/21/26 11:13, Zhenzhong Duan wrote:
> Below oops triggers when kill QEMU process:
> 
>    Oops: general protection fault, probably for non-canonical address 0x7fffffff844eaaa7: 0000 [#1] SMP NOPTI
>    Call Trace:
>     <TASK>
>     do_raw_spin_lock+0xaa/0xc0
>     _raw_spin_lock_irqsave+0x21/0x40
>     domain_remove_dev_pasid+0x52/0x160
>     intel_nested_set_dev_pasid+0x1b9/0x1e0
>     __iommu_set_group_pasid+0x56/0x120
>     pci_dev_reset_iommu_done+0xe3/0x180
>     pcie_flr+0x65/0x160
>     __pci_reset_function_locked+0x5b/0x120
>     vfio_pci_core_close_device+0x63/0xe0 [vfio_pci_core]
>     vfio_df_close+0x4f/0xa0
>     vfio_df_unbind_iommufd+0x2d/0x60
>     vfio_device_fops_release+0x3e/0x40
>     __fput+0xe5/0x2c0
>     task_work_run+0x58/0xa0
>     do_exit+0x2c8/0x600
>     do_group_exit+0x2f/0xa0
>     get_signal+0x863/0x8c0
>     arch_do_signal_or_restart+0x24/0x100
>     exit_to_user_mode_loop+0x87/0x380
>     do_syscall_64+0x2ff/0x11e0
>     entry_SYSCALL_64_after_hwframe+0x76/0x7e
> 
> The global static blocked domain is a dummy domain without corresponding
> dmar_domain structure, accessing beyond iommu_domain structure triggers
> oops easily. Fix it by return early in domain_remove_dev_pasid() like
> identity domain.
> 
> Fixes: 7d0c9da6c150 ("iommu/vt-d: Add set_dev_pasid callback for dma domain")
> Signed-off-by: Zhenzhong Duan<zhenzhong.duan@intel.com>
> ---
>   drivers/iommu/intel/iommu.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)

Queued for v7.1-rc. Thanks!

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2026-05-08  8:00 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-04-21  3:13 [PATCH] iommu/vt-d: Fix oops due to out of scope access Zhenzhong Duan
2026-04-21  8:04 ` Tian, Kevin
2026-04-21  9:44   ` Duan, Zhenzhong
2026-05-08  7:58 ` Baolu Lu

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

Powered by JetHome