Subject: Re: [PATCH -next RFC 2/3] block: factor out the setting of
 GD_NEED_PART_SCAN
To:     Hannes Reinecke <hare@suse.de>, Yu Kuai <yukuai1@huaweicloud.com>,
        hch@infradead.org, jack@suse.cz, axboe@kernel.dk
Cc:     linux-block@vger.kernel.org, linux-kernel@vger.kernel.org,
        yi.zhang@huawei.com, yangerkun@huawei.com,
        "yukuai (C)" <yukuai3@huawei.com>
References: <20230212092641.2394146-1-yukuai1@huaweicloud.com>
 <20230212092641.2394146-3-yukuai1@huaweicloud.com>
 <9b0b9059-c0f6-df5b-863a-987970f3567d@suse.de>
From:   Yu Kuai <yukuai1@huaweicloud.com>
Message-ID: <8eeacd36-6f35-230c-1c93-074c63eee08f@huaweicloud.com>
Date:   Mon, 13 Feb 2023 10:41:32 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:60.0) Gecko/20100101
 Thunderbird/60.8.0
MIME-Version: 1.0
In-Reply-To: <9b0b9059-c0f6-df5b-863a-987970f3567d@suse.de>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk

Hi,

在 2023/02/12 19:36, Hannes Reinecke 写道:
> On 2/12/23 10:26, Yu Kuai wrote:
>> From: Yu Kuai <yukuai3@huawei.com>
>>
>> In order to prevent scan partition for a device that is opened
>> exclusively by someone else, new conditions will be added to
>> disk_scan_partitions() in the next patch. Hence if device is opened
>> exclusively between bdev_add() and disk_scan_partitions(), the first
>> partition scan will fail unexpected. This patch factor out the setting
>> of GD_NEED_PART_SCAN to prevent the problem.
>>
>> Signed-off-by: Yu Kuai <yukuai3@huawei.com>
>> ---
>>   block/genhd.c | 2 +-
>>   block/ioctl.c | 1 +
>>   2 files changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/block/genhd.c b/block/genhd.c
>> index 075d8da284f5..c0d1220bd798 100644
>> --- a/block/genhd.c
>> +++ b/block/genhd.c
>> @@ -367,7 +367,6 @@ int disk_scan_partitions(struct gendisk *disk, 
>> fmode_t mode)
>>       if (disk->open_partitions)
>>           return -EBUSY;
>> -    set_bit(GD_NEED_PART_SCAN, &disk->state);
>>       bdev = blkdev_get_by_dev(disk_devt(disk), mode, NULL);
>>       if (IS_ERR(bdev))
>>           return PTR_ERR(bdev);
>> @@ -493,6 +492,7 @@ int __must_check device_add_disk(struct device 
>> *parent, struct gendisk *disk,
>>           if (ret)
>>               goto out_unregister_bdi;
>> +        set_bit(GD_NEED_PART_SCAN, &disk->state);
>>           bdev_add(disk->part0, ddev->devt);
>>           if (get_capacity(disk))
>>               disk_scan_partitions(disk, FMODE_READ);
> Usual caveat:
> What happens if the flag is already set here?
> Wouldn't that imply that another scan is underway?

I think this is not true, currently set the state just means we're going
to scan, and if some checking failed, the state will be left. I don't
have a good idea how to fix this yet...

Thanks,
Kuai

> And wouldn't it be better to use 'test_and_set()'?
> 
> 
>> diff --git a/block/ioctl.c b/block/ioctl.c
>> index 6dd49d877584..0eefcdb936a0 100644
>> --- a/block/ioctl.c
>> +++ b/block/ioctl.c
>> @@ -528,6 +528,7 @@ static int blkdev_common_ioctl(struct block_device 
>> *bdev, fmode_t mode,
>>               return -EACCES;
>>           if (bdev_is_partition(bdev))
>>               return -EINVAL;
>> +        set_bit(GD_NEED_PART_SCAN, &bdev->bd_disk->state);
>>           return disk_scan_partitions(bdev->bd_disk, mode & ~FMODE_EXCL);
>>       case BLKTRACESTART:
>>       case BLKTRACESTOP:
> Similar here.
> 
> Cheers,
> 
> Hannes