Date:   Thu, 9 Jul 2020 14:54:58 -0700
From:   Eric Biggers <ebiggers@...nel.org>
To:     Satya Tangirala <satyat@...gle.com>
Cc:     linux-fscrypt@...r.kernel.org, linux-fsdevel@...r.kernel.org,
        linux-f2fs-devel@...ts.sourceforge.net, linux-ext4@...r.kernel.org
Subject: Re: [PATCH 1/5] fscrypt: Add functions for direct I/O support

On Thu, Jul 09, 2020 at 07:47:47PM +0000, Satya Tangirala via Linux-f2fs-devel wrote:
> From: Eric Biggers <ebiggers@...gle.com>
> 
> Introduce fscrypt_dio_supported() to check whether a direct I/O request
> is unsupported due to encryption constraints, and
> fscrypt_limit_dio_pages() to check how many pages may be added to a bio
> being prepared for direct I/O.
> 
> The IV_INO_LBLK_32 fscrypt policy introduces the possibility that DUNs
> in logically continuous file blocks might wrap from 0xffffffff to 0.
> Bios in which the DUN wraps around like this cannot be submitted. This
> is especially difficult to handle when block_size != PAGE_SIZE, since in
> that case the DUN can wrap in the middle of a page.
> 
> For now, we add direct I/O support while using IV_INO_LBLK_32 policies
> only for the case when block_size == PAGE_SIZE. When IV_INO_LBLK_32
> policy is used, fscrypt_dio_supported() rejects the bio when
> block_size != PAGE_SIZE. fscrypt_limit_dio_pages() returns the number of
> pages that may be added to the bio without causing the DUN to wrap
> around within the bio.

This commit message is a bit outdated, since the latest version of
"fscrypt: add inline encryption support" already makes IV_INO_LBLK_32
with block_size != PAGE_SIZE fall back to filesystem-layer encryption,
and hence it won't allow direct I/O.

> 
> Signed-off-by: Eric Biggers <ebiggers@...gle.com>
> Signed-off-by: Satya Tangirala <satyat@...gle.com>

Can you mention any changes you made, e.g.:

Signed-off-by: Eric Biggers <ebiggers@...gle.com>
[ST: split original change into separate patches, and updated to account
 for inline encryption no longer being allowed with IV_INO_LBLK_32 and
 blocksize != PAGE_SIZE]
Signed-off-by: Satya Tangirala <satyat@...gle.com>

> +/**
> + * fscrypt_limit_dio_pages() - limit I/O pages to avoid discontiguous DUNs
> + * @inode: the file on which I/O is being done
> + * @pos: the file position (in bytes) at which the I/O is being done
> + * @nr_pages: the number of pages we want to submit starting at @pos
> + *
> + * For direct I/O: limit the number of pages that will be submitted in the bio
> + * targeting @pos, in order to avoid crossing a data unit number (DUN)
> + * discontinuity.  This is only needed for certain IV generation methods.
> + *
> + * This assumes block_size == PAGE_SIZE; see fscrypt_dio_supported().

The note about block_size == PAGE_SIZE here is outdated.

I was also struggling a bit to decide what to name this function.  Note
that it's not really direct I/O specific.  Also, fs/iomap/direct-io.c
needs it but fs/direct-io.c does not.

What this function really does is batch together the mergeability checks
for a logical range.

Maybe the comment could explain this better, and maybe the function
should be called "fscrypt_limit_io_pages()" instead.

> + * Return: the actual number of pages that can be submitted
> + */
> +int fscrypt_limit_dio_pages(const struct inode *inode, loff_t pos, int nr_pages)
> +{
> +	const struct fscrypt_info *ci = inode->i_crypt_info;
> +	u32 dun;
> +
> +	if (!fscrypt_inode_uses_inline_crypto(inode))
> +		return nr_pages;
> +
> +	if (nr_pages <= 1)
> +		return nr_pages;
> +
> +	if (!(fscrypt_policy_flags(&ci->ci_policy) &
> +	      FSCRYPT_POLICY_FLAG_IV_INO_LBLK_32))
> +		return nr_pages;
> +
> +	if (WARN_ON_ONCE(i_blocksize(inode) != PAGE_SIZE))
> +		return 1;
> +
> +	/* With IV_INO_LBLK_32, the DUN can wrap around from U32_MAX to 0. */
> +
> +	dun = ci->ci_hashed_ino + (pos >> inode->i_blkbits);
> +
> +	return min_t(u64, nr_pages, (u64)U32_MAX + 1 - dun);
> +}

- Eric

Powered by blists - more mailing lists