[PATCH v7] fetch-pack.c: use oidset to check existence of loose object
When fetching from a repository with large number of refs, because to
check existence of each refs in local repository to packed and loose
objects, 'git fetch' ends up doing a lot of lstat(2) to non-existing
loose form, which makes it slow.
Instead of making as many lstat(2) calls as the refs the remote side
advertised to see if these objects exist in the loose form, first
enumerate all the existing loose objects in hashmap beforehand and use
it to check existence of them if the number of refs is larger than the
number of loose objects.
With this patch, the number of lstat(2) calls in `git fetch` is reduced
from 411412 to 13794 for chromium repository, it has more than 48
remote refs.
I took time stat of `git fetch` when fetch-pack happens for chromium
repository 3 times on linux with SSD.
* with this patch
8.105s
8.309s
7.640s
avg: 8.018s
* master
12.287s
11.175s
12.227s
avg: 11.896s
On my MacBook Air which has slower lstat(2).
* with this patch
14.501s
* master
1m16.027s
`git fetch` on slow disk will be improved largely.
Signed-off-by: Takuto Ikuta <tik...@chromium.org>
---
cache.h | 2 ++
fetch-pack.c | 45 ++---
sha1_file.c | 3 +++
3 files changed, 47 insertions(+), 3 deletions(-)
diff --git a/cache.h b/cache.h
index d06932ed0..6a72f54d7 100644
--- a/cache.h
+++ b/cache.h
@@ -1773,6 +1773,8 @@ struct object_info {
#define OBJECT_INFO_SKIP_CACHED 4
/* Do not retry packed storage after checking packed and loose storage */
#define OBJECT_INFO_QUICK 8
+/* Do not check loose object */
+#define OBJECT_INFO_IGNORE_LOOSE 16
extern int sha1_object_info_extended(const unsigned char *, struct object_info
*, unsigned flags);
/*
diff --git a/fetch-pack.c b/fetch-pack.c
index d97461296..2ea358861 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c
@@ -711,6 +711,28 @@ static void mark_alternate_complete(struct object *obj)
mark_complete(>oid);
}
+struct loose_object_iter {
+ struct oidset *loose_object_set;
+ struct ref *refs;
+};
+
+/*
+ * If the number of refs is not larger than the number of loose objects,
+ * this function stops inserting.
+ */
+static int add_loose_objects_to_set(const struct object_id *oid,
+ const char *path,
+ void *data)
+{
+ struct loose_object_iter *iter = data;
+ oidset_insert(iter->loose_object_set, oid);
+ if (iter->refs == NULL)
+ return 1;
+
+ iter->refs = iter->refs->next;
+ return 0;
+}
+
static int everything_local(struct fetch_pack_args *args,
struct ref **refs,
struct ref **sought, int nr_sought)
@@ -719,16 +741,31 @@ static int everything_local(struct fetch_pack_args *args,
int retval;
int old_save_commit_buffer = save_commit_buffer;
timestamp_t cutoff = 0;
+ struct oidset loose_oid_set = OIDSET_INIT;
+ int use_oidset = 0;
+ struct loose_object_iter iter = {_oid_set, *refs};
+
+ /* Enumerate all loose objects or know refs are not so many. */
+ use_oidset = !for_each_loose_object(add_loose_objects_to_set,
+ , 0);
save_commit_buffer = 0;
for (ref = *refs; ref; ref = ref->next) {
struct object *o;
+ unsigned int flags = OBJECT_INFO_QUICK;
- if (!has_object_file_with_flags(>old_oid,
- OBJECT_INFO_QUICK))
- continue;
+ if (use_oidset &&
+ !oidset_contains(_oid_set, >old_oid)) {
+ /*
+* I know this does not exist in the loose form,
+* so check if it exists in a non-loose form.
