netdev - Re: [PATCH V1 mlx5-next 4/4] vfio/mlx5: Run the SAVE state command in an async mode

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <8d9dcc65-a754-ec6b-e1ed-1511efaf0b14@nvidia.com>
Date:   Tue, 10 May 2022 11:56:58 +0300
From:   Yishai Hadas <yishaih@...dia.com>
To:     Alex Williamson <alex.williamson@...hat.com>
CC:     <jgg@...dia.com>, <saeedm@...dia.com>, <kvm@...r.kernel.org>,
        <netdev@...r.kernel.org>, <kuba@...nel.org>, <leonro@...dia.com>,
        <maorg@...dia.com>, <cohuck@...hat.com>
Subject: Re: [PATCH V1 mlx5-next 4/4] vfio/mlx5: Run the SAVE state command in
 an async mode

On 09/05/2022 20:29, Alex Williamson wrote:
> On Sun, 8 May 2022 16:10:53 +0300
> Yishai Hadas <yishaih@...dia.com> wrote:
>> diff --git a/drivers/vfio/pci/mlx5/cmd.h b/drivers/vfio/pci/mlx5/cmd.h
>> index 2a20b7435393..d053d314b745 100644
>> --- a/drivers/vfio/pci/mlx5/cmd.h
>> +++ b/drivers/vfio/pci/mlx5/cmd.h
>> @@ -10,10 +10,20 @@
>>   #include <linux/vfio_pci_core.h>
>>   #include <linux/mlx5/driver.h>
>>   
>> +struct mlx5vf_async_data {
>> +	struct mlx5_async_work cb_work;
>> +	struct work_struct work;
>> +	int status;
>> +	u32 pdn;
>> +	u32 mkey;
>> +	void *out;
>> +};
>> +
>>   struct mlx5_vf_migration_file {
>>   	struct file *filp;
>>   	struct mutex lock;
>>   	bool disabled;
>> +	u8 is_err:1;
> Convert @disabled to bit field as well to pack these?

OK

>
> ...
>> @@ -558,6 +592,13 @@ static int mlx5vf_pci_probe(struct pci_dev *pdev,
>>   		return -ENOMEM;
>>   	vfio_pci_core_init_device(&mvdev->core_device, pdev, &mlx5vf_pci_ops);
>>   	mlx5vf_cmd_set_migratable(mvdev);
>> +	if (mvdev->migrate_cap) {
>> +		mvdev->cb_wq = alloc_ordered_workqueue("mlx5vf_wq", 0);
>> +		if (!mvdev->cb_wq) {
>> +			ret = -ENOMEM;
>> +			goto out_free;
>> +		}
>> +	}
> Should this be rolled into mlx5vf_cmd_set_migratable(), updating the
> function to return -errno?

This can be done, however, I would still keep the function as void as 
you previously suggested.

In case the WQ somehow couldn't be created it just means that migratable 
functionality couldn't be activated and its cap won't be set.

>>   	ret = vfio_pci_core_register_device(&mvdev->core_device);
>>   	if (ret)
>>   		goto out_free;
>> @@ -566,8 +607,11 @@ static int mlx5vf_pci_probe(struct pci_dev *pdev,
>>   	return 0;
>>   
>>   out_free:
>> -	if (mvdev->migrate_cap)
>> +	if (mvdev->migrate_cap) {
>>   		mlx5vf_cmd_remove_migratable(mvdev);
>> +		if (mvdev->cb_wq)
>> +			destroy_workqueue(mvdev->cb_wq);
>> +	}
>>   	vfio_pci_core_uninit_device(&mvdev->core_device);
>>   	kfree(mvdev);
>>   	return ret;
>> @@ -578,8 +622,10 @@ static void mlx5vf_pci_remove(struct pci_dev *pdev)
>>   	struct mlx5vf_pci_core_device *mvdev = dev_get_drvdata(&pdev->dev);
>>   
>>   	vfio_pci_core_unregister_device(&mvdev->core_device);
>> -	if (mvdev->migrate_cap)
>> +	if (mvdev->migrate_cap) {
>>   		mlx5vf_cmd_remove_migratable(mvdev);
>> +		destroy_workqueue(mvdev->cb_wq);
>> +	}
>>   	vfio_pci_core_uninit_device(&mvdev->core_device);
>>   	kfree(mvdev);
>>   }
> This looks like more evidence for expanding remove_migratable(),
> rolling this in as well.  If this workqueue were setup in
> set_migratable() we'd not need the special condition to test if cb_wq
> is NULL while migrate_cap is set.  Thanks,
>
> Alex
>
Makes sense, will be part of V2.

Yishai