diff mbox

commit 6d2a78e783416ba99e36beb1d4395b785b34e867 avoids dm integrity support

Message ID 49D06571.70903@ct.jp.nec.com (mailing list archive)
State Not Applicable, archived
Headers show

Commit Message

Kiyoshi Ueda March 30, 2009, 6:23 a.m. UTC 
Hi Martin, Jens, Alasdair,

I found this commit (6d2a78e783416ba99e36beb1d4395b785b34e867) in the recent
Linus's git makes every block device share single mempool for integrity cloning.
I think that it avoids integrity support for dm, because having only one mempool
in the kernel may cause a deadlock due to no memory while a bio goes down to
a device stack.

So dm needs to prepare own mempool for each device and pass it to
bio_integrity_clone() instead of the shared one.


commit 6d2a78e783416ba99e36beb1d4395b785b34e867
Author: Martin K. Petersen <martin.petersen@oracle.com>
Date:   Tue Mar 10 08:27:39 2009 +0100

    block: add private bio_set for bio integrity allocations
    
    The integrity bio allocation needs its own bio_set to avoid violating
    the mempool allocation rules and risking deadlocks.
    
    Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
    Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

Thanks,
Kiyoshi Ueda

--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

Comments

Martin K. Petersen March 30, 2009, 6:39 p.m. UTC | #1 
>>>>> "Kiyoshi" == Kiyoshi Ueda <k-ueda@ct.jp.nec.com> writes:

Kiyoshi> I found this commit (6d2a78e783416ba99e36beb1d4395b785b34e867)
Kiyoshi> in the recent Linus's git makes every block device share single
Kiyoshi> mempool for integrity cloning.

Yeah, until this patch went in I had the integrity stuff hanging off of
the bio_set to avoid the deadlocks you mention.  When this issue came up
a few weeks ago Jens suggested the changes in the commit above:

	http://marc.info/?l=linux-kernel&m=123662652229483

I just talked to him and we're going to back this patch for now.  Short
term I guess we'll have to settle for a separate pool in the bio_set for
integrity vectors.
Jens Axboe March 30, 2009, 6:59 p.m. UTC | #2 
On Mon, Mar 30 2009, Martin K. Petersen wrote:
> >>>>> "Kiyoshi" == Kiyoshi Ueda <k-ueda@ct.jp.nec.com> writes:
> 
> Kiyoshi> I found this commit (6d2a78e783416ba99e36beb1d4395b785b34e867)
> Kiyoshi> in the recent Linus's git makes every block device share single
> Kiyoshi> mempool for integrity cloning.
> 
> Yeah, until this patch went in I had the integrity stuff hanging off of
> the bio_set to avoid the deadlocks you mention.  When this issue came up
> a few weeks ago Jens suggested the changes in the commit above:
> 
> 	http://marc.info/?l=linux-kernel&m=123662652229483

Well, then suggestions there talk about slab reuse, it still needs
private mempools. The forward progress reference failed to take stacked
drivers into account, I didn't realize that they need to allocate
integrity data again as well. It's a shame, since the commit in question
is otherwise a nice cleanup of the integrity stuff. Apparently the patch
wasn't reviewed well enough, I'll take the blame on that.

> I just talked to him and we're going to back this patch for now.  Short
> term I guess we'll have to settle for a separate pool in the bio_set for
> integrity vectors.

Perhaps it would be cleaner to make the integrity allocation more
explicit in the supported paths, instead of hiding it in bio_set? Dunno,
haven't thought much about it, just an alternative approach
Martin K. Petersen March 31, 2009, 5:21 a.m. UTC | #3 
>>>>> "Jens" == Jens Axboe <jens.axboe@oracle.com> writes:

Jens> The forward progress reference failed to take stacked drivers into
Jens> account, I didn't realize that they need to allocate integrity
Jens> data again as well.

Yup.  We need a bio integrity struct as well as a vector to describe the
integrity pages.

Ideally I'd like to avoid cloning the integrity bio_vec altogether.  The
only reason I do it now is because I have to keep the integrity vector
in sync with the data vector when that gets sliced and diced.  Plus
there's the suck of partial completion.

If we never changed bio_vecs this wouldn't be an issue.  One option
would be to add an offset parameter to the bio.  That way we could
completely avoid cloning bio_vecs.  That would mean a bit more
complexity in building scatterlists at the bottom of the pile but we'd
do fewer memory allocations.

