Re: [f2fs-dev] [PATCH v5 2/2] f2fs: add F2FS_IOC_SET_COMPRESS_OPTION ioctl

[Eric Biggers <ebiggers@kernel.org>]

On Thu, Oct 29, 2020 at 03:29:17PM +0800, Chao Yu wrote:
> > +static int f2fs_ioc_set_compress_option(struct file *filp, unsigned long arg)
> > +{
> > +	struct inode *inode = file_inode(filp);
> > +	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
> > +	struct f2fs_comp_option option;
> > +	int ret = 0;
> > +
> > +	if (!f2fs_sb_has_compression(sbi))
> > +		return -EOPNOTSUPP;
> > +
> > +	if (!(filp->f_mode & FMODE_WRITE))
> > +		return -EBADF;
> > +
> > +	if (copy_from_user(&option, (struct f2fs_comp_option __user *)arg,
> > +				sizeof(option)))
> > +		return -EFAULT;
> > +
> > +	if (!f2fs_compressed_file(inode) ||
> > +			option.log_cluster_size < MIN_COMPRESS_LOG_SIZE ||
> > +			option.log_cluster_size > MAX_COMPRESS_LOG_SIZE ||
> > +			option.algorithm >= COMPRESS_MAX)
> > +		return -EINVAL;
> > +
> > +	file_start_write(filp);
> > +	inode_lock(inode);
> > +
> > +	if (f2fs_is_mmap_file(inode) || get_dirty_pages(inode)) {
> > +		ret = -EBUSY;
> > +		goto out;
> > +	}
> > +
> > +	if (inode->i_size != 0) {
> > +		ret = -EFBIG;
> > +		goto out;
> > +	}
> 
> Hmm...
> 
> Shouldn't it be:
> 
> if (algorithm >= COMPRESS_MAX) {
> 	ret = -ENOPKG;
> 	goto out;
> }
> 
> if (!f2fs_cops[algorithm])
> 	f2fs_warn(...);

Note that my intent with recommending ENOPKG was for it to be returned in the
!f2fs_cops[algorithm] case, similar to how opening an encrypted file when the
encryption algorithm is recognized but not supported by the kernel returns
ENOPKG.  For a truly unrecognized algorithm (algorithm >= COMPRESS_MAX), EINVAL
would probably be more appropriate.  So if !f2fs_cops[algorithm] is now allowed,
then ENOPKG should no longer be among the error codes this ioctl returns.

- Eric