On Wed, 15 Dec 2010, Sage Weil wrote:
> From: Henry C Chang <henry_c_chang@xxxxxxxxxxxxxxxxxxx>
>
> The user buffer may be 512-byte aligned, not page-aligned. We were
> assuming the buffer was page-aligned and only accounting for
> non-page-aligned io offsets.
>
> Signed-off-by: Henry C Chang <henry_c_chang@xxxxxxxxxxxxxxxxxxx>
This is unchanged, except for some minor type changes (u64 to unsigned
long). Unfortunately I can't seem to find the test I was using for the
other directio alignment bugs (which varied io but not buffer alignment).
Anyway, it's in the master branch. If it looks okay I can send this to
Linus for 2.6.37.
sage
> Signed-off-by: Sage Weil <sage@xxxxxxxxxxxx>
> ---
> fs/ceph/file.c | 31 +++++++++++++++++++------------
> 1 files changed, 19 insertions(+), 12 deletions(-)
>
> diff --git a/fs/ceph/file.c b/fs/ceph/file.c
> index 8d79b89..e860d8f 100644
> --- a/fs/ceph/file.c
> +++ b/fs/ceph/file.c
> @@ -282,7 +282,8 @@ int ceph_release(struct inode *inode, struct file *file)
> static int striped_read(struct inode *inode,
> u64 off, u64 len,
> struct page **pages, int num_pages,
> - int *checkeof, bool align_to_pages)
> + int *checkeof, bool align_to_pages,
> + unsigned long buf_align)
> {
> struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
> struct ceph_inode_info *ci = ceph_inode(inode);
> @@ -307,7 +308,7 @@ static int striped_read(struct inode *inode,
>
> more:
> if (align_to_pages)
> - page_align = (pos - io_align) & ~PAGE_MASK;
> + page_align = (pos - io_align + buf_align) & ~PAGE_MASK;
> else
> page_align = pos & ~PAGE_MASK;
> this_len = left;
> @@ -376,16 +377,18 @@ static ssize_t ceph_sync_read(struct file *file, char __user *data,
> struct inode *inode = file->f_dentry->d_inode;
> struct page **pages;
> u64 off = *poff;
> - int num_pages = calc_pages_for(off, len);
> - int ret;
> + int num_pages, ret;
>
> dout("sync_read on file %p %llu~%u %s\n", file, off, len,
> (file->f_flags & O_DIRECT) ? "O_DIRECT" : "");
>
> - if (file->f_flags & O_DIRECT)
> + if (file->f_flags & O_DIRECT) {
> + num_pages = calc_pages_for((unsigned long)data, len);
> pages = ceph_get_direct_page_vector(data, num_pages);
> - else
> + } else {
> + num_pages = calc_pages_for(off, len);
> pages = ceph_alloc_page_vector(num_pages, GFP_NOFS);
> + }
> if (IS_ERR(pages))
> return PTR_ERR(pages);
>
> @@ -400,7 +403,8 @@ static ssize_t ceph_sync_read(struct file *file, char __user *data,
> goto done;
>
> ret = striped_read(inode, off, len, pages, num_pages, checkeof,
> - file->f_flags & O_DIRECT);
> + file->f_flags & O_DIRECT,
> + (unsigned long)data & ~PAGE_MASK);
>
> if (ret >= 0 && (file->f_flags & O_DIRECT) == 0)
> ret = ceph_copy_page_vector_to_user(pages, data, off, ret);
> @@ -456,6 +460,7 @@ static ssize_t ceph_sync_write(struct file *file, const char __user *data,
> int do_sync = 0;
> int check_caps = 0;
> int page_align, io_align;
> + unsigned long buf_align;
> int ret;
> struct timespec mtime = CURRENT_TIME;
>
> @@ -471,6 +476,7 @@ static ssize_t ceph_sync_write(struct file *file, const char __user *data,
> pos = *offset;
>
> io_align = pos & ~PAGE_MASK;
> + buf_align = (unsigned long)data & ~PAGE_MASK;
>
> ret = filemap_write_and_wait_range(inode->i_mapping, pos, pos + left);
> if (ret < 0)
> @@ -496,12 +502,15 @@ static ssize_t ceph_sync_write(struct file *file, const char __user *data,
> */
> more:
> len = left;
> - if (file->f_flags & O_DIRECT)
> + if (file->f_flags & O_DIRECT) {
> /* write from beginning of first page, regardless of
> io alignment */
> - page_align = (pos - io_align) & ~PAGE_MASK;
> - else
> + page_align = (pos - io_align + buf_align) & ~PAGE_MASK;
> + num_pages = calc_pages_for((unsigned long)data, len);
> + } else {
> page_align = pos & ~PAGE_MASK;
> + num_pages = calc_pages_for(pos, len);
> + }
> req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
> ceph_vino(inode), pos, &len,
> CEPH_OSD_OP_WRITE, flags,
> @@ -512,8 +521,6 @@ more:
> if (!req)
> return -ENOMEM;
>
> - num_pages = calc_pages_for(pos, len);
> -
> if (file->f_flags & O_DIRECT) {
> pages = ceph_get_direct_page_vector(data, num_pages);
> if (IS_ERR(pages)) {
> --
> 1.7.1
>
> --
> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
>
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
