On Tue, Mar 11, 2025 at 11:41:44AM +0900, Shin'ichiro Kawasaki wrote:

However, blk_mq_add_to_batch() callers do not pass negative error
values. Instead, they pass status codes defined in various ways:

- NVMe PCI and Apple drivers pass NVMe status code
- virtio_blk driver passes the virtblk request header status byte
- null_blk driver passes blk_status_t

The __force cast in null_blk should have been a big fat warning..

To correct the ioerror check within blk_mq_add_to_batch(), make all
callers to uniformly pass the argument as blk_status_t. Modify the
callers to translate their specific status codes into blk_status_t. For
this translation, export the helper function nvme_error_status(). Adjust
blk_mq_add_to_batch() to translate blk_status_t back into the error
number for the appropriate check.

This still looks a bit ugly because of all the conversions to a
blk_status_t just to convert it back to a errno just to check for
a non-zero value (blk_status_to_errno can't return a positive value).

I suspect simply passing a "bool is_error" might actually be cleaner
than that, combined with a proper kerneldoc comment for
blk_mq_add_to_batch explaining how to set it?