[PATCH iwl-net v2 1/2] iavf: fix ASQ command buffer leak on init failure

[PATCH iwl-net v2 1/2] iavf: fix ASQ command buffer leak on init failure

[xuanqiang.luo]

From: Xuanqiang Luo <luoxuanqiang@kylinos.cn>

iavf_alloc_adminq_asq_ring() allocates cmd_buf before the remaining ASQ
resources. If iavf_alloc_asq_bufs() or iavf_config_asq_regs() fails, the
unwind path elides cmd_buf while freeing the other allocations.

The ASQ count is not set until initialization succeeds, so the shutdown
path cannot reclaim the buffer. Free cmd_buf in the common unwind path.

Fixes: d358aa9a7a2d ("i40evf: init code and hardware support")
Cc: stable@vger.kernel.org
Signed-off-by: Xuanqiang Luo <luoxuanqiang@kylinos.cn>
---
 drivers/net/ethernet/intel/iavf/iavf_adminq.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/net/ethernet/intel/iavf/iavf_adminq.c b/drivers/net/ethernet/intel/iavf/iavf_adminq.c
index 6937b7dd44cbb..40f76f9507f4b 100644
--- a/drivers/net/ethernet/intel/iavf/iavf_adminq.c
+++ b/drivers/net/ethernet/intel/iavf/iavf_adminq.c
@@ -60,6 +60,7 @@ static enum iavf_status iavf_alloc_adminq_arq_ring(struct iavf_hw *hw)
  **/
 static void iavf_free_adminq_asq(struct iavf_hw *hw)
 {
+	iavf_free_virt_mem(hw, &hw->aq.asq.cmd_buf);
 	iavf_free_dma_mem(hw, &hw->aq.asq.desc_buf);
 }
 
-- 
2.43.0

RE: [PATCH iwl-net v2 1/2] iavf: fix ASQ command buffer leak on init failure

[Jagielski, Jedrzej]

From: xuanqiang.luo@linux.dev <xuanqiang.luo@linux.dev> 
Sent: Wednesday, July 15, 2026 10:26 AM

>From: Xuanqiang Luo <luoxuanqiang@kylinos.cn>
>
>iavf_alloc_adminq_asq_ring() allocates cmd_buf before the remaining ASQ
>resources. If iavf_alloc_asq_bufs() or iavf_config_asq_regs() fails, the
>unwind path elides cmd_buf while freeing the other allocations.
>
>The ASQ count is not set until initialization succeeds, so the shutdown
>path cannot reclaim the buffer. Free cmd_buf in the common unwind path.
>
>Fixes: d358aa9a7a2d ("i40evf: init code and hardware support")
>Cc: stable@vger.kernel.org
>Signed-off-by: Xuanqiang Luo <luoxuanqiang@kylinos.cn>
>---
> drivers/net/ethernet/intel/iavf/iavf_adminq.c | 1 +
> 1 file changed, 1 insertion(+)
>
>diff --git a/drivers/net/ethernet/intel/iavf/iavf_adminq.c b/drivers/net/ethernet/intel/iavf/iavf_adminq.c
>index 6937b7dd44cbb..40f76f9507f4b 100644
>--- a/drivers/net/ethernet/intel/iavf/iavf_adminq.c
>+++ b/drivers/net/ethernet/intel/iavf/iavf_adminq.c
>@@ -60,6 +60,7 @@ static enum iavf_status iavf_alloc_adminq_arq_ring(struct iavf_hw *hw)
>  **/
> static void iavf_free_adminq_asq(struct iavf_hw *hw)
> {
>+	iavf_free_virt_mem(hw, &hw->aq.asq.cmd_buf);
> 	iavf_free_dma_mem(hw, &hw->aq.asq.desc_buf);
> }
> 
>-- 
>2.43.0

Looks fine, thanks!

Reviewed-by: Jedrzej Jagielski <jedrzej.jagielski@intel.com>

One note for the future - please be aware that there is minimal time period to be
waited before resubmitting new patch revision, which is at least 24h for netdev/IWL
mailing lists

Re: [PATCH iwl-net v2 1/2] iavf: fix ASQ command buffer leak on init failure

[luoxuanqiang]