I've been tinkering with that approach this evening.  I almost have it
working for the integrity vector.  Doing it for bios is obviously a much
bigger task.


Jens> Perhaps it would be cleaner to make the integrity allocation more
Jens> explicit in the supported paths, instead of hiding it in bio_set?
Jens> Dunno, haven't thought much about it, just an alternative approach

DM is the only subsystem that manually clones things.  MD uses bio_clone
and has no idea that the integrity fluff is there.  In any case I really
think the bio_set approach is a nicer interface than making every
stacking driver special-case integrity allocations.
diff mbox

Patch

diff --git a/fs/bio-integrity.c b/fs/bio-integrity.c
index fe2b1aa..31c46a2 100644
--- a/fs/bio-integrity.c
+++ b/fs/bio-integrity.c
@@ -26,23 +26,23 @@ 
 #include <linux/workqueue.h>
 
 static struct kmem_cache *bio_integrity_slab __read_mostly;
+static mempool_t *bio_integrity_pool;
+static struct bio_set *integrity_bio_set;
 static struct workqueue_struct *kintegrityd_wq;
 
 /**
- * bio_integrity_alloc_bioset - Allocate integrity payload and attach it to bio
+ * bio_integrity_alloc - Allocate integrity payload and attach it to bio
  * @bio:	bio to attach integrity metadata to
  * @gfp_mask:	Memory allocation mask
  * @nr_vecs:	Number of integrity metadata scatter-gather elements
- * @bs:		bio_set to allocate from
  *
  * Description: This function prepares a bio for attaching integrity
  * metadata.  nr_vecs specifies the maximum number of pages containing
  * integrity metadata that can be attached.
  */
-struct bio_integrity_payload *bio_integrity_alloc_bioset(struct bio *bio,
-							 gfp_t gfp_mask,
-							 unsigned int nr_vecs,
-							 struct bio_set *bs)
+struct bio_integrity_payload *bio_integrity_alloc(struct bio *bio,
+						  gfp_t gfp_mask,
+						  unsigned int nr_vecs)
 {
 	struct bio_integrity_payload *bip;
 	struct bio_vec *iv;
@@ -50,7 +50,7 @@  struct bio_integrity_payload *bio_integrity_alloc_bioset(struct bio *bio,
 
 	BUG_ON(bio == NULL);
 
-	bip = mempool_alloc(bs->bio_integrity_pool, gfp_mask);
+	bip = mempool_alloc(bio_integrity_pool, gfp_mask);
 	if (unlikely(bip == NULL)) {
 		printk(KERN_ERR "%s: could not alloc bip\n", __func__);
 		return NULL;
@@ -58,10 +58,10 @@  struct bio_integrity_payload *bio_integrity_alloc_bioset(struct bio *bio,
 
 	memset(bip, 0, sizeof(*bip));
 
-	iv = bvec_alloc_bs(gfp_mask, nr_vecs, &idx, bs);
+	iv = bvec_alloc_bs(gfp_mask, nr_vecs, &idx, integrity_bio_set);
 	if (unlikely(iv == NULL)) {
 		printk(KERN_ERR "%s: could not alloc bip_vec\n", __func__);
-		mempool_free(bip, bs->bio_integrity_pool);
+		mempool_free(bip, bio_integrity_pool);
 		return NULL;
 	}
 
@@ -72,35 +72,16 @@  struct bio_integrity_payload *bio_integrity_alloc_bioset(struct bio *bio,
 
 	return bip;
 }
-EXPORT_SYMBOL(bio_integrity_alloc_bioset);
-
-/**
- * bio_integrity_alloc - Allocate integrity payload and attach it to bio
- * @bio:	bio to attach integrity metadata to
- * @gfp_mask:	Memory allocation mask
- * @nr_vecs:	Number of integrity metadata scatter-gather elements
- *
- * Description: This function prepares a bio for attaching integrity
- * metadata.  nr_vecs specifies the maximum number of pages containing
- * integrity metadata that can be attached.
- */
-struct bio_integrity_payload *bio_integrity_alloc(struct bio *bio,
-						  gfp_t gfp_mask,
-						  unsigned int nr_vecs)
-{
-	return bio_integrity_alloc_bioset(bio, gfp_mask, nr_vecs, fs_bio_set);
-}
 EXPORT_SYMBOL(bio_integrity_alloc);
 
