Move generation, graph_pos to a slab
cover Move generation, graph_pos to a slab
The struct commit is used in many contexts. However, members generation and graph_pos are only used for commit-graph related operations and otherwise waste memory.
This wastage would have been more pronounced as transistion to generation number v2, which uses 64-bit generation number instead of current 32-bits.
The third patch ("commit: convert commit->graph_pos to a slab", 2020-06-04) is currently failing diff-submodule related tests (t4041, t4059 and t4060) for gcc [1]. I am going to send a second version soon, fixing that.
[1]: https://travis-ci.com/github/abhishekkumar2718/git/jobs/343441189
Abhishek Kumar (3):
commit: introduce helpers for generation slab
commit: convert commit->generation to a slab
commit: convert commit->graph_pos to a slab
alloc.c | 2 - blame.c | 2 +- bloom.c | 6 +- commit-graph.c | 116 +++++++++++++++++++++------- commit-graph.h | 8 ++ commit-reach.c | 50 ++++++------ commit.c | 6 +- commit.h | 6 -- contrib/coccinelle/generation.cocci | 12 +++ contrib/coccinelle/graph_pos.cocci | 12 +++ revision.c | 16 ++-- 11 files changed, 158 insertions(+), 78 deletions(-) create mode 100644 contrib/coccinelle/generation.cocci create mode 100644 contrib/coccinelle/graph_pos.cocci
Signature · 1 line
2.27.0
PATCH 1/3 commit: introduce helpers for generation slab
The struct member generation refers to "generation number" (or more broadly, a reachablity index value) used by commit-graph to reduce time taken to walk commits. However, generation is not useful in other contexts and bloats the struct.
Let's move it to a commit-slab and shrink the struct by four bytes.
--- commit-graph.c | 27 +++++++++++++++++++++++++++ commit-graph.h | 5 +++++ commit.h | 3 --- 3 files changed, 32 insertions(+), 3 deletions(-)
commit-graph.c +27 -0
@@ -87,6 +87,33 @@ static int commit_pos_cmp(const void *va, const void *vb)commit_pos_at (&commit_pos, b); } +define_commit_slab (generation_slab, uint32_t); +static struct generation_slab generation_slab =COMMIT_SLAB_INIT (1 , generation_slab); + +uint32_t generation (const struct commit *c) +{ +uint32_t *gen =generation_slab_peek (&generation_slab, c); + +return gen ? *gen :GENERATION_NUMBER_INFINITY ; +} + +static void set_generation (const struct commit *c,const uint32_t generation) +{ +unsigned int i = generation_slab.slab_count; +uint32_t *gen =generation_slab_at (&generation_slab, c); + +/* +* commit-slab initializes with zero, overwrite this with +* GENERATION_NUMBER_INFINITY +*/ +for (; i < generation_slab.slab_count; ++i) { +memset (generation_slab.slab[i],GENERATION_NUMBER_INFINITY , + generation_slab.slab_size *sizeof (uint32_t )); + } + + *gen = generation; +} +static int commit_gen_cmp (const void *va,const void *vb) {const struct commit *a = *(const struct commit **)va;
commit-graph.h +5 -0
@@ -8,6 +8,10 @@#include "object-store.h" #include "oidset.h" +#define GENERATION_NUMBER_INFINITY 0xFFFFFFFF +#define GENERATION_NUMBER_MAX 0x3FFFFFFF +#define GENERATION_NUMBER_ZERO 0 +#define GIT_TEST_COMMIT_GRAPH "GIT_TEST_COMMIT_GRAPH"#define GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD "GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD"#define GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS "GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS"
@@ -137,4 +141,5 @@ void free_commit_graph(struct commit_graph *);*/ void disable_commit_graph (struct repository *r); +uint32_t generation (const struct commit *c);#endif
commit.h +0 -3
@@ -11,9 +11,6 @@#include "commit-slab.h" #define COMMIT_NOT_FROM_GRAPH 0xFFFFFFFF -#define GENERATION_NUMBER_INFINITY 0xFFFFFFFF -#define GENERATION_NUMBER_MAX 0x3FFFFFFF -#define GENERATION_NUMBER_ZERO 0struct commit_list {struct commit *item;
Signature · 1 line
2.27.0
Re: [GSoC Patch 1/3] commit: introduce helpers for generation slab
Quoted 21 lines · Abhishek Kumar
The struct member generation refers to "generation number" (or more broadly, a reachablity index value) used by commit-graph to reduce time taken to walk commits. However, generation is not useful in other contexts and bloats the struct.
Let's move it to a commit-slab and shrink the struct by four bytes.
Signed-off-by Abhishek Kumar <[email protected]>--- commit-graph.c | 27 +++++++++++++++++++++++++++ commit-graph.h | 5 +++++ commit.h | 3 --- 3 files changed, 32 insertions(+), 3 deletions(-)commit-graph.c +0
-0@@ -87,6 +87,33 @@ static int commit_pos_cmp(const void *va, const void *vb)commit_pos_at (&commit_pos, b); }
commit-graph.c static int commit_pos_cmp(const void *va, const void *vb) · line 98
Quoted 10 lines
commit-graph.c +9
-0+define_commit_slab (generation_slab, uint32_t); +static struct generation_slab generation_slab =COMMIT_SLAB_INIT (1 , generation_slab); + +uint32_t generation (const struct commit *c) +{ +uint32_t *gen =generation_slab_peek (&generation_slab, c); + +return gen ? *gen :GENERATION_NUMBER_INFINITY ; +}
This is good: if we don't have the value, then use INFINITY. In the header file, perhaps include a warning comment that a caller _must_ first parse the commit or else we have no guarantee that the generation slab is populated. This matches the current expectations before accessing the generation member.
commit-graph.c static int commit_pos_cmp(const void *va, const void *vb) · line 112
Quoted 13 lines
commit-graph.c +13
-0+static void set_generation (const struct commit *c,const uint32_t generation) +{ +unsigned int i = generation_slab.slab_count; +uint32_t *gen =generation_slab_at (&generation_slab, c); + +/* +* commit-slab initializes with zero, overwrite this with +* GENERATION_NUMBER_INFINITY +*/ +for (; i < generation_slab.slab_count; ++i) { +memset (generation_slab.slab[i],GENERATION_NUMBER_INFINITY , + generation_slab.slab_size *sizeof (uint32_t )); + }
Here is an example where combining the graph_pos and generation slabs into one would be helpful. The only reason the generation would be INFINITY is if graph_pos is COMMIT_NOT_FROM_GRAPH. If the two values are side-by-side, we could just check graph_pos first and return INFINITY instead of paying this initialization cost as the slab grows.
I would also like to avoid initializing the slab if there is no commit-graph present. I wonder if we can populate the slab while parsing the commit-graph and check here if the slab is NULL before doing any other logic? (I'm not sure if this is possible, but it would be nice.)
Quoted 26 lines
commit-graph.h +5
-0@@ -8,6 +8,10 @@#include "object-store.h" #include "oidset.h" +#define GENERATION_NUMBER_INFINITY 0xFFFFFFFF +#define GENERATION_NUMBER_MAX 0x3FFFFFFF +#define GENERATION_NUMBER_ZERO 0 +#define GIT_TEST_COMMIT_GRAPH "GIT_TEST_COMMIT_GRAPH"#define GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD "GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD"#define GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS "GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS"@@ -137,4 +141,5 @@ void free_commit_graph(struct commit_graph *);*/ void disable_commit_graph (struct repository *r); +uint32_t generation (const struct commit *c);#endif commit.h +0
-0
commit.h line 16
Quoted 6 lines
commit.h +0
-3@@ -11,9 +11,6 @@#include "commit-slab.h" #define COMMIT_NOT_FROM_GRAPH 0xFFFFFFFF -#define GENERATION_NUMBER_INFINITY 0xFFFFFFFF -#define GENERATION_NUMBER_MAX 0x3FFFFFFF -#define GENERATION_NUMBER_ZERO 0
I appreciate that you are able to relocate these constants to a more appropriate location.
Thanks,
-Stolee
Re: [GSoC Patch 1/3] commit: introduce helpers for generation slab
In reply to Abhishek KumarQuoted 32 lines · Abhishek Kumar
The struct member generation refers to "generation number" (or more broadly, a reachablity index value) used by commit-graph to reduce time taken to walk commits. However, generation is not useful in other contexts and bloats the struct.
Let's move it to a commit-slab and shrink the struct by four bytes.
Signed-off-by Abhishek Kumar <[email protected]>--- commit-graph.c | 27 +++++++++++++++++++++++++++ commit-graph.h | 5 +++++ commit.h | 3 --- 3 files changed, 32 insertions(+), 3 deletions(-)commit-graph.c +10
-0@@ -87,6 +87,33 @@ static int commit_pos_cmp(const void *va, const void *vb)commit_pos_at (&commit_pos, b); } +define_commit_slab (generation_slab, uint32_t); +static struct generation_slab generation_slab =COMMIT_SLAB_INIT (1 , generation_slab); + +uint32_t generation (const struct commit *c) +{ +uint32_t *gen =generation_slab_peek (&generation_slab, c); + +return gen ? *gen :GENERATION_NUMBER_INFINITY ; +} +
commit-graph.c static int commit_pos_cmp(const void *va, const void *vb) · line 109
Quoted 10 lines
commit-graph.c +10
-0+static void set_generation (const struct commit *c,const uint32_t generation) +{ +unsigned int i = generation_slab.slab_count; +uint32_t *gen =generation_slab_at (&generation_slab, c); + +/* +* commit-slab initializes with zero, overwrite this with +* GENERATION_NUMBER_INFINITY +*/ +for (; i < generation_slab.slab_count; ++i) {
Style: favor post-increment over pre-increment when there is no difference, especially when updating the loop control in for() loop.
Quoted 15 lines
commit-graph.c +7
-0+memset (generation_slab.slab[i],GENERATION_NUMBER_INFINITY , + generation_slab.slab_size *sizeof (uint32_t )); + } + + *gen = generation; +} +static int commit_gen_cmp (const void *va,const void *vb) {const struct commit *a = *(const struct commit **)va;commit-graph.h +0
-0
commit-graph.h line 14
Quoted 7 lines
commit-graph.h +4
-0@@ -8,6 +8,10 @@#include "object-store.h" #include "oidset.h" +#define GENERATION_NUMBER_INFINITY 0xFFFFFFFF +#define GENERATION_NUMBER_MAX 0x3FFFFFFF +#define GENERATION_NUMBER_ZERO 0 +
Makes sense to move it from commit.h, I guess.
Quoted 23 lines
commit-graph.h +1
-0#define GIT_TEST_COMMIT_GRAPH "GIT_TEST_COMMIT_GRAPH"#define GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD "GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD"#define GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS "GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS"@@ -137,4 +141,5 @@ void free_commit_graph(struct commit_graph *);*/ void disable_commit_graph (struct repository *r); +uint32_t generation (const struct commit *c);#endif commit.h +0
-3@@ -11,9 +11,6 @@#include "commit-slab.h" #define COMMIT_NOT_FROM_GRAPH 0xFFFFFFFF -#define GENERATION_NUMBER_INFINITY 0xFFFFFFFF -#define GENERATION_NUMBER_MAX 0x3FFFFFFF -#define GENERATION_NUMBER_ZERO 0struct commit_list {struct commit *item;
Re: [GSoC Patch 1/3] commit: introduce helpers for generation slab
In reply to Abhishek KumarQuoted 1 line · Abhishek Kumar
The struct member generation refers to "generation number" (or more
Again, a minor thing: 'generation'.
Quoted 5 lines
broadly, a reachablity index value) used by commit-graph to reduce time taken to walk commits. However, generation is not useful in other contexts and bloats the struct.
Let's move it to a commit-slab and shrink the struct by four bytes.
It looks like the description is from earlier version of the commit, before it was split -- because this commit does not remove 'generation' member from the 'struct commit', actually.
This commit is about creating helper functions.
Quoted 13 lines
Signed-off-by Abhishek Kumar <[email protected]>--- commit-graph.c | 27 +++++++++++++++++++++++++++ commit-graph.h | 5 +++++ commit.h | 3 --- 3 files changed, 32 insertions(+), 3 deletions(-)commit-graph.c +0
-0
commit-graph.c static int commit_pos_cmp(const void *va, const void *vb) · line 91
Quoted 5 lines
commit-graph.c +2
-0@@ -87,6 +87,33 @@ static int commit_pos_cmp(const void *va, const void *vb)commit_pos_at (&commit_pos, b); } +define_commit_slab (generation_slab, uint32_t); +static struct generation_slab generation_slab =COMMIT_SLAB_INIT (1 , generation_slab);
All right, we need this for the following helper functions to work.
We might want to encapsulate all commit-graph data together, in a single struct (e.g. as 'struct commit_graph_data' instead of uint32_t here).
On the other hand other data is stored on slab often as separate scalar data (contains_cache, commit_seen, indegree_slab, author_date_slab, commit_base, commit_pos), but not always; sometimes it is a struct (bloom_filter_slab, buffer_slab, commit_rev_name), sometimes it is an array (commit_depth, ref_bitmap, commit_weight), and sometimes it is an array/list of structs or pointer to struct (commit_names, commit_name_slab, saved_parents, blame_suspects, commit_todo_item).
commit-graph.c static int commit_pos_cmp(const void *va, const void *vb) · line 98
Quoted 7 lines
commit-graph.c +7
-0+ +uint32_t generation (const struct commit *c) +{ +uint32_t *gen =generation_slab_peek (&generation_slab, c); + +return gen ? *gen :GENERATION_NUMBER_INFINITY ; +}
All right, this is a synthetic getter using the fact that commits outside the commit-graph should get GENERATION_NUMBER_INFINITY (because [effective] commit-graph is closed under reachability, is full DAG).
Should we have something like that for 'graph_pos' and COMMIT_NOT_FROM_GRAPH?
Quoted 7 lines
commit-graph.c +7
-0+ +static void set_generation (const struct commit *c,const uint32_t generation) +{ +unsigned int i = generation_slab.slab_count; +uint32_t *gen =generation_slab_at (&generation_slab, c); + +/*
commit-graph.c static int commit_pos_cmp(const void *va, const void *vb) · line 115
Quoted 10 lines
commit-graph.c +10
-0+* commit-slab initializes with zero, overwrite this with +* GENERATION_NUMBER_INFINITY +*/ +for (; i < generation_slab.slab_count; ++i) { +memset (generation_slab.slab[i],GENERATION_NUMBER_INFINITY , + generation_slab.slab_size *sizeof (uint32_t )); + } + + *gen = generation; +}
All right. I wonder if putting 'generation' and 'graph_pos' on the slab together, gathered in 'struct commit_graph_data' would make this helper more complex...
Quoted 30 lines
commit-graph.c +1
-0+static int commit_gen_cmp (const void *va,const void *vb) {const struct commit *a = *(const struct commit **)va;commit-graph.h +5
-0@@ -8,6 +8,10 @@#include "object-store.h" #include "oidset.h" +#define GENERATION_NUMBER_INFINITY 0xFFFFFFFF +#define GENERATION_NUMBER_MAX 0x3FFFFFFF +#define GENERATION_NUMBER_ZERO 0 +#define GIT_TEST_COMMIT_GRAPH "GIT_TEST_COMMIT_GRAPH"#define GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD "GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD"#define GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS "GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS"@@ -137,4 +141,5 @@ void free_commit_graph(struct commit_graph *);*/ void disable_commit_graph (struct repository *r); +uint32_t generation (const struct commit *c);#endif commit.h +0
-0
commit.h line 16
Quoted 9 lines
commit.h +0
-3@@ -11,9 +11,6 @@#include "commit-slab.h" #define COMMIT_NOT_FROM_GRAPH 0xFFFFFFFF -#define GENERATION_NUMBER_INFINITY 0xFFFFFFFF -#define GENERATION_NUMBER_MAX 0x3FFFFFFF -#define GENERATION_NUMBER_ZERO 0struct commit_list {struct commit *item;
Why this change?
