diff mbox series

[v4,3/5] dm: add support for passing through inline crypto support

Message ID 20210201051019.1174983-4-satyat@google.com (mailing list archive)
State New, archived
Headers show
Series add support for inline encryption to device mapper | expand

Commit Message

Satya Tangirala Feb. 1, 2021, 5:10 a.m. UTC
Update the device-mapper core to support exposing the inline crypto
support of the underlying device(s) through the device-mapper device.

This works by creating a "passthrough keyslot manager" for the dm
device, which declares support for encryption settings which all
underlying devices support.  When a supported setting is used, the bio
cloning code handles cloning the crypto context to the bios for all the
underlying devices.  When an unsupported setting is used, the blk-crypto
fallback is used as usual.

Crypto support on each underlying device is ignored unless the
corresponding dm target opts into exposing it.  This is needed because
for inline crypto to semantically operate on the original bio, the data
must not be transformed by the dm target.  Thus, targets like dm-linear
can expose crypto support of the underlying device, but targets like
dm-crypt can't.  (dm-crypt could use inline crypto itself, though.)

A DM device's table can only be changed if the "new" inline encryption
capabilities are a (*not* necessarily strict) superset of the "old" inline
encryption capabilities.  Attempts to make changes to the table that result
in some inline encryption capability becoming no longer supported will be
rejected.

For the sake of clarity, key eviction from underlying devices will be
handled in a future patch.

Co-developed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Satya Tangirala <satyat@google.com>
---
 drivers/md/dm-core.h          |   5 ++
 drivers/md/dm-table.c         | 157 ++++++++++++++++++++++++++++++++++
 drivers/md/dm.c               |  18 +++-
 include/linux/device-mapper.h |  11 +++
 4 files changed, 190 insertions(+), 1 deletion(-)

Comments

Eric Biggers Feb. 10, 2021, 8:17 p.m. UTC | #1
On Mon, Feb 01, 2021 at 05:10:17AM +0000, Satya Tangirala wrote:
> Update the device-mapper core to support exposing the inline crypto
> support of the underlying device(s) through the device-mapper device.
> 
> This works by creating a "passthrough keyslot manager" for the dm
> device, which declares support for encryption settings which all
> underlying devices support.  When a supported setting is used, the bio
> cloning code handles cloning the crypto context to the bios for all the
> underlying devices.  When an unsupported setting is used, the blk-crypto
> fallback is used as usual.
> 
> Crypto support on each underlying device is ignored unless the
> corresponding dm target opts into exposing it.  This is needed because
> for inline crypto to semantically operate on the original bio, the data
> must not be transformed by the dm target.  Thus, targets like dm-linear
> can expose crypto support of the underlying device, but targets like
> dm-crypt can't.  (dm-crypt could use inline crypto itself, though.)
> 
> A DM device's table can only be changed if the "new" inline encryption
> capabilities are a (*not* necessarily strict) superset of the "old" inline
> encryption capabilities.  Attempts to make changes to the table that result
> in some inline encryption capability becoming no longer supported will be
> rejected.
> 
> For the sake of clarity, key eviction from underlying devices will be
> handled in a future patch.
> 
> Co-developed-by: Eric Biggers <ebiggers@google.com>
> Signed-off-by: Eric Biggers <ebiggers@google.com>
> Signed-off-by: Satya Tangirala <satyat@google.com>

I don't see any obvious issues with this latest version.  I assume you've tested
it on real hardware?

If it's needed despite my Co-developed-by, feel free to add:

Reviewed-by: Eric Biggers <ebiggers@google.com>

A few nits about comments, in case you resend:

> diff --git a/drivers/md/dm-core.h b/drivers/md/dm-core.h
> index 086d293c2b03..bf3e66f39a4a 100644
> --- a/drivers/md/dm-core.h
> +++ b/drivers/md/dm-core.h
> @@ -13,6 +13,7 @@
>  #include <linux/ktime.h>
>  #include <linux/genhd.h>
>  #include <linux/blk-mq.h>
> +#include <linux/keyslot-manager.h>
>  
>  #include <trace/events/block.h>
>  
> @@ -162,6 +163,10 @@ struct dm_table {
>  	void *event_context;
>  
>  	struct dm_md_mempools *mempools;
> +
> +#ifdef CONFIG_BLK_INLINE_ENCRYPTION
> +	struct blk_keyslot_manager *ksm;
> +#endif
>  };

It might be helpful if there was a brief comment here that explained that this
field is only set temporarily while the table is being set up, and it gets set
to NULL after the capabilities have been transferred to the request_queue.
I.e., it's not something that stays around here while the dm device is active.

