diff mbox series

powerpc/iommu: Fix the missing iommu_group_put() during platform domain attach

Message ID 170784021983.6249.10039296655906636112.stgit@linux.ibm.com
State New
Headers
Series powerpc/iommu: Fix the missing iommu_group_put() during platform domain attach |

Commit Message

Shivaprasad G Bhat Feb. 13, 2024, 4:05 p.m. UTC 
  The function spapr_tce_platform_iommu_attach_dev() is missing to call
iommu_group_put() when the domain is already set. This refcount leak
shows up with BUG_ON() during DLPAR remove operation as,

  KernelBug: Kernel bug in state 'None': kernel BUG at arch/powerpc/platforms/pseries/iommu.c:100!
  Oops: Exception in kernel mode, sig: 5 [#1]
  LE PAGE_SIZE=64K MMU=Radix SMP NR_CPUS=8192 NUMA pSeries
  <snip>
  Hardware name: IBM,9080-HEX POWER10 (raw) 0x800200 0xf000006 of:IBM,FW1060.00 (NH1060_016) hv:phyp pSeries
  NIP:  c0000000000ff4d4 LR: c0000000000ff4cc CTR: 0000000000000000
  REGS: c0000013aed5f840 TRAP: 0700   Tainted: G          I         (6.8.0-rc3-autotest-g99bd3cb0d12e)
  MSR:  8000000000029033 <SF,EE,ME,IR,DR,RI,LE>  CR: 44002402  XER: 20040000
  CFAR: c000000000a0d170 IRQMASK: 0
  GPR00: c0000000000ff4cc c0000013aed5fae0 c000000001512700 c0000013aa362138
  GPR04: 0000000000000000 0000000000000000 0000000000000000 0000000119c8afd0
  GPR08: 0000000000000000 c000001284442b00 0000000000000001 0000000000001003
  GPR12: 0000000300000000 c0000018ffff2f00 0000000000000000 0000000000000000
  GPR16: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
  GPR20: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
  GPR24: c0000013aed5fc40 0000000000000002 0000000000000000 c000000002757d90
  GPR28: c0000000000ff440 c000000002757cb8 c00000183799c1a0 c0000013aa362b00
  NIP [c0000000000ff4d4] iommu_reconfig_notifier+0x94/0x200
  LR [c0000000000ff4cc] iommu_reconfig_notifier+0x8c/0x200
  Call Trace:
  [c0000013aed5fae0] [c0000000000ff4cc] iommu_reconfig_notifier+0x8c/0x200 (unreliable)
  [c0000013aed5fb10] [c0000000001a27b0] notifier_call_chain+0xb8/0x19c
  [c0000013aed5fb70] [c0000000001a2a78] blocking_notifier_call_chain+0x64/0x98
  [c0000013aed5fbb0] [c000000000c4a898] of_reconfig_notify+0x44/0xdc
  [c0000013aed5fc20] [c000000000c4add4] of_detach_node+0x78/0xb0
  [c0000013aed5fc70] [c0000000000f96a8] ofdt_write.part.0+0x86c/0xbb8
  [c0000013aed5fce0] [c00000000069b4bc] proc_reg_write+0xf4/0x150
  [c0000013aed5fd10] [c0000000005bfeb4] vfs_write+0xf8/0x488
  [c0000013aed5fdc0] [c0000000005c0570] ksys_write+0x84/0x140
  [c0000013aed5fe10] [c000000000033358] system_call_exception+0x138/0x330
  [c0000013aed5fe50] [c00000000000d05c] system_call_vectored_common+0x15c/0x2ec
  --- interrupt: 3000 at 0x20000433acb4
  <snip>
  ---[ end trace 0000000000000000 ]---

The patch adds the missing iommu_group_put() call.

