On 2023-08-11 5:08 AM, Christoph Hellwig wrote:
> nbd_clear_sock_ioctl kills the socket and with that the block
> device. Instead of just invalidating file system buffers,
> mark the device as dead, which will also invalidate the buffers
> as part of the proper shutdown sequence. This also includes
> invalidating partitions if there are any.
>
> Signed-off-by: Christoph Hellwig <hch@xxxxxx>
> ---
> drivers/block/nbd.c | 8 +++-----
> 1 file changed, 3 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c
> index 8576d696c7a221..42e0159bb258fa 100644
> --- a/drivers/block/nbd.c
> +++ b/drivers/block/nbd.c
> @@ -1434,12 +1434,10 @@ static int nbd_start_device_ioctl(struct nbd_device *nbd)
> return ret;
> }
>
> -static void nbd_clear_sock_ioctl(struct nbd_device *nbd,
> - struct block_device *bdev)
> +static void nbd_clear_sock_ioctl(struct nbd_device *nbd)
> {
> + blk_mark_disk_dead(nbd->disk);
> nbd_clear_sock(nbd);
> - __invalidate_device(bdev, true);
> - nbd_bdev_reset(nbd);
This change breaks nbd-client, which calls the NBD_CLEAR_SOCK ioctl during
device setup and socket reconnection. After merging this series (bisected to
511fb5bafed1), all NBD devices are immediately dead on arrival:
[ 14.605849] nbd0: detected capacity change from 0 to 4194304
[ 14.606211] Buffer I/O error on dev nbd0, logical block 0, async page read
[ 14.619101] Buffer I/O error on dev nbd0, logical block 0, async page read
[ 14.630490] nbd0: unable to read partition table
I wonder if disk_force_media_change() is the right thing to call here instead.
Regards,
Samuel
> if (test_and_clear_bit(NBD_RT_HAS_CONFIG_REF,
> &nbd->config->runtime_flags))
> nbd_config_put(nbd);
> @@ -1465,7 +1463,7 @@ static int __nbd_ioctl(struct block_device *bdev, struct nbd_device *nbd,
> case NBD_DISCONNECT:
> return nbd_disconnect(nbd);
> case NBD_CLEAR_SOCK:
> - nbd_clear_sock_ioctl(nbd, bdev);
> + nbd_clear_sock_ioctl(nbd);
> return 0;
> case NBD_SET_SOCK:
> return nbd_add_socket(nbd, arg, false);
