On Mon, Apr 18, 2016 at 2:07 AM, Christoph Hellwig <hch@xxxxxx> wrote: > On Sun, Apr 17, 2016 at 10:25:09PM +0800, Ming Lei wrote: >> The patch itself is correct, and the idea is good. >> >> But it can be simpler or more readable by always chaining bios into >> the same parent bio, which is submitted as the last one. > > Which means we now messed up our I/O order to not be sequential > for no good reason. >From I/O order view, I don't see any difference among the two approachs, and all are like the following, and both are similar with the way in linus tree too. - submit(sector: s[0], size:sz[0]) - submit(sector: s[1], size:sz[1]) ... - submit(sector:s[n-1], size:sz[n-1]) - submit_wait(sector:s[n], size:sz[n]) And there is always the euquation: s[i+1] = s[i] + sz[i] for i = 0, ..., n - 1, and s[n] + sz[n] = sector + nr_sects. Thanks, Ming Lei -- To unsubscribe from this list: send the line "unsubscribe linux-block" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
