| Message ID | 20201202064811.100688-1-wqu@suse.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | btrfs: preparation patches for subpage support | expand |
On Wed, Dec 02, 2020 at 02:47:56PM +0800, Qu Wenruo wrote: > This is the rebased preparation branch for all patches not yet merged into > misc-next. > > It can be fetched from github (with experimental sector aligned data write > support) > https://github.com/adam900710/linux/tree/subpage > > This patchset includes all the unmerged preparation patches for subpage > support. > > The patchset is sent without the main core for subpage support, as > myself has proven that, big patchset bombarding won't really make > reviewers happy, but only make the author happy (for a very short time). > > Thanks for the hard work from David, there are only 15 patches unmerged. > (With 2 new small patches to address u32 u64 problem) > > Patch 01~02: bio_offset related fixes. Make bio_offset to be u32. > Patch 03: Refactor metadata submission for later metadata write > support. > Patch 04~08: Metadata related refactor. > Patch 09~10: Data related refactor > Patch 11~15: Scrub related refactor and cleanup > > For the scrub patch, there was a discussion with David, about whether we > should use sector size as the unit for metadata scrub. > > His idea is, sector size should be the minimal unit for DATA, not > metadata. This indicates there is a undefined "minimal unit" of access. > > But my argument is, sector size is the minimal unit for all btrfs > access, current btrfs has an undefined "data size", and that "data size" > must equal to sectorsize for current btrfs implementation. > > Thus for "data size" < nodesize case, we should first add support for > "data size" > sectorsize first. > > Thus I kept the scrub patch untouched, since IMHO sector size is still > the minimal unit to access, thus iterating using sectorsize is > completely sane. > > Changelog: > v1: > - Separate prep patches from the huge subpage patchset > > - Rebased to misc-next > > - Add more commit message for patch "btrfs: extent_io: remove the > extent_start/extent_len for end_bio_extent_readpage()" > With one runtime example to explain why we are doing the same thing. > > - Fix the assert_spin_lock() usage > What we really want is lockdep_assert_held() > > - Re-iterate the reason why some extent io tests are invalid > This is especially important since later patches will reduce > extent_buffer::pages[] to bare minimal, killing the ability to > handle certain invalid extent buffers. > > - Use sectorsize_bits for division > During the convert, we should only use sectorsize_bits for division, > this solves the hassle on 32bit system to do division. > But we should not use sectorsize_bits no brain, as bit shift is not > straight forward as multiple/division. > > - Address the comments for btrfs_lookup_bio_sums() cleanup patchset > From naming to macro usages, all of those comments should further > improve the readability. > > v2: > - Remove new extent_io tree features > Now we won't utilize extent io tree for subpage support, thus new > features along with some aggressive refactor is no longer needed. > > - Reduce extent_io tree operations to reduce endio time latency > Although extent_io tree can do a lot of things like page status, but > it has obvious overhead, namingly search btree. > So keep the original behavior by only calling extent_io operation in a > big extent, to reduce latency > > v3: > - Rebased to latest misc-next > Now only 15 patches to submit. > > - Add two new patches to address u32 and u64 problems > The root problem is the on-disk format is abusing u64 for its length. > We have to draw a line between where we should convert to u32. > Currently for bio_offset and extent_len, we can safely use u32. > Just to be extra safe, added more ASSERT() for this. > > - Put BTRFS_MAX_METADATA_BLOCKSIZE into uapi > To avoid circle including "ctree.h" > > - Add more changelog for the patch enabling subpage scrub > > > Qu Wenruo (15): > btrfs: rename bio_offset of extent_submit_bio_start_t to > opt_file_offset > btrfs: pass bio_offset to check_data_csum() directly > btrfs: inode: make btrfs_verify_data_csum() follow sector size > btrfs: extent_io: extract the btree page submission code into its own > helper function > btrfs: extent_io: calculate inline extent buffer page size based on > page size > btrfs: extent_io: don't allow tree block to cross page boundary for > subpage support > btrfs: extent_io: update num_extent_pages() to support subpage sized > extent buffer > btrfs: handle sectorsize < PAGE_SIZE case for extent buffer accessors > btrfs: file-item: remove the btrfs_find_ordered_sum() call in > btrfs_lookup_bio_sums() > btrfs: file-item: refactor btrfs_lookup_bio_sums() to handle > out-of-order bvecs > btrfs: scrub: reduce the width for extent_len/stripe_len from 64 bits > to 32 bits > btrfs: scrub: always allocate one full page for one sector for RAID56 > btrfs: scrub: support subpage tree block scrub > btrfs: scrub: support subpage data scrub > btrfs: scrub: allow scrub to work with subpage sectorsize With a few minor fixups it's in misc-next, thanks.
