aops.c

linux 内核源代码

	if (is_mft && sync) {
do_mirror:
		for (i = 0; i < nr_bhs; i++) {
			unsigned long mft_no;
			unsigned int ofs;

			/*
			 * Skip buffers which are not at the beginning of
			 * records.
			 */
			if (i % bhs_per_rec)
				continue;
			tbh = bhs[i];
			/* Skip removed buffers (and hence records). */
			if (!tbh)
				continue;
			ofs = bh_offset(tbh);
			/* Get the mft record number. */
			mft_no = (((s64)page->index << PAGE_CACHE_SHIFT) + ofs)
					>> rec_size_bits;
			if (mft_no < vol->mftmirr_size)
				ntfs_sync_mft_mirror(vol, mft_no,
						(MFT_RECORD*)(kaddr + ofs),
						sync);
		}
		if (!sync)
			goto do_wait;
	}
	/* Remove the mst protection fixups again. */
	for (i = 0; i < nr_bhs; i++) {
		if (!(i % bhs_per_rec)) {
			tbh = bhs[i];
			if (!tbh)
				continue;
			post_write_mst_fixup((NTFS_RECORD*)(kaddr +
					bh_offset(tbh)));
		}
	}
	flush_dcache_page(page);
unm_done:
	/* Unlock any locked inodes. */
	while (nr_locked_nis-- > 0) {
		ntfs_inode *tni, *base_tni;
		
		tni = locked_nis[nr_locked_nis];
		/* Get the base inode. */
		mutex_lock(&tni->extent_lock);
		if (tni->nr_extents >= 0)
			base_tni = tni;
		else {
			base_tni = tni->ext.base_ntfs_ino;
			BUG_ON(!base_tni);
		}
		mutex_unlock(&tni->extent_lock);
		ntfs_debug("Unlocking %s inode 0x%lx.",
				tni == base_tni ? "base" : "extent",
				tni->mft_no);
		mutex_unlock(&tni->mrec_lock);
		atomic_dec(&tni->count);
		iput(VFS_I(base_tni));
	}
	SetPageUptodate(page);
	kunmap(page);
done:
	if (unlikely(err && err != -ENOMEM)) {
		/*
		 * Set page error if there is only one ntfs record in the page.
		 * Otherwise we would loose per-record granularity.
		 */
		if (ni->itype.index.block_size == PAGE_CACHE_SIZE)
			SetPageError(page);
		NVolSetErrors(vol);
	}
	if (page_is_dirty) {
		ntfs_debug("Page still contains one or more dirty ntfs "
				"records.  Redirtying the page starting at "
				"record 0x%lx.", page->index <<
				(PAGE_CACHE_SHIFT - rec_size_bits));
		redirty_page_for_writepage(wbc, page);
		unlock_page(page);
	} else {
		/*
		 * Keep the VM happy.  This must be done otherwise the
		 * radix-tree tag PAGECACHE_TAG_DIRTY remains set even though
		 * the page is clean.
		 */
		BUG_ON(PageWriteback(page));
		set_page_writeback(page);
		unlock_page(page);
		end_page_writeback(page);
	}
	if (likely(!err))
		ntfs_debug("Done.");
	return err;
}

/**
 * ntfs_writepage - write a @page to the backing store
 * @page:	page cache page to write out
 * @wbc:	writeback control structure
 *
 * This is called from the VM when it wants to have a dirty ntfs page cache
 * page cleaned.  The VM has already locked the page and marked it clean.
 *
 * For non-resident attributes, ntfs_writepage() writes the @page by calling
 * the ntfs version of the generic block_write_full_page() function,
 * ntfs_write_block(), which in turn if necessary creates and writes the
 * buffers associated with the page asynchronously.
 *
 * For resident attributes, OTOH, ntfs_writepage() writes the @page by copying
 * the data to the mft record (which at this stage is most likely in memory).
 * The mft record is then marked dirty and written out asynchronously via the
 * vfs inode dirty code path for the inode the mft record belongs to or via the
 * vm page dirty code path for the page the mft record is in.
 *
 * Based on ntfs_readpage() and fs/buffer.c::block_write_full_page().
 *
 * Return 0 on success and -errno on error.
 */
static int ntfs_writepage(struct page *page, struct writeback_control *wbc)
{
	loff_t i_size;
	struct inode *vi = page->mapping->host;
	ntfs_inode *base_ni = NULL, *ni = NTFS_I(vi);
	char *addr;
	ntfs_attr_search_ctx *ctx = NULL;
	MFT_RECORD *m = NULL;
	u32 attr_len;
	int err;