Best,
Signature · 1 line
Jakub Narębski
PATCH 2/3 commit: convert commit->generation to a slab
In reply to Abhishek KumarIn this commit, we will use the generation slab helpers introduced in last commit and replace existing uses of commit->generation using 'contrib/coccinelle/generation.cocci'
--- alloc.c | 1 - blame.c | 2 +- commit-graph.c | 39 +++++++++++----------- commit-reach.c | 50 ++++++++++++++--------------- commit.c | 4 +-- commit.h | 1 - contrib/coccinelle/generation.cocci | 12 +++++++ revision.c | 16 ++++----- 8 files changed, 68 insertions(+), 57 deletions(-) create mode 100644 contrib/coccinelle/generation.cocci
alloc.c +0 -1
@@ -109,7 +109,6 @@ void init_commit_node(struct repository *r, struct commit *c) c->object.type =OBJ_COMMIT ; c->index =alloc_commit_index (r); c->graph_pos =COMMIT_NOT_FROM_GRAPH ; - c->generation =GENERATION_NUMBER_INFINITY ; }void *alloc_commit_node (struct repository *r)
blame.c +1 -1
@@ -1272,7 +1272,7 @@ static int maybe_changed_path(struct repository *r,if (!bd)return 1 ; -if (origin->commit->generation ==GENERATION_NUMBER_INFINITY ) +if (generation (origin->commit) ==GENERATION_NUMBER_INFINITY )return 1 ; filter =get_bloom_filter (r, origin->commit,0 );
commit-graph.c +20 -19
@@ -120,9 +120,9 @@ static int commit_gen_cmp(const void *va, const void *vb)const struct commit *b = *(const struct commit **)vb;/* lower generation commits first */ -if (a->generation < b->generation) +if (generation (a) <generation (b))return -1 ; -else if (a->generation > b->generation) +else if (generation (a) >generation (b))return 1 ;/* use date as a heuristic when generations are equal */
@@ -712,7 +712,7 @@ static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, lex_index = pos - g->num_commits_in_base; commit_data = g->chunk_commit_data +GRAPH_DATA_WIDTH * lex_index; item->graph_pos = pos; - item->generation =get_be32 (commit_data + g->hash_len +8 ) >>2 ; +set_generation (item,get_be32 (commit_data + g->hash_len +8 ) >>2 ); }static inline void set_commit_tree (struct commit *c,struct tree *t)
@@ -754,7 +754,7 @@ static int fill_commit_in_graph(struct repository *r, date_low =get_be32 (commit_data + g->hash_len +12 ); item->date = (timestamp_t)((date_high <<32 ) | date_low); - item->generation =get_be32 (commit_data + g->hash_len +8 ) >>2 ; +set_generation (item,get_be32 (commit_data + g->hash_len +8 ) >>2 ); pptr = &item->parents;
@@ -1048,7 +1048,7 @@ static void write_graph_chunk_data(struct hashfile *f, int hash_len,else packedDate[0 ] =0 ; - packedDate[0 ] |=htonl ((*list)->generation <<2 ); + packedDate[0 ] |=htonl (generation ((*list)) <<2 ); packedDate[1 ] =htonl ((*list)->date);hashwrite (f, packedDate,8 );
@@ -1280,8 +1280,8 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx) ctx->commits.nr);for (i =0 ; i < ctx->commits.nr; i++) {display_progress (ctx->progress, i +1 ); -if (ctx->commits.list[i]->generation !=GENERATION_NUMBER_INFINITY && - ctx->commits.list[i]->generation !=GENERATION_NUMBER_ZERO ) +if (generation (ctx->commits.list[i]) !=GENERATION_NUMBER_INFINITY && +generation (ctx->commits.list[i]) !=GENERATION_NUMBER_ZERO )continue ;commit_list_insert (ctx->commits.list[i], &list);
@@ -1292,22 +1292,23 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)uint32_t max_generation =0 ;for (parent = current->parents; parent; parent = parent->next) { -if (parent->item->generation ==GENERATION_NUMBER_INFINITY || - parent->item->generation ==GENERATION_NUMBER_ZERO ) { +if (generation (parent->item) ==GENERATION_NUMBER_INFINITY || +generation (parent->item) ==GENERATION_NUMBER_ZERO ) { all_parents_computed =0 ;commit_list_insert (parent->item, &list);break ; - }else if (parent->item->generation > max_generation) { - max_generation = parent->item->generation; + }else if (generation (parent->item) > max_generation) { + max_generation =generation (parent->item); } }if (all_parents_computed) { - current->generation = max_generation +1 ; +set_generation (current, max_generation +1 );pop_commit (&list); -if (current->generation >GENERATION_NUMBER_MAX ) - current->generation =GENERATION_NUMBER_MAX ; +if (generation (current) >GENERATION_NUMBER_MAX ) +set_generation (current, +GENERATION_NUMBER_MAX ); } } }
@@ -2314,8 +2315,8 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)oid_to_hex (&graph_parents->item->object.oid), oid_to_hex(&odb_parents->item->object.oid)); -if (graph_parents->item->generation > max_generation) - max_generation = graph_parents->item->generation; +if (generation (graph_parents->item) > max_generation) + max_generation =generation (graph_parents->item); graph_parents = graph_parents->next; odb_parents = odb_parents->next;
@@ -2325,7 +2326,7 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)graph_report (_ ("commit-graph parent list for commit %s terminates early" ),oid_to_hex (&cur_oid)); -if (!graph_commit->generation) { +if (!generation (graph_commit)) {if (generation_zero ==GENERATION_NUMBER_EXISTS )graph_report (_ ("commit-graph has generation number zero for commit %s, but non-zero elsewhere" ),oid_to_hex (&cur_oid));
@@ -2345,10 +2346,10 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)if (max_generation ==GENERATION_NUMBER_MAX ) max_generation--; -if (graph_commit->generation != max_generation +1 ) +if (generation (graph_commit) != max_generation +1 )graph_report (_ ("commit-graph generation for commit %s is %u != %u" ),oid_to_hex (&cur_oid), - graph_commit->generation, +generation (graph_commit), max_generation +1 );if (graph_commit->date != odb_commit->date)
commit-reach.c +25 -25
138 lines not on this page · raw message
commit.c +2 -2
@@ -731,9 +731,9 @@ int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, voidconst struct commit *a = a_, *b = b_;/* newer commits first */ -if (a->generation < b->generation) +if (generation (a) <generation (b))return 1 ; -else if (a->generation > b->generation) +else if (generation (a) >generation (b))return -1 ;/* use date as a heuristic when generations are equal */
commit.h +0 -1
@@ -34,7 +34,6 @@ struct commit {*/ struct tree *maybe_tree;uint32_t graph_pos; -uint32_t generation;unsigned int index; };
contrib/coccinelle/generation.cocci new file +12 -0
@@ -0,0 +1,12 @@ +@@ +struct commit *c; +expression E; +@@ +- c->generation = E ++ set_generation(c, E) + +@@ +struct commit *c; +@@ +- c->generation ++ generation(c)
revision.c +8 -8
@@ -720,7 +720,7 @@ static int check_maybe_different_in_bloom_filter(struct rev_info *revs,if (!revs->repo->objects->commit_graph)return -1 ; -if (commit->generation ==GENERATION_NUMBER_INFINITY ) +if (generation (commit) ==GENERATION_NUMBER_INFINITY )return -1 ; filter =get_bloom_filter (revs->repo, commit,0 );
@@ -3314,7 +3314,7 @@ static void explore_to_depth(struct rev_info *revs,struct topo_walk_info *info = revs->topo_walk_info;struct commit *c;while ((c =prio_queue_peek (&info->explore_queue)) && - c->generation >= gen_cutoff) +generation (c) >= gen_cutoff)explore_walk_step (revs); }
@@ -3330,7 +3330,7 @@ static void indegree_walk_step(struct rev_info *revs)if (parse_commit_gently (c,1 ) <0 )return ; -explore_to_depth (revs, c->generation); +explore_to_depth (revs,generation (c));for (p = c->parents; p; p = p->next) {struct commit *parent = p->item;
@@ -3354,7 +3354,7 @@ static void compute_indegrees_to_depth(struct rev_info *revs,struct topo_walk_info *info = revs->topo_walk_info;struct commit *c;while ((c =prio_queue_peek (&info->indegree_queue)) && - c->generation >= gen_cutoff) +generation (c) >= gen_cutoff)indegree_walk_step (revs); }
@@ -3414,8 +3414,8 @@ static void init_topo_walk(struct rev_info *revs)test_flag_and_insert (&info->explore_queue, c,TOPO_WALK_EXPLORED );test_flag_and_insert (&info->indegree_queue, c,TOPO_WALK_INDEGREE ); -if (c->generation < info->min_generation) - info->min_generation = c->generation; +if (generation (c) < info->min_generation) + info->min_generation =generation (c); *(indegree_slab_at (&info->indegree, c)) =1 ;
@@ -3473,8 +3473,8 @@ static void expand_topo_walk(struct rev_info *revs, struct commit *commit)if (parse_commit_gently (parent,1 ) <0 )continue ; -if (parent->generation < info->min_generation) { - info->min_generation = parent->generation; +if (generation (parent) < info->min_generation) { + info->min_generation =generation (parent);compute_indegrees_to_depth (revs, info->min_generation); }
Signature · 1 line
2.27.0
Re: [GSoC Patch 2/3] commit: convert commit->generation to a slab
commit-graph.c static void write_graph_chunk_data(struct hashfile *f, int hash_len, · line 1051
Quoted 9 lines · Abhishek Kumar
In this commit, we will use the generation slab helpers introduced in last commit and replace existing uses of
commit->generationusing 'contrib/coccinelle/generation.cocci'commit-graph.c +1
-1@@ -1048,7 +1048,7 @@ static void write_graph_chunk_data(struct hashfile *f, int hash_len,else packedDate[0 ] =0 ; - packedDate[0 ] |=htonl ((*list)->generation <<2 ); + packedDate[0 ] |=htonl (generation ((*list)) <<2 );
nit: We no longer need the extra parens around *list.
revision.c static void init_topo_walk(struct rev_info *revs) · line 3418
Quoted 8 lines
revision.c +2
-2@@ -3414,8 +3414,8 @@ static void init_topo_walk(struct rev_info *revs)test_flag_and_insert (&info->explore_queue, c,TOPO_WALK_EXPLORED );test_flag_and_insert (&info->indegree_queue, c,TOPO_WALK_INDEGREE ); -if (c->generation < info->min_generation) - info->min_generation = c->generation; +if (generation (c) < info->min_generation) + info->min_generation =generation (c);
A pattern I've noticed in several places is that the struct member is accessed multiple times in the same method body, and this is auto-converted to multiple method calls. However, these values are fixed, so it would be better to store the value as a local variable and reuse that variable instead.
This is one of the shortcomings of the Coccinelle transformation, so you'll need to manually inspect each of the diff fragments to see if we can reduce the number of method calls. It might be helpful to do that as a follow-up, so we can see that this patch is generated by the Coccinelle script, and then a later patch can be scrutinized more carefully when you are doing manual code manipulation.
Thanks,
-Stolee
Re: [GSoC Patch 2/3] commit: convert commit->generation to a slab
In reply to Abhishek KumarQuoted 34 lines · Abhishek Kumar
In this commit, we will use the generation slab helpers introduced in last commit and replace existing uses of
commit->generationusing 'contrib/coccinelle/generation.cocci'Signed-off-by Abhishek Kumar <[email protected]>--- alloc.c | 1 - blame.c | 2 +- commit-graph.c | 39 +++++++++++----------- commit-reach.c | 50 ++++++++++++++--------------- commit.c | 4 +-- commit.h | 1 - contrib/coccinelle/generation.cocci | 12 +++++++ revision.c | 16 ++++----- 8 files changed, 68 insertions(+), 57 deletions(-) create mode 100644 contrib/coccinelle/generation.coccialloc.c +0
-1@@ -109,7 +109,6 @@ void init_commit_node(struct repository *r, struct commit *c) c->object.type =OBJ_COMMIT ; c->index =alloc_commit_index (r); c->graph_pos =COMMIT_NOT_FROM_GRAPH ; - c->generation =GENERATION_NUMBER_INFINITY ; }void *alloc_commit_node (struct repository *r)blame.c +0
-0
blame.c static int maybe_changed_path(struct repository *r, · line 1275
Quoted 5 lines
blame.c +1
-1@@ -1272,7 +1272,7 @@ static int maybe_changed_path(struct repository *r,if (!bd)return 1 ; -if (origin->commit->generation ==GENERATION_NUMBER_INFINITY ) +if (generation (origin->commit) ==GENERATION_NUMBER_INFINITY )
Hmmmm, as C is not all that object-oriented that lets us say "commit objects have generation() method", a plain vanilla function whose name is generation() is a bit overly vague. The field name this helper function replaces, .generation, is very localized to a commit "object" and does not have such a problem.
We probably need to choose a better name in the previous step to fix it.
Re: [GSoC Patch 2/3] commit: convert commit->generation to a slab
In reply to Abhishek KumarQuoted 5 lines · Abhishek Kumar
In this commit, we will use the generation slab helpers introduced in last commit and replace existing uses of
commit->generationusing 'contrib/coccinelle/generation.cocci'Signed-off-by Abhishek Kumar <[email protected]>
Quoted 12 lines
--- alloc.c | 1 - blame.c | 2 +- commit-graph.c | 39 +++++++++++----------- commit-reach.c | 50 ++++++++++++++--------------- commit.c | 4 +-- commit.h | 1 - contrib/coccinelle/generation.cocci | 12 +++++++ revision.c | 16 ++++----- 8 files changed, 68 insertions(+), 57 deletions(-) create mode 100644 contrib/coccinelle/generation.cocci
For easier review I have changed the order of files, grouping together different categories of changes.
First category is removing commit->generation and related changes:
commit.h struct commit { · line 39
Quoted 12 lines
commit.h +0
-1@@ -34,7 +34,6 @@ struct commit {*/ struct tree *maybe_tree;uint32_t graph_pos; -uint32_t generation;unsigned int index; };
This is quite straightforward.
alloc.c void init_commit_node(struct repository *r, struct commit *c) · line 114
Quoted 12 lines
alloc.c +0
-1@@ -109,7 +109,6 @@ void init_commit_node(struct repository *r, struct commit *c) c->object.type =OBJ_COMMIT ; c->index =alloc_commit_index (r); c->graph_pos =COMMIT_NOT_FROM_GRAPH ; - c->generation =GENERATION_NUMBER_INFINITY ; }void *alloc_commit_node (struct repository *r)
But this change might need a more detailed write-up in the commit message.
If I understand it correctly, this is function is used by generic commit allocator, and given commit object may not need generation number. That is why the default value of GENERATION_NUMBER_INFINITY is handled by new "getters" and "setters", i.e. generation() and set_generation() -- which are called only if commit-graph is present, I think.
The generation() returns GENERATION_NUMBER_INFINITY for commits not in generation_slab, assuming that for all commits in the commit graph it would be filled by commit-graph parsing -- just like init_commit_node() would initialize commit->generation to this value.
Because <slabname>_at() (including generation_slab_at()) extends array if necessary, we want all data on generation_slab to be correctly initialized to GENERATION_NUMBER_INFINITY, just like init_commit_node() would do it, because generation() "getter" cannot. The commit might be present on generation_slab just because allocation is done in chunks (slabs).
Second category is semantic patch that generates the rest of changes (which could be adjusted manually in subsequent commits).
Quoted 8 lines
contrib/coccinelle/generation.cocci new file +2
-0@@ -0,0 +1,12 @@ +@@ +struct commit *c;
contrib/coccinelle/generation.cocci line 12
Quoted 10 lines
contrib/coccinelle/generation.cocci new file +10
-0+expression E; +@@ +- c->generation = E ++ set_generation(c, E) + +@@ +struct commit *c; +@@ +- c->generation ++ generation(c)
I wonder if Coccinelle is able to automatically discard extra parentheses (the problem noticed by Stolee in his reply) with the following chunk:
+@@
+struct commit *c;
+@@
+- (c)->generation
++ generation(c)
Third category is all the changes that are just straight mechanical changes being the result of applying the Coccinelle patch.
Quoted 74 lines
blame.c +1
-1@@ -1272,7 +1272,7 @@ static int maybe_changed_path(struct repository *r,if (!bd)return 1 ; -if (origin->commit->generation ==GENERATION_NUMBER_INFINITY ) +if (generation (origin->commit) ==GENERATION_NUMBER_INFINITY )return 1 ; filter =get_bloom_filter (r, origin->commit,0 );commit-graph.c +8
-9@@ -120,9 +120,9 @@ static int commit_gen_cmp(const void *va, const void *vb)const struct commit *b = *(const struct commit **)vb;/* lower generation commits first */ -if (a->generation < b->generation) +if (generation (a) <generation (b))return -1 ; -else if (a->generation > b->generation) +else if (generation (a) >generation (b))return 1 ;/* use date as a heuristic when generations are equal */ @@ -712,7 +712,7 @@ static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, lex_index = pos - g->num_commits_in_base; commit_data = g->chunk_commit_data +GRAPH_DATA_WIDTH * lex_index; item->graph_pos = pos; - item->generation =get_be32 (commit_data + g->hash_len +8 ) >>2 ; +set_generation (item,get_be32 (commit_data + g->hash_len +8 ) >>2 ); }static inline void set_commit_tree (struct commit *c,struct tree *t)@@ -754,7 +754,7 @@ static int fill_commit_in_graph(struct repository *r, date_low =get_be32 (commit_data + g->hash_len +12 ); item->date = (timestamp_t)((date_high <<32 ) | date_low); - item->generation =get_be32 (commit_data + g->hash_len +8 ) >>2 ; +set_generation (item,get_be32 (commit_data + g->hash_len +8 ) >>2 ); pptr = &item->parents;@@ -1048,7 +1048,7 @@ static void write_graph_chunk_data(struct hashfile *f, int hash_len,else packedDate[0 ] =0 ; - packedDate[0 ] |=htonl ((*list)->generation <<2 ); + packedDate[0 ] |=htonl (generation ((*list)) <<2 ); packedDate[1 ] =htonl ((*list)->date);hashwrite (f, packedDate,8 );@@ -1280,8 +1280,8 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx) ctx->commits.nr);for (i =0 ; i < ctx->commits.nr; i++) {display_progress (ctx->progress, i +1 ); -if (ctx->commits.list[i]->generation !=GENERATION_NUMBER_INFINITY && - ctx->commits.list[i]->generation !=GENERATION_NUMBER_ZERO ) +if (generation (ctx->commits.list[i]) !=GENERATION_NUMBER_INFINITY && +generation (ctx->commits.list[i]) !=GENERATION_NUMBER_ZERO )continue ;commit_list_insert (ctx->commits.list[i], &list);@@ -1292,22 +1292,23 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)uint32_t max_generation =0 ;for (parent = current->parents; parent; parent = parent->next) { -if (parent->item->generation ==GENERATION_NUMBER_INFINITY || - parent->item->generation ==GENERATION_NUMBER_ZERO ) { + if (generation(parent->item) == GENERATION_NUMBER_INFINITY ||
commit-graph.c static void compute_generation_numbers(struct write_commit_graph_context *ctx) · line 1303
Quoted 10 lines
commit-graph.c +3
-2+ generation(parent->item) == GENERATION_NUMBER_ZERO) { all_parents_computed =0 ;commit_list_insert (parent->item, &list);break ; - }else if (parent->item->generation > max_generation) { - max_generation = parent->item->generation; + } else if (generation(parent->item) > max_generation) { + max_generation =generation (parent->item); } }
Here generation(parent->item) is called three times, which is probably cost effective to save the value to the local variable (as Stolee noticed).
Quoted 57 lines
commit-graph.c +9
-8if (all_parents_computed) { - current->generation = max_generation +1 ; +set_generation (current, max_generation +1 );pop_commit (&list); -if (current->generation >GENERATION_NUMBER_MAX ) - current->generation =GENERATION_NUMBER_MAX ; +if (generation (current) >GENERATION_NUMBER_MAX ) +set_generation (current, +GENERATION_NUMBER_MAX ); } } }@@ -2314,8 +2315,8 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)oid_to_hex (&graph_parents->item->object.oid), oid_to_hex(&odb_parents->item->object.oid)); -if (graph_parents->item->generation > max_generation) - max_generation = graph_parents->item->generation; +if (generation (graph_parents->item) > max_generation) + max_generation =generation (graph_parents->item); graph_parents = graph_parents->next; odb_parents = odb_parents->next;@@ -2325,7 +2326,7 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)graph_report (_ ("commit-graph parent list for commit %s terminates early" ),oid_to_hex (&cur_oid)); -if (!graph_commit->generation) { +if (!generation (graph_commit)) {if (generation_zero ==GENERATION_NUMBER_EXISTS )graph_report (_ ("commit-graph has generation number zero for commit %s, but non-zero elsewhere" ),oid_to_hex (&cur_oid));@@ -2345,10 +2346,10 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)if (max_generation ==GENERATION_NUMBER_MAX ) max_generation--; -if (graph_commit->generation != max_generation +1 ) +if (generation (graph_commit) != max_generation +1 )graph_report (_ ("commit-graph generation for commit %s is %u != %u" ),oid_to_hex (&cur_oid), - graph_commit->generation, +generation (graph_commit), max_generation +1 );if (graph_commit->date != odb_commit->date)commit-reach.c +1
-1@@ -59,13 +59,13 @@ static struct commit_list *paint_down_to_common(struct repository *r,struct commit_list *parents;int flags; -if (min_generation && commit->generation > last_gen) +if (min_generation &&generation (commit) > last_gen)
commit-reach.c static struct commit_list *paint_down_to_common(struct repository *r, · line 69
Quoted 10 lines
commit-reach.c +3
-3BUG ("bad generation skip %8x > %8x at %s" , - commit->generation, last_gen, +generation (commit), last_gen,oid_to_hex (&commit->object.oid)); - last_gen = commit->generation; + last_gen =generation (commit); -if (commit->generation < min_generation) +if (generation (commit) < min_generation)break ;
Here generation(commit) is called three times (two times in the exceptional case).
