| Message ID | 1358353508-5369-13-git-send-email-benoit@irqsave.net |
|---|---|
| State | New |
| Headers | show |
On 01/16/2013 09:25 AM, Benoît Canet wrote: > --- > block/qcow2-dedup.c | 13 +++++++++++++ > block/qcow2.h | 1 + > 2 files changed, 14 insertions(+) > > diff --git a/block/qcow2-dedup.c b/block/qcow2-dedup.c > index db23b71..4305746 100644 > --- a/block/qcow2-dedup.c > +++ b/block/qcow2-dedup.c > @@ -1311,3 +1311,16 @@ void qcow2_dedup_close(BlockDriverState *bs) > { > qcow2_dedup_free(bs); > } > + > +#define GTREE_NODE_SIZE sizeof(int) * 5 Improperly parenthesized. Also, this feels like a magic number, is there an actual sizeof(struct) you could use instead of hand-computing how much is used per node? > + > +void qcow2_dedup_update_metrics(BlockDriverState *bs) > +{ > + BDRVQcowState *s = bs->opaque; > + > + uint64_t nb_hashs = s->dedup_metrics.ram_hash_creations - > + s->dedup_metrics.ram_hash_deletions; > + > + s->dedup_metrics.ram_usage = nb_hashs * GTREE_NODE_SIZE * 2; But you got lucky that order of operations didn't care about the missing () here.
Le Wednesday 16 Jan 2013 à 13:10:12 (-0700), Eric Blake a écrit : > On 01/16/2013 09:25 AM, Benoît Canet wrote: > > --- > > block/qcow2-dedup.c | 13 +++++++++++++ > > block/qcow2.h | 1 + > > 2 files changed, 14 insertions(+) > > > > diff --git a/block/qcow2-dedup.c b/block/qcow2-dedup.c > > index db23b71..4305746 100644 > > --- a/block/qcow2-dedup.c > > +++ b/block/qcow2-dedup.c > > @@ -1311,3 +1311,16 @@ void qcow2_dedup_close(BlockDriverState *bs) > > { > > qcow2_dedup_free(bs); > > } > > + > > +#define GTREE_NODE_SIZE sizeof(int) * 5 > > Improperly parenthesized. Also, this feels like a magic number, is > there an actual sizeof(struct) you could use instead of hand-computing > how much is used per node? No the glib implementation totally hide it's structures.
On 01/17/2013 04:11 AM, Benoît Canet wrote: > Le Wednesday 16 Jan 2013 à 13:10:12 (-0700), Eric Blake a écrit : >> On 01/16/2013 09:25 AM, Benoît Canet wrote: >>> --- >>> block/qcow2-dedup.c | 13 +++++++++++++ >>> block/qcow2.h | 1 + >>> 2 files changed, 14 insertions(+) >>> >>> diff --git a/block/qcow2-dedup.c b/block/qcow2-dedup.c >>> index db23b71..4305746 100644 >>> --- a/block/qcow2-dedup.c >>> +++ b/block/qcow2-dedup.c >>> @@ -1311,3 +1311,16 @@ void qcow2_dedup_close(BlockDriverState *bs) >>> { >>> qcow2_dedup_free(bs); >>> } >>> + >>> +#define GTREE_NODE_SIZE sizeof(int) * 5 >> >> Improperly parenthesized. Also, this feels like a magic number, is >> there an actual sizeof(struct) you could use instead of hand-computing >> how much is used per node? > > No the glib implementation totally hide it's structures. Also, are you sure that this value is correct on both 32-bit and 64-bit machines, or does a gtree node include a void* which changes how much memory it uses based on platform? Even adding a comment pointing to the internal glib implementation that you were copying from would be helpful, so that it doesn't feel quite so magic.
diff --git a/block/qcow2-dedup.c b/block/qcow2-dedup.c index db23b71..4305746 100644 --- a/block/qcow2-dedup.c +++ b/block/qcow2-dedup.c @@ -1311,3 +1311,16 @@ void qcow2_dedup_close(BlockDriverState *bs) { qcow2_dedup_free(bs); } + +#define GTREE_NODE_SIZE sizeof(int) * 5 + +void qcow2_dedup_update_metrics(BlockDriverState *bs) +{ + BDRVQcowState *s = bs->opaque; + + uint64_t nb_hashs = s->dedup_metrics.ram_hash_creations - + s->dedup_metrics.ram_hash_deletions; + + s->dedup_metrics.ram_usage = nb_hashs * GTREE_NODE_SIZE * 2; + s->dedup_metrics.ram_usage += nb_hashs * sizeof(QCowHashNode); +} diff --git a/block/qcow2.h b/block/qcow2.h index 0729ff2..d8e8539 100644 --- a/block/qcow2.h +++ b/block/qcow2.h @@ -510,5 +510,6 @@ void qcow2_dedup_refcount_half_max_reached(BlockDriverState *bs, bool qcow2_dedup_is_running(BlockDriverState *bs); int qcow2_dedup_init(BlockDriverState *bs); void qcow2_dedup_close(BlockDriverState *bs); +void qcow2_dedup_update_metrics(BlockDriverState *bs); #endif