retry_writepage:
	BUG_ON(!PageLocked(page));
	i_size = i_size_read(vi);
	/* Is the page fully outside i_size? (truncate in progress) */
	if (unlikely(page->index >= (i_size + PAGE_CACHE_SIZE - 1) >>
			PAGE_CACHE_SHIFT)) {
		/*
		 * The page may have dirty, unmapped buffers.  Make them
		 * freeable here, so the page does not leak.
		 */
		block_invalidatepage(page, 0);
		unlock_page(page);
		ntfs_debug("Write outside i_size - truncated?");
		return 0;
	}
	/*
	 * Only $DATA attributes can be encrypted and only unnamed $DATA
	 * attributes can be compressed.  Index root can have the flags set but
	 * this means to create compressed/encrypted files, not that the
	 * attribute is compressed/encrypted.  Note we need to check for
	 * AT_INDEX_ALLOCATION since this is the type of both directory and
	 * index inodes.
	 */
	if (ni->type != AT_INDEX_ALLOCATION) {
		/* If file is encrypted, deny access, just like NT4. */
		if (NInoEncrypted(ni)) {
			unlock_page(page);
			BUG_ON(ni->type != AT_DATA);
			ntfs_debug("Denying write access to encrypted file.");
			return -EACCES;
		}
		/* Compressed data streams are handled in compress.c. */
		if (NInoNonResident(ni) && NInoCompressed(ni)) {
			BUG_ON(ni->type != AT_DATA);
			BUG_ON(ni->name_len);
			// TODO: Implement and replace this with
			// return ntfs_write_compressed_block(page);
			unlock_page(page);
			ntfs_error(vi->i_sb, "Writing to compressed files is "
					"not supported yet.  Sorry.");
			return -EOPNOTSUPP;
		}
		// TODO: Implement and remove this check.
		if (NInoNonResident(ni) && NInoSparse(ni)) {
			unlock_page(page);
			ntfs_error(vi->i_sb, "Writing to sparse files is not "
					"supported yet.  Sorry.");
			return -EOPNOTSUPP;
		}
	}
	/* NInoNonResident() == NInoIndexAllocPresent() */
	if (NInoNonResident(ni)) {
		/* We have to zero every time due to mmap-at-end-of-file. */
		if (page->index >= (i_size >> PAGE_CACHE_SHIFT)) {
			/* The page straddles i_size. */
			unsigned int ofs = i_size & ~PAGE_CACHE_MASK;
			zero_user_page(page, ofs, PAGE_CACHE_SIZE - ofs,
					KM_USER0);
		}
		/* Handle mst protected attributes. */
		if (NInoMstProtected(ni))
			return ntfs_write_mst_block(page, wbc);
		/* Normal, non-resident data stream. */
		return ntfs_write_block(page, wbc);
	}
	/*
	 * Attribute is resident, implying it is not compressed, encrypted, or
	 * mst protected.  This also means the attribute is smaller than an mft
	 * record and hence smaller than a page, so can simply return error on
	 * any pages with index above 0.  Note the attribute can actually be
	 * marked compressed but if it is resident the actual data is not
	 * compressed so we are ok to ignore the compressed flag here.
	 */
	BUG_ON(page_has_buffers(page));
	BUG_ON(!PageUptodate(page));
	if (unlikely(page->index > 0)) {
		ntfs_error(vi->i_sb, "BUG()! page->index (0x%lx) > 0.  "
				"Aborting write.", page->index);
		BUG_ON(PageWriteback(page));
		set_page_writeback(page);
		unlock_page(page);
		end_page_writeback(page);
		return -EIO;
	}
	if (!NInoAttr(ni))
		base_ni = ni;
	else
		base_ni = ni->ext.base_ntfs_ino;
	/* Map, pin, and lock the mft record. */
	m = map_mft_record(base_ni);
	if (IS_ERR(m)) {
		err = PTR_ERR(m);
		m = NULL;
		ctx = NULL;
		goto err_out;
	}
	/*
	 * If a parallel write made the attribute non-resident, drop the mft
	 * record and retry the writepage.
	 */
	if (unlikely(NInoNonResident(ni))) {
		unmap_mft_record(base_ni);
		goto retry_writepage;
	}
	ctx = ntfs_attr_get_search_ctx(base_ni, m);
	if (unlikely(!ctx)) {
		err = -ENOMEM;
		goto err_out;
	}
	err = ntfs_attr_lookup(ni->type, ni->name, ni->name_len,
			CASE_SENSITIVE, 0, NULL, 0, ctx);
	if (unlikely(err))
		goto err_out;
	/*
	 * Keep the VM happy.  This must be done otherwise the radix-tree tag
	 * PAGECACHE_TAG_DIRTY remains set even though the page is clean.
	 */
	BUG_ON(PageWriteback(page));
	set_page_writeback(page);
	unlock_page(page);
	attr_len = le32_to_cpu(ctx->attr->data.resident.value_length);
	i_size = i_size_read(vi);
	if (unlikely(attr_len > i_size)) {
		/* Race with shrinking truncate or a failed truncate. */
		attr_len = i_size;
		/*
		 * If the truncate failed, fix it up now.  If a concurrent
		 * truncate, we do its job, so it does not have to do anything.
		 */
		err = ntfs_resident_attr_value_resize(ctx->mrec, ctx->attr,
				attr_len);
		/* Shrinking cannot fail. */
		BUG_ON(err);
	}
	addr = kmap_atomic(page, KM_USER0);
	/* Copy the data from the page to the mft record. */
	memcpy((u8*)ctx->attr +
			le16_to_cpu(ctx->attr->data.resident.value_offset),
			addr, attr_len);
	/* Zero out of bounds area in the page cache page. */
	memset(addr + attr_len, 0, PAGE_CACHE_SIZE - attr_len);
	kunmap_atomic(addr, KM_USER0);
	flush_dcache_page(page);
	flush_dcache_mft_record_page(ctx->ntfs_ino);
	/* We are done with the page. */
	end_page_writeback(page);
	/* Finally, mark the mft record dirty, so it gets written back. */
	mark_mft_record_dirty(ctx->ntfs_ino);
	ntfs_attr_put_search_ctx(ctx);
	unmap_mft_record(base_ni);
	return 0;
err_out:
	if (err == -ENOMEM) {
		ntfs_warning(vi->i_sb, "Error allocating memory. Redirtying "
				"page so we try again later.");
		/*
		 * Put the page back on mapping->dirty_pages, but leave its
		 * buffers' dirty state as-is.
		 */
		redirty_page_for_writepage(wbc, page);
		err = 0;
	} else {
		ntfs_error(vi->i_sb, "Resident attribute write failed with "
				"error %i.", err);
		SetPageError(page);
		NVolSetErrors(ni->vol);
	}
	unlock_page(page);
	if (ctx)
		ntfs_attr_put_search_ctx(ctx);
	if (m)
		unmap_mft_record(base_ni);
	return err;
}