Fixes: a8ca9fc9134c ("powerpc/iommu: Do not do platform domain attach atctions after probe")
Signed-off-by: Shivaprasad G Bhat <sbhat@linux.ibm.com>
---
 arch/powerpc/kernel/iommu.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Comments

Jason Gunthorpe Feb. 13, 2024, 5:21 p.m. UTC | #1 
On Tue, Feb 13, 2024 at 10:05:22AM -0600, Shivaprasad G Bhat wrote:
> The function spapr_tce_platform_iommu_attach_dev() is missing to call
> iommu_group_put() when the domain is already set. This refcount leak
> shows up with BUG_ON() during DLPAR remove operation as,
> 
>   KernelBug: Kernel bug in state 'None': kernel BUG at arch/powerpc/platforms/pseries/iommu.c:100!
>   Oops: Exception in kernel mode, sig: 5 [#1]
>   LE PAGE_SIZE=64K MMU=Radix SMP NR_CPUS=8192 NUMA pSeries
>   <snip>
>   Hardware name: IBM,9080-HEX POWER10 (raw) 0x800200 0xf000006 of:IBM,FW1060.00 (NH1060_016) hv:phyp pSeries
>   NIP:  c0000000000ff4d4 LR: c0000000000ff4cc CTR: 0000000000000000
>   REGS: c0000013aed5f840 TRAP: 0700   Tainted: G          I         (6.8.0-rc3-autotest-g99bd3cb0d12e)
>   MSR:  8000000000029033 <SF,EE,ME,IR,DR,RI,LE>  CR: 44002402  XER: 20040000
>   CFAR: c000000000a0d170 IRQMASK: 0
>   GPR00: c0000000000ff4cc c0000013aed5fae0 c000000001512700 c0000013aa362138
>   GPR04: 0000000000000000 0000000000000000 0000000000000000 0000000119c8afd0
>   GPR08: 0000000000000000 c000001284442b00 0000000000000001 0000000000001003
>   GPR12: 0000000300000000 c0000018ffff2f00 0000000000000000 0000000000000000
>   GPR16: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
>   GPR20: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
>   GPR24: c0000013aed5fc40 0000000000000002 0000000000000000 c000000002757d90
>   GPR28: c0000000000ff440 c000000002757cb8 c00000183799c1a0 c0000013aa362b00
>   NIP [c0000000000ff4d4] iommu_reconfig_notifier+0x94/0x200
>   LR [c0000000000ff4cc] iommu_reconfig_notifier+0x8c/0x200
>   Call Trace:
>   [c0000013aed5fae0] [c0000000000ff4cc] iommu_reconfig_notifier+0x8c/0x200 (unreliable)
>   [c0000013aed5fb10] [c0000000001a27b0] notifier_call_chain+0xb8/0x19c
>   [c0000013aed5fb70] [c0000000001a2a78] blocking_notifier_call_chain+0x64/0x98
>   [c0000013aed5fbb0] [c000000000c4a898] of_reconfig_notify+0x44/0xdc
>   [c0000013aed5fc20] [c000000000c4add4] of_detach_node+0x78/0xb0
>   [c0000013aed5fc70] [c0000000000f96a8] ofdt_write.part.0+0x86c/0xbb8
>   [c0000013aed5fce0] [c00000000069b4bc] proc_reg_write+0xf4/0x150
>   [c0000013aed5fd10] [c0000000005bfeb4] vfs_write+0xf8/0x488
>   [c0000013aed5fdc0] [c0000000005c0570] ksys_write+0x84/0x140
>   [c0000013aed5fe10] [c000000000033358] system_call_exception+0x138/0x330
>   [c0000013aed5fe50] [c00000000000d05c] system_call_vectored_common+0x15c/0x2ec
>   --- interrupt: 3000 at 0x20000433acb4
>   <snip>
>   ---[ end trace 0000000000000000 ]---
> 
> The patch adds the missing iommu_group_put() call.
> 
> Fixes: a8ca9fc9134c ("powerpc/iommu: Do not do platform domain attach atctions after probe")
> Signed-off-by: Shivaprasad G Bhat <sbhat@linux.ibm.com>
> ---
>  arch/powerpc/kernel/iommu.c |    4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)

Doh, that is a weird splat for this but thanks for finding it

Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>

Jason
Venkat Rao Bagalkote Feb. 14, 2024, 5:34 a.m. UTC | #2 
Thanks for the patch. Applied this patch and verified and issue is fixed.

