direct.c

Linux Kernel 2.6.9 for OMAP1710

			memcpy(&first_verf.verifier, &wdata.verf.verifier,
								VERF_SIZE);
		if (wdata.verf.committed != NFS_FILE_SYNC) {
			need_commit = 1;
			if (memcmp(&first_verf.verifier,
					&wdata.verf.verifier, VERF_SIZE))
				goto sync_retry;
		}

                tot_bytes += result;
                wdata.args.offset += result;
		wdata.args.pgbase += result;
		curpage += wdata.args.pgbase >> PAGE_SHIFT;
		wdata.args.pgbase &= ~PAGE_MASK;
		request -= result;
	} while (request != 0);

	/*
	 * Commit data written so far, even in the event of an error
	 */
	if (need_commit) {
		wdata.args.count = tot_bytes;
		wdata.args.offset = file_offset;

		lock_kernel();
		result = NFS_PROTO(inode)->commit(&wdata);
		unlock_kernel();

		if (result < 0 || memcmp(&first_verf.verifier,
						&wdata.verf.verifier,
						VERF_SIZE) != 0)
			goto sync_retry;
	}
	result = tot_bytes;

out:
	nfs_end_data_update_defer(inode);

	return result;

sync_retry:
	wdata.args.stable = NFS_FILE_SYNC;
	goto retry;
}

/**
 * nfs_direct_write - For each iov segment, map the user's buffer
 *                    then generate write and commit RPCs.
 * @inode: target inode
 * @ctx: target file open context
 * @iov: array of vectors that define I/O buffer
 * file_offset: offset in file to begin the operation
 * nr_segs: size of iovec array
 *
 * Upon return, generic_file_direct_IO invalidates any cached pages
 * that non-direct readers might access, so they will pick up these
 * writes immediately.
 */
static int nfs_direct_write(struct inode *inode, struct nfs_open_context *ctx,
		const struct iovec *iov, loff_t file_offset,
		unsigned long nr_segs)
{
	ssize_t tot_bytes = 0;
	unsigned long seg = 0;

	while ((seg < nr_segs) && (tot_bytes >= 0)) {
		ssize_t result;
		int page_count;
		struct page **pages;
		const struct iovec *vec = &iov[seg++];
		unsigned long user_addr = (unsigned long) vec->iov_base;
		size_t size = vec->iov_len;

                page_count = nfs_get_user_pages(WRITE, user_addr, size, &pages);
                if (page_count < 0) {
                        nfs_free_user_pages(pages, 0, 0);
			if (tot_bytes > 0)
				break;
                        return page_count;
                }

		result = nfs_direct_write_seg(inode, ctx, user_addr, size,
				file_offset, pages, page_count);
		nfs_free_user_pages(pages, page_count, 0);

		if (result <= 0) {
			if (tot_bytes > 0)
				break;
			return result;
		}
		tot_bytes += result;
		file_offset += result;
		if (result < size)
			break;
	}
	return tot_bytes;
}

/**
 * nfs_direct_IO - NFS address space operation for direct I/O
 * rw: direction (read or write)
 * @iocb: target I/O control block
 * @iov: array of vectors that define I/O buffer
 * file_offset: offset in file to begin the operation
 * nr_segs: size of iovec array
 *
 */
ssize_t
nfs_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov,
		loff_t file_offset, unsigned long nr_segs)
{
	ssize_t result = -EINVAL;
	struct file *file = iocb->ki_filp;
	struct nfs_open_context *ctx;
	struct dentry *dentry = file->f_dentry;
	struct inode *inode = dentry->d_inode;

	/*
	 * No support for async yet
	 */
	if (!is_sync_kiocb(iocb))
		return result;

	ctx = (struct nfs_open_context *)file->private_data;
	switch (rw) {
	case READ:
		dprintk("NFS: direct_IO(read) (%s) off/no(%Lu/%lu)\n",
				dentry->d_name.name, file_offset, nr_segs);

		result = nfs_direct_read(inode, ctx, iov,
						file_offset, nr_segs);
		break;
	case WRITE:
		dprintk("NFS: direct_IO(write) (%s) off/no(%Lu/%lu)\n",
				dentry->d_name.name, file_offset, nr_segs);

		result = nfs_direct_write(inode, ctx, iov,
						file_offset, nr_segs);
		break;
	default:
		break;
	}
	return result;
}