+*/
+ flags |= OBJECT_INFO_IGNORE_LOOSE;
+ }
+ if (!has_object_file_with_flags(>old_oid, flags))
+ continue;
o = parse_object(>old_oid);
if (!o)
continue;
@@ -744,6 +781,8 @@ static int everything_local(struct fetch_pack_args *args,
}
}
+ oidset_clear(_oid_set);
+
if (!args->no_dependents) {
if (!args->deepen) {
for_each_ref(mark_complete_oid, NULL);
diff --git a/sha1_file.c b/sha1_file.c
index 1b94f39c4..c0a197947 100644
--- a/sha1_file.c
+++ b/sha1_file.c
@@ -1262,6 +1262,9 @@ int sha1_object_info_extended(const unsigned char *sha1,
struct object_info *oi,
if (find_pack_entry(real, ))
break;
+ if (flags & OBJECT_INFO_IGNORE_LOOSE)
+ return -1;
+
/* Most likely it's a loose object. */
if (!sha1_loose_object_info(real, oi, flags))
return 0;
--
2.16.2
Re: [PATCH v3] fetch-pack.c: use oidset to check existence of loose object
2018年3月14日(水) 2:53 Junio C Hamano <gits...@pobox.com>: > Takuto Ikuta <tik...@chromium.org> writes: > >> During fetch, everything_local() tries to mark common part by > >> walking the refs the other side advertised upon the first contact, > >> so it is correct that the number of checks is not reduced in a fetch > >> that does not fetch many refs, but the number of remote-tracking refs > >> you have has no effect, so I doubt such a rephrasing would make the > >> description more understandable. "When fetching from a repository > >> with large number of refs" is probably what you want to say, no? > > > > For refs existing in local repository, everything_local looks to be able to find > > corresponding object from packed or loose objects. And if it exists, > > I think, cost of lstat(2) is relatively smaller than other operations. > > But for remote refs, everything_local fails to find it from packed > > object (this check is fast) > > and it tries to find loose object by using lstat(2), and this fails and slow. > > My patch is to skip this lstat(2) to non-existing ref objects for repositories > > having large number of remote refs. > This still does not make sense to me, and I suspect that I am > misreading you. In what sense are you using the word "repository" > and "remote refs"? Yes, I think I did not understand the terms correctly. I meant "repository" for repository in remote servers, and "remote refs" for refs shown by `git ls-remote`. > Imagine this situation. I have a local repository A, I fetch from a > remote repository B but in my repository A, I do *not* use > remote-tracking refs to remember what the last values of refs at > repository B. Now when I try to fetch a single ref from B into A, > many refs B advertises point at objects A has never heard about, and > that triggers many lstat(2) that yields ENOENT that is slow. Your > patch is to optimize this so that we learn these objects do not > exist locally without running many lstat(2) to objects and helps > repositories (like my repository A) when fetching from a repository > with large number of refs (like the repository B). It does not > matter how many "remote refs" receiving repository (e.g. my A) has, > and it does not matter how many "remote refs" sending repository > (e.g. my B) has---whether it is refs/remotes/origin/foo > (i.e. "remote") or refs/heads/foo (i.e. "local"), a ref at B that > points at an object that is missing at A will cause the same > lstat(2) at A without your change. > That is why I think "When fetching from a repository with large > number of refs" is what you meant, not "fetching into a repository > with large number of remote refs" nor "fetching from a repository > with large number of remote refs". Thank you for explanation. And yes, that is exactly I want to do in my patch. Fixed description in v6. Thanks.
[PATCH v6] fetch-pack.c: use oidset to check existence of loose object
When fetching from a repository with large number of refs, because to
check existence of each refs in local repository to packed and loose
objects, 'git fetch' ends up doing a lot of lstat(2) to non-existing
loose form, which makes it slow.
Instead of making as many lstat(2) calls as the refs the remote side
advertised to see if these objects exist in the loose form, first
enumerate all the existing loose objects in hashmap beforehand and use
it to check existence of them if the number of refs is larger than the
number of loose objects.
With this patch, the number of lstat(2) calls in `git fetch` is reduced
from 411412 to 13794 for chromium repository, it has more than 48
remote refs.
I took time stat of `git fetch` when fetch-pack happens for chromium
repository 3 times on linux with SSD.
* with this patch
8.105s
8.309s
7.640s
avg: 8.018s
* master
12.287s
11.175s
12.227s
avg: 11.896s
On my MacBook Air which has slower lstat(2).
* with this patch
14.501s
* master
1m16.027s
`git fetch` on slow disk will be improved largely.
