| Message ID | 20170507000552.20847-6-eblake@redhat.com |
|---|---|
| State | New |
| Headers | show |
On 07.05.2017 02:05, Eric Blake wrote: > Although it doesn't add all that much type safety (this is C, after > all), it does add a bit of legibility to use the name QCow2ClusterType > instead of a plain int. > > In particular, qcow2_get_cluster_offset() has an overloaded return > type; a QCow2ClusterType on success, and -errno on failure; keeping > the cluster type in a separate variable makes it slightly easier for > the next patch to make further computations based on the type. > > Suggested-by: Max Reitz <mreitz@redhat.com> > Signed-off-by: Eric Blake <eblake@redhat.com> > > --- > v13: new patch > --- > block/qcow2.h | 6 +++--- > block/qcow2-cluster.c | 17 +++++++++-------- > block/qcow2-refcount.c | 2 +- > 3 files changed, 13 insertions(+), 12 deletions(-) [...] > diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c > index f3bfce6..ed78a30 100644 > --- a/block/qcow2-cluster.c > +++ b/block/qcow2-cluster.c Above this hunk, there is count_contiguous_clusters() with a variable "first_cluster_type" that could be a QCow2ClusterType as well. > @@ -340,7 +340,7 @@ static int count_contiguous_clusters(int nb_clusters, int cluster_size, > */ > static int count_contiguous_clusters_unallocated(int nb_clusters, > uint64_t *l2_table, > - int wanted_type) > + QCow2ClusterType wanted_type) > { > int i; > And some lines below this (in this function), there is a "type" variable that (c|sh)ould be a QCow2ClusterType, too. Although it's quite a functional change, I would be willing to change both when applying, if you allowed me to. (Once again, "no good deed shall go unpunished", as Markus likes to say :-)) Max
On 08.05.2017 17:43, Max Reitz wrote: > On 07.05.2017 02:05, Eric Blake wrote: >> Although it doesn't add all that much type safety (this is C, after >> all), it does add a bit of legibility to use the name QCow2ClusterType >> instead of a plain int. >> >> In particular, qcow2_get_cluster_offset() has an overloaded return >> type; a QCow2ClusterType on success, and -errno on failure; keeping >> the cluster type in a separate variable makes it slightly easier for >> the next patch to make further computations based on the type. >> >> Suggested-by: Max Reitz <mreitz@redhat.com> >> Signed-off-by: Eric Blake <eblake@redhat.com> >> >> --- >> v13: new patch >> --- >> block/qcow2.h | 6 +++--- >> block/qcow2-cluster.c | 17 +++++++++-------- >> block/qcow2-refcount.c | 2 +- >> 3 files changed, 13 insertions(+), 12 deletions(-) > > [...] > >> diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c >> index f3bfce6..ed78a30 100644 >> --- a/block/qcow2-cluster.c >> +++ b/block/qcow2-cluster.c > > Above this hunk, there is count_contiguous_clusters() with a variable > "first_cluster_type" that could be a QCow2ClusterType as well. > >> @@ -340,7 +340,7 @@ static int count_contiguous_clusters(int nb_clusters, int cluster_size, >> */ >> static int count_contiguous_clusters_unallocated(int nb_clusters, >> uint64_t *l2_table, >> - int wanted_type) >> + QCow2ClusterType wanted_type) >> { >> int i; >> > > And some lines below this (in this function), there is a "type" variable > that (c|sh)ould be a QCow2ClusterType, too. I see you're doing this in the next patch... Well, that's not really where it belongs, though. > Although it's quite a functional change, I would be willing to change > both when applying, if you allowed me to. I guess I'm extending my offer to handle the rebase fallout of the next patch, too...? Max > (Once again, "no good deed shall go unpunished", as Markus likes to say :-)) > > Max >
On 05/08/2017 10:43 AM, Max Reitz wrote: > On 07.05.2017 02:05, Eric Blake wrote: >> Although it doesn't add all that much type safety (this is C, after >> all), it does add a bit of legibility to use the name QCow2ClusterType >> instead of a plain int. >> >> In particular, qcow2_get_cluster_offset() has an overloaded return >> type; a QCow2ClusterType on success, and -errno on failure; keeping >> the cluster type in a separate variable makes it slightly easier for >> the next patch to make further computations based on the type. >> >> Suggested-by: Max Reitz <mreitz@redhat.com> >> Signed-off-by: Eric Blake <eblake@redhat.com> >> >> --- >> v13: new patch >> --- >> block/qcow2.h | 6 +++--- >> block/qcow2-cluster.c | 17 +++++++++-------- >> block/qcow2-refcount.c | 2 +- >> 3 files changed, 13 insertions(+), 12 deletions(-) > > [...] > >> diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c >> index f3bfce6..ed78a30 100644 >> --- a/block/qcow2-cluster.c >> +++ b/block/qcow2-cluster.c > > Above this hunk, there is count_contiguous_clusters() with a variable > "first_cluster_type" that could be a QCow2ClusterType as well. Hmm, I guess I didn't catch them all. Yes, that one definitely qualified. > >> @@ -340,7 +340,7 @@ static int count_contiguous_clusters(int nb_clusters, int cluster_size, >> */ >> static int count_contiguous_clusters_unallocated(int nb_clusters, >> uint64_t *l2_table, >> - int wanted_type) >> + QCow2ClusterType wanted_type) >> { >> int i; >> > > And some lines below this (in this function), there is a "type" variable > that (c|sh)ould be a QCow2ClusterType, too. That one I blame on poor rebasing on my side. > > Although it's quite a functional change, I would be willing to change > both when applying, if you allowed me to. > > (Once again, "no good deed shall go unpunished", as Markus likes to say :-)) Yes, both changes are appropriate, and I'm fine with you making the tweaks as part of putting it on your tree (and the corresponding fallout on 6/12).