Quoted 203 lines
commit-reach.c +21
-21flags = commit->object.flags & (PARENT1 |PARENT2 |STALE );@@ -176,7 +176,7 @@ static int remove_redundant(struct repository *r, struct commit **array, int cntrepo_parse_commit (r, array[i]);for (i =0 ; i < cnt; i++) {struct commit_list *common; -uint32_t min_generation = array[i]->generation; +uint32_t min_generation =generation (array[i]);if (redundant[i])continue ;@@ -186,8 +186,8 @@ static int remove_redundant(struct repository *r, struct commit **array, int cnt filled_index[filled] = j; work[filled++] = array[j]; -if (array[j]->generation < min_generation) - min_generation = array[j]->generation; +if (generation (array[j]) < min_generation) + min_generation =generation (array[j]); } common =paint_down_to_common (r, array[i], filled, work, min_generation);@@ -323,16 +323,16 @@ int repo_in_merge_bases_many(struct repository *r, struct commit *commit,for (i =0 ; i < nr_reference; i++) {if (repo_parse_commit (r, reference[i]))return ret; -if (reference[i]->generation < min_generation) - min_generation = reference[i]->generation; +if (generation (reference[i]) < min_generation) + min_generation =generation (reference[i]); } -if (commit->generation > min_generation) +if (generation (commit) > min_generation)return ret; bases =paint_down_to_common (r, commit, nr_reference, reference, - commit->generation); +generation (commit));if (commit->object.flags &PARENT2 ) ret =1 ;clear_commit_marks (commit, all_flags);@@ -467,7 +467,7 @@ static enum contains_result contains_test(struct commit *candidate,/* Otherwise, we don't know; prepare to recurse */ parse_commit_or_die (candidate); -if (candidate->generation < cutoff) +if (generation (candidate) < cutoff)return CONTAINS_NO ;return CONTAINS_UNKNOWN ;@@ -492,8 +492,8 @@ static enum contains_result contains_tag_algo(struct commit *candidate,for (p = want; p; p = p->next) {struct commit *c = p->item;load_commit_graph_info (the_repository, c); -if (c->generation < cutoff) - cutoff = c->generation; +if (generation (c) < cutoff) + cutoff =generation (c); } result =contains_test (candidate, want, cache, cutoff);@@ -544,9 +544,9 @@ static int compare_commits_by_gen(const void *_a, const void *_b)const struct commit *a = *(const struct commit *const *)_a;const struct commit *b = *(const struct commit *const *)_b; -if (a->generation < b->generation) +if (generation (a) <generation (b))return -1 ; -if (a->generation > b->generation) +if (generation (a) >generation (b))return 1 ;return 0 ; }@@ -585,7 +585,7 @@ int can_all_from_reach_with_flag(struct object_array *from, list[nr_commits] = (struct commit *)from_one;if (parse_commit (list[nr_commits]) || - list[nr_commits]->generation < min_generation) { +generation (list[nr_commits]) < min_generation) { result =0 ; goto cleanup; }@@ -621,7 +621,7 @@ int can_all_from_reach_with_flag(struct object_array *from,if (parse_commit (parent->item) || parent->item->date < min_commit_date || - parent->item->generation < min_generation) +generation (parent->item) < min_generation)continue ;commit_list_insert (parent->item, &stack);@@ -665,8 +665,8 @@ int can_all_from_reach(struct commit_list *from, struct commit_list *to,if (from_iter->item->date < min_commit_date) min_commit_date = from_iter->item->date; -if (from_iter->item->generation < min_generation) - min_generation = from_iter->item->generation; +if (generation (from_iter->item) < min_generation) + min_generation =generation (from_iter->item); } from_iter = from_iter->next;@@ -677,8 +677,8 @@ int can_all_from_reach(struct commit_list *from, struct commit_list *to,if (to_iter->item->date < min_commit_date) min_commit_date = to_iter->item->date; -if (to_iter->item->generation < min_generation) - min_generation = to_iter->item->generation; +if (generation (to_iter->item) < min_generation) + min_generation =generation (to_iter->item); } to_iter->item->object.flags |=PARENT2 ;@@ -721,8 +721,8 @@ struct commit_list *get_reachable_subset(struct commit **from, int nr_from,struct commit *c = *item;parse_commit (c); -if (c->generation < min_generation) - min_generation = c->generation; +if (generation (c) < min_generation) + min_generation =generation (c);if (!(c->object.flags &PARENT1 )) { c->object.flags |=PARENT1 ;@@ -755,7 +755,7 @@ struct commit_list *get_reachable_subset(struct commit **from, int nr_from,parse_commit (p); -if (p->generation < min_generation) +if (generation (p) < min_generation)continue ;if (p->object.flags &PARENT2 )commit.c +2
-2@@ -731,9 +731,9 @@ int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, voidconst struct commit *a = a_, *b = b_;/* newer commits first */ -if (a->generation < b->generation) +if (generation (a) <generation (b))return 1 ; -else if (a->generation > b->generation) +else if (generation (a) >generation (b))return -1 ;/* use date as a heuristic when generations are equal */ revision.c +6
-6@@ -720,7 +720,7 @@ static int check_maybe_different_in_bloom_filter(struct rev_info *revs,if (!revs->repo->objects->commit_graph)return -1 ; -if (commit->generation ==GENERATION_NUMBER_INFINITY ) +if (generation (commit) ==GENERATION_NUMBER_INFINITY )return -1 ; filter =get_bloom_filter (revs->repo, commit,0 );@@ -3314,7 +3314,7 @@ static void explore_to_depth(struct rev_info *revs,struct topo_walk_info *info = revs->topo_walk_info;struct commit *c;while ((c =prio_queue_peek (&info->explore_queue)) && - c->generation >= gen_cutoff) +generation (c) >= gen_cutoff)explore_walk_step (revs); }@@ -3330,7 +3330,7 @@ static void indegree_walk_step(struct rev_info *revs)if (parse_commit_gently (c,1 ) <0 )return ; -explore_to_depth (revs, c->generation); +explore_to_depth (revs,generation (c));for (p = c->parents; p; p = p->next) {struct commit *parent = p->item;@@ -3354,7 +3354,7 @@ static void compute_indegrees_to_depth(struct rev_info *revs,struct topo_walk_info *info = revs->topo_walk_info;struct commit *c;while ((c =prio_queue_peek (&info->indegree_queue)) && - c->generation >= gen_cutoff) +generation (c) >= gen_cutoff)indegree_walk_step (revs); }@@ -3414,8 +3414,8 @@ static void init_topo_walk(struct rev_info *revs)test_flag_and_insert (&info->explore_queue, c,TOPO_WALK_EXPLORED );test_flag_and_insert (&info->indegree_queue, c,TOPO_WALK_INDEGREE ); -if (c->generation < info->min_generation) - info->min_generation = c->generation; +if (generation (c) < info->min_generation) + info->min_generation =generation (c); *(indegree_slab_at (&info->indegree, c)) =1 ;
revision.c static void expand_topo_walk(struct rev_info *revs, struct commit *commit) · line 3479
Quoted 9 lines
revision.c +2
-2@@ -3473,8 +3473,8 @@ static void expand_topo_walk(struct rev_info *revs, struct commit *commit)if (parse_commit_gently (parent,1 ) <0 )continue ; -if (parent->generation < info->min_generation) { - info->min_generation = parent->generation; +if (generation (parent) < info->min_generation) { + info->min_generation =generation (parent);compute_indegrees_to_depth (revs, info->min_generation); }
Best,
--
Jakub Narębski
PATCH 3/3 commit: convert commit->graph_pos to a slab
In reply to Abhishek KumarThe member graph_pos refers to the integer position used to identify a commit in commit-graph files. However, graph_pos is not useful in other contexts and bloats the struct.
Let's move it to a commit-slab and shrink the struct by four bytes.
Existing references to graph_pos are replaced using 'contrib/coccinelle/graph_pos.cocci'.
--- alloc.c | 1 - bloom.c | 6 ++-- commit-graph.c | 50 +++++++++++++++++++++++------- commit-graph.h | 3 ++ commit.c | 2 +- commit.h | 2 -- contrib/coccinelle/graph_pos.cocci | 12 +++++++ 7 files changed, 58 insertions(+), 18 deletions(-) create mode 100644 contrib/coccinelle/graph_pos.cocci
alloc.c +0 -1
@@ -108,7 +108,6 @@ void init_commit_node(struct repository *r, struct commit *c) { c->object.type =OBJ_COMMIT ; c->index =alloc_commit_index (r); - c->graph_pos =COMMIT_NOT_FROM_GRAPH ; }void *alloc_commit_node (struct repository *r)
bloom.c +3 -3
@@ -34,14 +34,14 @@ static int load_bloom_filter_from_graph(struct commit_graph *g, {uint32_t lex_pos, start_index, end_index; -while (c->graph_pos < g->num_commits_in_base) +while (graph_pos (c) < g->num_commits_in_base) g = g->base_graph;/* The commit graph commit 'c' lives in doesn't carry bloom filters. */ if (!g->chunk_bloom_indexes)return 0 ; - lex_pos = c->graph_pos - g->num_commits_in_base; + lex_pos =graph_pos (c) - g->num_commits_in_base; end_index =get_be32 (g->chunk_bloom_indexes +4 * lex_pos);
@@ -188,7 +188,7 @@ struct bloom_filter *get_bloom_filter(struct repository *r,if (!filter->data) {load_commit_graph_info (r, c); -if (c->graph_pos !=COMMIT_NOT_FROM_GRAPH && +if (graph_pos (c) !=COMMIT_NOT_FROM_GRAPH && r->objects->commit_graph->chunk_bloom_indexes) {if (load_bloom_filter_from_graph (r->objects->commit_graph, filter, c))return filter;
commit-graph.c +39 -11
113 lines not on this page · raw message
commit-graph.h +3 -0
@@ -8,6 +8,7 @@#include "object-store.h" #include "oidset.h" +#define COMMIT_NOT_FROM_GRAPH 0xFFFFFFFF#define GENERATION_NUMBER_INFINITY 0xFFFFFFFF#define GENERATION_NUMBER_MAX 0x3FFFFFFF#define GENERATION_NUMBER_ZERO 0
@@ -142,4 +143,6 @@ void free_commit_graph(struct commit_graph *);void disable_commit_graph (struct repository *r);uint32_t generation (const struct commit *c); + +uint32_t graph_pos (const struct commit *c);#endif
commit.c +1 -1
@@ -339,7 +339,7 @@ struct tree *repo_get_commit_tree(struct repository *r,if (commit->maybe_tree || !commit->object.parsed)return commit->maybe_tree; -if (commit->graph_pos !=COMMIT_NOT_FROM_GRAPH ) +if (graph_pos (commit) !=COMMIT_NOT_FROM_GRAPH )return get_commit_tree_in_graph (r, commit);return NULL ;
commit.h +0 -2
@@ -10,8 +10,6 @@#include "pretty.h" #include "commit-slab.h" -#define COMMIT_NOT_FROM_GRAPH 0xFFFFFFFF -struct commit_list {struct commit *item;struct commit_list *next;
contrib/coccinelle/graph_pos.cocci new file +12 -0
@@ -0,0 +1,12 @@ +@@ +struct commit *c; +expression E; +@@ +- c->graph_pos = E ++ set_graph_pos(c, E) + +@@ +struct commit *c; +@@ +- c->graph_pos ++ graph_pos(c)
Signature · 1 line
2.27.0
Re: [GSoC Patch 3/3] commit: convert commit->graph_pos to a slab
Quoted 10 lines · Abhishek Kumar
The member
graph_posrefers to the integer position used to identify a commit in commit-graph files. However,graph_posis not useful in other contexts and bloats the struct.Let's move it to a commit-slab and shrink the struct by four bytes.
Existing references to
graph_posare replaced using 'contrib/coccinelle/graph_pos.cocci'.Signed-off-by Abhishek Kumar <[email protected]>
Quoted 10 lines
--- alloc.c | 1 - bloom.c | 6 ++-- commit-graph.c | 50 +++++++++++++++++++++++------- commit-graph.h | 3 ++ commit.c | 2 +- commit.h | 2 -- contrib/coccinelle/graph_pos.cocci | 12 +++++++ 7 files changed, 58 insertions(+), 18 deletions(-) create mode 100644 contrib/coccinelle/graph_pos.cocci
I have reordered the chunks to make it easier to review.
Quoted 18 lines
commit.h +0
-2@@ -10,8 +10,6 @@#include "pretty.h" #include "commit-slab.h" -#define COMMIT_NOT_FROM_GRAPH 0xFFFFFFFF -struct commit_list {struct commit *item;struct commit_list *next;commit-graph.h +0
-0
commit-graph.h line 14
Quoted 7 lines
commit-graph.h +1
-0@@ -8,6 +8,7 @@#include "object-store.h" #include "oidset.h" +#define COMMIT_NOT_FROM_GRAPH 0xFFFFFFFF#define GENERATION_NUMBER_INFINITY 0xFFFFFFFF#define GENERATION_NUMBER_MAX 0x3FFFFFFF#define GENERATION_NUMBER_ZERO 0
Those two chunks move COMMIT_NOT_FROM_GRAPH from commit.h to commit-graph.h, because it is no longer needed in init_commit_node() from alloc.c. On the other hand the remaining commit-graph #define-s were moved in first commit in series.
What I DO NOT SEE in this commit is actual *removal* of graph_pos field from the struct commit, i.e.:
diff --git a/commit.h b/commit.h
index cc610400d5..01e1c4c3eb 100644
--- a/commit.h
+++ b/commit.h
@@ -34,6 +34,5 @@ struct commit {
*/
struct tree *maybe_tree;
- uint32_t graph_pos;
unsigned int index;
};alloc.c void init_commit_node(struct repository *r, struct commit *c) · line 113
Quoted 12 lines
alloc.c +0
-1@@ -108,7 +108,6 @@ void init_commit_node(struct repository *r, struct commit *c) { c->object.type =OBJ_COMMIT ; c->index =alloc_commit_index (r); - c->graph_pos =COMMIT_NOT_FROM_GRAPH ; }void *alloc_commit_node (struct repository *r)
This removes commit->graph_pos initialization from init_commit_node(), and thus from alloc_commit_node(); the handling of COMMIT_NOT_FROM_GRAPH is moved to setter and getter "methods", see below.
Quoted 39 lines
commit-graph.c +28
-0@@ -87,6 +87,34 @@ static int commit_pos_cmp(const void *va, const void *vb)commit_pos_at (&commit_pos, b); } +define_commit_slab (graph_pos_slab, uint32_t); +static struct graph_pos_slab graph_pos_slab =COMMIT_SLAB_INIT (1 , graph_pos_slab); + +uint32_t graph_pos (const struct commit *c) +{ +uint32_t *pos =graph_pos_slab_peek (&graph_pos_slab, c); + +return pos ? *pos :COMMIT_NOT_FROM_GRAPH ; +} + +static void set_graph_pos (const struct commit *c,const uint32_t position) +{ +unsigned int i = graph_pos_slab.slab_count; +uint32_t *pos =graph_pos_slab_at (&graph_pos_slab, c); + +/* +* commit-slab initializes with zero, overwrite this with +* COMMIT_NOT_FROM_GRAPH +*/ +for (; i < graph_pos_slab.slab_count; ++i) + { +memset (graph_pos_slab.slab[i],COMMIT_NOT_FROM_GRAPH , + graph_pos_slab.slab_size *sizeof (uint32_t )); + } + + *pos = position; +} +define_commit_slab (generation_slab, uint32_t);static struct generation_slab generation_slab =COMMIT_SLAB_INIT (1 , generation_slab);
[...]
commit-graph.c void free_commit_graph(struct commit_graph *); · line 148
Quoted 7 lines
commit-graph.c +2
-0@@ -142,4 +143,6 @@ void free_commit_graph(struct commit_graph *);void disable_commit_graph (struct repository *r);uint32_t generation (const struct commit *c); + +uint32_t graph_pos (const struct commit *c);#endif
I wonder why those helper functions: graph_pos() and set_graph_pos() were not introduced in the 1st patch of this series, together with generation() and set_generation().
The same comments as for previous patch apply: if graph_pos was not explicitely set, we want for it to be COMMIT_NOT_FROM_GRAPH (like init_commit_node() did before this change). If it is not on commit-slab, it is not set -- this is handled by graph_pos() function. However we allocate memory on slab in chunks, and set_graph_pos() ensures that those extra allocated graph_pos values on commit-slab are properly initialized to COMMIT_NOT_FROM_GRAPH, like init_commit_node() did.
Note that we always have COMMIT_NOT_FROM_GRAPH for graph_pos if and only if there is GENERATION_NUMBER_INFINITY for generation, so perhaps putting those two together in struct commit_graph_info would make sense. But whether doing more work in "setter" set_*() functions or doing extra conditional in "getter" *() would give better performance needs (micro-)benchmarking.
Quoted 8 lines
contrib/coccinelle/graph_pos.cocci new file +2
-0@@ -0,0 +1,12 @@ +@@ +struct commit *c;
contrib/coccinelle/graph_pos.cocci line 12
Quoted 10 lines
contrib/coccinelle/graph_pos.cocci new file +10
-0+expression E; +@@ +- c->graph_pos = E ++ set_graph_pos(c, E) + +@@ +struct commit *c; +@@ +- c->graph_pos ++ graph_pos(c)
This is semantic patch that generates the rest of changes. (If any of those needs manual improvement, it is better left for a separate "manual fixup" patch, in my opinion.)
Quoted 11 lines
bloom.c +1
-1@@ -34,14 +34,14 @@ static int load_bloom_filter_from_graph(struct commit_graph *g, {uint32_t lex_pos, start_index, end_index; -while (c->graph_pos < g->num_commits_in_base) +while (graph_pos (c) < g->num_commits_in_base) g = g->base_graph;
bloom.c static int load_bloom_filter_from_graph(struct commit_graph *g, · line 47
Quoted 10 lines
bloom.c +1
-1/* The commit graph commit 'c' lives in doesn't carry bloom filters. */ if (!g->chunk_bloom_indexes)return 0 ; - lex_pos = c->graph_pos - g->num_commits_in_base; + lex_pos =graph_pos (c) - g->num_commits_in_base; end_index =get_be32 (g->chunk_bloom_indexes +4 * lex_pos);
Here graph_pos(c) is used twice.
It probably needs to be checked (with benchmark) if it matters.
Quoted 43 lines
bloom.c +1
-1@@ -188,7 +188,7 @@ struct bloom_filter *get_bloom_filter(struct repository *r,if (!filter->data) {load_commit_graph_info (r, c); -if (c->graph_pos !=COMMIT_NOT_FROM_GRAPH && +if (graph_pos (c) !=COMMIT_NOT_FROM_GRAPH && r->objects->commit_graph->chunk_bloom_indexes) {if (load_bloom_filter_from_graph (r->objects->commit_graph, filter, c))return filter;commit-graph.c +3
-3@@ -697,7 +725,7 @@ static struct commit_list **insert_parent_or_die(struct repository *r, c =lookup_commit (r, &oid);if (!c)die (_ ("could not find commit %s" ),oid_to_hex (&oid)); - c->graph_pos = pos; +set_graph_pos (c, pos);return &commit_list_insert (c, pptr)->next; }@@ -711,7 +739,7 @@ static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, lex_index = pos - g->num_commits_in_base; commit_data = g->chunk_commit_data +GRAPH_DATA_WIDTH * lex_index; - item->graph_pos = pos; +set_graph_pos (item, pos);set_generation (item,get_be32 (commit_data + g->hash_len +8 ) >>2 ); }@@ -741,7 +769,7 @@ static int fill_commit_in_graph(struct repository *r,* Store the "full" position, but then use the * "local" position for the rest of the calculation. */ - item->graph_pos = pos; +set_graph_pos (item, pos); lex_index = pos - g->num_commits_in_base; commit_data = g->chunk_commit_data + (g->hash_len +16 ) * lex_index;
commit-graph.c static int fill_commit_in_graph(struct repository *r, · line 819
Quoted 8 lines
commit-graph.c +2
-2@@ -786,8 +814,8 @@ static int fill_commit_in_graph(struct repository *r,static int find_commit_in_graph (struct commit *item,struct commit_graph *g,uint32_t *pos) { -if (item->graph_pos !=COMMIT_NOT_FROM_GRAPH ) { - *pos = item->graph_pos; +if (graph_pos (item) !=COMMIT_NOT_FROM_GRAPH ) { + *pos =graph_pos (item);return 1 ;
Here graph_pos(item) is used twice.