> +/*
> + * Constructs and returns a keyslot manager that represents the crypto
> + * capabilities of the devices described by the dm_table. However, if the
> + * constructed keyslot manager does not support a superset of the crypto
> + * capabilities supported by the current keyslot manager of the mapped_device,
> + * it returns an error instead, since we don't support restricting crypto
> + * capabilities on table changes. Finally, if the constructed keyslot manager
> + * doesn't actually support any crypto modes at all, it just returns NULL.
> + */
> +static int
> +dm_table_construct_keyslot_manager(struct dm_table *t)

This doesn't "return" the keyslot manager anymore, but rather assigns it to
t->ksm.  It would also be helpful if the comment explicitly mentioned that the
goal is to find the capabilities that all the devices have in common.

E.g. "Initializes t->ksm with a keyslot manager that represents the common set
of crypto capabilities of the devices described by the dm_table.".

> diff --git a/include/linux/device-mapper.h b/include/linux/device-mapper.h
> index 61a66fb8ebb3..d2142f5a82a7 100644
> --- a/include/linux/device-mapper.h
> +++ b/include/linux/device-mapper.h
> @@ -257,6 +257,12 @@ struct target_type {
>  #define DM_TARGET_NOWAIT		0x00000080
>  #define dm_target_supports_nowait(type) ((type)->features & DM_TARGET_NOWAIT)
>  
> +/*
> + *
> + */
> +#define DM_TARGET_PASSES_CRYPTO		0x00000100
> +#define dm_target_passes_crypto(type) ((type)->features & DM_TARGET_PASSES_CRYPTO)

The above comment isn't very useful :-)

- Eric
Satya Tangirala Feb. 11, 2021, 10:58 p.m. UTC | #2
On Wed, Feb 10, 2021 at 12:17:30PM -0800, Eric Biggers wrote:
> On Mon, Feb 01, 2021 at 05:10:17AM +0000, Satya Tangirala wrote:
> > Update the device-mapper core to support exposing the inline crypto
> > support of the underlying device(s) through the device-mapper device.
> > 
> > This works by creating a "passthrough keyslot manager" for the dm
> > device, which declares support for encryption settings which all
> > underlying devices support.  When a supported setting is used, the bio
> > cloning code handles cloning the crypto context to the bios for all the
> > underlying devices.  When an unsupported setting is used, the blk-crypto
> > fallback is used as usual.
> > 
> > Crypto support on each underlying device is ignored unless the
> > corresponding dm target opts into exposing it.  This is needed because
> > for inline crypto to semantically operate on the original bio, the data
> > must not be transformed by the dm target.  Thus, targets like dm-linear
> > can expose crypto support of the underlying device, but targets like
> > dm-crypt can't.  (dm-crypt could use inline crypto itself, though.)
> > 
> > A DM device's table can only be changed if the "new" inline encryption
> > capabilities are a (*not* necessarily strict) superset of the "old" inline
> > encryption capabilities.  Attempts to make changes to the table that result
> > in some inline encryption capability becoming no longer supported will be
> > rejected.
> > 
> > For the sake of clarity, key eviction from underlying devices will be
> > handled in a future patch.
> > 
> > Co-developed-by: Eric Biggers <ebiggers@google.com>
> > Signed-off-by: Eric Biggers <ebiggers@google.com>
> > Signed-off-by: Satya Tangirala <satyat@google.com>
> 
> I don't see any obvious issues with this latest version.  I assume you've tested
> it on real hardware?
> 
> If it's needed despite my Co-developed-by, feel free to add:
> 
> Reviewed-by: Eric Biggers <ebiggers@google.com>
> 
> A few nits about comments, in case you resend:
> 
Thanks! I addressed the nits, added acked/reviewed-bys and resent v5.
diff mbox series

Patch

diff --git a/drivers/md/dm-core.h b/drivers/md/dm-core.h
index 086d293c2b03..bf3e66f39a4a 100644
--- a/drivers/md/dm-core.h
+++ b/drivers/md/dm-core.h
@@ -13,6 +13,7 @@ 
 #include <linux/ktime.h>
 #include <linux/genhd.h>
 #include <linux/blk-mq.h>
+#include <linux/keyslot-manager.h>
 
