Messages in this thread | | | Date | Thu, 30 Jan 2014 11:26:04 -0800 | Subject | Re: [GIT PULL] Core block IO changes for 3.14 | From | Linus Torvalds <> |
| |
On Tue, Jan 28, 2014 at 4:45 PM, Jaegeuk Kim <jaegeuk.kim@samsung.com> wrote: > > In the case of f2fs, could you please check the following code changes?
So I ended up doing the merge differently wrt the f2fs_write_end_io() function.
As far as I can tell, we can just initialize the f2fs_sb_info directly and unconditionally using the first bio_vec entry, giving us the simpler
struct f2fs_sb_info *sbi = F2FS_SB(bio->bi_io_vec->bv_page->mapping->host->i_sb);
which is still ugly, but whatever. I'm not quite seeing why f2fs doesn't just set bio->private to the superblock pointer and avoid this whole complex and long pointer chasing, but there's probably some reason.
Anyway, it compiles for me, and looks simpler than the alternate resolutions, but maybe there is some reason I miss why you guys did it the way you did. And while I could test that it compiles and looks sane, I don't have a f2fs to actually *test* it on, and maybe I screwed up royally.
I'll push it out once I've done the rest of the allmodconfig build-test.
Linus
| |