Quoted 6 lines
commit-graph.c +0
-0}else {struct commit_graph *cur_g = g;@@ -843,11 +871,11 @@ static struct tree *load_tree_for_commit(struct repository *r,struct object_id oid;const unsigned char *commit_data;
commit-graph.c static struct tree *load_tree_for_commit(struct repository *r, · line 881
Quoted 10 lines
commit-graph.c +2
-2-while (c->graph_pos < g->num_commits_in_base) +while (graph_pos (c) < g->num_commits_in_base) g = g->base_graph; commit_data = g->chunk_commit_data + -GRAPH_DATA_WIDTH * (c->graph_pos - g->num_commits_in_base); +GRAPH_DATA_WIDTH * (graph_pos (c) - g->num_commits_in_base);hashcpy (oid.hash, commit_data);set_commit_tree (c,lookup_tree (r, &oid));
Here graph_pos(c) is used twice.
Quoted 41 lines
commit-graph.c +4
-4@@ -861,7 +889,7 @@ static struct tree *get_commit_tree_in_graph_one(struct repository *r, {if (c->maybe_tree)return c->maybe_tree; -if (c->graph_pos ==COMMIT_NOT_FROM_GRAPH ) +if (graph_pos (c) ==COMMIT_NOT_FROM_GRAPH )BUG ("get_commit_tree_in_graph_one called from non-commit-graph commit" );return load_tree_for_commit (r, g, (struct commit *)c);@@ -1247,7 +1275,7 @@ static void close_reachable(struct write_commit_graph_context *ctx)continue ;if (ctx->split) {if ((!parse_commit (commit) && - commit->graph_pos ==COMMIT_NOT_FROM_GRAPH ) || +graph_pos (commit) ==COMMIT_NOT_FROM_GRAPH ) || flags ==COMMIT_GRAPH_SPLIT_REPLACE )add_missing_parents (ctx, commit); }else if (!parse_commit_no_graph (commit))@@ -1493,7 +1521,7 @@ static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)if (ctx->split) {struct commit *c =lookup_commit (ctx->r, &ctx->oids.list[i]); -if (!c || c->graph_pos !=COMMIT_NOT_FROM_GRAPH ) +if (!c ||graph_pos (c) !=COMMIT_NOT_FROM_GRAPH )continue ; }@@ -1527,7 +1555,7 @@ static void copy_oids_to_commits(struct write_commit_graph_context *ctx) ctx->commits.list[ctx->commits.nr] =lookup_commit (ctx->r, &ctx->oids.list[i]);if (ctx->split && flags !=COMMIT_GRAPH_SPLIT_REPLACE && - ctx->commits.list[ctx->commits.nr]->graph_pos !=COMMIT_NOT_FROM_GRAPH ) +graph_pos (ctx->commits.list[ctx->commits.nr]) !=COMMIT_NOT_FROM_GRAPH )continue ;if (ctx->split && flags ==COMMIT_GRAPH_SPLIT_REPLACE )commit.c +0
-0
commit.c struct tree *repo_get_commit_tree(struct repository *r, · line 345
Quoted 8 lines
commit.c +1
-1@@ -339,7 +339,7 @@ struct tree *repo_get_commit_tree(struct repository *r,if (commit->maybe_tree || !commit->object.parsed)return commit->maybe_tree; -if (commit->graph_pos !=COMMIT_NOT_FROM_GRAPH ) +if (graph_pos (commit) !=COMMIT_NOT_FROM_GRAPH )return get_commit_tree_in_graph (r, commit);return NULL ;
Best,
Signature · 1 line
Jakub Narębski
Quoted 7 lines · Abhishek Kumar
The struct commit is used in many contexts. However, members generation and
graph_posare only used for commit-graph related operations and otherwise waste memory.This wastage would have been more pronounced as transistion to generation number v2, which uses 64-bit generation number instead of current 32-bits.
Thanks! This is an important step, and will already improve performance in subtle ways.
Quoted 11 lines
The third patch ("commit: convert
commit->graph_posto a slab", 2020-06-04) is currently failing diff-submodule related tests (t4041, t4059 and t4060) for gcc [1]. I am going to send a second version soon, fixing that.[1]: https://travis-ci.com/github/abhishekkumar2718/git/jobs/343441189
Abhishek Kumar (3):
commit: introduce helpers for generation slab
commit: convertcommit->generationto a slab
commit: convertcommit->graph_posto a slab
If we have a commit-graph file, then we have graph_pos and generation both coming from that file. Perhaps it would be better to combine the data into a single slab that stores a "struct commit_graph_data" or something?
This would change only the slab definitions, since you already do a good job of wrapping the slab access in methods.
Quoted 6 lines
alloc.c | 2 - blame.c | 2 +- bloom.c | 6 +- commit-graph.c | 116 +++++++++++++++++++++------- commit-graph.h | 8 ++ commit-reach.c | 50 ++++++------
Quoted 8 lines
commit.c | 6 +- commit.h | 6 -- contrib/coccinelle/generation.cocci | 12 +++ contrib/coccinelle/graph_pos.cocci | 12 +++ revision.c | 16 ++-- 11 files changed, 158 insertions(+), 78 deletions(-) create mode 100644 contrib/coccinelle/generation.cocci create mode 100644 contrib/coccinelle/graph_pos.cocci
I appreciate the Coccinelle scripts to help identify automatic fixes for other topics in-flight. However, I wonder if they would be better placed inside the existing commit.cocci file?
Thanks,
-Stolee
Quoted 4 lines · Derrick Stolee
If we have a commit-graph file, then we have
graph_posand generation both coming from that file. Perhaps it would be better to combine the data into a single slab that stores a "structcommit_graph_data" or something?
Excellent.
Quoted 11 lines · Derrick Stolee
Quoted 7 lines · Abhishek Kumar
The struct commit is used in many contexts. However, members generation and
graph_posare only used for commit-graph related operations and otherwise waste memory.This wastage would have been more pronounced as transistion to generation number v2, which uses 64-bit generation number instead of current 32-bits.
Thanks! This is an important step, and will already improve performance in subtle ways.
While the reduced memory footprint of each commit object might improve performance, accessing graph position and generation numbers in a commit-slab is more expensive than direct field accesses in 'struct commit' instances. Consequently, these patches increase the runtime of 'git merge-base --is-ancestor HEAD~50000 HEAD' in the linux repository from 0.630s to 0.940s.
Quoted 7 lines
Quoted 2 lines
create mode 100644 contrib/coccinelle/generation.cocci create mode 100644 contrib/coccinelle/graph_pos.cocciI appreciate the Coccinelle scripts to help identify automatic fixes for other topics in-flight. However, I wonder if they would be better placed inside the existing
commit.coccifile?
We add Coccinelle scripts to avoid undesirable code patterns entering our code base. That, however, is not the case here: this is a one-time conversion, and at the end of this series 'struct commit' won't have a 'generation' field anymore, so once it's merged the compiler will catch any new 'commit->generation' accesses. Therefore I don't think that these Coccinelle scripts should be added at all.
Quoted 13 lines · SZEDER Gábor
Quoted 11 lines · Derrick Stolee
Quoted 7 lines · Abhishek Kumar
The struct commit is used in many contexts. However, members generation and
graph_posare only used for commit-graph related operations and otherwise waste memory.This wastage would have been more pronounced as transistion to generation number v2, which uses 64-bit generation number instead of current 32-bits.
Thanks! This is an important step, and will already improve performance in subtle ways.
Quoted 7 lines
While the reduced memory footprint of each commit object might improve performance, accessing graph position and generation numbers in a commit-slab is more expensive than direct field accesses in 'struct commit' instances. Consequently, these patches increase the runtime of 'git merge-base --is-ancestor HEAD~50000 HEAD' in the linux repository from 0.630s to 0.940s.
Thank you for checking performance. Performance penalty was something we had discussed here [1].
Caching the commit slab results in local variables helped wonderfully in v2 [2]. For example, the runtime of 'git merge-base --is-ancestor HEAD~50000 HEAD' in the linux repository increased from 0.762 to 0.767s. Since this is a change of <1%, it is *no longer* a performance regression in my opinion.
[1]: https://lore.kernel.org/git/[email protected]/
[2]: https://lore.kernel.org/git/[email protected]/
Quoted 9 lines
Quoted 8 lines
Quoted 3 lines
create mode 100644 contrib/coccinelle/generation.cocci create mode 100644 contrib/coccinelle/graph_pos.cocciI appreciate the Coccinelle scripts to help identify automatic fixes for other topics in-flight. However, I wonder if they would be better placed inside the existing
commit.coccifile?
Quoted 7 lines
We add Coccinelle scripts to avoid undesirable code patterns entering our code base. That, however, is not the case here: this is a one-time conversion, and at the end of this series 'struct commit' won't have a 'generation' field anymore, so once it's merged the compiler will catch any new '
commit->generation' accesses. Therefore I don't think that these Coccinelle scripts should be added at all.
Alright, that makes sense to me. Will remove in a subsequent version.
Thanks
Abhishek
Quoted 22 lines · Abhishek Kumar
Quoted 13 lines · SZEDER Gábor
Quoted 11 lines · Derrick Stolee
Quoted 7 lines · Abhishek Kumar
The struct commit is used in many contexts. However, members generation and
graph_posare only used for commit-graph related operations and otherwise waste memory.This wastage would have been more pronounced as transistion to generation number v2, which uses 64-bit generation number instead of current 32-bits.
Thanks! This is an important step, and will already improve performance in subtle ways.
Quoted 7 lines
While the reduced memory footprint of each commit object might improve performance, accessing graph position and generation numbers in a commit-slab is more expensive than direct field accesses in 'struct commit' instances. Consequently, these patches increase the runtime of 'git merge-base --is-ancestor HEAD~50000 HEAD' in the linux repository from 0.630s to 0.940s.
Quoted 7 lines
Thank you for checking performance. Performance penalty was something we had discussed here [1].
Caching the commit slab results in local variables helped wonderfully in v2 [2]. For example, the runtime of 'git merge-base --is-ancestor HEAD~50000 HEAD' in the linux repository increased from 0.762 to 0.767s. Since this is a change of <1%, it is *no longer* a performance regression in my opinion.
Interesting, I measured 0.870s with v2, still a notable increase from 0.630s.
Quoted 26 lines · SZEDER Gábor
Quoted 25 lines · Abhishek Kumar
Quoted 13 lines · SZEDER Gábor
Quoted 11 lines · Derrick Stolee
Quoted 7 lines · Abhishek Kumar
The struct commit is used in many contexts. However, members generation and
graph_posare only used for commit-graph related operations and otherwise waste memory.This wastage would have been more pronounced as transistion to generation number v2, which uses 64-bit generation number instead of current 32-bits.
Thanks! This is an important step, and will already improve performance in subtle ways.
Quoted 7 lines
While the reduced memory footprint of each commit object might improve performance, accessing graph position and generation numbers in a commit-slab is more expensive than direct field accesses in 'struct commit' instances. Consequently, these patches increase the runtime of 'git merge-base --is-ancestor HEAD~50000 HEAD' in the linux repository from 0.630s to 0.940s.
Thank you for checking performance. Performance penalty was something we had discussed here [1].
Quoted 7 lines
Quoted 4 lines
Caching the commit slab results in local variables helped wonderfully in v2 [2]. For example, the runtime of 'git merge-base --is-ancestor HEAD~50000 HEAD' in the linux repository increased from 0.762 to 0.767s. Since this is a change of <1%, it is *no longer* a performance regression in my opinion.
Interesting, I measured 0.870s with v2, still a notable increase from 0.630s.
This is an interesting point. The --is-ancestor is critical to the performance issue (as measured on my machine).
For "git merge-base HEAD~50000 HEAD" on the Linux repo, I get
v2.27.0:
real 0m0.515s user 0m0.467s sys 0m0.048s
v2 series:
real 0m0.534s user 0m0.481s sys 0m0.053s
With "--is-ancestor" I see the following:
v2.27.0:
real 0m0.591s user 0m0.539s sys 0m0.052s
v2 series:
real 0m0.773s user 0m0.733s sys 0m0.040s
The --is-ancestor option [1] says
Check if the first <commit> is an ancestor of the second <commit>, and exit with status 0 if true, or with status 1 if not. Errors are signaled by a non-zero status that is not 1.
[1] https://git-scm.com/docs/git-merge-base#Documentation/git-merge-base.txt---is-ancestor
This _should_ be faster than "git branch --contains HEAD~50000", but it is much much slower:
$ time git branch --contains HEAD~50000 real 0m0.068s user 0m0.061s sys 0m0.008s
So, there is definitely something going on that slows the "--is-ancestor" path in this case. But, the solution is not to halt the current patch (which likely has memory footprint benefits when dealing with a lot of tree and blob objects) and instead fix the underlying algorithm.
Let's add that to the list of things to do.
Quoted 8 lines
Quoted 7 lines
Quoted 2 lines
create mode 100644 contrib/coccinelle/generation.cocci create mode 100644 contrib/coccinelle/graph_pos.cocciI appreciate the Coccinelle scripts to help identify automatic fixes for other topics in-flight. However, I wonder if they would be better placed inside the existing
commit.coccifile?
Quoted 6 lines
We add Coccinelle scripts to avoid undesirable code patterns entering our code base. That, however, is not the case here: this is a one-time conversion, and at the end of this series 'struct commit' won't have a 'generation' field anymore, so once it's merged the compiler will catch any new '
commit->generation' accesses. Therefore I don't think that these Coccinelle scripts should be added at all.
I disagree. We _also_ add Coccinelle scripts when doing one-time refactors to avoid logical merge conflicts with other topics in flight. If someone else is working on a parallel topic that adds references to graph_pos or generation member, then the scripts provide an easy way for the maintainer to update those references in the merge commit. Alternatively, the contributor could rebase on top of this series and run the scripts themselves to fix their patches before submission.
For example, this was done carefully in the sha->object_id conversion using contrib/coccinelle/object_id.cocci.
Thanks,
-Stolee
Quoted 41 lines · Derrick Stolee
Quoted 26 lines · SZEDER Gábor
Quoted 25 lines · Abhishek Kumar
Quoted 13 lines · SZEDER Gábor
Quoted 11 lines · Derrick Stolee
Quoted 7 lines · Abhishek Kumar
The struct commit is used in many contexts. However, members generation and
graph_posare only used for commit-graph related operations and otherwise waste memory.This wastage would have been more pronounced as transistion to generation number v2, which uses 64-bit generation number instead of current 32-bits.
Thanks! This is an important step, and will already improve performance in subtle ways.
Quoted 7 lines
While the reduced memory footprint of each commit object might improve performance, accessing graph position and generation numbers in a commit-slab is more expensive than direct field accesses in 'struct commit' instances. Consequently, these patches increase the runtime of 'git merge-base --is-ancestor HEAD~50000 HEAD' in the linux repository from 0.630s to 0.940s.
Thank you for checking performance. Performance penalty was something we had discussed here [1].
Quoted 7 lines
Quoted 4 lines
Caching the commit slab results in local variables helped wonderfully in v2 [2]. For example, the runtime of 'git merge-base --is-ancestor HEAD~50000 HEAD' in the linux repository increased from 0.762 to 0.767s. Since this is a change of <1%, it is *no longer* a performance regression in my opinion.
Interesting, I measured 0.870s with v2, still a notable increase from 0.630s.
This is an interesting point. The --is-ancestor is critical to the performance issue (as measured on my machine).
For "git merge-base HEAD~50000 HEAD" on the Linux repo, I get
v2.27.0:
Quoted 8 lines
real 0m0.515s user 0m0.467s sys 0m0.048sv2 series:
real 0m0.534s user 0m0.481s sys 0m0.053s
I, too, see similarly small differences in this case.
Quoted 29 lines
With "--is-ancestor" I see the following:
v2.27.0:
real 0m0.591s user 0m0.539s sys 0m0.052sv2 series:
real 0m0.773s user 0m0.733s sys 0m0.040sThe --is-ancestor option [1] says
Check if the first <commit> is an ancestor of the second <commit>, and exit with status 0 if true, or with status 1 if not. Errors are signaled by a non-zero status that is not 1.
[1] https://git-scm.com/docs/git-merge-base#Documentation/git-merge-base.txt---is-ancestor
This _should_ be faster than "git branch --contains HEAD~50000", but it is much much slower:
$ time git branch --contains HEAD~50000 real 0m0.068s user 0m0.061s sys 0m0.008s
Quoted 5 lines
So, there is definitely something going on that slows the "--is-ancestor" path in this case. But, the solution is not to halt the current patch (which likely has memory footprint benefits when dealing with a lot of tree and blob objects) and instead fix the underlying algorithm.
Other, more common cases are affected as well, notably the simple 'git rev-list --topo-order':
performance: 1.226479734 s: git command: /home/szeder/src/git/BUILDS/v2.27.0/bin/git rev-list --topo-order HEAD max RSS: 162400k performance: 1.741309536 s: git command: /home/szeder/src/git/git rev-list --topo-order HEAD max RSS: 169556k
Is the supposed memory footprint reduction that large to justify this runtime increase?
Quoted 1 line
Let's add that to the list of things to do.
And to the commit messages.
Quoted 24 lines
Quoted 8 lines
Quoted 7 lines
Quoted 2 lines
create mode 100644 contrib/coccinelle/generation.cocci create mode 100644 contrib/coccinelle/graph_pos.cocciI appreciate the Coccinelle scripts to help identify automatic fixes for other topics in-flight. However, I wonder if they would be better placed inside the existing
commit.coccifile?Quoted 6 lines
We add Coccinelle scripts to avoid undesirable code patterns entering our code base. That, however, is not the case here: this is a one-time conversion, and at the end of this series 'struct commit' won't have a 'generation' field anymore, so once it's merged the compiler will catch any new '
commit->generation' accesses. Therefore I don't think that these Coccinelle scripts should be added at all.I disagree. We _also_ add Coccinelle scripts when doing one-time refactors to avoid logical merge conflicts with other topics in flight. If someone else is working on a parallel topic that adds references to
graph_posor generation member, then the scripts provide an easy way for the maintainer to update those references in the merge commit. Alternatively, the contributor could rebase on top of this series and run the scripts themselves to fix their patches before submission.
Quoted 2 lines
For example, this was done carefully in the
sha->object_idconversion usingcontrib/coccinelle/object_id.cocci.
'object_id.cocci' is not about sha->object_id conversions, but about avoiding undesirable code patterns, e.g. we prefer oideq() over !oidcmp(), and the compiler, of course, can't help to catch that. Coccinelle scripts used for actual sha->object_id transformations were not added to 'object_id.cocci', but were recorded only in the commit messages for reference, see e.g. 9b56149996 (merge-recursive: convert struct merge_file_info to object_id, 2016-06-24) and a couple of its ancestors.
Quoted 31 lines · SZEDER Gábor
Quoted 30 lines · Abhishek Kumar
Quoted 14 lines · SZEDER Gábor
Quoted 6 lines · Derrick Stolee
Quoted 4 lines · Abhishek Kumar
The struct commit is used in many contexts. However, members generation and
graph_posare only used for commit-graph related operations and otherwise waste memory.Quoted 6 lines
Quoted 3 lines
This wastage would have been more pronounced as transistion to generation number v2, which uses 64-bit generation number instead of current 32-bits.
Thanks! This is an important step, and will already improve performance in subtle ways.
Quoted 6 lines
While the reduced memory footprint of each commit object might improve performance, accessing graph position and generation numbers in a commit-slab is more expensive than direct field accesses in 'struct commit' instances. Consequently, these patches increase the runtime of 'git merge-base --is-ancestor HEAD~50000 HEAD' in the linux repository from 0.630s to 0.940s.
Thank you for checking performance. Performance penalty was something we had discussed here [1].
