On 10/17/23 2:47 PM, Bart Van Assche wrote: > Hi Jens, > > UFS vendors need the data lifetime information to achieve good performance. > Without this information there is significantly higher write amplification due > to garbage collection. Hence this patch series that add support in F2FS and > also in the block layer for data lifetime information. The SCSI disk (sd) > driver is modified such that it passes write hint information to SCSI devices > via the GROUP NUMBER field. > > Please consider this patch series for the next merge window. My main hesitation with this is that there's a big gap between what makes theoretical sense and practical sense. When we previously tried this, turns out devices retained the data temperature on media, as expected, but tossed it out when data was GC'ed. That made it more of a benchmarking case than anything else. How do we know that things are better now? In previous postings I've seen you point at some papers, but I'm mostly concerned with practical use cases and devices. Are there any results, at all, from that? Or is this a case of vendors asking for something to check some marketing boxes or have value add? I can take a closer look once this is fully understood. Not adding something like this without proper justification. I'm also really against growing struct bio just for this. Why is patch 2 not just using the ioprio field at least? -- Jens Axboe