/**
 * nfs_file_direct_read - file direct read operation for NFS files
 * @iocb: target I/O control block
 * @buf: user's buffer into which to read data
 * count: number of bytes to read
 * pos: byte offset in file where reading starts
 *
 * We use this function for direct reads instead of calling
 * generic_file_aio_read() in order to avoid gfar's check to see if
 * the request starts before the end of the file.  For that check
 * to work, we must generate a GETATTR before each direct read, and
 * even then there is a window between the GETATTR and the subsequent
 * READ where the file size could change.  So our preference is simply
 * to do all reads the application wants, and the server will take
 * care of managing the end of file boundary.
 * 
 * This function also eliminates unnecessarily updating the file's
 * atime locally, as the NFS server sets the file's atime, and this
 * client must read the updated atime from the server back into its
 * cache.
 */
ssize_t
nfs_file_direct_read(struct kiocb *iocb, char __user *buf, size_t count, loff_t pos)
{
	ssize_t retval = -EINVAL;
	loff_t *ppos = &iocb->ki_pos;
	struct file *file = iocb->ki_filp;
	struct nfs_open_context *ctx =
			(struct nfs_open_context *) file->private_data;
	struct dentry *dentry = file->f_dentry;
	struct address_space *mapping = file->f_mapping;
	struct inode *inode = mapping->host;
	struct iovec iov = {
		.iov_base = buf,
		.iov_len = count,
	};

	dprintk("nfs: direct read(%s/%s, %lu@%lu)\n",
		dentry->d_parent->d_name.name, dentry->d_name.name,
		(unsigned long) count, (unsigned long) pos);

	if (!is_sync_kiocb(iocb))
		goto out;
	if (count < 0)
		goto out;
	retval = -EFAULT;
	if (!access_ok(VERIFY_WRITE, iov.iov_base, iov.iov_len))
		goto out;
	retval = 0;
	if (!count)
		goto out;

	if (mapping->nrpages) {
		retval = filemap_fdatawrite(mapping);
		if (retval == 0)
			retval = filemap_fdatawait(mapping);
		if (retval)
			goto out;
	}

	retval = nfs_direct_read(inode, ctx, &iov, pos, 1);
	if (retval > 0)
		*ppos = pos + retval;

out:
	return retval;
}

/**
 * nfs_file_direct_write - file direct write operation for NFS files
 * @iocb: target I/O control block
 * @buf: user's buffer from which to write data
 * count: number of bytes to write
 * pos: byte offset in file where writing starts
 *
 * We use this function for direct writes instead of calling
 * generic_file_aio_write() in order to avoid taking the inode
 * semaphore and updating the i_size.  The NFS server will set
 * the new i_size and this client must read the updated size
 * back into its cache.  We let the server do generic write
 * parameter checking and report problems.
 *
 * We also avoid an unnecessary invocation of generic_osync_inode(),
 * as it is fairly meaningless to sync the metadata of an NFS file.
 *
 * We eliminate local atime updates, see direct read above.
 *
 * We avoid unnecessary page cache invalidations for normal cached
 * readers of this file.
 *
 * Note that O_APPEND is not supported for NFS direct writes, as there
 * is no atomic O_APPEND write facility in the NFS protocol.
 */
ssize_t
nfs_file_direct_write(struct kiocb *iocb, const char __user *buf, size_t count, loff_t pos)
{
	ssize_t retval = -EINVAL;
	loff_t *ppos = &iocb->ki_pos;
	unsigned long limit = current->rlim[RLIMIT_FSIZE].rlim_cur;
	struct file *file = iocb->ki_filp;
	struct nfs_open_context *ctx =
			(struct nfs_open_context *) file->private_data;
	struct dentry *dentry = file->f_dentry;
	struct address_space *mapping = file->f_mapping;
	struct inode *inode = mapping->host;
	struct iovec iov = {
		.iov_base = (char __user *)buf,
		.iov_len = count,
	};

	dfprintk(VFS, "nfs: direct write(%s/%s(%ld), %lu@%lu)\n",
		dentry->d_parent->d_name.name, dentry->d_name.name,
		inode->i_ino, (unsigned long) count, (unsigned long) pos);

	if (!is_sync_kiocb(iocb))
		goto out;
	if (count < 0)
		goto out;
        if (pos < 0)
		goto out;
	retval = -EFAULT;
	if (!access_ok(VERIFY_READ, iov.iov_base, iov.iov_len))
		goto out;
        if (file->f_error) {
                retval = file->f_error;
                file->f_error = 0;
                goto out;
        }
	retval = -EFBIG;
	if (limit != RLIM_INFINITY) {
		if (pos >= limit) {
			send_sig(SIGXFSZ, current, 0);
			goto out;
		}
		if (count > limit - (unsigned long) pos)
			count = limit - (unsigned long) pos;
	}
	retval = 0;
	if (!count)
		goto out;

	if (mapping->nrpages) {
		retval = filemap_fdatawrite(mapping);
		if (retval == 0)
			retval = filemap_fdatawait(mapping);
		if (retval)
			goto out;
	}

	retval = nfs_direct_write(inode, ctx, &iov, pos, 1);
	if (mapping->nrpages)
		invalidate_inode_pages2(mapping);
	if (retval > 0)
		*ppos = pos + retval;

out:
	return retval;
}