 /**
  * bio_integrity_free - Free bio integrity payload
  * @bio:	bio containing bip to be freed
- * @bs:		bio_set this bio was allocated from
  *
  * Description: Used to free the integrity portion of a bio. Usually
  * called from bio_free().
  */
-void bio_integrity_free(struct bio *bio, struct bio_set *bs)
+void bio_integrity_free(struct bio *bio)
 {
 	struct bio_integrity_payload *bip = bio->bi_integrity;
 
@@ -111,8 +92,8 @@  void bio_integrity_free(struct bio *bio, struct bio_set *bs)
 	    && bip->bip_buf != NULL)
 		kfree(bip->bip_buf);
 
-	bvec_free_bs(bs, bip->bip_vec, bip->bip_pool);
-	mempool_free(bip, bs->bio_integrity_pool);
+	bvec_free_bs(integrity_bio_set, bip->bip_vec, bip->bip_pool);
+	mempool_free(bip, bio_integrity_pool);
 
 	bio->bi_integrity = NULL;
 }
@@ -686,19 +667,17 @@  EXPORT_SYMBOL(bio_integrity_split);
  * @bio:	New bio
  * @bio_src:	Original bio
  * @gfp_mask:	Memory allocation mask
- * @bs:		bio_set to allocate bip from
  *
  * Description:	Called to allocate a bip when cloning a bio
  */
-int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
-			gfp_t gfp_mask, struct bio_set *bs)
+int bio_integrity_clone(struct bio *bio, struct bio *bio_src, gfp_t gfp_mask)
 {
 	struct bio_integrity_payload *bip_src = bio_src->bi_integrity;
 	struct bio_integrity_payload *bip;
 
 	BUG_ON(bip_src == NULL);
 
-	bip = bio_integrity_alloc_bioset(bio, gfp_mask, bip_src->bip_vcnt, bs);
+	bip = bio_integrity_alloc(bio, gfp_mask, bip_src->bip_vcnt);
 
 	if (bip == NULL)
 		return -EIO;
@@ -714,37 +693,25 @@  int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
 }
 EXPORT_SYMBOL(bio_integrity_clone);
 
-int bioset_integrity_create(struct bio_set *bs, int pool_size)
+static int __init bio_integrity_init(void)
 {
-	bs->bio_integrity_pool = mempool_create_slab_pool(pool_size,
-							  bio_integrity_slab);
-	if (!bs->bio_integrity_pool)
-		return -1;
-
-	return 0;
-}
-EXPORT_SYMBOL(bioset_integrity_create);
+	kintegrityd_wq = create_workqueue("kintegrityd");
 
-void bioset_integrity_free(struct bio_set *bs)
-{
-	if (bs->bio_integrity_pool)
-		mempool_destroy(bs->bio_integrity_pool);
-}
-EXPORT_SYMBOL(bioset_integrity_free);
+	if (!kintegrityd_wq)
+		panic("Failed to create kintegrityd\n");
 
-void __init bio_integrity_init_slab(void)
-{
 	bio_integrity_slab = KMEM_CACHE(bio_integrity_payload,
 					SLAB_HWCACHE_ALIGN|SLAB_PANIC);
-}
 
-static int __init integrity_init(void)
-{
-	kintegrityd_wq = create_workqueue("kintegrityd");
+	bio_integrity_pool = mempool_create_slab_pool(BIO_POOL_SIZE,
+						      bio_integrity_slab);
+	if (!bio_integrity_pool)
+		panic("bio_integrity: can't allocate bip pool\n");
 
-	if (!kintegrityd_wq)
-		panic("Failed to create kintegrityd\n");
+	integrity_bio_set = bioset_create(BIO_POOL_SIZE, 0);
+	if (!integrity_bio_set)
+		panic("bio_integrity: can't allocate bio_set\n");
 
 	return 0;
 }
-subsys_initcall(integrity_init);
+subsys_initcall(bio_integrity_init);
diff --git a/fs/bio.c b/fs/bio.c
index 9cc1430..a040cde 100644
--- a/fs/bio.c
+++ b/fs/bio.c
@@ -248,7 +248,7 @@  void bio_free(struct bio *bio, struct bio_set *bs)
 		bvec_free_bs(bs, bio->bi_io_vec, BIO_POOL_IDX(bio));
 
 	if (bio_integrity(bio))
-		bio_integrity_free(bio, bs);
+		bio_integrity_free(bio);
 
