[PATCH] UBIFS: read crypto_lookup from datanodes to znodes

Artem Bityutskiy dedekind1 at gmail.com
Tue May 15 09:03:42 EDT 2012


On Tue, 2012-05-15 at 11:01 +0200, Joel Reardon wrote:
> Crypto_lookup values are now oppertunistically read from the data node header whenever a data
> node is read from flash and cached in the TNC. The value will be zero until it
> is read from flash. (If it is needed before it happens to be loaded, i.e., to
> delete a node, then it will be forced read.)
> 
> It was tested by setting cryptolookup values for new datanodes by their lnum
> and offset on flash. This was later asserted in the TNC that they were either
> zero, or always equal. On mounting, the TNC crypto lookup values were
> confirmed to be zero and later, after reading the corresponding files,
> confirmed they were loaded and corresponded to the location on flash.
> 
> Signed-off-by: Joel Reardon <reardonj at inf.ethz.ch>
> ---
>  fs/ubifs/gc.c          |    8 +++++++-
>  fs/ubifs/journal.c     |    9 ++++++---
>  fs/ubifs/tnc.c         |   40 ++++++++++++++++++++++++++++++++++++++++
>  fs/ubifs/ubifs-media.h |    3 ++-
>  4 files changed, 55 insertions(+), 5 deletions(-)
> 
> diff --git a/fs/ubifs/gc.c b/fs/ubifs/gc.c
> index f146447..16df2f4 100644
> --- a/fs/ubifs/gc.c
> +++ b/fs/ubifs/gc.c
> @@ -322,15 +322,21 @@ static int move_node(struct ubifs_info *c, struct ubifs_scan_leb *sleb,
>  		     struct ubifs_scan_node *snod, struct ubifs_wbuf *wbuf)
>  {
>  	int err, new_lnum = wbuf->lnum, new_offs = wbuf->offs + wbuf->used;
> +	long long crypto_lookup = 0;
> 
>  	cond_resched();
>  	err = ubifs_wbuf_write_nolock(wbuf, snod->node, snod->len);
>  	if (err)
>  		return err;
> 
> +	if (key_type(c, &snod->key) == UBIFS_DATA_KEY) {
> +		struct ubifs_data_node *dn =
> +			(struct ubifs_data_node *) snod->node;

We do not cast void *.

> +		crypto_lookup = le64_to_cpu(dn->crypto_lookup);
> +	}
>  	err = ubifs_tnc_replace(c, &snod->key, sleb->lnum,
>  				snod->offs, new_lnum, new_offs,
> -				snod->len, 0);
> +				snod->len, crypto_lookup);
>  	list_del(&snod->list);
>  	kfree(snod);
>  	return err;
> diff --git a/fs/ubifs/journal.c b/fs/ubifs/journal.c
> index 2072b9a..3f7aa05 100644
> --- a/fs/ubifs/journal.c
> +++ b/fs/ubifs/journal.c
> @@ -89,7 +89,6 @@ static inline void zero_dent_node_unused(struct ubifs_dent_node *dent)
>   */
>  static inline void zero_data_node_unused(struct ubifs_data_node *data)
>  {
> -	data->padding0 = 0;
>  	memset(data->padding1, 0, 2);
>  }
> 
> @@ -735,6 +734,7 @@ int ubifs_jnl_write_data(struct ubifs_info *c, const struct inode *inode,
> 
>  	dlen = UBIFS_DATA_NODE_SZ + out_len;
>  	data->compr_type = cpu_to_le16(compr_type);
> +	data->crypto_lookup = cpu_to_le64(0);
> 
>  	/* Make reservation before allocating sequence numbers */
>  	err = make_reservation(c, DATAHD, dlen);
> @@ -742,12 +742,14 @@ int ubifs_jnl_write_data(struct ubifs_info *c, const struct inode *inode,
>  		goto out_free;
> 
>  	err = write_node(c, DATAHD, data, dlen, &lnum, &offs);
> +
>  	if (err)

Please, do not add junk newlines.

>  		goto out_release;
>  	ubifs_wbuf_add_ino_nolock(&c->jheads[DATAHD].wbuf, key_inum(c, key));
>  	release_head(c, DATAHD);
> 
> -	err = ubifs_tnc_add(c, key, lnum, offs, dlen, 0);
> +	err = ubifs_tnc_add(c, key, lnum, offs, dlen,
> +			    le64_to_cpu(data->crypto_lookup));
>  	if (err)
>  		goto out_ro;
> 
> @@ -1226,7 +1228,8 @@ int ubifs_jnl_truncate(struct ubifs_info *c, const struct inode *inode,
> 
>  	if (dlen) {
>  		sz = offs + UBIFS_INO_NODE_SZ + UBIFS_TRUN_NODE_SZ;
> -		err = ubifs_tnc_add(c, &key, lnum, sz, dlen, 0);
> +		err = ubifs_tnc_add(c, &key, lnum, sz, dlen,
> +				    le64_to_cpu(dn->crypto_lookup));
>  		if (err)
>  			goto out_ro;
>  	}
> diff --git a/fs/ubifs/tnc.c b/fs/ubifs/tnc.c
> index b222a72..ac0d03c 100644
> --- a/fs/ubifs/tnc.c
> +++ b/fs/ubifs/tnc.c
> @@ -1431,6 +1431,35 @@ static int maybe_leb_gced(struct ubifs_info *c, int lnum, int gc_seq1)
>  }
> 
>  /**
> + * tnc_set_crypto_lookup - set the crypto_lookup field in a zbranch.
> + * @c: UBIFS file-system description object
> + * @zbr: the crypto_lookup field is set in this zbranch
> + * @node: a pointer to a node containing the zbranch.
> + *
> + * Crypto_lookup values are stored on flash memory inside the data node.
> + * When the system is running, they should be oppertunistically stored in

What do you mean by "oppertunistically stored" ? Should be
"opportunistically", but my concern is that we _always_ initialize this
when read, not by opportunity.

> + * memory inside the zbranch. This function is called whenever a node
> + * is read from flash memory. If @node is a data node (therefore containing a
> + * @crypto_lookup field), then:
> + * if @zbr has an unset @crypto_lookup, set it to the value from @node
> + * if @zbr has already a @crypto_lookup, assert they are the same.
> + */
> +static void tnc_set_crypto_lookup(struct ubifs_info *c,
> +				  struct ubifs_zbranch *zbr, void *node)
> +{
> +	if (key_type(c, &(zbr->key)) == UBIFS_DATA_KEY) {
> +		struct ubifs_data_node *dn =
> +			(struct ubifs_data_node *) node;

No cast.
> +		if (zbr->crypto_lookup != 0) {

No braces.

> +			ubifs_assert(zbr->crypto_lookup ==
> +				     le64_to_cpu(dn->crypto_lookup));
> +		} else {

Please, introduce a temporary variable for crypto_lookup instead.

-- 
Best Regards,
Artem Bityutskiy
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part
URL: <http://lists.infradead.org/pipermail/linux-mtd/attachments/20120515/344614bb/attachment.sig>


More information about the linux-mtd mailing list