diff --git a/block/qcow2.h b/block/qcow2.h index 8731f24..c148bbc 100644 --- a/block/qcow2.h +++ b/block/qcow2.h @@ -349,12 +349,12 @@ typedef struct QCowL2Meta QLIST_ENTRY(QCowL2Meta) next_in_flight; } QCowL2Meta; -enum { +typedef enum QCow2ClusterType { QCOW2_CLUSTER_UNALLOCATED, QCOW2_CLUSTER_NORMAL, QCOW2_CLUSTER_COMPRESSED, QCOW2_CLUSTER_ZERO -}; +} QCow2ClusterType; typedef enum QCow2MetadataOverlap { QCOW2_OL_MAIN_HEADER_BITNR = 0, @@ -443,7 +443,7 @@ static inline uint64_t qcow2_max_refcount_clusters(BDRVQcow2State *s) return QCOW_MAX_REFTABLE_SIZE >> s->cluster_bits; } -static inline int qcow2_get_cluster_type(uint64_t l2_entry) +static inline QCow2ClusterType qcow2_get_cluster_type(uint64_t l2_entry) { if (l2_entry & QCOW_OFLAG_COMPRESSED) { return QCOW2_CLUSTER_COMPRESSED; diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c index f3bfce6..ed78a30 100644 --- a/block/qcow2-cluster.c +++ b/block/qcow2-cluster.c @@ -340,7 +340,7 @@ static int count_contiguous_clusters(int nb_clusters, int cluster_size, */ static int count_contiguous_clusters_unallocated(int nb_clusters, uint64_t *l2_table, - int wanted_type) + QCow2ClusterType wanted_type) { int i; @@ -500,6 +500,7 @@ int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset, int l1_bits, c; unsigned int offset_in_cluster; uint64_t bytes_available, bytes_needed, nb_clusters; + QCow2ClusterType type; int ret; offset_in_cluster = offset_into_cluster(s, offset); @@ -522,13 +523,13 @@ int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset, l1_index = offset >> l1_bits; if (l1_index >= s->l1_size) { - ret = QCOW2_CLUSTER_UNALLOCATED; + type = QCOW2_CLUSTER_UNALLOCATED; goto out; } l2_offset = s->l1_table[l1_index] & L1E_OFFSET_MASK; if (!l2_offset) { - ret = QCOW2_CLUSTER_UNALLOCATED; + type = QCOW2_CLUSTER_UNALLOCATED; goto out; } @@ -557,8 +558,8 @@ int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset, * true */ assert(nb_clusters <= INT_MAX); - ret = qcow2_get_cluster_type(*cluster_offset); - switch (ret) { + type = qcow2_get_cluster_type(*cluster_offset); + switch (type) { case QCOW2_CLUSTER_COMPRESSED: /* Compressed clusters can only be processed one by one */ c = 1; @@ -633,7 +634,7 @@ out: assert(bytes_available - offset_in_cluster <= UINT_MAX); *bytes = bytes_available - offset_in_cluster; - return ret; + return type; fail: qcow2_cache_put(bs, s->l2_table_cache, (void **)&l2_table); @@ -891,7 +892,7 @@ static int count_cow_clusters(BDRVQcow2State *s, int nb_clusters, for (i = 0; i < nb_clusters; i++) { uint64_t l2_entry = be64_to_cpu(l2_table[l2_index + i]); - int cluster_type = qcow2_get_cluster_type(l2_entry); + QCow2ClusterType cluster_type = qcow2_get_cluster_type(l2_entry); switch(cluster_type) { case QCOW2_CLUSTER_NORMAL: @@ -1757,7 +1758,7 @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table, for (j = 0; j < s->l2_size; j++) { uint64_t l2_entry = be64_to_cpu(l2_table[j]); int64_t offset = l2_entry & L2E_OFFSET_MASK; - int cluster_type = qcow2_get_cluster_type(l2_entry); + QCow2ClusterType cluster_type = qcow2_get_cluster_type(l2_entry); bool preallocated = offset != 0; if (cluster_type != QCOW2_CLUSTER_ZERO) { diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c index 908dbe5..e639b34 100644 --- a/block/qcow2-refcount.c +++ b/block/qcow2-refcount.c @@ -1640,7 +1640,7 @@ static int check_oflag_copied(BlockDriverState *bs, BdrvCheckResult *res, for (j = 0; j < s->l2_size; j++) { uint64_t l2_entry = be64_to_cpu(l2_table[j]); uint64_t data_offset = l2_entry & L2E_OFFSET_MASK; - int cluster_type = qcow2_get_cluster_type(l2_entry); + QCow2ClusterType cluster_type = qcow2_get_cluster_type(l2_entry); if ((cluster_type == QCOW2_CLUSTER_NORMAL) || ((cluster_type == QCOW2_CLUSTER_ZERO) && (data_offset != 0))) {
Although it doesn't add all that much type safety (this is C, after all), it does add a bit of legibility to use the name QCow2ClusterType instead of a plain int. In particular, qcow2_get_cluster_offset() has an overloaded return type; a QCow2ClusterType on success, and -errno on failure; keeping the cluster type in a separate variable makes it slightly easier for the next patch to make further computations based on the type. Suggested-by: Max Reitz <mreitz@redhat.com> Signed-off-by: Eric Blake <eblake@redhat.com> --- v13: new patch --- block/qcow2.h | 6 +++--- block/qcow2-cluster.c | 17 +++++++++-------- block/qcow2-refcount.c | 2 +- 3 files changed, 13 insertions(+), 12 deletions(-)