在 2008-08-25一的 16:50 +0530，Aneesh Kumar K.V写道：
> delayed allocation allocate blocks during writepages. That also
> means we cannot handle block allocation failures. Switch to
> non - delalloc when we are running low on free blocks.
> Delayed allocation need to do aggressive meta-data block reservation
> considering that the requested blocks can all be discontiguous.
> Switching to non-delalloc avoids that. Also we can satisfy
> partial write in non-delalloc mode.
>
> Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@xxxxxxxxxxxxxxxxxx>
> ---
> fs/ext4/inode.c | 48 ++++++++++++++++++++++++++++++++++++++++++++++--
> 1 files changed, 46 insertions(+), 2 deletions(-)
>
> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> index 3f3ecc0..d923a14 100644
> --- a/fs/ext4/inode.c
> +++ b/fs/ext4/inode.c
> @@ -2482,6 +2482,29 @@ static int ext4_da_writepages(struct address_space *mapping,
> return ret;
> }
>
> +#define FALL_BACK_TO_NONDELALLOC 1
> +static int ext4_nonda_switch(struct super_block *sb)
> +{
> + s64 free_blocks, dirty_blocks;
> + struct ext4_sb_info *sbi = EXT4_SB(sb);
> +
> + /*
> + * switch to non delalloc mode if we are running low
> + * on free block. The free block accounting via percpu
> + * counters can get slightly wrong with FBC_BATCH getting
> + * accumulated on each CPU without updating global counters
> + * Delalloc need an accurate free block accounting. So switch
> + * to non delalloc when we are near to error range.
> + */
> + free_blocks = percpu_counter_read_positive(&sbi->s_freeblocks_counter);
> + dirty_blocks = percpu_counter_read_positive(&sbi->s_dirtyblocks_counter);
> + if ( 2 * free_blocks < 3 * dirty_blocks) {
> + /* free block count is less that 150% of dirty blocks */
> + return 1;
> + }
In the case the free_blocks is below the EXT4_FREEBLOCKS_WATERMARK, we
should turn back to nondelalloc mode, even if there is no dirty_blocks.
> + return 0;
> +}
> +
> static int ext4_da_write_begin(struct file *file, struct address_space *mapping,
> loff_t pos, unsigned len, unsigned flags,
> struct page **pagep, void **fsdata)
> @@ -2496,6 +2519,13 @@ static int ext4_da_write_begin(struct file *file, struct address_space *mapping,
> index = pos >> PAGE_CACHE_SHIFT;
> from = pos & (PAGE_CACHE_SIZE - 1);
> to = from + len;
> +
> + if (ext4_nonda_switch(inode->i_sb)) {
> + *fsdata = (void *)FALL_BACK_TO_NONDELALLOC;
> + return ext4_write_begin(file, mapping, pos,
> + len, flags, pagep, fsdata);
> + }
> + *fsdata = (void *)0;
We probably should add a warning if *fsdata is non 0, instead of forcing
it reset to 0 unconditionally.
> retry:
> /*
> * With delayed allocation, we don't log the i_disksize update
> @@ -2564,6 +2594,19 @@ static int ext4_da_write_end(struct file *file,
> handle_t *handle = ext4_journal_current_handle();
> loff_t new_i_size;
> unsigned long start, end;
> + int write_mode = (int)fsdata;
> +
> + if (write_mode == FALL_BACK_TO_NONDELALLOC) {
> + if (ext4_should_order_data(inode)) {
> + return ext4_ordered_write_end(file, mapping, pos,
> + len, copied, page, fsdata);
> + } else if (ext4_should_writeback_data(inode)) {
> + return ext4_writeback_write_end(file, mapping, pos,
> + len, copied, page, fsdata);
> + } else {
> + BUG();
Shouldn't we warnining user that we can't fall back to journalled mode
instead, let it continue with delalloc mode, instead of BUG() the
system?
> + }
> + }
>
> start = pos & (PAGE_CACHE_SIZE - 1);
> end = start + copied -1;
> @@ -4901,6 +4944,7 @@ int ext4_page_mkwrite(struct vm_area_struct *vma, struct page *page)
> loff_t size;
> unsigned long len;
> int ret = -EINVAL;
> + void *fsdata;
> struct file *file = vma->vm_file;
> struct inode *inode = file->f_path.dentry->d_inode;
> struct address_space *mapping = inode->i_mapping;
> @@ -4939,11 +4983,11 @@ int ext4_page_mkwrite(struct vm_area_struct *vma, struct page *page)
> * on the same page though
> */
> ret = mapping->a_ops->write_begin(file, mapping, page_offset(page),
> - len, AOP_FLAG_UNINTERRUPTIBLE, &page, NULL);
> + len, AOP_FLAG_UNINTERRUPTIBLE, &page, &fsdata);
> if (ret < 0)
> goto out_unlock;
> ret = mapping->a_ops->write_end(file, mapping, page_offset(page),
> - len, len, page, NULL);
> + len, len, page, fsdata);
> if (ret < 0)
> goto out_unlock;
> ret = 0;
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