在 2026/7/15 17:28, Jagielski, Jedrzej 写道:
> From: xuanqiang.luo@linux.dev <xuanqiang.luo@linux.dev>
> Sent: Wednesday, July 15, 2026 10:26 AM
>
>> From: Xuanqiang Luo <luoxuanqiang@kylinos.cn>
>>
>> iavf_alloc_adminq_asq_ring() allocates cmd_buf before the remaining ASQ
>> resources. If iavf_alloc_asq_bufs() or iavf_config_asq_regs() fails, the
>> unwind path elides cmd_buf while freeing the other allocations.
>>
>> The ASQ count is not set until initialization succeeds, so the shutdown
>> path cannot reclaim the buffer. Free cmd_buf in the common unwind path.
>>
>> Fixes: d358aa9a7a2d ("i40evf: init code and hardware support")
>> Cc: stable@vger.kernel.org
>> Signed-off-by: Xuanqiang Luo <luoxuanqiang@kylinos.cn>
>> ---
>> drivers/net/ethernet/intel/iavf/iavf_adminq.c | 1 +
>> 1 file changed, 1 insertion(+)
>>
>> diff --git a/drivers/net/ethernet/intel/iavf/iavf_adminq.c b/drivers/net/ethernet/intel/iavf/iavf_adminq.c
>> index 6937b7dd44cbb..40f76f9507f4b 100644
>> --- a/drivers/net/ethernet/intel/iavf/iavf_adminq.c
>> +++ b/drivers/net/ethernet/intel/iavf/iavf_adminq.c
>> @@ -60,6 +60,7 @@ static enum iavf_status iavf_alloc_adminq_arq_ring(struct iavf_hw *hw)
>>   **/
>> static void iavf_free_adminq_asq(struct iavf_hw *hw)
>> {
>> +	iavf_free_virt_mem(hw, &hw->aq.asq.cmd_buf);
>> 	iavf_free_dma_mem(hw, &hw->aq.asq.desc_buf);
>> }
>>
>> -- 
>> 2.43.0
> Looks fine, thanks!
>
> Reviewed-by: Jedrzej Jagielski <jedrzej.jagielski@intel.com>
>
> One note for the future - please be aware that there is minimal time period to be
> waited before resubmitting new patch revision, which is at least 24h for netdev/IWL
> mailing lists

Thanks for the reminder!

I also received a notification from netdev-bot, and I'll keep this in
mind for future revisions.

RE: [Intel-wired-lan] [PATCH iwl-net v2 1/2] iavf: fix ASQ command buffer leak on init failure

[Loktionov, Aleksandr]

> -----Original Message-----
> From: Intel-wired-lan <intel-wired-lan-bounces@osuosl.org> On Behalf
> Of xuanqiang.luo@linux.dev
> Sent: Wednesday, July 15, 2026 10:26 AM
> To: Nguyen, Anthony L <anthony.l.nguyen@intel.com>; Kitszel,
> Przemyslaw <przemyslaw.kitszel@intel.com>; Jagielski; Jagielski,
> Jedrzej <jedrzej.jagielski@intel.com>; intel-wired-
> lan@lists.osuosl.org
> Cc: Andrew Lunn <andrew+netdev@lunn.ch>; Mitch Williams
> <mitch.a.williams@intel.com>; Greg Rose <gregory.v.rose@intel.com>;
> Sudheer Mogilappagari <sudheer.mogilappagari@intel.com>;
> netdev@vger.kernel.org; Xuanqiang Luo <luoxuanqiang@kylinos.cn>;
> stable@vger.kernel.org
> Subject: [Intel-wired-lan] [PATCH iwl-net v2 1/2] iavf: fix ASQ
> command buffer leak on init failure
> 
> From: Xuanqiang Luo <luoxuanqiang@kylinos.cn>
> 
> iavf_alloc_adminq_asq_ring() allocates cmd_buf before the remaining
> ASQ resources. If iavf_alloc_asq_bufs() or iavf_config_asq_regs()
> fails, the unwind path elides cmd_buf while freeing the other
> allocations.
> 
> The ASQ count is not set until initialization succeeds, so the
> shutdown path cannot reclaim the buffer. Free cmd_buf in the common
> unwind path.
> 
> Fixes: d358aa9a7a2d ("i40evf: init code and hardware support")
> Cc: stable@vger.kernel.org
> Signed-off-by: Xuanqiang Luo <luoxuanqiang@kylinos.cn>
> ---
>  drivers/net/ethernet/intel/iavf/iavf_adminq.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/net/ethernet/intel/iavf/iavf_adminq.c
> b/drivers/net/ethernet/intel/iavf/iavf_adminq.c
> index 6937b7dd44cbb..40f76f9507f4b 100644
> --- a/drivers/net/ethernet/intel/iavf/iavf_adminq.c
> +++ b/drivers/net/ethernet/intel/iavf/iavf_adminq.c
> @@ -60,6 +60,7 @@ static enum iavf_status
> iavf_alloc_adminq_arq_ring(struct iavf_hw *hw)
>   **/
>  static void iavf_free_adminq_asq(struct iavf_hw *hw)  {
> +	iavf_free_virt_mem(hw, &hw->aq.asq.cmd_buf);
>  	iavf_free_dma_mem(hw, &hw->aq.asq.desc_buf);  }
> 
> --
> 2.43.0

Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>

[PATCH iwl-net v2 2/2] iavf: fix QoS capabilities memory leak

[xuanqiang.luo]

From: Xuanqiang Luo <luoxuanqiang@kylinos.cn>

Commit 4c1a457cb8b0 ("iavf: add support to exchange qos capabilities")
allocates adapter->qos_caps during probe, but iavf_remove() does not
free it. This leaks the allocation whenever an iavf device is removed.

