On Mon, May 26, 2008 at 07:21:24PM +0200, Jan Kara wrote:
> On Wed 21-05-08 13:51:09, Aneesh Kumar K.V wrote:
[....snip.....]
> > {
> > @@ -3837,7 +3850,7 @@ int ext4_page_mkwrite(struct vm_area_struct *vma, struct page *page)
> > if (ext4_should_writeback_data(inode))
> > ret = __ext4_writeback_writepage(page, &wbc);
> > else if (ext4_should_order_data(inode))
> > - ret = __ext4_ordered_writepage(page, &wbc);
> > + ret = __ext4_ordered_alloc_and_writepage(page, &wbc, 1);
> > else
> > ret = __ext4_journalled_writepage(page, &wbc);
> > /* Page got unlocked in writepage */
> >
> >
> >
> > ie we call __ext4_ordered_alloc_and_writepage with alloc = 1 only in
> > case of page_mkwrite. All the other case we should have all the buffer
> > heads mapped. Otherwise we will try to allocate new blocks which starts
> > a new transaction holding page lock.
> When do we try to allocate new blocks in writepage now? ext4_page_mkwrite()
> should have done the allocation before writepage() was called so there
> should be no need to allocate anything... But maybe I miss something.
That's what i also meant by the above changes. The block are allocated
only in ext4_page_mkwrite and not during writepage. So calling
ext4_*_writepage during mkwrite confuse quiet a lot. Instead i was
trying to make it explicit by making page_mkwrite call
ext4_ordered_alloc_and_writepage and by adding BUG() in writepage
callback if it ever get called by an unmapped buffer.
I have got another question now related to page_mkwrite. AFAIU writepage
writeout dirty buffer_heads. It also looks at whether the pages are
dirty or not. In the page_mkwrite callback both are not true. ie we call
set_page_dirty from do_wp_page after calling page_mkwrite. I haven't
verified whether the above is correct or not. Just thinking reading the
code.
>
> > > -static int ext4_writeback_writepage(struct page *page,
> > > +static int __ext4_writeback_writepage(struct page *page,
> > > struct writeback_control *wbc)
> > > {
> > > struct inode *inode = page->mapping->host;
> > > +
> > > + if (test_opt(inode->i_sb, NOBH))
> > > + return nobh_writepage(page, ext4_get_block, wbc);
> > > + else
> > > + return block_write_full_page(page, ext4_get_block, wbc);
> > > +}
> > > +
> > > +
> > > +static int ext4_writeback_writepage(struct page *page,
> > > + struct writeback_control *wbc)
> > > +{
> > > + if (!ext4_journal_current_handle())
> > > + return __ext4_writeback_writepage(page, wbc);
> > > +
> > > + redirty_page_for_writepage(wbc, page);
> > > + unlock_page(page);
> > > + return 0;
> > > +}
> > > +
> > > +static int __ext4_journalled_writepage(struct page *page,
> > > + struct writeback_control *wbc)
> > > +{
> > > + struct address_space *mapping = page->mapping;
> > > + struct inode *inode = mapping->host;
> > > + struct buffer_head *page_bufs;
> > > handle_t *handle = NULL;
> > > int ret = 0;
> > > int err;
> > >
> > > - if (ext4_journal_current_handle())
> > > - goto out_fail;
> > > + ret = block_prepare_write(page, 0, PAGE_CACHE_SIZE, ext4_get_block);
> > > + if (ret != 0)
> > > + goto out_unlock;
> > > +
> > > + page_bufs = page_buffers(page);
> > > + walk_page_buffers(handle, page_bufs, 0, PAGE_CACHE_SIZE, NULL,
> > > + bget_one);
> > > + /* As soon as we unlock the page, it can go away, but we have
> > > + * references to buffers so we are safe */
> > > + unlock_page(page);
> > >
> > > handle = ext4_journal_start(inode, ext4_writepage_trans_blocks(inode));
> > > if (IS_ERR(handle)) {
> > > ret = PTR_ERR(handle);
> > > - goto out_fail;
> > > + goto out;
> > > }
> > >
> > > - if (test_opt(inode->i_sb, NOBH) && ext4_should_writeback_data(inode))
> > > - ret = nobh_writepage(page, ext4_get_block, wbc);
> > > - else
> > > - ret = block_write_full_page(page, ext4_get_block, wbc);
> > > + ret = walk_page_buffers(handle, page_bufs, 0,
> > > + PAGE_CACHE_SIZE, NULL, do_journal_get_write_access);
> > >
> > > + err = walk_page_buffers(handle, page_bufs, 0,
> > > + PAGE_CACHE_SIZE, NULL, write_end_fn);
> > > + if (ret == 0)
> > > + ret = err;
> > > err = ext4_journal_stop(handle);
> > > if (!ret)
> > > ret = err;
> > > - return ret;
> > >
> > > -out_fail:
> > > - redirty_page_for_writepage(wbc, page);
> > > + walk_page_buffers(handle, page_bufs, 0,
> > > + PAGE_CACHE_SIZE, NULL, bput_one);
> > > + EXT4_I(inode)->i_state |= EXT4_STATE_JDATA;
> > > + goto out;
> > > +
> > > +out_unlock:
> > > unlock_page(page);
> > > +out:
> > > return ret;
> > > }
> > >
> > > static int ext4_journalled_writepage(struct page *page,
> > > struct writeback_control *wbc)
> > > {
> > > - struct inode *inode = page->mapping->host;
> > > - handle_t *handle = NULL;
> > > - int ret = 0;
> > > - int err;
> > > -
> > > if (ext4_journal_current_handle())
> > > goto no_write;
> > >
> > > - handle = ext4_journal_start(inode, ext4_writepage_trans_blocks(inode));
> > > - if (IS_ERR(handle)) {
> > > - ret = PTR_ERR(handle);
> > > - goto no_write;
> > > - }
> > > -
> > > if (!page_has_buffers(page) || PageChecked(page)) {
> >
> >
> > This will never happen with writepage right ? And we don't call
> > ext4_journalled_writepage from page_mkwrite. So is this needed ?
> > If not __ext4_journalled_writepage can handle everything in a single
> > transaction right and assume that it is called within a transaction.
> I'm not sure I understand you. PageChecked() can happen from writepage
> call path. We set PageChecked() when we do set_page_dirty() as far as I
> remember... Basically, we use this flag to decide whether writepage should
> do checkpointing or write into the journal.
What i meant by the above question was can ext4_journalled_writepage get
called with page_buffers == NULL
So the check if (!page_has_buffers(page)) can go away right ?
I have posted some changes after this at
http://article.gmane.org/gmane.comp.file-systems.ext4/6768
Message-Id: <1211391859-17399-1-git-send-email-aneesh.kumar@xxxxxxxxxxxxxxxxxx>
-aneesh
--
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
