[PATCH v7] iomap: make inline data support more flexible
Joseph Qi
joseph.qi at linux.alibaba.com
Mon Jul 26 18:08:21 AEST 2021
On 7/24/21 1:41 AM, Gao Xiang wrote:
> Add support for reading inline data content into the page cache from
> nonzero page-aligned file offsets. This enables the EROFS tailpacking
> mode where the last few bytes of the file are stored right after the
> inode.
>
> The buffered write path remains untouched since EROFS cannot be used
> for testing. It'd be better to be implemented if upcoming real users
> care and provide a real pattern rather than leave untested dead code
> around.
>
> Cc: Christoph Hellwig <hch at lst.de>
> Cc: Darrick J. Wong <djwong at kernel.org>
> Cc: Matthew Wilcox <willy at infradead.org>
> Cc: Andreas Gruenbacher <andreas.gruenbacher at gmail.com>
> Tested-by: Huang Jianan <huangjianan at oppo.com> # erofs
> Signed-off-by: Gao Xiang <hsiangkao at linux.alibaba.com>
Looks good to me.
Reviewed-by: Joseph Qi <joseph.qi at linux.alibaba.com>
> ---
> v6: https://lore.kernel.org/r/20210722031729.51628-1-hsiangkao@linux.alibaba.com
> changes since v6:
> - based on Christoph's reply;
> - update commit message suggested by Darrick;
> - disable buffered write path until some real fs users.
>
> fs/iomap/buffered-io.c | 42 ++++++++++++++++++++++++++----------------
> fs/iomap/direct-io.c | 10 ++++++----
> include/linux/iomap.h | 14 ++++++++++++++
> 3 files changed, 46 insertions(+), 20 deletions(-)
>
> diff --git a/fs/iomap/buffered-io.c b/fs/iomap/buffered-io.c
> index 87ccb3438bec..f351e1f9e3f6 100644
> --- a/fs/iomap/buffered-io.c
> +++ b/fs/iomap/buffered-io.c
> @@ -205,25 +205,29 @@ struct iomap_readpage_ctx {
> struct readahead_control *rac;
> };
>
> -static void
> -iomap_read_inline_data(struct inode *inode, struct page *page,
> - struct iomap *iomap)
> +static int iomap_read_inline_data(struct inode *inode, struct page *page,
> + struct iomap *iomap, loff_t pos)
> {
> - size_t size = i_size_read(inode);
> + size_t size = iomap->length + iomap->offset - pos;
> void *addr;
>
> if (PageUptodate(page))
> - return;
> + return PAGE_SIZE;
>
> - BUG_ON(page_has_private(page));
> - BUG_ON(page->index);
> - BUG_ON(size > PAGE_SIZE - offset_in_page(iomap->inline_data));
> + /* inline data must start page aligned in the file */
> + if (WARN_ON_ONCE(offset_in_page(pos)))
> + return -EIO;
> + if (WARN_ON_ONCE(!iomap_inline_data_size_valid(iomap)))
> + return -EIO;
> + if (WARN_ON_ONCE(page_has_private(page)))
> + return -EIO;
>
> addr = kmap_atomic(page);
> - memcpy(addr, iomap->inline_data, size);
> + memcpy(addr, iomap_inline_buf(iomap, pos), size);
> memset(addr + size, 0, PAGE_SIZE - size);
> kunmap_atomic(addr);
> SetPageUptodate(page);
> + return PAGE_SIZE;
> }
>
> static inline bool iomap_block_needs_zeroing(struct inode *inode,
> @@ -246,11 +250,8 @@ iomap_readpage_actor(struct inode *inode, loff_t pos, loff_t length, void *data,
> unsigned poff, plen;
> sector_t sector;
>
> - if (iomap->type == IOMAP_INLINE) {
> - WARN_ON_ONCE(pos);
> - iomap_read_inline_data(inode, page, iomap);
> - return PAGE_SIZE;
> - }
> + if (iomap->type == IOMAP_INLINE)
> + return iomap_read_inline_data(inode, page, iomap, pos);
>
> /* zero post-eof blocks as the page may be mapped */
