On 9/10/25 10:29 AM, Bart Van Assche wrote: > On 9/9/25 5:53 PM, Damien Le Moal wrote: >> On 9/10/25 3:26 AM, Bart Van Assche wrote: >>> Splitting a bio into an empty flush bio and a non-flush data bio is >>> only necessary if there are multiple underlying devices. Do not split >>> REQ_PREFLUSH bios if there is only one underlying device. This patch >>> preserves the order of REQ_PREFLUSH writes if there is only one >>> underlying device and if multiple writes are queued. This patch fixes >>> a write error observed with zoned storage devices and with write >>> pipelining enabled. >> >> Write pipelining is not upstream. > > That doesn't matter in this case because this patch also improves > performance without write pipelining, even for conventional (non zoned) storage > devices. Then please rewrite the commit message to clearly state that and add Fixes tags as appropriate. Do not mention things that are not upstream please. > BTW, I'm still waiting for a review from you for this patch series that > was posted two weeks ago: "[PATCH v24 00/18] Improve write performance > for zoned UFS devices" > (https://lore.kernel.org/linux-block/20250827212937.2759348-1-bvanassche@xxxxxxx/) Yes, I still have this series in my inbox. I have been very busy and have lots of conferences coming up so the review is going to be slow. -- Damien Le Moal Western Digital Research