Signed-off-by: Takuto Ikuta <tik...@chromium.org>
---
cache.h | 2 ++
fetch-pack.c | 45 ++---
sha1_file.c | 3 +++
3 files changed, 47 insertions(+), 3 deletions(-)
diff --git a/cache.h b/cache.h
index d06932ed0..6a72f54d7 100644
--- a/cache.h
+++ b/cache.h
@@ -1773,6 +1773,8 @@ struct object_info {
#define OBJECT_INFO_SKIP_CACHED 4
/* Do not retry packed storage after checking packed and loose storage */
#define OBJECT_INFO_QUICK 8
+/* Do not check loose object */
+#define OBJECT_INFO_IGNORE_LOOSE 16
extern int sha1_object_info_extended(const unsigned char *, struct object_info
*, unsigned flags);
/*
diff --git a/fetch-pack.c b/fetch-pack.c
index d97461296..92b9bb4d9 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c
@@ -711,6 +711,28 @@ static void mark_alternate_complete(struct object *obj)
mark_complete(>oid);
}
+struct loose_object_iter {
+ struct oidset *loose_object_set;
+ struct ref *refs;
+};
+
+/*
+ * If the number of refs is not larger than the number of loose objects,
+ * this function stops inserting and returns false.
+ */
+static int add_loose_objects_to_set(const struct object_id *oid,
+ const char *path,
+ void *data)
+{
+ struct loose_object_iter *iter = data;
+ oidset_insert(iter->loose_object_set, oid);
+ if (iter->refs == NULL)
+ return 1;
+
+ iter->refs = iter->refs->next;
+ return 0;
+}
+
static int everything_local(struct fetch_pack_args *args,
struct ref **refs,
struct ref **sought, int nr_sought)
@@ -719,16 +741,31 @@ static int everything_local(struct fetch_pack_args *args,
int retval;
int old_save_commit_buffer = save_commit_buffer;
timestamp_t cutoff = 0;
+ struct oidset loose_oid_set = OIDSET_INIT;
+ int use_oidset = 0;
+ struct loose_object_iter iter = {_oid_set, *refs};
+
+ /* Enumerate all loose objects or know refs are not so many. */
+ use_oidset = !for_each_loose_object(add_loose_objects_to_set,
+ , 0);
save_commit_buffer = 0;
for (ref = *refs; ref; ref = ref->next) {
struct object *o;
+ unsigned int flags = OBJECT_INFO_QUICK;
- if (!has_object_file_with_flags(>old_oid,
- OBJECT_INFO_QUICK))
- continue;
+ if (use_oidset &&
+ !oidset_contains(_oid_set, >old_oid)) {
+ /*
+* I know this does not exist in the loose form,
+* so check if it exists in a non-loose form.
+*/
+ flags |= OBJECT_INFO_IGNORE_LOOSE;
+ }
+ if (!has_object_file_with_flags(>old_oid, flags))
+ continue;
o = parse_object(>old_oid);
if (!o)
continue;
@@ -744,6 +781,8 @@ static int everything_local(struct fetch_pack_args *args,
}
}
+ oidset_clear(_oid_set);
+
if (!args->no_dependents) {
if (!args->deepen) {
for_each_ref(mark_complete_oid, NULL);
diff --git a/sha1_file.c b/sha1_file.c
index 1b94f39c4..c0a197947 100644
--- a/sha1_file.c
+++ b/sha1_file.c
@@ -1262,6 +1262,9 @@ int sha1_object_info_extended(const unsigned char *sha1,
struct object_info *oi,
if (find_pack_entry(real, ))
break;
+ if (flags & OBJECT_INFO_IGNORE_LOOSE)
+ return -1;
+
/* Most likely it's a loose object. */
if (!sha1_loose_objec
Re: [PATCH v3] fetch-pack.c: use oidset to check existence of loose object
2018-03-10 3:00 GMT+09:00 Junio C Hamano <gits...@pobox.com>: > Takuto Ikuta <tik...@chromium.org> writes: > >> Yes, I just wanted to say 'git fetch' invokes fetch-pack. >> fetch-pack is skipped when running git fetch repeatedly while >> remote has no update by quickfetch. So I disabled it to see the >> performance of fetch-pack. In chromium repository, master branch >> is updated several times in an hour, so git fetch invokes fetch-pack >> in such frequency. > > I do understand that if you run "git fetch" against the same place > in quick succession three times, the first one may cost a lot (not > just that it has to do the everything_local() computation that you > care about, it also has to actually do the network transfer), while > the second and third ones that are done before anything new happens > on the other end will not involve everything_local() overhead thanks > to quickfetch. > > A "fetch" that is run against a remote that has nothing new, but > still triggers everything_local() only because quickfetch is > disabled, is an artificial case that has no relevance to the real > world, I suspect, because the quickfetch optimization is to solve > the "there is nothing to be done, still do_fetch_pack() spends so > much cycles only to realize that it does not have anything to do, > why?" issue. > I changed not to say "disabling quickfetch" in v5 description. Actually, I'd like to take several stats on linux because the perf difference is small and it can be affected by network and quickfetch. > Isn't running the "git fetch" command with the "--dry-run" option > many times in quick succession a lot closer to what you really want > to measure, I wonder? That way, your first fetch won't be touching > the state of the local side to affect your second and subsequent > fetches. > Looks no? Even when I use --dry-run, second