Renamed to RWF_UNCACHED, fixed a bug, and added support for writes. For the latter, they work like O_DIRECT writes in that they wait for completion, then drop the range from the cache. https://git.kernel.dk/cgit/linux-block/commit/?h=buffered-uncached&id=0a1d5f95b15679f902012dc058b5020cf1bc7a6c …
Why do writes work that way, rather than leaving the page cache state unmodified? Haven't thought this through, but intuitively I can see use-cases for both, leaving the set of cache pages unmodified and removing the range.
-
-
We need to wait for writes to complete for invalidation to work. Always open to suggestions, though, but how does "leaving the set of cache pages unmodified" differ from just a normal buffered write?
-
It'd not leave a range present in the page cache when not present before the write.
- 2 more replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.