Hi Jens,
except for the DASD case under discussion the last users of ioctl_by_bdev
are the file system drivers that want to query CDROM information using
ioctls. This series switches them to use function calls directly into
the CDROM midlayer instead, which implies:
- adding a cdrom_device_info pointer to the gendisk, so that file systems
can find it without going to the low-level driver first
- ensuring that the CDROM midlayer (which isn't a lot of code) is built
in if the file systems are built in so that they can actually call the
exported functions
Changes since v1:
- fix up the no-CDROM error case in isofs_get_last_session to return 0
instead of -EINVAL.
Jens,
can you pick this up for the for-5.8/block tree? We're about ready
to kill ioctl_by_bdev after this series.
On 4/25/20 1:56 AM, Christoph Hellwig wrote:
> Hi Jens,
>
> except for the DASD case under discussion the last users of ioctl_by_bdev
> are the file system drivers that want to query CDROM information using
> ioctls. This series switches them to use function calls directly into
> the CDROM midlayer instead, which implies:
>
> - adding a cdrom_device_info pointer to the gendisk, so that file systems
> can find it without going to the low-level driver first
> - ensuring that the CDROM midlayer (which isn't a lot of code) is built
> in if the file systems are built in so that they can actually call the
> exported functions
>
> Changes since v1:
> - fix up the no-CDROM error case in isofs_get_last_session to return 0
> instead of -EINVAL.
Applied for 5.8, thanks.
--
Jens Axboe