run did not invoke fetch-pack. >>> In any case, do_fetch_pack() tries to see if all of the tip commits >>> we are going to fetch exist locally, so when you are trying a fetch >>> that grabs huge number of refs (by the way, it means that the first >>> sentence of the proposed log message is not quite true---it is "When >>> fetching a large number of refs", as it does not matter how many >>> refs _we_ have, no?), everything_local() ends up making repeated >>> calls to has_object_file_with_flags() to all of the refs. >> >> I fixed description by changing 'refs' to 'remote refs'. In my understanding, >> git tries to check existence of remote refs even if we won't fetch such refs. > > During fetch, everything_local() tries to mark common part by > walking the refs the other side advertised upon the first contact, > so it is correct that the number of checks is not reduced in a fetch > that does not fetch many refs, but the number of remote-tracking refs > you have has no effect, so I doubt such a rephrasing would make the > description more understandable. "When fetching from a repository > with large number of refs" is probably what you want to say, no? > For refs existing in local repository, everything_local looks to be able to find corresponding object from packed or loose objects. And if it exists, I think, cost of lstat(2) is relatively smaller than other operations. But for remote refs, everything_local fails to find it from packed object (this check is fast) and it tries to find loose object by using lstat(2), and this fails and slow. My patch is to skip this lstat(2) to non-existing ref objects for repositories having large number of remote refs. 2018-03-10 4:41 GMT+09:00 Junio C Hamano <gits...@pobox.com>: > Junio C Hamano <gits...@pobox.com> writes: > >>> Yes. But I think the default limit for the number of loose objects, 7000, >>> gives us small overhead when we do enumeration of all objects. >> >> Hmph, I didn't see the code that does the estimation of loose object >> count before starting to enumerate, though. > > Another thing the code could do to avoid negative consequences on > projects that look quite different from yours (e.g. the other side > does not have insane number of refs, but there are locally quite a > few loose objects) is to count how many entries are on *refs list > before we decide to enumerate all loose objects. When the refs list > is relatively shorter than the estimated number of loose objects > (you can actually do the estimation based on sampling, or just rely > on your assumed 7k), it may be a win _not_ to trigger the new code > you are adding to this codepath with this patch. I would imagine > that the simplest implementaion may just count > >
[PATCH v5] fetch-pack.c: use oidset to check existence of loose object
In repository having large number of remote refs, because to check
existence of each refs in local repository to packed and loose objects,
'git fetch' ends up doing a lot of lstat(2) to non-existing loose form,
which makes it slow.
Instead of making as many lstat(2) calls as the refs the remote side
advertised to see if these objects exist in the loose form, first
enumerate all the existing loose objects in hashmap beforehand and use
it to check existence of them if the number of refs is larger than the
number of loose objects.
With this patch, the number of lstat(2) calls in `git fetch` is reduced
from 411412 to 13794 for chromium repository, it has more than 48
remote refs.
I took time stat of `git fetch` when fetch-pack happens for chromium
repository 3 times on linux with SSD.
* with this patch
8.105s
8.309s
7.640s
avg: 8.018s
* master
12.287s
11.175s
12.227s
avg: 11.896s
On my MacBook Air which has slower lstat(2).
* with this patch
14.501s
* master
1m16.027s
`git fetch` on slow disk will be improved largely.
Signed-off-by: Takuto Ikuta <tik...@chromium.org>
---
cache.h | 2 ++
fetch-pack.c | 45 ++---
sha1_file.c | 3 +++
3 files changed, 47 insertions(+), 3 deletions(-)
diff --git a/cache.h b/cache.h
index d06932ed0..6a72f54d7 100644
--- a/cache.h
+++ b/cache.h
@@ -1773,6 +1773,8 @@ struct object_info {
#define OBJECT_INFO_SKIP_CACHED 4
/* Do not retry packed storage after checking packed and loose storage */
#define OBJECT_INFO_QUICK 8
+/* Do not check loose object */
+#define OBJECT_INFO_IGNORE_LOOSE 16
extern int sha1_object_info_extended(const unsigned char *, struct object_info
*, unsigned flags);
/*
diff --git a/fetch-pack.c b/fetch-pack.c
index d97461296..92b9bb4d9 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c
@@ -711,6 +711,28 @@ static void mark_alternate_complete(struct object *obj)
mark_complete(>oid);
}
+struct loose_object_iter {
+ struct oidset *loose_object_set;
+ struct ref *refs;
+};
+
+/*
+ * If the number of refs is not larger than the number of loose objects,
+ * this function stops inserting and returns false.
