Hi Jens, >> The fact that you didn't even compile this not withstanding, why >> not just move the existing check from blk_mq_free_request() to >> __blk_mq_free_request()? Could you please review the change again when you have a chance, I made the change that you suggested earlier. I apologize for the broken patch submitted earlier, we did test the change on our production host and able to confirm the change did address the active request leaking issue. It was totally my fault for not proofreading the change before submitting it to the upstream. Best, Tian