This issue way originally reported in the below mail.

https://marc.info/?l=linux-kernel&m=170737160630106&w=2


Tested-by: Venkat Rao Bagalkote <venkat88@linux.vnet.ibm.com>

On 13/02/24 10:51 pm, Jason Gunthorpe wrote:
> On Tue, Feb 13, 2024 at 10:05:22AM -0600, Shivaprasad G Bhat wrote:
>> The function spapr_tce_platform_iommu_attach_dev() is missing to call
>> iommu_group_put() when the domain is already set. This refcount leak
>> shows up with BUG_ON() during DLPAR remove operation as,
>>
>>    KernelBug: Kernel bug in state 'None': kernel BUG at arch/powerpc/platforms/pseries/iommu.c:100!
>>    Oops: Exception in kernel mode, sig: 5 [#1]
>>    LE PAGE_SIZE=64K MMU=Radix SMP NR_CPUS=8192 NUMA pSeries
>>    <snip>
>>    Hardware name: IBM,9080-HEX POWER10 (raw) 0x800200 0xf000006 of:IBM,FW1060.00 (NH1060_016) hv:phyp pSeries
>>    NIP:  c0000000000ff4d4 LR: c0000000000ff4cc CTR: 0000000000000000
>>    REGS: c0000013aed5f840 TRAP: 0700   Tainted: G          I         (6.8.0-rc3-autotest-g99bd3cb0d12e)
>>    MSR:  8000000000029033 <SF,EE,ME,IR,DR,RI,LE>  CR: 44002402  XER: 20040000
>>    CFAR: c000000000a0d170 IRQMASK: 0
>>    GPR00: c0000000000ff4cc c0000013aed5fae0 c000000001512700 c0000013aa362138
>>    GPR04: 0000000000000000 0000000000000000 0000000000000000 0000000119c8afd0
>>    GPR08: 0000000000000000 c000001284442b00 0000000000000001 0000000000001003
>>    GPR12: 0000000300000000 c0000018ffff2f00 0000000000000000 0000000000000000
>>    GPR16: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
>>    GPR20: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
>>    GPR24: c0000013aed5fc40 0000000000000002 0000000000000000 c000000002757d90
>>    GPR28: c0000000000ff440 c000000002757cb8 c00000183799c1a0 c0000013aa362b00
>>    NIP [c0000000000ff4d4] iommu_reconfig_notifier+0x94/0x200
>>    LR [c0000000000ff4cc] iommu_reconfig_notifier+0x8c/0x200
>>    Call Trace:
>>    [c0000013aed5fae0] [c0000000000ff4cc] iommu_reconfig_notifier+0x8c/0x200 (unreliable)
>>    [c0000013aed5fb10] [c0000000001a27b0] notifier_call_chain+0xb8/0x19c
>>    [c0000013aed5fb70] [c0000000001a2a78] blocking_notifier_call_chain+0x64/0x98
>>    [c0000013aed5fbb0] [c000000000c4a898] of_reconfig_notify+0x44/0xdc
>>    [c0000013aed5fc20] [c000000000c4add4] of_detach_node+0x78/0xb0
>>    [c0000013aed5fc70] [c0000000000f96a8] ofdt_write.part.0+0x86c/0xbb8
>>    [c0000013aed5fce0] [c00000000069b4bc] proc_reg_write+0xf4/0x150
>>    [c0000013aed5fd10] [c0000000005bfeb4] vfs_write+0xf8/0x488
>>    [c0000013aed5fdc0] [c0000000005c0570] ksys_write+0x84/0x140
>>    [c0000013aed5fe10] [c000000000033358] system_call_exception+0x138/0x330
>>    [c0000013aed5fe50] [c00000000000d05c] system_call_vectored_common+0x15c/0x2ec
>>    --- interrupt: 3000 at 0x20000433acb4
>>    <snip>
>>    ---[ end trace 0000000000000000 ]---
>>
>> The patch adds the missing iommu_group_put() call.
>>
>> Fixes: a8ca9fc9134c ("powerpc/iommu: Do not do platform domain attach atctions after probe")
>> Signed-off-by: Shivaprasad G Bhat <sbhat@linux.ibm.com>
>> ---
>>   arch/powerpc/kernel/iommu.c |    4 +++-
>>   1 file changed, 3 insertions(+), 1 deletion(-)
> Doh, that is a weird splat for this but thanks for finding it
>
> Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
>
> Jason
>
Michael Ellerman Feb. 14, 2024, 12:53 p.m. UTC | #3 
Venkat Rao Bagalkote <venkat88@linux.vnet.ibm.com> writes:
> Thanks for the patch. Applied this patch and verified and issue is fixed.
>
> This issue way originally reported in the below mail.
>
> https://marc.info/?l=linux-kernel&m=170737160630106&w=2

