On Fri, Jun 21, 2024 at 10:27:20AM -0700, Darrick J. Wong wrote: > That sounds like a good approach for ext4 which will likely be feeding > you folios from the pagecache anyway. If Andrey sticks with the buffer > cache that I wrote, then the virtual address is a slab allocation which > is already mapped. ... yet another reasons to not be pointleslly different code and create special cases and extra code in every layer of the stack..
