On 7/2/20 7:48 AM, Ming Lei wrote: > The built-in flush request shares tag with the request inserted to flush > machinery, turns out its .end_io callback has to touch the built-in > flush request's internal tag or tag. > > On the other hand, we have to make sure that driver tag is released > from __blk_mq_end_request(), since this request may not be completed > via blk_mq_complete_request(). > > Given we have moved blk_mq_put_driver_tag() out of header file, fix this > issue by releasing driver tag before calling .end_io(). As I wrote yesterday, I reverted the three patches: https://git.kernel.dk/cgit/linux-block/log/?h=for-5.9/block so you'll need to send a new and updated series. Please fold in the 'unused variable' and related fixes as well. -- Jens Axboe