 #include <trace/events/block.h>
 
@@ -162,6 +163,10 @@  struct dm_table {
 	void *event_context;
 
 	struct dm_md_mempools *mempools;
+
+#ifdef CONFIG_BLK_INLINE_ENCRYPTION
+	struct blk_keyslot_manager *ksm;
+#endif
 };
 
 static inline struct completion *dm_get_completion_from_kobject(struct kobject *kobj)
diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index 4acf2342f7ad..b37f69343923 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -187,6 +187,8 @@  static void free_devices(struct list_head *devices, struct mapped_device *md)
 	}
 }
 
+static void dm_table_destroy_keyslot_manager(struct dm_table *t);
+
 void dm_table_destroy(struct dm_table *t)
 {
 	unsigned int i;
@@ -215,6 +217,8 @@  void dm_table_destroy(struct dm_table *t)
 
 	dm_free_md_mempools(t->mempools);
 
+	dm_table_destroy_keyslot_manager(t);
+
 	kfree(t);
 }
 
@@ -1203,6 +1207,152 @@  static int dm_table_register_integrity(struct dm_table *t)
 	return 0;
 }
 
+#ifdef CONFIG_BLK_INLINE_ENCRYPTION
+
+struct dm_keyslot_manager {
+	struct blk_keyslot_manager ksm;
+	struct mapped_device *md;
+};
+
+static int device_intersect_crypto_modes(struct dm_target *ti,
+					 struct dm_dev *dev, sector_t start,
+					 sector_t len, void *data)
+{
+	struct blk_keyslot_manager *parent = data;
+	struct blk_keyslot_manager *child = bdev_get_queue(dev->bdev)->ksm;
+
+	blk_ksm_intersect_modes(parent, child);
+	return 0;
+}
+
+void dm_destroy_keyslot_manager(struct blk_keyslot_manager *ksm)
+{
+	struct dm_keyslot_manager *dksm = container_of(ksm,
+						       struct dm_keyslot_manager,
+						       ksm);
+
+	if (!ksm)
+		return;
+
+	blk_ksm_destroy(ksm);
+	kfree(dksm);
+}
+
+static void dm_table_destroy_keyslot_manager(struct dm_table *t)
+{
+	dm_destroy_keyslot_manager(t->ksm);
+	t->ksm = NULL;
+}
+
+/*
+ * Constructs and returns a keyslot manager that represents the crypto
+ * capabilities of the devices described by the dm_table. However, if the
+ * constructed keyslot manager does not support a superset of the crypto
+ * capabilities supported by the current keyslot manager of the mapped_device,
+ * it returns an error instead, since we don't support restricting crypto
+ * capabilities on table changes. Finally, if the constructed keyslot manager
+ * doesn't actually support any crypto modes at all, it just returns NULL.
+ */
+static int
+dm_table_construct_keyslot_manager(struct dm_table *t)
+{
+	struct dm_keyslot_manager *dksm;
+	struct blk_keyslot_manager *ksm;
+	struct dm_target *ti;
+	unsigned int i;
+	bool ksm_is_empty = true;
+
+	dksm = kmalloc(sizeof(*dksm), GFP_KERNEL);
+	if (!dksm)
+		return -ENOMEM;
+	dksm->md = t->md;
+
+	ksm = &dksm->ksm;
+	blk_ksm_init_passthrough(ksm);
+	ksm->max_dun_bytes_supported = UINT_MAX;
+	memset(ksm->crypto_modes_supported, 0xFF,
+	       sizeof(ksm->crypto_modes_supported));
+
+	for (i = 0; i < dm_table_get_num_targets(t); i++) {
+		ti = dm_table_get_target(t, i);
+
+		if (!dm_target_passes_crypto(ti->type)) {
+			blk_ksm_intersect_modes(ksm, NULL);
+			break;
+		}
+		if (!ti->type->iterate_devices)
+			continue;
+		ti->type->iterate_devices(ti, device_intersect_crypto_modes,
+					  ksm);
+	}
+
+	if (t->md->queue && !blk_ksm_is_superset(ksm, t->md->queue->ksm)) {
+		DMWARN("Inline encryption capabilities of new DM table were more restrictive than the old table's. This is not supported!");
+		dm_destroy_keyslot_manager(ksm);
+		return -EINVAL;
+	}
+
+	/*
+	 * If the new KSM doesn't actually support any crypto modes, we may as
+	 * well represent it with a NULL ksm.
+	 */
+	ksm_is_empty = true;
+	for (i = 0; i < ARRAY_SIZE(ksm->crypto_modes_supported); i++) {
+		if (ksm->crypto_modes_supported[i]) {
+			ksm_is_empty = false;
+			break;
+		}
+	}
+
+	if (ksm_is_empty) {
+		dm_destroy_keyslot_manager(ksm);
+		ksm = NULL;
+	}
+
+	t->ksm = ksm;
+
+	return 0;
+}
+
+static void dm_update_keyslot_manager(struct request_queue *q,
+				      struct dm_table *t)
+{
+	if (!t->ksm)
+		return;
+
+	/* Make the ksm less restrictive */
+	if (!q->ksm) {
+		blk_ksm_register(t->ksm, q);
+	} else {
+		blk_ksm_update_capabilities(q->ksm, t->ksm);
+		dm_destroy_keyslot_manager(t->ksm);
+	}
+	t->ksm = NULL;
+}
+
+#else /* CONFIG_BLK_INLINE_ENCRYPTION */
+
+static int
+dm_table_construct_keyslot_manager(struct dm_table *t)
+{
+	return 0;
+}
+
+void dm_destroy_keyslot_manager(struct blk_keyslot_manager *ksm)
+{
+}
+
+static void dm_table_destroy_keyslot_manager(struct dm_table *t)
+{
+}
+
+static void dm_update_keyslot_manager(struct request_queue *q,
+				      struct dm_table *t)
+{
+}
+
+#endif /* !CONFIG_BLK_INLINE_ENCRYPTION */
+
 /*
  * Prepares the table for use by building the indices,
  * setting the type, and allocating mempools.
@@ -1229,6 +1379,12 @@  int dm_table_complete(struct dm_table *t)
 		return r;
 	}
 