Free adapter->qos_caps in iavf_remove().

Fixes: 4c1a457cb8b0 ("iavf: add support to exchange qos capabilities")
Cc: stable@vger.kernel.org
Reviewed-by: Jedrzej Jagielski <jedrzej.jagielski@intel.com>
Signed-off-by: Xuanqiang Luo <luoxuanqiang@kylinos.cn>
---
 drivers/net/ethernet/intel/iavf/iavf_main.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/net/ethernet/intel/iavf/iavf_main.c b/drivers/net/ethernet/intel/iavf/iavf_main.c
index 29b8403a066bc..c7f69a9040588 100644
--- a/drivers/net/ethernet/intel/iavf/iavf_main.c
+++ b/drivers/net/ethernet/intel/iavf/iavf_main.c
@@ -5589,6 +5589,7 @@ static void iavf_remove(struct pci_dev *pdev)
 	iounmap(hw->hw_addr);
 	pci_release_regions(pdev);
 	kfree(adapter->vf_res);
+	kfree(adapter->qos_caps);
 	spin_lock_bh(&adapter->mac_vlan_list_lock);
 	/* If we got removed before an up/down sequence, we've got a filter
 	 * hanging out there that we need to get rid of.
-- 
2.43.0

RE: [Intel-wired-lan] [PATCH iwl-net v2 2/2] iavf: fix QoS capabilities memory leak

[Loktionov, Aleksandr]

> -----Original Message-----
> From: Intel-wired-lan <intel-wired-lan-bounces@osuosl.org> On Behalf
> Of xuanqiang.luo@linux.dev
> Sent: Wednesday, July 15, 2026 10:26 AM
> To: Nguyen, Anthony L <anthony.l.nguyen@intel.com>; Kitszel,
> Przemyslaw <przemyslaw.kitszel@intel.com>; Jagielski; Jagielski,
> Jedrzej <jedrzej.jagielski@intel.com>; intel-wired-
> lan@lists.osuosl.org
> Cc: Andrew Lunn <andrew+netdev@lunn.ch>; Mitch Williams
> <mitch.a.williams@intel.com>; Greg Rose <gregory.v.rose@intel.com>;
> Sudheer Mogilappagari <sudheer.mogilappagari@intel.com>;
> netdev@vger.kernel.org; Xuanqiang Luo <luoxuanqiang@kylinos.cn>;
> stable@vger.kernel.org
> Subject: [Intel-wired-lan] [PATCH iwl-net v2 2/2] iavf: fix QoS
> capabilities memory leak
> 
> From: Xuanqiang Luo <luoxuanqiang@kylinos.cn>
> 
> Commit 4c1a457cb8b0 ("iavf: add support to exchange qos capabilities")
> allocates adapter->qos_caps during probe, but iavf_remove() does not
> free it. This leaks the allocation whenever an iavf device is removed.
> 
> Free adapter->qos_caps in iavf_remove().
> 
> Fixes: 4c1a457cb8b0 ("iavf: add support to exchange qos capabilities")
> Cc: stable@vger.kernel.org
> Reviewed-by: Jedrzej Jagielski <jedrzej.jagielski@intel.com>
> Signed-off-by: Xuanqiang Luo <luoxuanqiang@kylinos.cn>
> ---
>  drivers/net/ethernet/intel/iavf/iavf_main.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/net/ethernet/intel/iavf/iavf_main.c
> b/drivers/net/ethernet/intel/iavf/iavf_main.c
> index 29b8403a066bc..c7f69a9040588 100644
> --- a/drivers/net/ethernet/intel/iavf/iavf_main.c
> +++ b/drivers/net/ethernet/intel/iavf/iavf_main.c
> @@ -5589,6 +5589,7 @@ static void iavf_remove(struct pci_dev *pdev)
>  	iounmap(hw->hw_addr);
>  	pci_release_regions(pdev);
>  	kfree(adapter->vf_res);
> +	kfree(adapter->qos_caps);
>  	spin_lock_bh(&adapter->mac_vlan_list_lock);
>  	/* If we got removed before an up/down sequence, we've got a
> filter
>  	 * hanging out there that we need to get rid of.
> --
> 2.43.0

Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>