Caching the commit slab results in local variables helped wonderfully in v2 [2]. For example, the runtime of 'git merge-base --is-ancestor HEAD~50000 HEAD' in the linux repository increased from 0.762 to 0.767s. Since this is a change of <1%, it is *no longer* a performance regression in my opinion.
Quoted 5 lines
Quoted 2 lines
[1]: https://lore.kernel.org/git/[email protected]/
[2]: https://lore.kernel.org/git/[email protected]/Interesting, I measured 0.870s with v2, still a notable increase from 0.630s [a change of +38%].
I wonder what might be the cause for this difference. Is it difference in hardware (faster memory, larger CPU cache?), difference in operating system, or difference in position of HEAD?
On one hand it is large relative difference. On the other hand it is almost unnoticeable absolute difference of 0.25s.
I also wonder how the performance changes (with moving commit-graph data to the slab) for commands that do not use this data, like e.g.:
$ git -o core.commitGraph=false merge-base --is-ancestor HEAD~50000 HEAD
or
$ git gc
Sidenote: I think the performance changes should be mentioned at least in the cover letter for the series, if not in commit message(s).
Best,
Signature · 1 line
Jakub Narębski
Quoted 3 lines · Abhishek Kumar
The struct commit is used in many contexts. However, members generation and
graph_posare only used for commit-graph related operations and otherwise waste memory.
Very minor nitpick: this sentence would read better if the names of generation and graph_pos fields (but especially the 'generation') were quoted.
Quoted 4 lines
This wastage would have been more pronounced as transistion to generation number v2, which uses 64-bit generation number instead of current 32-bits.
Good. Moving reachability index value into a commit slab was one of prerequisites to switching to the generation number v2, see [2]
[2]: https://public-inbox.org/git/[email protected]/t/#u
The other prerequisite was proper handling of commit-graph format change, either by using "metadata chunk" as more flexible replacement of mishandled format version field in the commit-graph file header, or as proposed in [3] (and subsequent posts), removing "CDAT" chunk and replacing it with "CDA2" chunk.
[3]: https://public-inbox.org/git/[email protected]/t/#u
Also, we should probably stop mishandling the format version field, that is do not error out [4] when commit-graph version of the file does not match version supported by git code running the command, but just simply not use the commit-graph (like it is done for Bloom filter chunks).
[4]: https://github.com/git/git/blob/master/commit-graph.c#L253
Quoted 13 lines
The third patch ("commit: convert
commit->graph_posto a slab", 2020-06-04) is currently failing diff-submodule related tests (t4041, t4059 and t4060) for gcc [1]. I am going to send a second version soon, fixing that.[1]: https://travis-ci.com/github/abhishekkumar2718/git/jobs/343441189
Abhishek Kumar (3):
commit: introduce helpers for generation slab
commit: convertcommit->generationto a slab
commit: convertcommit->graph_posto a slab
Quoted 10 lines
alloc.c | 2 - blame.c | 2 +- bloom.c | 6 +- commit-graph.c | 116 +++++++++++++++++++++------- commit-graph.h | 8 ++ commit-reach.c | 50 ++++++------ commit.c | 6 +- commit.h | 6 -- contrib/coccinelle/generation.cocci | 12 +++ contrib/coccinelle/graph_pos.cocci | 12 +++
It is nice to see the use of Coccinelle scripts.
Quoted 4 lines
revision.c | 16 ++-- 11 files changed, 158 insertions(+), 78 deletions(-) create mode 100644 contrib/coccinelle/generation.cocci create mode 100644 contrib/coccinelle/graph_pos.cocci
Best,
Signature · 1 line
Jakub Narębski
v2 cover Move generation, graph_pos to a slab
In reply to Abhishek KumarThe struct commit is used in many contexts. However, members generation and graph_pos are only used for commit graph related operations and otherwise waste memory.
This wastage would have been more pronounced as we transition to generation number v2, which uses 64-bite generation number instead of current 32-bits.
Abhishek Kumar (4):
commit-graph: introduce commit_graph_data_slab
commit: move members graph_pos, generation to a slab
commit-graph: use generation directly when writing commit-graph
commit-graph: minimize commit_graph_data_slab access
alloc.c | 2 - blame.c | 2 +- bloom.c | 7 +- commit-graph.c | 127 ++++++++++++++++++++++++-------- commit-graph.h | 10 +++ commit-reach.c | 69 ++++++++++------- commit.c | 8 +- contrib/coccinelle/commit.cocci | 18 +++++ revision.c | 20 +++-- 9 files changed, 190 insertions(+), 73 deletions(-)
Signature · 38 lines
2.27.0 Thanks to Dr. Stolee, Dr. Narebski and Junio for their excellent suggestions. Changes in v2: - Introduce struct commit_graph_data. - Merge `graph_pos`, `generation` slabs into a single, `commit_graph_data` slab. - Use graph position for an intermediate check for generation, saving the cost of initializing generation numbers. - Add an follow-up patch caching results of slab access in local variables. - Move coccinelle transformation to commit.coccinelle instead of creating new scripts. - Elaborate on removing default values from init_commit_node(). - Revert moving macro constants (e.g. COMMIT_NOT_FROM_GRAPH, GENERATION_NUMBER_ZERO) from commit.h to commit-graph.h About the failing diff-submodule related tests, I came up with a plausible explanation but could be wrong on this: Commit slabs rely on uniqueness of commit->index to access data. But submodules are repositories on their own, alloc_commit_index(), which relies on repository->parsed_objects->commit_count no longer returns unique values. A commit belong to super repo and another belonging to submodule might have the same index but different generation and graph positions. This could be fixed by defining commit index as maximum of commit index of all repositories + 1 but I have no idea how that would impact other code. Thoughts on this? Regards Abhishek
PATCH v2 1/4 commit-graph: introduce commit_graph_data_slab
The struct commit is used in many contexts. However, members generation and graph_pos are only used for commit-graph related operations and otherwise waste memory.
As they are often accessed together, let's introduce struct commit_graph_data and move them to a commit_graph_data slab.
--- commit-graph.c | 49 +++++++++++++++++++++++++++++++++++++++++++++++++ commit-graph.h | 10 ++++++++++ 2 files changed, 59 insertions(+)
commit-graph.c +49 -0
@@ -87,6 +87,55 @@ static int commit_pos_cmp(const void *va, const void *vb)commit_pos_at (&commit_pos, b); } +define_commit_slab(commit_graph_data_slab, struct commit_graph_data); +static struct commit_graph_data_slab commit_graph_data_slab = +COMMIT_SLAB_INIT (1 , commit_graph_data_slab); + +uint32_t commit_graph_position (const struct commit *c) +{ +struct commit_graph_data *data = +commit_graph_data_slab_peek (&commit_graph_data_slab, c); + +return data ? data->graph_pos :COMMIT_NOT_FROM_GRAPH ; +} + +uint32_t commit_graph_generation (const struct commit *c) +{ +struct commit_graph_data *data = +commit_graph_data_slab_peek (&commit_graph_data_slab, c); + +if (!data) +return GENERATION_NUMBER_INFINITY ; +if (data->graph_pos ==COMMIT_NOT_FROM_GRAPH ) +return GENERATION_NUMBER_INFINITY ; + +return data->generation; +} + +static struct commit_graph_data *commit_graph_data_at (const struct commit *c) +{ +uint32_t i = commit_graph_data_slab.slab_count, j; +uint32_t slab_size = commit_graph_data_slab.slab_size; +struct commit_graph_data *data = +commit_graph_data_slab_at (&commit_graph_data_slab, c); + +/* +* commit-slab initializes elements with zero, overwrite this with +* COMMIT_NOT_FROM_GRAPH for graph_pos. +* +* We avoid the cost of initializing `generation` as generation +* number would be GENERATION_NUMBER_INFINITY if graph position +* is COMMIT_NOT_FROM_GRAPH. +*/ +for (; i < commit_graph_data_slab.slab_count; i++) { +for (j =0 ; j < slab_size; j++) { + commit_graph_data_slab.slab[i][j].graph_pos =COMMIT_NOT_FROM_GRAPH ; + } + } + +return data; +} +static int commit_gen_cmp (const void *va,const void *vb) {const struct commit *a = *(const struct commit **)va;
commit-graph.h +10 -0
@@ -137,4 +137,14 @@ void free_commit_graph(struct commit_graph *);*/ void disable_commit_graph (struct repository *r); +struct commit_graph_data { +uint32_t graph_pos; +uint32_t generation; +}; + +/* +* Commits should be parsed before accessing generation, graph positions. +*/ +uint32_t commit_graph_generation (const struct commit *); +uint32_t commit_graph_position (const struct commit *);#endif
Signature · 1 line
2.27.0
Re: [GSOC Patch v2 1/4] commit-graph: introduce commit_graph_data_slab
Quoted 60 lines · Abhishek Kumar
The struct commit is used in many contexts. However, members
generationandgraph_posare only used for commit-graph related operations and otherwise waste memory.As they are often accessed together, let's introduce struct
commit_graph_dataand move them to acommit_graph_dataslab.Signed-off-by Abhishek Kumar <[email protected]>--- commit-graph.c | 49 +++++++++++++++++++++++++++++++++++++++++++++++++ commit-graph.h | 10 ++++++++++ 2 files changed, 59 insertions(+)commit-graph.c +39
-0@@ -87,6 +87,55 @@ static int commit_pos_cmp(const void *va, const void *vb)commit_pos_at (&commit_pos, b); } +define_commit_slab(commit_graph_data_slab, struct commit_graph_data); +static struct commit_graph_data_slab commit_graph_data_slab = +COMMIT_SLAB_INIT (1 , commit_graph_data_slab); + +uint32_t commit_graph_position (const struct commit *c) +{ +struct commit_graph_data *data = +commit_graph_data_slab_peek (&commit_graph_data_slab, c); + +return data ? data->graph_pos :COMMIT_NOT_FROM_GRAPH ; +} + +uint32_t commit_graph_generation (const struct commit *c) +{ +struct commit_graph_data *data = +commit_graph_data_slab_peek (&commit_graph_data_slab, c); + +if (!data) +return GENERATION_NUMBER_INFINITY ; +if (data->graph_pos ==COMMIT_NOT_FROM_GRAPH ) +return GENERATION_NUMBER_INFINITY ; + +return data->generation; +} + +static struct commit_graph_data *commit_graph_data_at (const struct commit *c) +{ +uint32_t i = commit_graph_data_slab.slab_count, j; +uint32_t slab_size = commit_graph_data_slab.slab_size; +struct commit_graph_data *data = +commit_graph_data_slab_at (&commit_graph_data_slab, c); + +/* +* commit-slab initializes elements with zero, overwrite this with +* COMMIT_NOT_FROM_GRAPH for graph_pos. +* +* We avoid the cost of initializing `generation` as generation +* number would be GENERATION_NUMBER_INFINITY if graph position +* is COMMIT_NOT_FROM_GRAPH.
commit-graph.c static int commit_pos_cmp(const void *va, const void *vb) · line 138
Quoted 10 lines
commit-graph.c +10
-0+*/ +for (; i < commit_graph_data_slab.slab_count; i++) { +for (j =0 ; j < slab_size; j++) { + commit_graph_data_slab.slab[i][j].graph_pos =COMMIT_NOT_FROM_GRAPH ; + } + } + +return data; +} +
It looks like this function ('commit_graph_data_at') is unused in this patch. No big deal, especially since I can see that you are using it in the second patch, but I think that you should remove it from this patch and instead introduce it there.
In fact, this causes a build error (with 'make DEVELOPER=1') because of '-Wunused-function'. It's good practice to "git rebase -x 'make DEVELOPER=1' origin/master" before sending.
Quoted 23 lines
commit-graph.c +0
-0static int commit_gen_cmp (const void *va,const void *vb) {const struct commit *a = *(const struct commit **)va;commit-graph.h +10
-1@@ -137,4 +137,14 @@ void free_commit_graph(struct commit_graph *);*/ void disable_commit_graph (struct repository *r); +struct commit_graph_data { +uint32_t graph_pos; +uint32_t generation; +}; + +/* +* Commits should be parsed before accessing generation, graph positions. +*/ +uint32_t commit_graph_generation (const struct commit *); +uint32_t commit_graph_position (const struct commit *);#endif --
Quoted 1 line
2.27.0
Thanks,
Taylor
PATCH v2 2/4 commit: move members graph_pos, generation to a slab
In reply to Abhishek KumarWe remove members graph_pos and generation from the struct commit. The default assignments in init_commit_node() are no longer valid, which is fine as the slab helpers return appropriate default values and the assignments are removed.
We will replace existing use of commit->generation and commit->graph_pos by commit_graph_data slab helpers using `contrib/coccinelle/commit.cocci'.
--- alloc.c | 2 -- blame.c | 2 +- bloom.c | 6 ++-- commit-graph.c | 60 ++++++++++++++++----------------- commit-graph.h | 2 +- commit-reach.c | 50 +++++++++++++-------------- commit.c | 6 ++-- contrib/coccinelle/commit.cocci | 18 ++++++++++ revision.c | 16 ++++----- 9 files changed, 89 insertions(+), 73 deletions(-)
alloc.c +0 -2
@@ -108,8 +108,6 @@ void init_commit_node(struct repository *r, struct commit *c) { c->object.type =OBJ_COMMIT ; c->index =alloc_commit_index (r); - c->graph_pos =COMMIT_NOT_FROM_GRAPH ; - c->generation =GENERATION_NUMBER_INFINITY ; }void *alloc_commit_node (struct repository *r)
blame.c +1 -1
@@ -1272,7 +1272,7 @@ static int maybe_changed_path(struct repository *r,if (!bd)return 1 ; -if (origin->commit->generation ==GENERATION_NUMBER_INFINITY ) +if (commit_graph_generation (origin->commit) ==GENERATION_NUMBER_INFINITY )return 1 ; filter =get_bloom_filter (r, origin->commit,0 );
bloom.c +3 -3
@@ -34,14 +34,14 @@ static int load_bloom_filter_from_graph(struct commit_graph *g, {uint32_t lex_pos, start_index, end_index; -while (c->graph_pos < g->num_commits_in_base) +while (commit_graph_position (c) < g->num_commits_in_base) g = g->base_graph;/* The commit graph commit 'c' lives in doesn't carry bloom filters. */ if (!g->chunk_bloom_indexes)return 0 ; - lex_pos = c->graph_pos - g->num_commits_in_base; + lex_pos =commit_graph_position (c) - g->num_commits_in_base; end_index =get_be32 (g->chunk_bloom_indexes +4 * lex_pos);
@@ -188,7 +188,7 @@ struct bloom_filter *get_bloom_filter(struct repository *r,if (!filter->data) {load_commit_graph_info (r, c); -if (c->graph_pos !=COMMIT_NOT_FROM_GRAPH && +if (commit_graph_position (c) !=COMMIT_NOT_FROM_GRAPH && r->objects->commit_graph->chunk_bloom_indexes) {if (load_bloom_filter_from_graph (r->objects->commit_graph, filter, c))return filter;
commit-graph.c +30 -30
177 lines not on this page · raw message
commit-graph.h +1 -1
@@ -142,7 +142,7 @@ struct commit_graph_data {uint32_t generation; }; -/* +/* * Commits should be parsed before accessing generation, graph positions. */ uint32_t commit_graph_generation (const struct commit *);
commit-reach.c +25 -25
138 lines not on this page · raw message
commit.c +3 -3
@@ -339,7 +339,7 @@ struct tree *repo_get_commit_tree(struct repository *r,if (commit->maybe_tree || !commit->object.parsed)return commit->maybe_tree; -if (commit->graph_pos !=COMMIT_NOT_FROM_GRAPH ) +if (commit_graph_position (commit) !=COMMIT_NOT_FROM_GRAPH )return get_commit_tree_in_graph (r, commit);return NULL ;
@@ -731,9 +731,9 @@ int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, voidconst struct commit *a = a_, *b = b_;/* newer commits first */ -if (a->generation < b->generation) +if (commit_graph_generation (a) <commit_graph_generation (b))return 1 ; -else if (a->generation > b->generation) +else if (commit_graph_generation (a) >commit_graph_generation (b))return -1 ;/* use date as a heuristic when generations are equal */
contrib/coccinelle/commit.cocci +18 -0
@@ -32,3 +32,21 @@ expression c; - c->maybe_tree + repo_get_commit_tree(specify_the_right_repo_here, c) ...>} + +@@ +struct commit *c; +expression E; +@@ +( +- c->generation = E; ++ commit_graph_data_at(c)->generation = E; +| +- c->graph_pos = E; ++ commit_graph_data_at(c)->graph_pos = E; +| +- c->generation ++ commit_graph_generation(c) +| +- c->graph_pos ++ commit_graph_position(c) +)
revision.c +8 -8
@@ -720,7 +720,7 @@ static int check_maybe_different_in_bloom_filter(struct rev_info *revs,if (!revs->repo->objects->commit_graph)return -1 ; -if (commit->generation ==GENERATION_NUMBER_INFINITY ) +if (commit_graph_generation (commit) ==GENERATION_NUMBER_INFINITY )return -1 ; filter =get_bloom_filter (revs->repo, commit,0 );
@@ -3314,7 +3314,7 @@ static void explore_to_depth(struct rev_info *revs,struct topo_walk_info *info = revs->topo_walk_info;struct commit *c;while ((c =prio_queue_peek (&info->explore_queue)) && - c->generation >= gen_cutoff) +commit_graph_generation (c) >= gen_cutoff)explore_walk_step (revs); }
@@ -3330,7 +3330,7 @@ static void indegree_walk_step(struct rev_info *revs)if (parse_commit_gently (c,1 ) <0 )return ; -explore_to_depth (revs, c->generation); +explore_to_depth (revs,commit_graph_generation (c));for (p = c->parents; p; p = p->next) {struct commit *parent = p->item;
@@ -3354,7 +3354,7 @@ static void compute_indegrees_to_depth(struct rev_info *revs,struct topo_walk_info *info = revs->topo_walk_info;struct commit *c;while ((c =prio_queue_peek (&info->indegree_queue)) && - c->generation >= gen_cutoff) +commit_graph_generation (c) >= gen_cutoff)indegree_walk_step (revs); }
@@ -3414,8 +3414,8 @@ static void init_topo_walk(struct rev_info *revs)test_flag_and_insert (&info->explore_queue, c,TOPO_WALK_EXPLORED );test_flag_and_insert (&info->indegree_queue, c,TOPO_WALK_INDEGREE ); -if (c->generation < info->min_generation) - info->min_generation = c->generation; +if (commit_graph_generation (c) < info->min_generation) + info->min_generation =commit_graph_generation (c); *(indegree_slab_at (&info->indegree, c)) =1 ;
@@ -3473,8 +3473,8 @@ static void expand_topo_walk(struct rev_info *revs, struct commit *commit)if (parse_commit_gently (parent,1 ) <0 )continue ; -if (parent->generation < info->min_generation) { - info->min_generation = parent->generation; +if (commit_graph_generation (parent) < info->min_generation) { + info->min_generation =commit_graph_generation (parent);compute_indegrees_to_depth (revs, info->min_generation); }
Signature · 1 line
2.27.0
Re: [GSOC Patch v2 2/4] commit: move members graph_pos, generation to a slab
Quoted 36 lines · Abhishek Kumar
commit-graph.c +6
-7@@ -1302,8 +1302,8 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx) ctx->commits.nr);for (i =0 ; i < ctx->commits.nr; i++) {display_progress (ctx->progress, i +1 ); -if (ctx->commits.list[i]->generation !=GENERATION_NUMBER_INFINITY && - ctx->commits.list[i]->generation !=GENERATION_NUMBER_ZERO ) +if (commit_graph_generation (ctx->commits.list[i]) !=GENERATION_NUMBER_INFINITY && +commit_graph_generation (ctx->commits.list[i]) !=GENERATION_NUMBER_ZERO )continue ;commit_list_insert (ctx->commits.list[i], &list);@@ -1314,22 +1314,22 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)uint32_t max_generation =0 ;for (parent = current->parents; parent; parent = parent->next) { -if (parent->item->generation ==GENERATION_NUMBER_INFINITY || - parent->item->generation ==GENERATION_NUMBER_ZERO ) { +if (commit_graph_generation (parent->item) ==GENERATION_NUMBER_INFINITY || +commit_graph_generation (parent->item) ==GENERATION_NUMBER_ZERO ) { all_parents_computed =0 ;commit_list_insert (parent->item, &list);break ; - }else if (parent->item->generation > max_generation) { - max_generation = parent->item->generation; + }else if (commit_graph_generation (parent->item) > max_generation) { + max_generation =commit_graph_generation (parent->item); } }if (all_parents_computed) { - current->generation = max_generation +1 ;
commit-graph.c static void compute_generation_numbers(struct write_commit_graph_context *ctx) · line 1335
Quoted 10 lines
commit-graph.c +3
-2+commit_graph_data_at (current)->generation = max_generation +1 ;pop_commit (&list); -if (current->generation >GENERATION_NUMBER_MAX ) - current->generation =GENERATION_NUMBER_MAX ; +if (commit_graph_generation (current) >GENERATION_NUMBER_MAX ) +commit_graph_data_at (current)->generation =GENERATION_NUMBER_MAX ; } } }
Something about these conversions is not right, as they send compute_generation_numbers() into an endless loop, and 't5318-commit-graph.sh' hangs because of this.