Please use lore for links, in this case:

https://lore.kernel.org/all/274e0d2b-b5cc-475e-94e6-8427e88e271d@linux.vnet.ibm.com/

cheers

> On 13/02/24 10:51 pm, Jason Gunthorpe wrote:
>> On Tue, Feb 13, 2024 at 10:05:22AM -0600, Shivaprasad G Bhat wrote:
>>> The function spapr_tce_platform_iommu_attach_dev() is missing to call
>>> iommu_group_put() when the domain is already set. This refcount leak
>>> shows up with BUG_ON() during DLPAR remove operation as,
>>>
>>>    KernelBug: Kernel bug in state 'None': kernel BUG at arch/powerpc/platforms/pseries/iommu.c:100!
>>>    Oops: Exception in kernel mode, sig: 5 [#1]
>>>    LE PAGE_SIZE=64K MMU=Radix SMP NR_CPUS=8192 NUMA pSeries
>>>    <snip>
>>>    Hardware name: IBM,9080-HEX POWER10 (raw) 0x800200 0xf000006 of:IBM,FW1060.00 (NH1060_016) hv:phyp pSeries
>>>    NIP:  c0000000000ff4d4 LR: c0000000000ff4cc CTR: 0000000000000000
>>>    REGS: c0000013aed5f840 TRAP: 0700   Tainted: G          I         (6.8.0-rc3-autotest-g99bd3cb0d12e)
>>>    MSR:  8000000000029033 <SF,EE,ME,IR,DR,RI,LE>  CR: 44002402  XER: 20040000
>>>    CFAR: c000000000a0d170 IRQMASK: 0
>>>    GPR00: c0000000000ff4cc c0000013aed5fae0 c000000001512700 c0000013aa362138
>>>    GPR04: 0000000000000000 0000000000000000 0000000000000000 0000000119c8afd0
>>>    GPR08: 0000000000000000 c000001284442b00 0000000000000001 0000000000001003
>>>    GPR12: 0000000300000000 c0000018ffff2f00 0000000000000000 0000000000000000
>>>    GPR16: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
>>>    GPR20: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
>>>    GPR24: c0000013aed5fc40 0000000000000002 0000000000000000 c000000002757d90
>>>    GPR28: c0000000000ff440 c000000002757cb8 c00000183799c1a0 c0000013aa362b00
>>>    NIP [c0000000000ff4d4] iommu_reconfig_notifier+0x94/0x200
>>>    LR [c0000000000ff4cc] iommu_reconfig_notifier+0x8c/0x200
>>>    Call Trace:
>>>    [c0000013aed5fae0] [c0000000000ff4cc] iommu_reconfig_notifier+0x8c/0x200 (unreliable)
>>>    [c0000013aed5fb10] [c0000000001a27b0] notifier_call_chain+0xb8/0x19c
>>>    [c0000013aed5fb70] [c0000000001a2a78] blocking_notifier_call_chain+0x64/0x98
>>>    [c0000013aed5fbb0] [c000000000c4a898] of_reconfig_notify+0x44/0xdc
>>>    [c0000013aed5fc20] [c000000000c4add4] of_detach_node+0x78/0xb0
>>>    [c0000013aed5fc70] [c0000000000f96a8] ofdt_write.part.0+0x86c/0xbb8
>>>    [c0000013aed5fce0] [c00000000069b4bc] proc_reg_write+0xf4/0x150
>>>    [c0000013aed5fd10] [c0000000005bfeb4] vfs_write+0xf8/0x488
>>>    [c0000013aed5fdc0] [c0000000005c0570] ksys_write+0x84/0x140
>>>    [c0000013aed5fe10] [c000000000033358] system_call_exception+0x138/0x330
>>>    [c0000013aed5fe50] [c00000000000d05c] system_call_vectored_common+0x15c/0x2ec
>>>    --- interrupt: 3000 at 0x20000433acb4
>>>    <snip>
>>>    ---[ end trace 0000000000000000 ]---
>>>
>>> The patch adds the missing iommu_group_put() call.
>>>
>>> Fixes: a8ca9fc9134c ("powerpc/iommu: Do not do platform domain attach atctions after probe")
>>> Signed-off-by: Shivaprasad G Bhat <sbhat@linux.ibm.com>
>>> ---
>>>   arch/powerpc/kernel/iommu.c |    4 +++-
>>>   1 file changed, 3 insertions(+), 1 deletion(-)
>> Doh, that is a weird splat for this but thanks for finding it
>>
>> Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
>>
>> Jason
>>
Jason Gunthorpe Feb. 14, 2024, 12:58 p.m. UTC | #4 
On Wed, Feb 14, 2024 at 11:53:20PM +1100, Michael Ellerman wrote:
> Venkat Rao Bagalkote <venkat88@linux.vnet.ibm.com> writes:
> > Thanks for the patch. Applied this patch and verified and issue is fixed.
> >
> > This issue way originally reported in the below mail.
> >
> > https://marc.info/?l=linux-kernel&m=170737160630106&w=2
> 
> Please use lore for links, in this case:
> 
> https://lore.kernel.org/all/274e0d2b-b5cc-475e-94e6-8427e88e271d@linux.vnet.ibm.com/