This is the rebased preparation branch for all patches not yet merged into misc-next. It can be fetched from github (with experimental sector aligned data write support) https://github.com/adam900710/linux/tree/subpage This patchset includes all the unmerged preparation patches for subpage support. The patchset is sent without the main core for subpage support, as myself has proven that, big patchset bombarding won't really make reviewers happy, but only make the author happy (for a very short time). Thanks for the hard work from David, there are only 15 patches unmerged. (With 2 new small patches to address u32 u64 problem) Patch 01~02: bio_offset related fixes. Make bio_offset to be u32. Patch 03: Refactor metadata submission for later metadata write support. Patch 04~08: Metadata related refactor. Patch 09~10: Data related refactor Patch 11~15: Scrub related refactor and cleanup For the scrub patch, there was a discussion with David, about whether we should use sector size as the unit for metadata scrub. His idea is, sector size should be the minimal unit for DATA, not metadata. This indicates there is a undefined "minimal unit" of access. But my argument is, sector size is the minimal unit for all btrfs access, current btrfs has an undefined "data size", and that "data size" must equal to sectorsize for current btrfs implementation. Thus for "data size" < nodesize case, we should first add support for "data size" > sectorsize first. Thus I kept the scrub patch untouched, since IMHO sector size is still the minimal unit to access, thus iterating using sectorsize is completely sane. Changelog: v1: - Separate prep patches from the huge subpage patchset - Rebased to misc-next - Add more commit message for patch "btrfs: extent_io: remove the extent_start/extent_len for end_bio_extent_readpage()" With one runtime example to explain why we are doing the same thing. - Fix the assert_spin_lock() usage What we really want is lockdep_assert_held() - Re-iterate the reason why some extent io tests are invalid This is especially important since later patches will reduce extent_buffer::pages[] to bare minimal, killing the ability to handle certain invalid extent buffers. - Use sectorsize_bits for division During the convert, we should only use sectorsize_bits for division, this solves the hassle on 32bit system to do division. But we should not use sectorsize_bits no brain, as bit shift is not straight forward as multiple/division. - Address the comments for btrfs_lookup_bio_sums() cleanup patchset From naming to macro usages, all of those comments should further improve the readability. v2: - Remove new extent_io tree features Now we won't utilize extent io tree for subpage support, thus new features along with some aggressive refactor is no longer needed. - Reduce extent_io tree operations to reduce endio time latency Although extent_io tree can do a lot of things like page status, but it has obvious overhead, namingly search btree. So keep the original behavior by only calling extent_io operation in a big extent, to reduce latency v3: - Rebased to latest misc-next Now only 15 patches to submit. - Add two new patches to address u32 and u64 problems The root problem is the on-disk format is abusing u64 for its length. We have to draw a line between where we should convert to u32. Currently for bio_offset and extent_len, we can safely use u32. Just to be extra safe, added more ASSERT() for this. - Put BTRFS_MAX_METADATA_BLOCKSIZE into uapi To avoid circle including "ctree.h" - Add more changelog for the patch enabling subpage scrub Qu Wenruo (15): btrfs: rename bio_offset of extent_submit_bio_start_t to opt_file_offset btrfs: pass bio_offset to check_data_csum() directly btrfs: inode: make btrfs_verify_data_csum() follow sector size btrfs: extent_io: extract the btree page submission code into its own helper function btrfs: extent_io: calculate inline extent buffer page size based on page size btrfs: extent_io: don't allow tree block to cross page boundary for subpage support btrfs: extent_io: update num_extent_pages() to support subpage sized extent buffer btrfs: handle sectorsize < PAGE_SIZE case for extent buffer accessors btrfs: file-item: remove the btrfs_find_ordered_sum() call in btrfs_lookup_bio_sums() btrfs: file-item: refactor btrfs_lookup_bio_sums() to handle out-of-order bvecs btrfs: scrub: reduce the width for extent_len/stripe_len from 64 bits to 32 bits btrfs: scrub: always allocate one full page for one sector for RAID56 btrfs: scrub: support subpage tree block scrub btrfs: scrub: support subpage data scrub btrfs: scrub: allow scrub to work with subpage sectorsize fs/btrfs/compression.c | 5 +- fs/btrfs/ctree.c | 3 +- fs/btrfs/ctree.h | 48 ++++-- fs/btrfs/disk-io.c | 17 +- fs/btrfs/disk-io.h | 2 +- fs/btrfs/extent_io.c | 232 +++++++++++++++++----------- fs/btrfs/extent_io.h | 18 ++- fs/btrfs/file-item.c | 266 +++++++++++++++++++++----------- fs/btrfs/inode.c | 61 +++++--- fs/btrfs/ordered-data.c | 44 ------ fs/btrfs/ordered-data.h | 2 - fs/btrfs/scrub.c | 102 +++++++----- fs/btrfs/struct-funcs.c | 18 ++- include/uapi/linux/btrfs_tree.h | 4 + 14 files changed, 496 insertions(+), 326 deletions(-)