+ */
+static int add_loose_objects_to_set(const struct object_id *oid,
+ const char *path,
+ void *data)
+{
+ struct loose_object_iter *iter = data;
+ oidset_insert(iter->loose_object_set, oid);
+ if (iter->refs == NULL)
+ return 1;
+
+ iter->refs = iter->refs->next;
+ return 0;
+}
+
static int everything_local(struct fetch_pack_args *args,
struct ref **refs,
struct ref **sought, int nr_sought)
@@ -719,16 +741,31 @@ static int everything_local(struct fetch_pack_args *args,
int retval;
int old_save_commit_buffer = save_commit_buffer;
timestamp_t cutoff = 0;
+ struct oidset loose_oid_set = OIDSET_INIT;
+ int use_oidset = 0;
+ struct loose_object_iter iter = {_oid_set, *refs};
+
+ /* Enumerate all loose objects or know refs are not so many. */
+ use_oidset = !for_each_loose_object(add_loose_objects_to_set,
+ , 0);
save_commit_buffer = 0;
for (ref = *refs; ref; ref = ref->next) {
struct object *o;
+ unsigned int flags = OBJECT_INFO_QUICK;
- if (!has_object_file_with_flags(>old_oid,
- OBJECT_INFO_QUICK))
- continue;
+ if (use_oidset &&
+ !oidset_contains(_oid_set, >old_oid)) {
+ /*
+* I know this does not exist in the loose form,
+* so check if it exists in a non-loose form.
+*/
+ flags |= OBJECT_INFO_IGNORE_LOOSE;
+ }
+ if (!has_object_file_with_flags(>old_oid, flags))
+ continue;
o = parse_object(>old_oid);
if (!o)
continue;
@@ -744,6 +781,8 @@ static int everything_local(struct fetch_pack_args *args,
}
}
+ oidset_clear(_oid_set);
+
if (!args->no_dependents) {
if (!args->deepen) {
for_each_ref(mark_complete_oid, NULL);
diff --git a/sha1_file.c b/sha1_file.c
index 1b94f39c4..c0a197947 100644
--- a/sha1_file.c
+++ b/sha1_file.c
@@ -1262,6 +1262,9 @@ int sha1_object_info_extended(const unsigned char *sha1,
struct object_info *oi,
if (find_pack_entry(real, ))
break;
+ if (flags & OBJECT_INFO_IGNORE_LOOSE)
+ return -1;
+
/* Most likely it's a loose object. */
if (!sha1_loose_object_info(real, oi, flags))
return 0;
--
2.16.2
[PATCH v4] fetch-pack.c: use oidset to check existence of loose object
In repository having large number of remote refs, because to check
existence of each refs in local repository to packed and loose objects,
'git fetch' ends up doing a lot of lstat(2) to non-existing loose form,
which makes it slow.
Instead of making as many lstat(2) calls as the refs the remote side
advertised to see if these objects exist in the loose form, first
enumerate all the existing loose objects in hashmap beforehand and use
it to check existence of them if the number of refs is larger than the
number of loose objects.
With this patch, the number of lstat(2) calls in `git fetch` is reduced
from 411412 to 13794 for chromium repository, it has more than 48
remote refs.
I took time stat of `git fetch` disabling quickfetch, so that fetch-pack
runs, for chromium repository 3 time on linux with SSD.
* with this patch
8.105s
8.309s
7.640s
avg: 8.018s
* master
12.287s
11.175s
12.227s
avg: 11.896s
On my MacBook Air which has slower lstat(2).
* with this patch
14.501s
* master
1m16.027s
`git fetch` on slow disk will be improved largely.