Also if you are respinning you may prefer this

@@ -1285,14 +1285,15 @@ spapr_tce_platform_iommu_attach_dev(struct iommu_domain *platform_domain,
                                    struct device *dev)
 {
        struct iommu_domain *domain = iommu_get_domain_for_dev(dev);
-       struct iommu_group *grp = iommu_group_get(dev);
        struct iommu_table_group *table_group;
+       struct iommu_group *grp;
        int ret = -EINVAL;
 
        /* At first attach the ownership is already set */
        if (!domain)
                return 0;
 
+       grp = iommu_group_get(dev);
        if (!grp)
                return -ENODEV;

Which is sort of why this happened in the first place :)

Jason
Shivaprasad G Bhat Feb. 14, 2024, 6:13 p.m. UTC | #5 
On 2/14/24 18:28, Jason Gunthorpe wrote:
> On Wed, Feb 14, 2024 at 11:53:20PM +1100, Michael Ellerman wrote:
>> Venkat Rao Bagalkote <venkat88@linux.vnet.ibm.com> writes:
>>> Thanks for the patch. Applied this patch and verified and issue is fixed.
>>>
>>> This issue way originally reported in the below mail.
>>>
>>> https://marc.info/?l=linux-kernel&m=170737160630106&w=2
>> Please use lore for links, in this case:
>>
>> https://lore.kernel.org/all/274e0d2b-b5cc-475e-94e6-8427e88e271d@linux.vnet.ibm.com/
> Also if you are respinning you may prefer this
>
> @@ -1285,14 +1285,15 @@ spapr_tce_platform_iommu_attach_dev(struct iommu_domain *platform_domain,
>                                      struct device *dev)
>   {
>          struct iommu_domain *domain = iommu_get_domain_for_dev(dev);
> -       struct iommu_group *grp = iommu_group_get(dev);
>          struct iommu_table_group *table_group;
> +       struct iommu_group *grp;
>          int ret = -EINVAL;
>   
>          /* At first attach the ownership is already set */
>          if (!domain)
>                  return 0;
>   
> +       grp = iommu_group_get(dev);
>          if (!grp)
>                  return -ENODEV;
>
> Which is sort of why this happened in the first place :)

Right! Posted the v2 here

https://lore.kernel.org/linux-iommu/170793401503.7491.9431631474642074097.stgit@linux.ibm.com/