 	/*
 	 * If we have front padding, adjust the bio pointer before freeing
@@ -466,7 +466,7 @@  struct bio *bio_clone(struct bio *bio, gfp_t gfp_mask)
 	if (bio_integrity(bio)) {
 		int ret;
 
-		ret = bio_integrity_clone(b, bio, gfp_mask, fs_bio_set);
+		ret = bio_integrity_clone(b, bio, gfp_mask);
 
 		if (ret < 0) {
 			bio_put(b);
@@ -1529,7 +1529,6 @@  void bioset_free(struct bio_set *bs)
 	if (bs->bio_pool)
 		mempool_destroy(bs->bio_pool);
 
-	bioset_integrity_free(bs);
 	biovec_free_pools(bs);
 	bio_put_slab(bs);
 
@@ -1570,9 +1569,6 @@  struct bio_set *bioset_create(unsigned int pool_size, unsigned int front_pad)
 	if (!bs->bio_pool)
 		goto bad;
 
-	if (bioset_integrity_create(bs, pool_size))
-		goto bad;
-
 	if (!biovec_create_pools(bs, pool_size))
 		return bs;
 
@@ -1610,7 +1606,6 @@  static int __init init_bio(void)
 	if (!bio_slabs)
 		panic("bio: can't allocate bios\n");
 
-	bio_integrity_init_slab();
 	biovec_init_slabs();
 
 	fs_bio_set = bioset_create(BIO_POOL_SIZE, 0);
diff --git a/include/linux/bio.h b/include/linux/bio.h
index d8bd43b..b05b1d4 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -426,9 +426,6 @@  struct bio_set {
 	unsigned int front_pad;
 
 	mempool_t *bio_pool;
-#if defined(CONFIG_BLK_DEV_INTEGRITY)
-	mempool_t *bio_integrity_pool;
-#endif
 	mempool_t *bvec_pool;
 };
 
@@ -519,9 +516,8 @@  static inline int bio_has_data(struct bio *bio)
 
 #define bio_integrity(bio) (bio->bi_integrity != NULL)
 
-extern struct bio_integrity_payload *bio_integrity_alloc_bioset(struct bio *, gfp_t, unsigned int, struct bio_set *);
 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
-extern void bio_integrity_free(struct bio *, struct bio_set *);
+extern void bio_integrity_free(struct bio *);
 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
 extern int bio_integrity_enabled(struct bio *bio);
 extern int bio_integrity_set_tag(struct bio *, void *, unsigned int);
@@ -531,27 +527,21 @@  extern void bio_integrity_endio(struct bio *, int);
 extern void bio_integrity_advance(struct bio *, unsigned int);
 extern void bio_integrity_trim(struct bio *, unsigned int, unsigned int);
 extern void bio_integrity_split(struct bio *, struct bio_pair *, int);
-extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t, struct bio_set *);
-extern int bioset_integrity_create(struct bio_set *, int);
-extern void bioset_integrity_free(struct bio_set *);
-extern void bio_integrity_init_slab(void);
+extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
 
 #else /* CONFIG_BLK_DEV_INTEGRITY */
 
 #define bio_integrity(a)		(0)
-#define bioset_integrity_create(a, b)	(0)
 #define bio_integrity_prep(a)		(0)
 #define bio_integrity_enabled(a)	(0)
-#define bio_integrity_clone(a, b, c,d )	(0)
-#define bioset_integrity_free(a)	do { } while (0)
-#define bio_integrity_free(a, b)	do { } while (0)
+#define bio_integrity_clone(a, b, c)	(0)
+#define bio_integrity_free(a)		do { } while (0)
 #define bio_integrity_endio(a, b)	do { } while (0)
 #define bio_integrity_advance(a, b)	do { } while (0)
 #define bio_integrity_trim(a, b, c)	do { } while (0)
 #define bio_integrity_split(a, b, c)	do { } while (0)
 #define bio_integrity_set_tag(a, b, c)	do { } while (0)
 #define bio_integrity_get_tag(a, b, c)	do { } while (0)
-#define bio_integrity_init_slab(a)	do { } while (0)
 
 #endif /* CONFIG_BLK_DEV_INTEGRITY */