Re: [GSOC Patch v2 2/4] commit: move members graph_pos, generation to a slab
Quoted 40 lines · SZEDER Gábor
Quoted 39 lines · Abhishek Kumar
commit-graph.c +7
-7@@ -1302,8 +1302,8 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx) ctx->commits.nr);for (i =0 ; i < ctx->commits.nr; i++) {display_progress (ctx->progress, i +1 ); -if (ctx->commits.list[i]->generation !=GENERATION_NUMBER_INFINITY && - ctx->commits.list[i]->generation !=GENERATION_NUMBER_ZERO ) +if (commit_graph_generation (ctx->commits.list[i]) !=GENERATION_NUMBER_INFINITY && +commit_graph_generation (ctx->commits.list[i]) !=GENERATION_NUMBER_ZERO )continue ;commit_list_insert (ctx->commits.list[i], &list);@@ -1314,22 +1314,22 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)uint32_t max_generation =0 ;for (parent = current->parents; parent; parent = parent->next) { -if (parent->item->generation ==GENERATION_NUMBER_INFINITY || - parent->item->generation ==GENERATION_NUMBER_ZERO ) { +if (commit_graph_generation (parent->item) ==GENERATION_NUMBER_INFINITY || +commit_graph_generation (parent->item) ==GENERATION_NUMBER_ZERO ) { all_parents_computed =0 ;commit_list_insert (parent->item, &list);break ; - }else if (parent->item->generation > max_generation) { - max_generation = parent->item->generation; + }else if (commit_graph_generation (parent->item) > max_generation) { + max_generation =commit_graph_generation (parent->item); } }if (all_parents_computed) { - current->generation = max_generation +1 ; +commit_graph_data_at (current)->generation = max_generation +1 ;pop_commit (&list);
Quoted 11 lines
commit-graph.c static void compute_generation_numbers(struct write_commit_graph_context *ctx) · line 1335
Quoted 7 lines
commit-graph.c +2
-2-if (current->generation >GENERATION_NUMBER_MAX ) - current->generation =GENERATION_NUMBER_MAX ; +if (commit_graph_generation (current) >GENERATION_NUMBER_MAX ) +commit_graph_data_at (current)->generation =GENERATION_NUMBER_MAX ; } } }Something about these conversions is not right, as they send
compute_generation_numbers()into an endless loop, and 't5318-commit-graph.sh' hangs because of this.
Abhishek responded off-list, but it's worth having the discussion here, too.
While the next patch fixes the bug introduced here, we strive to have every patch compile and pass all tests on all platforms. It can be hard to verify that last "all platforms" condition, but we can run (most) tests on each of our patches using the following:
$ git rebase -x "make -j12 DEVELOPER=1 && (cd t && prove -j8 t[0-8]*.sh)" <base>
Thanks, Szeder, for finding this issue in the patch.
Looking at this patch and patch 3, I think you should just squash that patch into this one, since the code you are removing in patch 3 was added by this one. Add a paragraph in your commit message that details why we need to use commit_graph_data_at() directly in write_graph_chunk_data() and compute_generation_numbers().
Thanks,
-Stolee
PATCH v2 3/4 commit-graph: use generation directly when writing commit-graph
In reply to Abhishek Kumarcommit_graph_generation() returns GENERATION_NUMBER_INFINITY if the graph position for commit is COMMIT_NOT_FROM_GRAPH.
While this is true when reading from a commit graph, no graph positions are associated with a commit when writing a commit graph. Therefore, the helper incorrectly returns GENERATION_NUMBER_INFINITY despite having a finite generation number.
Let's fix this by using generation number directly when writing a commit graph.
--- commit-graph.c | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-)
commit-graph.c +8 -5
@@ -1070,7 +1070,7 @@ static void write_graph_chunk_data(struct hashfile *f, int hash_len,else packedDate[0 ] =0 ; - packedDate[0 ] |=htonl (commit_graph_generation ((*list)) <<2 ); + packedDate[0 ] |=htonl (commit_graph_data_at (*list)->generation <<2 ); packedDate[1 ] =htonl ((*list)->date);hashwrite (f, packedDate,8 );
@@ -1301,9 +1301,11 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)_ ("Computing commit graph generation numbers" ), ctx->commits.nr);for (i =0 ; i < ctx->commits.nr; i++) { +uint32_t generation =commit_graph_data_at (ctx->commits.list[i])->generation; +display_progress (ctx->progress, i +1 ); -if (commit_graph_generation (ctx->commits.list[i]) !=GENERATION_NUMBER_INFINITY && -commit_graph_generation (ctx->commits.list[i]) !=GENERATION_NUMBER_ZERO ) +if (generation !=GENERATION_NUMBER_INFINITY && + generation !=GENERATION_NUMBER_ZERO )continue ;commit_list_insert (ctx->commits.list[i], &list);
@@ -1314,8 +1316,9 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)uint32_t max_generation =0 ;for (parent = current->parents; parent; parent = parent->next) { -if (commit_graph_generation (parent->item) ==GENERATION_NUMBER_INFINITY || -commit_graph_generation (parent->item) ==GENERATION_NUMBER_ZERO ) { + +if (generation ==GENERATION_NUMBER_INFINITY || + generation ==GENERATION_NUMBER_ZERO ) { all_parents_computed =0 ;commit_list_insert (parent->item, &list);break ;
Signature · 1 line
2.27.0
Re: [GSOC Patch v2 3/4] commit-graph: use generation directly when writing commit-graph
Quoted 10 lines · Abhishek Kumar
commit_graph_generation()returnsGENERATION_NUMBER_INFINITYif the graph position for commit isCOMMIT_NOT_FROM_GRAPH.While this is true when reading from a commit graph, no graph positions are associated with a commit when writing a commit graph. Therefore, the helper incorrectly returns
GENERATION_NUMBER_INFINITYdespite having a finite generation number.Let's fix this by using generation number directly when writing a commit graph.
I think that to avoid having non-working patch (which can cause problems when bisecting), it would be a better idea to switch the order of patches 2 and 3. This way we won't have incorrect behaviour.
Quoted 11 lines
Signed-off-by Abhishek Kumar <[email protected]>--- commit-graph.c | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-)commit-graph.c +0
-0
commit-graph.c static void write_graph_chunk_data(struct hashfile *f, int hash_len, · line 1074
Quoted 6 lines
commit-graph.c +1
-1@@ -1070,7 +1070,7 @@ static void write_graph_chunk_data(struct hashfile *f, int hash_len,else packedDate[0 ] =0 ; - packedDate[0 ] |=htonl (commit_graph_generation ((*list)) <<2 ); + packedDate[0 ] |=htonl (commit_graph_data_at (*list)->generation <<2 );
All right.
Quoted 5 lines
commit-graph.c +0
-0packedDate[1 ] =htonl ((*list)->date);hashwrite (f, packedDate,8 );@@ -1301,9 +1301,11 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)_ ("Computing commit graph generation numbers" ), ctx->commits.nr);
commit-graph.c static void compute_generation_numbers(struct write_commit_graph_context *ctx) · line 1310
Quoted 10 lines
commit-graph.c +4
-2for (i =0 ; i < ctx->commits.nr; i++) { +uint32_t generation =commit_graph_data_at (ctx->commits.list[i])->generation; +display_progress (ctx->progress, i +1 ); -if (commit_graph_generation (ctx->commits.list[i]) !=GENERATION_NUMBER_INFINITY && -commit_graph_generation (ctx->commits.list[i]) !=GENERATION_NUMBER_ZERO ) +if (generation !=GENERATION_NUMBER_INFINITY && + generation !=GENERATION_NUMBER_ZERO )continue ;
All right; this also introduces local variable to avoid accessing the slab twice^W four times...
commit-graph.c static void compute_generation_numbers(struct write_commit_graph_context *ctx) · line 1324
Quoted 13 lines
commit-graph.c +3
-2commit_list_insert (ctx->commits.list[i], &list);@@ -1314,8 +1316,9 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)uint32_t max_generation =0 ;for (parent = current->parents; parent; parent = parent->next) { -if (commit_graph_generation (parent->item) ==GENERATION_NUMBER_INFINITY || -commit_graph_generation (parent->item) ==GENERATION_NUMBER_ZERO ) { + +if (generation ==GENERATION_NUMBER_INFINITY || + generation ==GENERATION_NUMBER_ZERO ) { all_parents_computed =0 ;commit_list_insert (parent->item, &list);break ;
... which is then used here.
Best,
Signature · 1 line
Jakub Narębski
Re: [GSOC Patch v2 3/4] commit-graph: use generation directly when writing commit-graph
Quoted 10 lines · Jakub Narębski
Quoted 8 lines · Abhishek Kumar
commit_graph_generation()returnsGENERATION_NUMBER_INFINITYif the graph position for commit isCOMMIT_NOT_FROM_GRAPH.While this is true when reading from a commit graph, no graph positions are associated with a commit when writing a commit graph. Therefore, the helper incorrectly returns
GENERATION_NUMBER_INFINITYdespite having a finite generation number.
Quoted 6 lines
Quoted 2 lines
Let's fix this by using generation number directly when writing a commit graph.
I think that to avoid having non-working patch (which can cause problems when bisecting), it would be a better idea to switch the order of patches 2 and 3. This way we won't have incorrect behaviour.
Yup... agreed (with the additional caveat that the unused function in the first patch also be moved into this new--larger--patch).
Quoted 55 lines
Quoted 11 lines
Signed-off-by Abhishek Kumar <[email protected]>--- commit-graph.c | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-)commit-graph.c +0
-0commit-graph.c static void write_graph_chunk_data(struct hashfile *f, int hash_len, · line 1074
Quoted 6 lines
commit-graph.c +1
-1@@ -1070,7 +1070,7 @@ static void write_graph_chunk_data(struct hashfile *f, int hash_len,else packedDate[0 ] =0 ; - packedDate[0 ] |=htonl (commit_graph_generation ((*list)) <<2 ); + packedDate[0 ] |=htonl (commit_graph_data_at (*list)->generation <<2 );All right.
Quoted 5 lines
commit-graph.c +0
-0packedDate[1 ] =htonl ((*list)->date);hashwrite (f, packedDate,8 );@@ -1301,9 +1301,11 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)_ ("Computing commit graph generation numbers" ), ctx->commits.nr);commit-graph.c static void compute_generation_numbers(struct write_commit_graph_context *ctx) · line 1310
Quoted 10 lines
commit-graph.c +4
-2for (i =0 ; i < ctx->commits.nr; i++) { +uint32_t generation =commit_graph_data_at (ctx->commits.list[i])->generation; +display_progress (ctx->progress, i +1 ); -if (commit_graph_generation (ctx->commits.list[i]) !=GENERATION_NUMBER_INFINITY && -commit_graph_generation (ctx->commits.list[i]) !=GENERATION_NUMBER_ZERO ) +if (generation !=GENERATION_NUMBER_INFINITY && + generation !=GENERATION_NUMBER_ZERO )continue ;All right; this also introduces local variable to avoid accessing the slab twice^W four times...
commit-graph.c static void compute_generation_numbers(struct write_commit_graph_context *ctx) · line 1324
Quoted 13 lines
commit-graph.c +3
-2commit_list_insert (ctx->commits.list[i], &list);@@ -1314,8 +1316,9 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)uint32_t max_generation =0 ;for (parent = current->parents; parent; parent = parent->next) { -if (commit_graph_generation (parent->item) ==GENERATION_NUMBER_INFINITY || -commit_graph_generation (parent->item) ==GENERATION_NUMBER_ZERO ) { + +if (generation ==GENERATION_NUMBER_INFINITY || + generation ==GENERATION_NUMBER_ZERO ) { all_parents_computed =0 ;commit_list_insert (parent->item, &list);break ;... which is then used here.
Quoted 3 lines
Best,
--
Jakub Narębski
Thanks,
Taylor
PATCH v2 4/4 commit-graph: minimize commit_graph_data_slab access
In reply to Abhishek KumarIn an earlier patch, multiple struct acccesses to graph_pos and generation were auto-converted to multiple method calls.
Since the values are fixed and commit-slab access costly, we would be better off with storing the values as a local variable and reuse it.
[1]: https://lore.kernel.org/git/[email protected]/
--- bloom.c | 5 +++-- commit-graph.c | 55 +++++++++++++++++++++++++++++----------------- commit-reach.c | 59 ++++++++++++++++++++++++++++++++------------------ commit.c | 6 +++-- revision.c | 12 ++++++---- 5 files changed, 88 insertions(+), 49 deletions(-)
bloom.c +3 -2
@@ -33,15 +33,16 @@ static int load_bloom_filter_from_graph(struct commit_graph *g, struct commit *c) {uint32_t lex_pos, start_index, end_index; +uint32_t graph_pos =commit_graph_position (c); -while (commit_graph_position (c) < g->num_commits_in_base) +while (graph_pos < g->num_commits_in_base) g = g->base_graph;/* The commit graph commit 'c' lives in doesn't carry bloom filters. */ if (!g->chunk_bloom_indexes)return 0 ; - lex_pos =commit_graph_position (c) - g->num_commits_in_base; + lex_pos = graph_pos - g->num_commits_in_base; end_index =get_be32 (g->chunk_bloom_indexes +4 * lex_pos);
commit-graph.c +35 -20
153 lines not on this page · raw message
commit-reach.c +38 -21
140 lines not on this page · raw message
commit.c +4 -2
@@ -729,11 +729,13 @@ int compare_commits_by_author_date(const void *a_, const void *b_,int compare_commits_by_gen_then_commit_date (const void *a_,const void *b_,void *unused) {const struct commit *a = a_, *b = b_; +const uint32_t generation_a =commit_graph_generation (a), + generation_b =commit_graph_generation (b);/* newer commits first */ -if (commit_graph_generation (a) <commit_graph_generation (b)) +if (generation_a < generation_b)return 1 ; -else if (commit_graph_generation (a) >commit_graph_generation (b)) +else if (generation_a > generation_b)return -1 ;/* use date as a heuristic when generations are equal */
revision.c +8 -4
@@ -3407,6 +3407,7 @@ static void init_topo_walk(struct rev_info *revs) info->min_generation =GENERATION_NUMBER_INFINITY ;for (list = revs->commits; list; list = list->next) {struct commit *c = list->item; +uint32_t generation;if (parse_commit_gently (c,1 ))continue ;
@@ -3414,8 +3415,9 @@ static void init_topo_walk(struct rev_info *revs)test_flag_and_insert (&info->explore_queue, c,TOPO_WALK_EXPLORED );test_flag_and_insert (&info->indegree_queue, c,TOPO_WALK_INDEGREE ); -if (commit_graph_generation (c) < info->min_generation) - info->min_generation =commit_graph_generation (c); + generation =commit_graph_generation (c); +if (generation < info->min_generation) + info->min_generation = generation; *(indegree_slab_at (&info->indegree, c)) =1 ;
@@ -3466,6 +3468,7 @@ static void expand_topo_walk(struct rev_info *revs, struct commit *commit)for (p = commit->parents; p; p = p->next) {struct commit *parent = p->item;int *pi; +uint32_t generation;if (parent->object.flags &UNINTERESTING )continue ;
@@ -3473,8 +3476,9 @@ static void expand_topo_walk(struct rev_info *revs, struct commit *commit)if (parse_commit_gently (parent,1 ) <0 )continue ; -if (commit_graph_generation (parent) < info->min_generation) { - info->min_generation =commit_graph_generation (parent); + generation =commit_graph_generation (parent); +if (generation < info->min_generation) { + info->min_generation = generation;compute_indegrees_to_depth (revs, info->min_generation); }
Signature · 1 line
2.27.0
Quoted 7 lines · Abhishek Kumar
The struct commit is used in many contexts. However, members
generationandgraph_posare only used for commit graph related operations and otherwise waste memory.This wastage would have been more pronounced as we transition to generation number v2, which uses 64-bite generation number instead of current 32-bits.
It would be nice (though not required) to have some specific data:
benchmarks with time and memory (RSS maybe?) of Git commands using commit-graph and those not using it, before and after this patch series. Maybe time to run the test suite, or the perf suite...
But this is not a show stopper, in my opinion.
Best,
Jakub Narębski
Quoted 16 lines
Abhishek Kumar (4):
commit-graph: introducecommit_graph_data_slab
commit: move membersgraph_pos, generation to a slab
commit-graph: use generation directly when writing commit-graph
commit-graph: minimizecommit_graph_data_slabaccessalloc.c | 2 - blame.c | 2 +- bloom.c | 7 +- commit-graph.c | 127 ++++++++++++++++++++++++-------- commit-graph.h | 10 +++ commit-reach.c | 69 ++++++++++------- commit.c | 8 +- contrib/coccinelle/commit.cocci | 18 +++++ revision.c | 20 +++-- 9 files changed, 190 insertions(+), 73 deletions(-)
Hi Abhishek,
I am so excited that you are working on this, and welcome to Git! This change will make a meaningful difference for us at GitHub (who use commit graphs extensively), and it sounds like they will even make a larger difference with your later changes.
I was out of the office last week, and so I hadn't gotten a chance to review the first version of this series, but I'll review v2 now...
Quoted 3 lines · Abhishek Kumar
The struct commit is used in many contexts. However, members
generationandgraph_posare only used for commit graph related operations and otherwise waste memory.
Thanks,
Taylor
v4 cover Move generation, graph_pos to a slab
In reply to Abhishek KumarThe struct commit is used in many contexts. However, members generation and graph_pos are only used for commit graph related operations and otherwise waste memory.
This wastage would have been more pronounced as we transition to generation nuber v2, which uses 64-bit generation number instead of current 32-bits.
While the overall test suite runs as fast as master
(series: 26m48s, master: 27m34s, faster by 2.87%), certain commands like git merge-base --is-ancestor were slowed by 40% as discovered by Szeder Gábor [1]. After minimizing commit-slab access, the slow down persists but is closer to 20%.
Derrick Stolee believes the slow down is attributable to the underlying algorithm rather than the slowness of commit-slab access [2] and we will follow-up in a later series.
