2017-05-19 23:19 GMT+03:00 Lionel Bouton <lionel-subscript...@bouton.name>: > I was too focused on other problems and having a fresh look at what I > wrote I'm embarrassed by what I read. > Used pages for a given amount of data should be (amount / PAGE_SIZE) + > ((amount % PAGE_SIZE) == 0 ? 0 : 1) this seems enough of a common thing > to compute that the kernel might have a macro defined for this.
If i understand the code correctly, the logic of comparing the size of input/output by bytes is enough (IMHO) for skipping the compression of non-compressible data, and as btrfs uses PAGE_SIZE as a data cluster size (and if i understand correctly it's minimum IO size), the logic of that can be improved in case when compressed data use 126978 < compressed_size < 131072. The easiest way to improve that case, i think, is making the compressed size bigger by PAGE_SIZE. JFYI: Once I've read on the list that btrfs don't compress data, if data are less then PAGE_SIZE because it's useless (i didn't find that in the code, so i think that i don't fully understand code of compress routine) After some time i got a idea that if btrfs determines store data compressed or not by comparing input/output size of data (i.e. if compressed size is bigger in compare to input data, don't store compressed version), this logic can be improved by also checking if compression profit is more then PAGE_SIZE, because if it's not, compressed data will pass check and comsume the same amount of space and as a result will not show any gain. -- Have a nice day, Timofey. -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
