[PATCH] bcache: untagle cache_aolloc

[PATCH] bcache: untagle cache_aolloc

[Johannes Thumshirn]

bcache's cache_alloc() function currenty has no way freeing memory if one of
the allocations fails. Untangle the if + allocation statement so we have
defined checkpoints to free previous allocations if one fails.

Signed-off-by: Johannes Thumshirn <jthumshirn@suse.de>
Tested-by: Coly Li <colyli@suse.de>
---
 drivers/md/bcache/super.c | 51 +++++++++++++++++++++++++++++++++++++----------
 1 file changed, 40 insertions(+), 11 deletions(-)

diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c
index f5dbb4e..b5703a9 100644
--- a/drivers/md/bcache/super.c
+++ b/drivers/md/bcache/super.c
@@ -1817,18 +1817,29 @@ static int cache_alloc(struct cache_sb *sb, struct cache *ca)
 
 	free = roundup_pow_of_two(ca->sb.nbuckets) >> 10;
 
-	if (!init_fifo(&ca->free[RESERVE_BTREE], 8, GFP_KERNEL) ||
-	    !init_fifo(&ca->free[RESERVE_PRIO], prio_buckets(ca), GFP_KERNEL) ||
-	    !init_fifo(&ca->free[RESERVE_MOVINGGC], free, GFP_KERNEL) ||
-	    !init_fifo(&ca->free[RESERVE_NONE], free, GFP_KERNEL) ||
-	    !init_fifo(&ca->free_inc,	free << 2, GFP_KERNEL) ||
-	    !init_heap(&ca->heap,	free << 3, GFP_KERNEL) ||
-	    !(ca->buckets	= vzalloc(sizeof(struct bucket) *
-					  ca->sb.nbuckets)) ||
-	    !(ca->prio_buckets	= kzalloc(sizeof(uint64_t) * prio_buckets(ca) *
-					  2, GFP_KERNEL)) ||
-	    !(ca->disk_buckets	= alloc_bucket_pages(GFP_KERNEL, ca)))
+	if (!init_fifo(&ca->free[RESERVE_BTREE], 8, GFP_KERNEL))
 		return -ENOMEM;
+	if (!init_fifo(&ca->free[RESERVE_PRIO], prio_buckets(ca), GFP_KERNEL))
+		goto free_btree_fifo;
+	if (!init_fifo(&ca->free[RESERVE_MOVINGGC], free, GFP_KERNEL))
+		goto free_prio_fifo;
+	if (!init_fifo(&ca->free[RESERVE_NONE], free, GFP_KERNEL))
+		goto free_movinggc_fifo;
+	if (!init_fifo(&ca->free_inc, free << 2, GFP_KERNEL))
+		goto free_none_fifo;
+	if (!init_heap(&ca->heap, free << 3, GFP_KERNEL))
+		goto free_inc_fifo;
+	ca->buckets = vzalloc(sizeof(struct bucket) * ca->sb.nbuckets);
+	if (!ca->buckets)
+		goto free_cache_heap;
+	ca->prio_buckets = kzalloc(sizeof(uint64_t) * prio_buckets(ca) * 2,
+				   GFP_KERNEL);
+	if (!ca->prio_buckets)
+		goto free_cache_buckets;
+
+	ca->disk_buckets = alloc_bucket_pages(GFP_KERNEL, ca);
+	if (!ca->disk_buckets)
+		goto free_prio_buckets;
 
 	ca->prio_last_buckets = ca->prio_buckets + prio_buckets(ca);
 
@@ -1836,6 +1847,24 @@ static int cache_alloc(struct cache_sb *sb, struct cache *ca)
 		atomic_set(&b->pin, 0);
 
 	return 0;
+
+free_prio_buckets:
+	kfree(ca->prio_buckets);
+free_cache_buckets:
+	vfree(ca->buckets);
+free_cache_heap:
+	free_heap(&ca->heap);
+free_inc_fifo:
+	free_fifo(&ca->free_inc);
+free_none_fifo:
+	free_fifo(&ca->free[RESERVE_NONE]);
+free_movinggc_fifo:
+	free_fifo(&ca->free[RESERVE_MOVINGGC]);
+free_prio_fifo:
+	free_fifo(&ca->free[RESERVE_PRIO]);
+free_btree_fifo:
+	free_fifo(&ca->free[RESERVE_BTREE]);
+	return -ENOMEM;
 }
 
 static int register_cache(struct cache_sb *sb, struct page *sb_page,
-- 
1.8.5.6

Re: [PATCH] bcache: untagle cache_aolloc

[Kent Overstreet]

On Mon, Jul 18, 2016 at 12:11:09PM +0200, Johannes Thumshirn wrote:
> bcache's cache_alloc() function currenty has no way freeing memory if one of
> the allocations fails. Untangle the if + allocation statement so we have
> defined checkpoints to free previous allocations if one fails.

nack. The existing error path handles failure midway through just fine.

Re: [PATCH] bcache: untagle cache_aolloc

[Johannes Thumshirn]

On Mon, Jul 18, 2016 at 02:13:33AM -0800, Kent Overstreet wrote:
> On Mon, Jul 18, 2016 at 12:11:09PM +0200, Johannes Thumshirn wrote:
> > bcache's cache_alloc() function currenty has no way freeing memory if one of
> > the allocations fails. Untangle the if + allocation statement so we have
> > defined checkpoints to free previous allocations if one fails.
> 
> nack. The existing error path handles failure midway through just fine.

Come on, the patch improves the readability of the if statement by some orders
of magnitude as well.

Are you OK with it if I change the subject/commit log?

	Johannes

-- 
Johannes Thumshirn                                          Storage
jthumshirn@suse.de                                +49 911 74053 689
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: Felix Imendörffer, Jane Smithard, Graham Norton
HRB 21284 (AG Nürnberg)
Key fingerprint = EC38 9CAB C2C4 F25D 8600 D0D0 0393 969D 2D76 0850

Re: [PATCH] bcache: untagle cache_aolloc

[Kent Overstreet]

On Mon, Jul 18, 2016 at 12:24:11PM +0200, Johannes Thumshirn wrote:
> On Mon, Jul 18, 2016 at 02:13:33AM -0800, Kent Overstreet wrote:
> > On Mon, Jul 18, 2016 at 12:11:09PM +0200, Johannes Thumshirn wrote:
> > > bcache's cache_alloc() function currenty has no way freeing memory if one of
> > > the allocations fails. Untangle the if + allocation statement so we have
> > > defined checkpoints to free previous allocations if one fails.
> > 
> > nack. The existing error path handles failure midway through just fine.
> 
> Come on, the patch improves the readability of the if statement by some orders
> of magnitude as well.
> 
> Are you OK with it if I change the subject/commit log?

No, it's just churn and I don't agree that it improves readability. On the
contrary, now the cleanup code has to be duplicated in two places - which
invites them getting out of sync and introducing bugs.