Thanks,

Shivaprasad

> Jason
Michael Ellerman Feb. 15, 2024, 12:57 p.m. UTC | #6 
On Tue, 13 Feb 2024 10:05:22 -0600, Shivaprasad G Bhat wrote:
> The function spapr_tce_platform_iommu_attach_dev() is missing to call
> iommu_group_put() when the domain is already set. This refcount leak
> shows up with BUG_ON() during DLPAR remove operation as,
> 
>   KernelBug: Kernel bug in state 'None': kernel BUG at arch/powerpc/platforms/pseries/iommu.c:100!
>   Oops: Exception in kernel mode, sig: 5 [#1]
>   LE PAGE_SIZE=64K MMU=Radix SMP NR_CPUS=8192 NUMA pSeries
>   <snip>
>   Hardware name: IBM,9080-HEX POWER10 (raw) 0x800200 0xf000006 of:IBM,FW1060.00 (NH1060_016) hv:phyp pSeries
>   NIP:  c0000000000ff4d4 LR: c0000000000ff4cc CTR: 0000000000000000
>   REGS: c0000013aed5f840 TRAP: 0700   Tainted: G          I         (6.8.0-rc3-autotest-g99bd3cb0d12e)
>   MSR:  8000000000029033 <SF,EE,ME,IR,DR,RI,LE>  CR: 44002402  XER: 20040000
>   CFAR: c000000000a0d170 IRQMASK: 0
>   GPR00: c0000000000ff4cc c0000013aed5fae0 c000000001512700 c0000013aa362138
>   GPR04: 0000000000000000 0000000000000000 0000000000000000 0000000119c8afd0
>   GPR08: 0000000000000000 c000001284442b00 0000000000000001 0000000000001003
>   GPR12: 0000000300000000 c0000018ffff2f00 0000000000000000 0000000000000000
>   GPR16: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
>   GPR20: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
>   GPR24: c0000013aed5fc40 0000000000000002 0000000000000000 c000000002757d90
>   GPR28: c0000000000ff440 c000000002757cb8 c00000183799c1a0 c0000013aa362b00
>   NIP [c0000000000ff4d4] iommu_reconfig_notifier+0x94/0x200
>   LR [c0000000000ff4cc] iommu_reconfig_notifier+0x8c/0x200
>   Call Trace:
>   [c0000013aed5fae0] [c0000000000ff4cc] iommu_reconfig_notifier+0x8c/0x200 (unreliable)
>   [c0000013aed5fb10] [c0000000001a27b0] notifier_call_chain+0xb8/0x19c
>   [c0000013aed5fb70] [c0000000001a2a78] blocking_notifier_call_chain+0x64/0x98
>   [c0000013aed5fbb0] [c000000000c4a898] of_reconfig_notify+0x44/0xdc
>   [c0000013aed5fc20] [c000000000c4add4] of_detach_node+0x78/0xb0
>   [c0000013aed5fc70] [c0000000000f96a8] ofdt_write.part.0+0x86c/0xbb8
>   [c0000013aed5fce0] [c00000000069b4bc] proc_reg_write+0xf4/0x150
>   [c0000013aed5fd10] [c0000000005bfeb4] vfs_write+0xf8/0x488
>   [c0000013aed5fdc0] [c0000000005c0570] ksys_write+0x84/0x140
>   [c0000013aed5fe10] [c000000000033358] system_call_exception+0x138/0x330
>   [c0000013aed5fe50] [c00000000000d05c] system_call_vectored_common+0x15c/0x2ec
>   --- interrupt: 3000 at 0x20000433acb4
>   <snip>
>   ---[ end trace 0000000000000000 ]---
> 
> [...]

Applied to powerpc/fixes.

[1/1] powerpc/iommu: Fix the missing iommu_group_put() during platform domain attach
      https://git.kernel.org/powerpc/c/0846dd77c8349ec92ca0079c9c71d130f34cb192

cheers
diff mbox series

Patch

diff --git a/arch/powerpc/kernel/iommu.c b/arch/powerpc/kernel/iommu.c
index d71eac3b2887..a9bebfd56b3b 100644
--- a/arch/powerpc/kernel/iommu.c
+++ b/arch/powerpc/kernel/iommu.c
@@ -1289,8 +1289,10 @@  spapr_tce_platform_iommu_attach_dev(struct iommu_domain *platform_domain,
 	struct iommu_table_group *table_group;
 
 	/* At first attach the ownership is already set */
-	if (!domain)
+	if (!domain) {
+		iommu_group_put(grp);
 		return 0;
+	}
 
 	table_group = iommu_group_get_iommudata(grp);
 	/*