#endif	/* NTFS_RW */

/**
 * ntfs_aops - general address space operations for inodes and attributes
 */
const struct address_space_operations ntfs_aops = {
	.readpage	= ntfs_readpage,	/* Fill page with data. */
	.sync_page	= block_sync_page,	/* Currently, just unplugs the
						   disk request queue. */
#ifdef NTFS_RW
	.writepage	= ntfs_writepage,	/* Write dirty page to disk. */
#endif /* NTFS_RW */
	.migratepage	= buffer_migrate_page,	/* Move a page cache page from
						   one physical page to an
						   other. */
};

/**
 * ntfs_mst_aops - general address space operations for mst protecteed inodes
 *		   and attributes
 */
const struct address_space_operations ntfs_mst_aops = {
	.readpage	= ntfs_readpage,	/* Fill page with data. */
	.sync_page	= block_sync_page,	/* Currently, just unplugs the
						   disk request queue. */
#ifdef NTFS_RW
	.writepage	= ntfs_writepage,	/* Write dirty page to disk. */
	.set_page_dirty	= __set_page_dirty_nobuffers,	/* Set the page dirty
						   without touching the buffers
						   belonging to the page. */
#endif /* NTFS_RW */
	.migratepage	= buffer_migrate_page,	/* Move a page cache page from
						   one physical page to an
						   other. */
};

#ifdef NTFS_RW

/**
 * mark_ntfs_record_dirty - mark an ntfs record dirty
 * @page:	page containing the ntfs record to mark dirty
 * @ofs:	byte offset within @page at which the ntfs record begins
 *
 * Set the buffers and the page in which the ntfs record is located dirty.
 *
 * The latter also marks the vfs inode the ntfs record belongs to dirty
 * (I_DIRTY_PAGES only).
 *
 * If the page does not have buffers, we create them and set them uptodate.
 * The page may not be locked which is why we need to handle the buffers under
 * the mapping->private_lock.  Once the buffers are marked dirty we no longer
 * need the lock since try_to_free_buffers() does not free dirty buffers.
 */
void mark_ntfs_record_dirty(struct page *page, const unsigned int ofs) {
	struct address_space *mapping = page->mapping;
	ntfs_inode *ni = NTFS_I(mapping->host);
	struct buffer_head *bh, *head, *buffers_to_free = NULL;
	unsigned int end, bh_size, bh_ofs;

	BUG_ON(!PageUptodate(page));
	end = ofs + ni->itype.index.block_size;
	bh_size = VFS_I(ni)->i_sb->s_blocksize;
	spin_lock(&mapping->private_lock);
	if (unlikely(!page_has_buffers(page))) {
		spin_unlock(&mapping->private_lock);
		bh = head = alloc_page_buffers(page, bh_size, 1);
		spin_lock(&mapping->private_lock);
		if (likely(!page_has_buffers(page))) {
			struct buffer_head *tail;

			do {
				set_buffer_uptodate(bh);
				tail = bh;
				bh = bh->b_this_page;
			} while (bh);
			tail->b_this_page = head;
			attach_page_buffers(page, head);
		} else
			buffers_to_free = bh;
	}
	bh = head = page_buffers(page);
	BUG_ON(!bh);
	do {
		bh_ofs = bh_offset(bh);
		if (bh_ofs + bh_size <= ofs)
			continue;
		if (unlikely(bh_ofs >= end))
			break;
		set_buffer_dirty(bh);
	} while ((bh = bh->b_this_page) != head);
	spin_unlock(&mapping->private_lock);
	__set_page_dirty_nobuffers(page);
	if (unlikely(buffers_to_free)) {
		do {
			bh = buffers_to_free->b_this_page;
			free_buffer_head(buffers_to_free);
			buffers_to_free = bh;
		} while (buffers_to_free);
	}
}

#endif /* NTFS_RW */