Abhishek Kumar (4):
object: drop parsed_object_pool->commit_count
commit-graph: introduce commit_graph_data_slab
commit: move members graph_pos, generation to a slab
commit-graph: minimize commit_graph_data_slab access
alloc.c | 18 +++-- alloc.h | 2 +- blame.c | 2 +- blob.c | 2 +- bloom.c | 7 +- builtin/commit-graph.c | 2 +- builtin/fsck.c | 2 +- commit-graph.c | 130 ++++++++++++++++++++++++-------- commit-graph.h | 10 +++ commit-reach.c | 69 ++++++++++------- commit.c | 12 +-- commit.h | 2 - contrib/coccinelle/commit.cocci | 18 +++++ object.c | 4 +- object.h | 3 +- refs.c | 2 +- revision.c | 20 +++-- t/helper/test-reach.c | 2 +- tag.c | 2 +- tree.c | 2 +- 20 files changed, 217 insertions(+), 94 deletions(-)
Signature · 23 lines
2.27.0 Changes in v4: - Fix segfault while initializing commit_graph_data_slab. - Fix the "commit index should have unique values" more cleanly Changes in v3: - Introduce alloc commit to fix the failing diff-submodule test. - Elaborate on perforamnce and the slow down in commit message Changes in v2: - Introduce struct commit_graph_data - Merge `graph_pos`, `generation` slabs into a single, `commit_graph_data` slab. - Use graph position for an intermediate check for generation, saving the cost of initializing generation numbers. - Add an follow-up patch caching results of slab access in local variables. - Move coccinelle transformation to commit.coccinelle instead of creating new scripts. - Elaborate on removing default values from init_commit_node(). - Revert moving macro constants (e.g. COMMIT_NOT_FROM_GRAPH, GENERATION_NUMBER_ZERO0 from commit.h to commit-graph.h
PATCH v4 1/4 object: drop parsed_object_pool->commit_count
14ba97f8 (alloc: allow arbitrary repositories for alloc functions, 2018-05-15) introduced parsed_object_pool->commit_count to keep count of commits per repository and was used to assign commit->index.
However, commit-slab code requires commit->index values to be unique and a global count would be correct, rather than a per-repo count.
Let's introduce a static counter variable, parsed_commits_count to keep track of parsed commits so far.
As commit_count has no use anymore, let's also drop it from the struct.
--- alloc.c | 16 +++++++++++----- alloc.h | 2 +- blob.c | 2 +- builtin/commit-graph.c | 2 +- builtin/fsck.c | 2 +- commit.c | 4 ++-- object.c | 4 ++-- object.h | 3 +-- refs.c | 2 +- t/helper/test-reach.c | 2 +- tag.c | 2 +- tree.c | 2 +- 12 files changed, 24 insertions(+), 19 deletions(-)
alloc.c +11 -5
@@ -99,15 +99,21 @@ void *alloc_object_node(struct repository *r)return obj; } -static unsigned int alloc_commit_index (struct repository *r) +/* +* The returned count is to be used as an index into commit slabs, +* that are *NOT* maintained per repository, and that is why a single +* global counter is used. +*/ +static unsigned int alloc_commit_index (void ) { -return r->parsed_objects->commit_count++; +static unsigned int parsed_commits_count; +return parsed_commits_count++; } -void init_commit_node (struct repository *r,struct commit *c) +void init_commit_node (struct commit *c) { c->object.type =OBJ_COMMIT ; - c->index =alloc_commit_index (r); + c->index =alloc_commit_index (); c->graph_pos =COMMIT_NOT_FROM_GRAPH ; c->generation =GENERATION_NUMBER_INFINITY ; }
@@ -115,7 +121,7 @@ void init_commit_node(struct repository *r, struct commit *c)void *alloc_commit_node (struct repository *r) {struct commit *c =alloc_node (r->parsed_objects->commit_state,sizeof (struct commit )); -init_commit_node (r, c); +init_commit_node (c);return c; }
alloc.h +1 -1
@@ -9,7 +9,7 @@ struct repository;void *alloc_blob_node (struct repository *r);void *alloc_tree_node (struct repository *r); -void init_commit_node (struct repository *r,struct commit *c); +void init_commit_node (struct commit *c);void *alloc_commit_node (struct repository *r);void *alloc_tag_node (struct repository *r);void *alloc_object_node (struct repository *r);
blob.c +1 -1
@@ -10,7 +10,7 @@ struct blob *lookup_blob(struct repository *r, const struct object_id *oid)struct object *obj =lookup_object (r, oid);if (!obj)return create_object (r, oid,alloc_blob_node (r)); -return object_as_type (r, obj,OBJ_BLOB ,0 ); +return object_as_type (obj,OBJ_BLOB ,0 ); }int parse_blob_buffer (struct blob *item,void *buffer,unsigned long size)
builtin/commit-graph.c +1 -1
@@ -154,7 +154,7 @@ static int read_one_commit(struct oidset *commits, struct progress *progress, NULL, 0);if (!result)return error (_ ("invalid object: %s" ), hash); -else if (object_as_type (the_repository, result,OBJ_COMMIT ,1 )) +else if (object_as_type (result,OBJ_COMMIT ,1 ))oidset_insert (commits, &result->oid);display_progress (progress,oidset_size (commits));
builtin/fsck.c +1 -1
@@ -241,7 +241,7 @@ static void mark_unreachable_referents(const struct object_id *oid)enum object_type type =oid_object_info (the_repository, &obj->oid,NULL );if (type >0 ) -object_as_type (the_repository, obj, type,0 ); +object_as_type (obj, type,0 ); } options.walk = mark_used;
commit.c +2 -2
@@ -37,7 +37,7 @@ struct commit *lookup_commit_reference_gently(struct repository *r,if (!obj)return NULL ; -return object_as_type (r, obj,OBJ_COMMIT , quiet); +return object_as_type (obj,OBJ_COMMIT , quiet); }struct commit *lookup_commit_reference (struct repository *r,const struct object_id *oid)
@@ -62,7 +62,7 @@ struct commit *lookup_commit(struct repository *r, const struct object_id *oid)struct object *obj =lookup_object (r, oid);if (!obj)return create_object (r, oid,alloc_commit_node (r)); -return object_as_type (r, obj,OBJ_COMMIT ,0 ); +return object_as_type (obj,OBJ_COMMIT ,0 ); }struct commit *lookup_commit_reference_by_name (const char *name)
object.c +2 -2
@@ -157,13 +157,13 @@ void *create_object(struct repository *r, const struct object_id *oid, void *o)return obj; } -void *object_as_type (struct repository *r,struct object *obj,enum object_type type,int quiet) +void *object_as_type (struct object *obj,enum object_type type,int quiet) {if (obj->type == type)return obj;else if (obj->type ==OBJ_NONE ) {if (type ==OBJ_COMMIT ) -init_commit_node (r, (struct commit *) obj); +init_commit_node ((struct commit *) obj);else obj->type = type;return obj;
object.h +1 -2
@@ -15,7 +15,6 @@ struct parsed_object_pool {struct alloc_state *commit_state;struct alloc_state *tag_state;struct alloc_state *object_state; -unsigned commit_count;/* parent substitutions from .git/info/grafts and .git/shallow */ struct commit_graft **grafts;
@@ -121,7 +120,7 @@ struct object *lookup_object(struct repository *r, const struct object_id *oid);void *create_object (struct repository *r,const struct object_id *oid,void *obj); -void *object_as_type (struct repository *r,struct object *obj,enum object_type type,int quiet); +void *object_as_type (struct object *obj,enum object_type type,int quiet);/* * Returns the object, having parsed it to find out what it is.
refs.c +1 -1
@@ -339,7 +339,7 @@ enum peel_status peel_object(const struct object_id *name, struct object_id *oidif (o->type ==OBJ_NONE ) {int type =oid_object_info (the_repository, name,NULL ); -if (type <0 || !object_as_type (the_repository, o, type,0 )) +if (type <0 || !object_as_type (o, type,0 ))return PEEL_INVALID ; }
t/helper/test-reach.c +1 -1
@@ -67,7 +67,7 @@ int cmd__reach(int ac, const char **av)die ("failed to load commit for input %s resulting in oid %s\n" , buf.buf,oid_to_hex (&oid)); - c =object_as_type (r, peeled,OBJ_COMMIT ,0 ); + c =object_as_type (peeled,OBJ_COMMIT ,0 );if (!c) die("failed to load commit for input %s resulting in oid %s\n",
tag.c +1 -1
@@ -103,7 +103,7 @@ struct tag *lookup_tag(struct repository *r, const struct object_id *oid)struct object *obj =lookup_object (r, oid);if (!obj)return create_object (r, oid,alloc_tag_node (r)); -return object_as_type (r, obj,OBJ_TAG ,0 ); +return object_as_type (obj,OBJ_TAG ,0 ); }static timestamp_t parse_tag_date (const char *buf,const char *tail)
tree.c +1 -1
@@ -200,7 +200,7 @@ struct tree *lookup_tree(struct repository *r, const struct object_id *oid)struct object *obj =lookup_object (r, oid);if (!obj)return create_object (r, oid,alloc_tree_node (r)); -return object_as_type (r, obj,OBJ_TREE ,0 ); +return object_as_type (obj,OBJ_TREE ,0 ); }int parse_tree_buffer (struct tree *item,void *buffer,unsigned long size)
Signature · 1 line
2.27.0
PATCH v4 2/4 commit-graph: introduce commit_graph_data_slab
In reply to Abhishek KumarThe struct commit is used in many contexts. However, members generation and graph_pos are only used for commit-graph related operations and otherwise waste memory.
This wastage would have been more pronounced as we transition to generation number v2, which uses 64-bit generation number instead of current 32-bits.
As they are often accessed together, let's introduce struct commit_graph_data and move them to a commit_graph_data slab.
While the overall test suite runs just as fast as master, (series: 26m48s, master: 27m34s, faster by 2.87%), certain commands like git merge-base --is-ancestor were slowed by 40% as discovered by Szeder Gábor [1]. After minimizing commit-slab access, the slow down persists but is closer to 20%.
Derrick Stolee believes the slow down is attributable to the underlying algorithm rather than the slowness of commit-slab access [2] and we will follow-up in a later series.
[1]: https://lore.kernel.org/git/[email protected]/
[2]: https://lore.kernel.org/git/[email protected]/
---
On linux.git with HEAD at 08bf1a27 (Merge tag 'powerpc-5.8-2' ofgit://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux,
2020-06-13):
git merge-base --is-ancestor HEAD~50000 HEAD
Time (master): 0.787s Time (series): 0.927s Change: 17.79% (slower)
Max RSS (master): 177694kb Max RSS (series): 177707kb Change: 0.01% (more)
git gc
Time (master): 3m55s Time (series): 3m38s Change: 7.23% (faster)
Max RSS (master): 4889868kb Max RSS (series): 4911960kb Change: 0.45% (more)
Earlier implementation of commit_graph_data_at() was incorrect, as we used to iterate from old slab count to new slab count - assuming all intermediate slabs are allocated. This is incorrect as the slabs are allocated only when there's a corresponding commit.
It now makes *two slab accesses* in the worst case, but it's okay since the worst case occurs once nearly every (512kb / 8b) commits.
commit-graph.c | 78 +++++++++++++++++++++++++++++++++++++++++++------- commit-graph.h | 10 +++++++ 2 files changed, 78 insertions(+), 10 deletions(-)
commit-graph.c +68 -10
@@ -87,6 +87,58 @@ static int commit_pos_cmp(const void *va, const void *vb)commit_pos_at (&commit_pos, b); } +define_commit_slab(commit_graph_data_slab, struct commit_graph_data); +static struct commit_graph_data_slab commit_graph_data_slab = +COMMIT_SLAB_INIT (1 , commit_graph_data_slab); + +uint32_t commit_graph_position (const struct commit *c) +{ +struct commit_graph_data *data = +commit_graph_data_slab_peek (&commit_graph_data_slab, c); + +return data ? data->graph_pos :COMMIT_NOT_FROM_GRAPH ; +} + +uint32_t commit_graph_generation (const struct commit *c) +{ +struct commit_graph_data *data = +commit_graph_data_slab_peek (&commit_graph_data_slab, c); + +if (!data) +return GENERATION_NUMBER_INFINITY ; +else if (data->graph_pos ==COMMIT_NOT_FROM_GRAPH ) +return GENERATION_NUMBER_INFINITY ; + +return data->generation; +} + +static struct commit_graph_data *commit_graph_data_at (const struct commit *c) +{ +unsigned int i, nth_slab; +struct commit_graph_data *data = +commit_graph_data_slab_peek (&commit_graph_data_slab, c); + +if (data) +return data; + + nth_slab = c->index / commit_graph_data_slab.slab_size; + data =commit_graph_data_slab_at (&commit_graph_data_slab, c); + +/* +* commit-slab initializes elements with zero, overwrite this with +* COMMIT_NOT_FROM_GRAPH for graph_pos. +* +* We avoid initializing generation with checking if graph position +* is not COMMIT_NOT_FROM_GRAPH. +*/ +for (i =0 ; i < commit_graph_data_slab.slab_size; i++) { + commit_graph_data_slab.slab[nth_slab][i].graph_pos = +COMMIT_NOT_FROM_GRAPH ; + } + +return data; +} +static int commit_gen_cmp (const void *va,const void *vb) {const struct commit *a = *(const struct commit **)va;
@@ -1020,7 +1072,7 @@ static void write_graph_chunk_data(struct hashfile *f, int hash_len,else packedDate[0 ] =0 ; - packedDate[0 ] |=htonl ((*list)->generation <<2 ); + packedDate[0 ] |=htonl (commit_graph_data_at (*list)->generation <<2 ); packedDate[1 ] =htonl ((*list)->date);hashwrite (f, packedDate,8 );
@@ -1251,9 +1303,11 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)_ ("Computing commit graph generation numbers" ), ctx->commits.nr);for (i =0 ; i < ctx->commits.nr; i++) { +uint32_t generation =commit_graph_data_at (ctx->commits.list[i])->generation; +display_progress (ctx->progress, i +1 ); -if (ctx->commits.list[i]->generation !=GENERATION_NUMBER_INFINITY && - ctx->commits.list[i]->generation !=GENERATION_NUMBER_ZERO ) +if (generation !=GENERATION_NUMBER_INFINITY && + generation !=GENERATION_NUMBER_ZERO )continue ;commit_list_insert (ctx->commits.list[i], &list);
@@ -1264,22 +1318,26 @@ static void compute_generation_numbers(struct write_commit_graph_context *ctx)uint32_t max_generation =0 ;for (parent = current->parents; parent; parent = parent->next) { -if (parent->item->generation ==GENERATION_NUMBER_INFINITY || - parent->item->generation ==GENERATION_NUMBER_ZERO ) { + generation =commit_graph_data_at (parent->item)->generation; + +if (generation ==GENERATION_NUMBER_INFINITY || + generation ==GENERATION_NUMBER_ZERO ) { all_parents_computed =0 ;commit_list_insert (parent->item, &list);break ; - }else if (parent->item->generation > max_generation) { - max_generation = parent->item->generation; + }else if (generation > max_generation) { + max_generation = generation; } }if (all_parents_computed) { - current->generation = max_generation +1 ; +struct commit_graph_data *data =commit_graph_data_at (current); + + data->generation = max_generation +1 ;pop_commit (&list); -if (current->generation >GENERATION_NUMBER_MAX ) - current->generation =GENERATION_NUMBER_MAX ; +if (data->generation >GENERATION_NUMBER_MAX ) + data->generation =GENERATION_NUMBER_MAX ; } } }
commit-graph.h +10 -0
@@ -135,4 +135,14 @@ void free_commit_graph(struct commit_graph *);*/ void disable_commit_graph (struct repository *r); +struct commit_graph_data { +uint32_t graph_pos; +uint32_t generation; +}; + +/* +* Commits should be parsed before accessing generation, graph positions. +*/ +uint32_t commit_graph_generation (const struct commit *); +uint32_t commit_graph_position (const struct commit *);#endif
Signature · 1 line
2.27.0
PATCH v4 3/4 commit: move members graph_pos, generation to a slab
In reply to Abhishek KumarWe remove members graph_pos and generation from the struct commit. The default assignments in init_commit_node() are no longer valid, which is fine as the slab helpers return appropriate default values and the assignments are removed.
We will replace existing use of commit->generation and commit->graph_pos by commit_graph_data_slab helpers using `contrib/coccinelle/commit.cocci'.