+	r = dm_table_construct_keyslot_manager(t);
+	if (r) {
+		DMERR("could not construct keyslot manager.");
+		return r;
+	}
+
 	r = dm_table_alloc_md_mempools(t, t->md);
 	if (r)
 		DMERR("unable to allocate mempools");
@@ -1891,6 +2047,7 @@  void dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
 	}
 #endif
 
+	dm_update_keyslot_manager(q, t);
 	blk_queue_update_readahead(q);
 }
 
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index 7bac564f3faa..ee0434764941 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -28,6 +28,7 @@ 
 #include <linux/refcount.h>
 #include <linux/part_stat.h>
 #include <linux/blk-crypto.h>
+#include <linux/keyslot-manager.h>
 
 #define DM_MSG_PREFIX "core"
 
@@ -1718,6 +1719,19 @@  static const struct dax_operations dm_dax_ops;
 
 static void dm_wq_work(struct work_struct *work);
 
+#ifdef CONFIG_BLK_INLINE_ENCRYPTION
+static void dm_queue_destroy_keyslot_manager(struct request_queue *q)
+{
+	dm_destroy_keyslot_manager(q->ksm);
+}
+
+#else /* CONFIG_BLK_INLINE_ENCRYPTION */
+
+static inline void dm_queue_destroy_keyslot_manager(struct request_queue *q)
+{
+}
+#endif /* !CONFIG_BLK_INLINE_ENCRYPTION */
+
 static void cleanup_mapped_device(struct mapped_device *md)
 {
 	if (md->wq)
@@ -1739,8 +1753,10 @@  static void cleanup_mapped_device(struct mapped_device *md)
 		put_disk(md->disk);
 	}
 
-	if (md->queue)
+	if (md->queue) {
+		dm_queue_destroy_keyslot_manager(md->queue);
 		blk_cleanup_queue(md->queue);
+	}
 
 	cleanup_srcu_struct(&md->io_barrier);
 
diff --git a/include/linux/device-mapper.h b/include/linux/device-mapper.h
index 61a66fb8ebb3..d2142f5a82a7 100644
--- a/include/linux/device-mapper.h
+++ b/include/linux/device-mapper.h
@@ -257,6 +257,12 @@  struct target_type {
 #define DM_TARGET_NOWAIT		0x00000080
 #define dm_target_supports_nowait(type) ((type)->features & DM_TARGET_NOWAIT)
 
+/*
+ *
+ */
+#define DM_TARGET_PASSES_CRYPTO		0x00000100
+#define dm_target_passes_crypto(type) ((type)->features & DM_TARGET_PASSES_CRYPTO)
+
 struct dm_target {
 	struct dm_table *table;
 	struct target_type *type;
@@ -533,6 +539,11 @@  void dm_table_run_md_queue_async(struct dm_table *t);
 struct dm_table *dm_swap_table(struct mapped_device *md,
 			       struct dm_table *t);
 
+/*
+ * Table keyslot manager functions
+ */
+void dm_destroy_keyslot_manager(struct blk_keyslot_manager *ksm);
+
 /*
  * A wrapper around vmalloc.
  */