linux-kernel - Re: [PATCH v7 07/11] iomap: fix iomap_dio

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <4c6e092d-5580-42c8-9932-b42995e914be@oracle.com>
Date: Tue, 11 Jun 2024 08:38:34 +0100
From: John Garry <john.g.garry@...cle.com>
To: "Pankaj Raghav (Samsung)" <kernel@...kajraghav.com>, david@...morbit.com,
        djwong@...nel.org, chandan.babu@...cle.com, brauner@...nel.org,
        akpm@...ux-foundation.org, willy@...radead.org
Cc: mcgrof@...nel.org, linux-mm@...ck.org, hare@...e.de,
        linux-kernel@...r.kernel.org, yang@...amperecomputing.com,
        Zi Yan <zi.yan@...t.com>, linux-xfs@...r.kernel.org,
        p.raghav@...sung.com, linux-fsdevel@...r.kernel.org, hch@....de,
        gost.dev@...sung.com, cl@...amperecomputing.com
Subject: Re: [PATCH v7 07/11] iomap: fix iomap_dio_zero() for fs bs > system
 page size

On 07/06/2024 15:58, Pankaj Raghav (Samsung) wrote:
> From: Pankaj Raghav <p.raghav@...sung.com>
> 
> iomap_dio_zero() will pad a fs block with zeroes if the direct IO size
> < fs block size. iomap_dio_zero() has an implicit assumption that fs block
> size < page_size. This is true for most filesystems at the moment.
> 
> If the block size > page size, this will send the contents of the page
> next to zero page(as len > PAGE_SIZE) to the underlying block device,
> causing FS corruption.
> 
> iomap is a generic infrastructure and it should not make any assumptions
> about the fs block size and the page size of the system.
> 
> Signed-off-by: Pankaj Raghav <p.raghav@...sung.com>
> Reviewed-by: Hannes Reinecke <hare@...e.de>
> ---
>   fs/internal.h          |  5 +++++
>   fs/iomap/buffered-io.c |  6 ++++++
>   fs/iomap/direct-io.c   | 26 ++++++++++++++++++++++++--
>   3 files changed, 35 insertions(+), 2 deletions(-)
> 
> diff --git a/fs/internal.h b/fs/internal.h
> index 84f371193f74..30217f0ff4c6 100644
> --- a/fs/internal.h
> +++ b/fs/internal.h
> @@ -35,6 +35,11 @@ static inline void bdev_cache_init(void)
>   int __block_write_begin_int(struct folio *folio, loff_t pos, unsigned len,
>   		get_block_t *get_block, const struct iomap *iomap);
>   
> +/*
> + * iomap/direct-io.c
> + */
> +int iomap_dio_init(void);
> +
>   /*
>    * char_dev.c
>    */
> diff --git a/fs/iomap/buffered-io.c b/fs/iomap/buffered-io.c
> index 49938419fcc7..9f791db473e4 100644
> --- a/fs/iomap/buffered-io.c
> +++ b/fs/iomap/buffered-io.c
> @@ -1990,6 +1990,12 @@ EXPORT_SYMBOL_GPL(iomap_writepages);
>   
>   static int __init iomap_init(void)
>   {
> +	int ret;
> +
> +	ret = iomap_dio_init();
> +	if (ret)
> +		return ret;
> +
>   	return bioset_init(&iomap_ioend_bioset, 4 * (PAGE_SIZE / SECTOR_SIZE),
>   			   offsetof(struct iomap_ioend, io_bio),
>   			   BIOSET_NEED_BVECS);

I suppose that it does not matter that zero_fs_block is leaked if this 
fails (or is it even leaked?), as I don't think that failing that 
bioset_init() call is handled at all.

> diff --git a/fs/iomap/direct-io.c b/fs/iomap/direct-io.c
> index f3b43d223a46..b95600b254a3 100644
> --- a/fs/iomap/direct-io.c
> +++ b/fs/iomap/direct-io.c
> @@ -27,6 +27,13 @@
>   #define IOMAP_DIO_WRITE		(1U << 30)
>   #define IOMAP_DIO_DIRTY		(1U << 31)
>   
> +/*
> + * Used for sub block zeroing in iomap_dio_zero()
> + */
> +#define ZERO_FSB_SIZE (65536)
> +#define ZERO_FSB_ORDER (get_order(ZERO_FSB_SIZE))
> +static struct page *zero_fs_block;
> +
>   struct iomap_dio {
>   	struct kiocb		*iocb;
>   	const struct iomap_dio_ops *dops;
> @@ -52,6 +59,16 @@ struct iomap_dio {
>   	};
>   };
>   
> +int iomap_dio_init(void)
> +{
> +	zero_fs_block = alloc_pages(GFP_KERNEL | __GFP_ZERO, ZERO_FSB_ORDER);
> +
> +	if (!zero_fs_block)
> +		return -ENOMEM;
> +
> +	return 0;
> +}
> +
>   static struct bio *iomap_dio_alloc_bio(const struct iomap_iter *iter,
>   		struct iomap_dio *dio, unsigned short nr_vecs, blk_opf_t opf)
>   {
> @@ -236,17 +253,22 @@ static void iomap_dio_zero(const struct iomap_iter *iter, struct iomap_dio *dio,
>   		loff_t pos, unsigned len)
>   {
>   	struct inode *inode = file_inode(dio->iocb->ki_filp);
> -	struct page *page = ZERO_PAGE(0);
>   	struct bio *bio;
>   
> +	/*
> +	 * Max block size supported is 64k
> +	 */
> +	WARN_ON_ONCE(len > ZERO_FSB_SIZE);

JFYI, As mentioned in 
https://lore.kernel.org/linux-xfs/20240429174746.2132161-1-john.g.garry@oracle.com/T/#m5354e2b2531a5552a8b8acd4a95342ed4d7500f2, 
we would like to support an arbitrary size. Maybe I will need to loop 
for zeroing sizes > 64K.

> +
>   	bio = iomap_dio_alloc_bio(iter, dio, 1, REQ_OP_WRITE | REQ_SYNC | REQ_IDLE);
>   	fscrypt_set_bio_crypt_ctx(bio, inode, pos >> inode->i_blkbits,
>   				  GFP_KERNEL);
> +
>   	bio->bi_iter.bi_sector = iomap_sector(&iter->iomap, pos);
>   	bio->bi_private = dio;
>   	bio->bi_end_io = iomap_dio_bio_end_io;
>   
> -	__bio_add_page(bio, page, len, 0);
> +	__bio_add_page(bio, zero_fs_block, len, 0);
>   	iomap_dio_submit_bio(iter, dio, bio, pos);
>   }
>