--- alloc.c | 2 -- blame.c | 2 +- bloom.c | 6 ++-- commit-graph.c | 40 +++++++++++++------------- commit-reach.c | 50 ++++++++++++++++----------------- commit.c | 6 ++-- commit.h | 2 -- contrib/coccinelle/commit.cocci | 18 ++++++++++++ revision.c | 16 +++++------ 9 files changed, 78 insertions(+), 64 deletions(-)
alloc.c +0 -2
@@ -114,8 +114,6 @@ void init_commit_node(struct commit *c) { c->object.type =OBJ_COMMIT ; c->index =alloc_commit_index (); - c->graph_pos =COMMIT_NOT_FROM_GRAPH ; - c->generation =GENERATION_NUMBER_INFINITY ; }void *alloc_commit_node (struct repository *r)
blame.c +1 -1
@@ -1272,7 +1272,7 @@ static int maybe_changed_path(struct repository *r,if (!bd)return 1 ; -if (origin->commit->generation ==GENERATION_NUMBER_INFINITY ) +if (commit_graph_generation (origin->commit) ==GENERATION_NUMBER_INFINITY )return 1 ; filter =get_bloom_filter (r, origin->commit,0 );
bloom.c +3 -3
@@ -34,14 +34,14 @@ static int load_bloom_filter_from_graph(struct commit_graph *g, {uint32_t lex_pos, start_index, end_index; -while (c->graph_pos < g->num_commits_in_base) +while (commit_graph_position (c) < g->num_commits_in_base) g = g->base_graph;/* The commit graph commit 'c' lives in doesn't carry bloom filters. */ if (!g->chunk_bloom_indexes)return 0 ; - lex_pos = c->graph_pos - g->num_commits_in_base; + lex_pos =commit_graph_position (c) - g->num_commits_in_base; end_index =get_be32 (g->chunk_bloom_indexes +4 * lex_pos);
@@ -193,7 +193,7 @@ struct bloom_filter *get_bloom_filter(struct repository *r,if (!filter->data) {load_commit_graph_info (r, c); -if (c->graph_pos !=COMMIT_NOT_FROM_GRAPH && +if (commit_graph_position (c) !=COMMIT_NOT_FROM_GRAPH && r->objects->commit_graph->chunk_bloom_indexes) {if (load_bloom_filter_from_graph (r->objects->commit_graph, filter, c))return filter;
commit-graph.c +20 -20
@@ -145,9 +145,9 @@ static int commit_gen_cmp(const void *va, const void *vb)const struct commit *b = *(const struct commit **)vb;/* lower generation commits first */ -if (a->generation < b->generation) +if (commit_graph_generation (a) <commit_graph_generation (b))return -1 ; -else if (a->generation > b->generation) +else if (commit_graph_generation (a) >commit_graph_generation (b))return 1 ;/* use date as a heuristic when generations are equal */
@@ -722,7 +722,7 @@ static struct commit_list **insert_parent_or_die(struct repository *r, c =lookup_commit (r, &oid);if (!c)die (_ ("could not find commit %s" ),oid_to_hex (&oid)); - c->graph_pos = pos; +commit_graph_data_at (c)->graph_pos = pos;return &commit_list_insert (c, pptr)->next; }
@@ -736,8 +736,8 @@ static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, lex_index = pos - g->num_commits_in_base; commit_data = g->chunk_commit_data +GRAPH_DATA_WIDTH * lex_index; - item->graph_pos = pos; - item->generation =get_be32 (commit_data + g->hash_len +8 ) >>2 ; +commit_graph_data_at (item)->graph_pos = pos; +commit_graph_data_at (item)->generation =get_be32 (commit_data + g->hash_len +8 ) >>2 ; }static inline void set_commit_tree (struct commit *c,struct tree *t)
@@ -766,7 +766,7 @@ static int fill_commit_in_graph(struct repository *r,* Store the "full" position, but then use the * "local" position for the rest of the calculation. */ - item->graph_pos = pos; +commit_graph_data_at (item)->graph_pos = pos; lex_index = pos - g->num_commits_in_base; commit_data = g->chunk_commit_data + (g->hash_len +16 ) * lex_index;
@@ -779,7 +779,7 @@ static int fill_commit_in_graph(struct repository *r, date_low =get_be32 (commit_data + g->hash_len +12 ); item->date = (timestamp_t)((date_high <<32 ) | date_low); - item->generation =get_be32 (commit_data + g->hash_len +8 ) >>2 ; +commit_graph_data_at (item)->generation =get_be32 (commit_data + g->hash_len +8 ) >>2 ; pptr = &item->parents;
@@ -811,8 +811,8 @@ static int fill_commit_in_graph(struct repository *r,static int find_commit_in_graph (struct commit *item,struct commit_graph *g,uint32_t *pos) { -if (item->graph_pos !=COMMIT_NOT_FROM_GRAPH ) { - *pos = item->graph_pos; +if (commit_graph_position (item) !=COMMIT_NOT_FROM_GRAPH ) { + *pos =commit_graph_position (item);return 1 ; }else {struct commit_graph *cur_g = g;
@@ -868,11 +868,11 @@ static struct tree *load_tree_for_commit(struct repository *r,struct object_id oid;const unsigned char *commit_data; -while (c->graph_pos < g->num_commits_in_base) +while (commit_graph_position (c) < g->num_commits_in_base) g = g->base_graph; commit_data = g->chunk_commit_data + -GRAPH_DATA_WIDTH * (c->graph_pos - g->num_commits_in_base); +GRAPH_DATA_WIDTH * (commit_graph_position (c) - g->num_commits_in_base);hashcpy (oid.hash, commit_data);set_commit_tree (c,lookup_tree (r, &oid));
@@ -886,7 +886,7 @@ static struct tree *get_commit_tree_in_graph_one(struct repository *r, {if (c->maybe_tree)return c->maybe_tree; -if (c->graph_pos ==COMMIT_NOT_FROM_GRAPH ) +if (commit_graph_position (c) ==COMMIT_NOT_FROM_GRAPH )BUG ("get_commit_tree_in_graph_one called from non-commit-graph commit" );return load_tree_for_commit (r, g, (struct commit *)c);
@@ -1271,7 +1271,7 @@ static void close_reachable(struct write_commit_graph_context *ctx)continue ;if (ctx->split) {if ((!parse_commit (commit) && - commit->graph_pos ==COMMIT_NOT_FROM_GRAPH ) || +commit_graph_position (commit) ==COMMIT_NOT_FROM_GRAPH ) || flags ==COMMIT_GRAPH_SPLIT_REPLACE )add_missing_parents (ctx, commit); }else if (!parse_commit_no_graph (commit))
@@ -1516,7 +1516,7 @@ static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)if (ctx->split) {struct commit *c =lookup_commit (ctx->r, &ctx->oids.list[i]); -if (!c || c->graph_pos !=COMMIT_NOT_FROM_GRAPH ) +if (!c ||commit_graph_position (c) !=COMMIT_NOT_FROM_GRAPH )continue ; }
@@ -1550,7 +1550,7 @@ static void copy_oids_to_commits(struct write_commit_graph_context *ctx) ctx->commits.list[ctx->commits.nr] =lookup_commit (ctx->r, &ctx->oids.list[i]);if (ctx->split && flags !=COMMIT_GRAPH_SPLIT_REPLACE && - ctx->commits.list[ctx->commits.nr]->graph_pos !=COMMIT_NOT_FROM_GRAPH ) +commit_graph_position (ctx->commits.list[ctx->commits.nr]) !=COMMIT_NOT_FROM_GRAPH )continue ;if (ctx->split && flags ==COMMIT_GRAPH_SPLIT_REPLACE )
@@ -2337,8 +2337,8 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)oid_to_hex (&graph_parents->item->object.oid), oid_to_hex(&odb_parents->item->object.oid)); -if (graph_parents->item->generation > max_generation) - max_generation = graph_parents->item->generation; +if (commit_graph_generation (graph_parents->item) > max_generation) + max_generation =commit_graph_generation (graph_parents->item); graph_parents = graph_parents->next; odb_parents = odb_parents->next;
@@ -2348,7 +2348,7 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)graph_report (_ ("commit-graph parent list for commit %s terminates early" ),oid_to_hex (&cur_oid)); -if (!graph_commit->generation) { +if (!commit_graph_generation (graph_commit)) {if (generation_zero ==GENERATION_NUMBER_EXISTS )graph_report (_ ("commit-graph has generation number zero for commit %s, but non-zero elsewhere" ),oid_to_hex (&cur_oid));
@@ -2368,10 +2368,10 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)if (max_generation ==GENERATION_NUMBER_MAX ) max_generation--; -if (graph_commit->generation != max_generation +1 ) +if (commit_graph_generation (graph_commit) != max_generation +1 )graph_report (_ ("commit-graph generation for commit %s is %u != %u" ),oid_to_hex (&cur_oid), - graph_commit->generation, +commit_graph_generation (graph_commit), max_generation +1 );if (graph_commit->date != odb_commit->date)
commit-reach.c +25 -25
138 lines not on this page · raw message
commit.c +3 -3
@@ -339,7 +339,7 @@ struct tree *repo_get_commit_tree(struct repository *r,if (commit->maybe_tree || !commit->object.parsed)return commit->maybe_tree; -if (commit->graph_pos !=COMMIT_NOT_FROM_GRAPH ) +if (commit_graph_position (commit) !=COMMIT_NOT_FROM_GRAPH )return get_commit_tree_in_graph (r, commit);return NULL ;
@@ -731,9 +731,9 @@ int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, voidconst struct commit *a = a_, *b = b_;/* newer commits first */ -if (a->generation < b->generation) +if (commit_graph_generation (a) <commit_graph_generation (b))return 1 ; -else if (a->generation > b->generation) +else if (commit_graph_generation (a) >commit_graph_generation (b))return -1 ;/* use date as a heuristic when generations are equal */
commit.h +0 -2
@@ -36,8 +36,6 @@ struct commit {* or get_commit_tree_oid(). */ struct tree *maybe_tree; -uint32_t graph_pos; -uint32_t generation;unsigned int index; };
contrib/coccinelle/commit.cocci +18 -0
@@ -32,3 +32,21 @@ expression c; - c->maybe_tree + repo_get_commit_tree(specify_the_right_repo_here, c) ...>} + +@@ +struct commit *c; +expression E; +@@ +( +- c->generation = E; ++ commit_graph_data_at(c)->generation = E; +| +- c->graph_pos = E; ++ commit_graph_data_at(c)->graph_pos = E; +| +- c->generation ++ commit_graph_generation(c) +| +- c->graph_pos ++ commit_graph_position(c) +)
revision.c +8 -8
@@ -725,7 +725,7 @@ static int check_maybe_different_in_bloom_filter(struct rev_info *revs,if (!revs->repo->objects->commit_graph)return -1 ; -if (commit->generation ==GENERATION_NUMBER_INFINITY ) +if (commit_graph_generation (commit) ==GENERATION_NUMBER_INFINITY )return -1 ; filter =get_bloom_filter (revs->repo, commit,0 );
@@ -3320,7 +3320,7 @@ static void explore_to_depth(struct rev_info *revs,struct topo_walk_info *info = revs->topo_walk_info;struct commit *c;while ((c =prio_queue_peek (&info->explore_queue)) && - c->generation >= gen_cutoff) +commit_graph_generation (c) >= gen_cutoff)explore_walk_step (revs); }
@@ -3336,7 +3336,7 @@ static void indegree_walk_step(struct rev_info *revs)if (parse_commit_gently (c,1 ) <0 )return ; -explore_to_depth (revs, c->generation); +explore_to_depth (revs,commit_graph_generation (c));for (p = c->parents; p; p = p->next) {struct commit *parent = p->item;
@@ -3360,7 +3360,7 @@ static void compute_indegrees_to_depth(struct rev_info *revs,struct topo_walk_info *info = revs->topo_walk_info;struct commit *c;while ((c =prio_queue_peek (&info->indegree_queue)) && - c->generation >= gen_cutoff) +commit_graph_generation (c) >= gen_cutoff)indegree_walk_step (revs); }
@@ -3420,8 +3420,8 @@ static void init_topo_walk(struct rev_info *revs)test_flag_and_insert (&info->explore_queue, c,TOPO_WALK_EXPLORED );test_flag_and_insert (&info->indegree_queue, c,TOPO_WALK_INDEGREE ); -if (c->generation < info->min_generation) - info->min_generation = c->generation; +if (commit_graph_generation (c) < info->min_generation) + info->min_generation =commit_graph_generation (c); *(indegree_slab_at (&info->indegree, c)) =1 ;
@@ -3479,8 +3479,8 @@ static void expand_topo_walk(struct rev_info *revs, struct commit *commit)if (parse_commit_gently (parent,1 ) <0 )continue ; -if (parent->generation < info->min_generation) { - info->min_generation = parent->generation; +if (commit_graph_generation (parent) < info->min_generation) { + info->min_generation =commit_graph_generation (parent);compute_indegrees_to_depth (revs, info->min_generation); }
Signature · 1 line
2.27.0
PATCH v4 4/4 commit-graph: minimize commit_graph_data_slab access
In reply to Abhishek KumarIn an earlier patch, multiple struct acccesses to graph_pos and generation were auto-converted to multiple method calls.
Since the values are fixed and commit-slab access costly, we would be better off with storing the values as a local variable and reusing it.
--- bloom.c | 5 +++-- commit-graph.c | 40 ++++++++++++++++++++++------------ commit-reach.c | 59 ++++++++++++++++++++++++++++++++------------------ commit.c | 6 +++-- revision.c | 12 ++++++---- 5 files changed, 79 insertions(+), 43 deletions(-)
bloom.c +3 -2
@@ -33,15 +33,16 @@ static int load_bloom_filter_from_graph(struct commit_graph *g, struct commit *c) {uint32_t lex_pos, start_index, end_index; +uint32_t graph_pos =commit_graph_position (c); -while (commit_graph_position (c) < g->num_commits_in_base) +while (graph_pos < g->num_commits_in_base) g = g->base_graph;/* The commit graph commit 'c' lives in doesn't carry bloom filters. */ if (!g->chunk_bloom_indexes)return 0 ; - lex_pos =commit_graph_position (c) - g->num_commits_in_base; + lex_pos = graph_pos - g->num_commits_in_base; end_index =get_be32 (g->chunk_bloom_indexes +4 * lex_pos);
commit-graph.c +26 -14
@@ -144,10 +144,12 @@ static int commit_gen_cmp(const void *va, const void *vb)const struct commit *a = *(const struct commit **)va;const struct commit *b = *(const struct commit **)vb; +uint32_t generation_a =commit_graph_generation (a); +uint32_t generation_b =commit_graph_generation (b);/* lower generation commits first */ -if (commit_graph_generation (a) <commit_graph_generation (b)) +if (generation_a < generation_b)return -1 ; -else if (commit_graph_generation (a) >commit_graph_generation (b)) +else if (generation_a > generation_b)return 1 ;/* use date as a heuristic when generations are equal */
@@ -729,6 +731,7 @@ static struct commit_list **insert_parent_or_die(struct repository *r,static void fill_commit_graph_info (struct commit *item,struct commit_graph *g,uint32_t pos) {const unsigned char *commit_data; +struct commit_graph_data *graph_data;uint32_t lex_index;while (pos < g->num_commits_in_base)
@@ -736,8 +739,10 @@ static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, lex_index = pos - g->num_commits_in_base; commit_data = g->chunk_commit_data +GRAPH_DATA_WIDTH * lex_index; -commit_graph_data_at (item)->graph_pos = pos; -commit_graph_data_at (item)->generation =get_be32 (commit_data + g->hash_len +8 ) >>2 ; + + graph_data =commit_graph_data_at (item); + graph_data->graph_pos = pos; + graph_data->generation =get_be32 (commit_data + g->hash_len +8 ) >>2 ; }static inline void set_commit_tree (struct commit *c,struct tree *t)
@@ -753,6 +758,7 @@ static int fill_commit_in_graph(struct repository *r,uint32_t *parent_data_ptr;uint64_t date_low, date_high;struct commit_list **pptr; +struct commit_graph_data *graph_data;const unsigned char *commit_data;uint32_t lex_index;
@@ -766,7 +772,8 @@ static int fill_commit_in_graph(struct repository *r,* Store the "full" position, but then use the * "local" position for the rest of the calculation. */ -commit_graph_data_at (item)->graph_pos = pos; + graph_data =commit_graph_data_at (item); + graph_data->graph_pos = pos; lex_index = pos - g->num_commits_in_base; commit_data = g->chunk_commit_data + (g->hash_len +16 ) * lex_index;
@@ -779,7 +786,7 @@ static int fill_commit_in_graph(struct repository *r, date_low =get_be32 (commit_data + g->hash_len +12 ); item->date = (timestamp_t)((date_high <<32 ) | date_low); -commit_graph_data_at (item)->generation =get_be32 (commit_data + g->hash_len +8 ) >>2 ; + graph_data->generation =get_be32 (commit_data + g->hash_len +8 ) >>2 ; pptr = &item->parents;
@@ -811,8 +818,9 @@ static int fill_commit_in_graph(struct repository *r,static int find_commit_in_graph (struct commit *item,struct commit_graph *g,uint32_t *pos) { -if (commit_graph_position (item) !=COMMIT_NOT_FROM_GRAPH ) { - *pos =commit_graph_position (item); +uint32_t graph_pos =commit_graph_position (item); +if (graph_pos !=COMMIT_NOT_FROM_GRAPH ) { + *pos = graph_pos;return 1 ; }else {struct commit_graph *cur_g = g;
@@ -867,12 +875,13 @@ static struct tree *load_tree_for_commit(struct repository *r, {struct object_id oid;const unsigned char *commit_data; +uint32_t graph_pos =commit_graph_position (c); -while (commit_graph_position (c) < g->num_commits_in_base) +while (graph_pos < g->num_commits_in_base) g = g->base_graph; commit_data = g->chunk_commit_data + -GRAPH_DATA_WIDTH * (commit_graph_position (c) - g->num_commits_in_base); +GRAPH_DATA_WIDTH * (graph_pos - g->num_commits_in_base);hashcpy (oid.hash, commit_data);set_commit_tree (c,lookup_tree (r, &oid));
@@ -2299,6 +2308,7 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)struct commit *graph_commit, *odb_commit;struct commit_list *graph_parents, *odb_parents;uint32_t max_generation =0 ; +uint32_t generation;display_progress (progress, i +1 );hashcpy (cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
@@ -2337,8 +2347,9 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)oid_to_hex (&graph_parents->item->object.oid), oid_to_hex(&odb_parents->item->object.oid)); -if (commit_graph_generation (graph_parents->item) > max_generation) - max_generation =commit_graph_generation (graph_parents->item); + generation =commit_graph_generation (graph_parents->item); +if (generation > max_generation) + max_generation = generation; graph_parents = graph_parents->next; odb_parents = odb_parents->next;
@@ -2368,10 +2379,11 @@ int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)if (max_generation ==GENERATION_NUMBER_MAX ) max_generation--; -if (commit_graph_generation (graph_commit) != max_generation +1 ) + generation =commit_graph_generation (graph_commit); +if (generation != max_generation +1 )graph_report (_ ("commit-graph generation for commit %s is %u != %u" ),oid_to_hex (&cur_oid), -commit_graph_generation (graph_commit), + generation, max_generation +1 );if (graph_commit->date != odb_commit->date)
commit-reach.c +38 -21
140 lines not on this page · raw message
commit.c +4 -2
@@ -729,11 +729,13 @@ int compare_commits_by_author_date(const void *a_, const void *b_,int compare_commits_by_gen_then_commit_date (const void *a_,const void *b_,void *unused) {const struct commit *a = a_, *b = b_; +const uint32_t generation_a =commit_graph_generation (a), + generation_b =commit_graph_generation (b);/* newer commits first */ -if (commit_graph_generation (a) <commit_graph_generation (b)) +if (generation_a < generation_b)return 1 ; -else if (commit_graph_generation (a) >commit_graph_generation (b)) +else if (generation_a > generation_b)return -1 ;/* use date as a heuristic when generations are equal */
revision.c +8 -4
@@ -3413,6 +3413,7 @@ static void init_topo_walk(struct rev_info *revs) info->min_generation =GENERATION_NUMBER_INFINITY ;for (list = revs->commits; list; list = list->next) {struct commit *c = list->item; +uint32_t generation;if (parse_commit_gently (c,1 ))continue ;
@@ -3420,8 +3421,9 @@ static void init_topo_walk(struct rev_info *revs)test_flag_and_insert (&info->explore_queue, c,TOPO_WALK_EXPLORED );test_flag_and_insert (&info->indegree_queue, c,TOPO_WALK_INDEGREE ); -if (commit_graph_generation (c) < info->min_generation) - info->min_generation =commit_graph_generation (c); + generation =commit_graph_generation (c); +if (generation < info->min_generation) + info->min_generation = generation; *(indegree_slab_at (&info->indegree, c)) =1 ;
@@ -3472,6 +3474,7 @@ static void expand_topo_walk(struct rev_info *revs, struct commit *commit)for (p = commit->parents; p; p = p->next) {struct commit *parent = p->item;int *pi; +uint32_t generation;if (parent->object.flags &UNINTERESTING )continue ;
@@ -3479,8 +3482,9 @@ static void expand_topo_walk(struct rev_info *revs, struct commit *commit)if (parse_commit_gently (parent,1 ) <0 )continue ; -if (commit_graph_generation (parent) < info->min_generation) { - info->min_generation =commit_graph_generation (parent); + generation =commit_graph_generation (parent); +if (generation < info->min_generation) { + info->min_generation = generation;compute_indegrees_to_depth (revs, info->min_generation); }
Signature · 1 line
2.27.0
Quoted 7 lines · Abhishek Kumar
The struct commit is used in many contexts. However, members
generationandgraph_posare only used for commit graph related operations and otherwise waste memory.This wastage would have been more pronounced as we transition to generation nuber v2, which uses 64-bit generation number instead of current 32-bits.
Thanks, Szeder (CC'd) for the quality review in the previous versions. I manually built and tested all of the patches here and verified they passed all tests.
I think this series is in good shape.
Thanks,
-Stolee
Quoted 9 lines · Derrick Stolee
Quoted 7 lines · Abhishek Kumar
The struct commit is used in many contexts. However, members
generationandgraph_posare only used for commit graph related operations and otherwise waste memory.This wastage would have been more pronounced as we transition to generation nuber v2, which uses 64-bit generation number instead of current 32-bits.
Quoted 5 lines
Thanks, Szeder (CC'd) for the quality review in the previous versions. I manually built and tested all of the patches here and verified they passed all tests.
I think this series is in good shape.
Thank you to all who are involved in this topic. Looking good.