> iop = iomap_page_create(inode, page);
> @@ -589,6 +590,15 @@ __iomap_write_begin(struct inode *inode, loff_t pos, unsigned len, int flags,
> return 0;
> }
>
> +static int iomap_write_begin_inline(struct inode *inode,
> + struct page *page, struct iomap *srcmap)
> +{
> + /* needs more work for the tailpacking case, disable for now */
> + if (WARN_ON_ONCE(srcmap->offset != 0))
> + return -EIO;
> + return iomap_read_inline_data(inode, page, srcmap, 0);
> +}
> +
> static int
> iomap_write_begin(struct inode *inode, loff_t pos, unsigned len, unsigned flags,
> struct page **pagep, struct iomap *iomap, struct iomap *srcmap)
> @@ -618,14 +628,14 @@ iomap_write_begin(struct inode *inode, loff_t pos, unsigned len, unsigned flags,
> }
>
> if (srcmap->type == IOMAP_INLINE)
> - iomap_read_inline_data(inode, page, srcmap);
> + status = iomap_write_begin_inline(inode, page, srcmap);
> else if (iomap->flags & IOMAP_F_BUFFER_HEAD)
> status = __block_write_begin_int(page, pos, len, NULL, srcmap);
> else
> status = __iomap_write_begin(inode, pos, len, flags, page,
> srcmap);
>
> - if (unlikely(status))
> + if (unlikely(status < 0))
> goto out_unlock;
>
> *pagep = page;
> diff --git a/fs/iomap/direct-io.c b/fs/iomap/direct-io.c
> index 9398b8c31323..a6aaea2764a5 100644
> --- a/fs/iomap/direct-io.c
> +++ b/fs/iomap/direct-io.c
> @@ -378,23 +378,25 @@ iomap_dio_inline_actor(struct inode *inode, loff_t pos, loff_t length,
> struct iomap_dio *dio, struct iomap *iomap)
> {
> struct iov_iter *iter = dio->submit.iter;
> + void *dst = iomap_inline_buf(iomap, pos);
> size_t copied;
>
> - BUG_ON(pos + length > PAGE_SIZE - offset_in_page(iomap->inline_data));
> + if (WARN_ON_ONCE(!iomap_inline_data_size_valid(iomap)))
> + return -EIO;
>
> if (dio->flags & IOMAP_DIO_WRITE) {
> loff_t size = inode->i_size;
>
> if (pos > size)
> - memset(iomap->inline_data + size, 0, pos - size);
> - copied = copy_from_iter(iomap->inline_data + pos, length, iter);
> + memset(iomap_inline_buf(iomap, size), 0, pos - size);
> + copied = copy_from_iter(dst, length, iter);
> if (copied) {
> if (pos + copied > size)
> i_size_write(inode, pos + copied);
> mark_inode_dirty(inode);
> }
> } else {
> - copied = copy_to_iter(iomap->inline_data + pos, length, iter);
> + copied = copy_to_iter(dst, length, iter);
> }
> dio->size += copied;
> return copied;
> diff --git a/include/linux/iomap.h b/include/linux/iomap.h
> index 479c1da3e221..56b118c6d05c 100644
> --- a/include/linux/iomap.h
> +++ b/include/linux/iomap.h
> @@ -97,6 +97,20 @@ iomap_sector(struct iomap *iomap, loff_t pos)
> return (iomap->addr + pos - iomap->offset) >> SECTOR_SHIFT;
> }
>
> +static inline void *iomap_inline_buf(const struct iomap *iomap, loff_t pos)
> +{
> + return iomap->inline_data - iomap->offset + pos;
> +}
> +
> +/*
> + * iomap->inline_data is a potentially kmapped page, ensure it never crosses a
> + * page boundary.
> + */
> +static inline bool iomap_inline_data_size_valid(const struct iomap *iomap)
> +{
> + return iomap->length <= PAGE_SIZE - offset_in_page(iomap->inline_data);
> +}
> +
> /*
> * When a filesystem sets page_ops in an iomap mapping it returns, page_prepare
> * and page_done will be called for each page written to. This only applies to
>
More information about the Linux-erofs
mailing list