Signed-off-by: Takuto Ikuta <tik...@chromium.org>
---
cache.h | 2 ++
fetch-pack.c | 45 ++---
sha1_file.c | 3 +++
3 files changed, 47 insertions(+), 3 deletions(-)
diff --git a/cache.h b/cache.h
index d06932ed0..6a72f54d7 100644
--- a/cache.h
+++ b/cache.h
@@ -1773,6 +1773,8 @@ struct object_info {
#define OBJECT_INFO_SKIP_CACHED 4
/* Do not retry packed storage after checking packed and loose storage */
#define OBJECT_INFO_QUICK 8
+/* Do not check loose object */
+#define OBJECT_INFO_IGNORE_LOOSE 16
extern int sha1_object_info_extended(const unsigned char *, struct object_info
*, unsigned flags);
/*
diff --git a/fetch-pack.c b/fetch-pack.c
index d97461296..92b9bb4d9 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c
@@ -711,6 +711,28 @@ static void mark_alternate_complete(struct object *obj)
mark_complete(>oid);
}
+struct loose_object_iter {
+ struct oidset *loose_object_set;
+ struct ref *refs;
+};
+
+/*
+ * If the number of refs is not larger than the number of loose objects,
+ * this function stops inserting and returns false.
+ */
+static int add_loose_objects_to_set(const struct object_id *oid,
+ const char *path,
+ void *data)
+{
+ struct loose_object_iter *iter = data;
+ oidset_insert(iter->loose_object_set, oid);
+ if (iter->refs == NULL)
+ return 1;
+
+ iter->refs = iter->refs->next;
+ return 0;
+}
+
static int everything_local(struct fetch_pack_args *args,
struct ref **refs,
struct ref **sought, int nr_sought)
@@ -719,16 +741,31 @@ static int everything_local(struct fetch_pack_args *args,
int retval;
int old_save_commit_buffer = save_commit_buffer;
timestamp_t cutoff = 0;
+ struct oidset loose_oid_set = OIDSET_INIT;
+ int use_oidset = 0;
+ struct loose_object_iter iter = {_oid_set, *refs};
+
+ /* Enumerate all loose objects or know refs are not so many. */
+ use_oidset = !for_each_loose_object(add_loose_objects_to_set,
+ , 0);
save_commit_buffer = 0;
for (ref = *refs; ref; ref = ref->next) {
struct object *o;
+ unsigned int flags = OBJECT_INFO_QUICK;
- if (!has_object_file_with_flags(>old_oid,
- OBJECT_INFO_QUICK))
- continue;
+ if (use_oidset &&
+ !oidset_contains(_oid_set, >old_oid)) {
+ /*
+* I know this does not exist in the loose form,
+* so check if it exists in a non-loose form.
+*/
+ flags |= OBJECT_INFO_IGNORE_LOOSE;
+ }
+ if (!has_object_file_with_flags(>old_oid, flags))
+ continue;
o = parse_object(>old_oid);
if (!o)
continue;
@@ -744,6 +781,8 @@ static int everything_local(struct fetch_pack_args *args,
}
}
+ oidset_clear(_oid_set);
+
if (!args->no_dependents) {
if (!args->deepen) {
for_each_ref(mark_complete_oid, NULL);
diff --git a/sha1_file.c b/sha1_file.c
index 1b94f39c4..c0a197947 100644
--- a/sha1_file.c
+++ b/sha1_file.c
@@ -1262,6 +1262,9 @@ int sha1_object_info_extended(const unsigned char *sha1,
struct object_info *oi,
if (find_pack_entry(real, ))
break;
+ if (flags & OBJECT_INFO_IGNORE_LOOSE)
+ return -1;
+
/* Most likely it's a loose object. */
if (!sha1_loose_objec
Re: [PATCH] fetch-pack.c: use oidset to check existence of loose object
2018-03-09 3:42 GMT+09:00 Junio C Hamano <gits...@pobox.com>:
> Takuto Ikuta <tik...@chromium.org> writes:
>> This patch stores existing loose objects in hashmap beforehand and use
>> it to check existence instead of using lstat.
>>
>> With this patch, the number of lstat calls in `git fetch` is reduced
>> from 411412 to 13794 for chromium repository.
>>
>> I took time stat of `git fetch` disabling quickfetch for chromium
>> repository 3 time on linux with SSD.
>
> Now you drop a clue that would help to fill in the blanks above, but
> I am not sure what the significance of your having to disable
> quickfetch in order to take measurements---it makes it sound as if
> it is an articificial problem that does not exist in real life
> (i.e. when quickfetch is not disabled), but I am getting the feeling
> that it is not what you wanted to say here.
>
Yes, I just wanted to say 'git fetch' invokes fetch-pack.
fetch-pack is skipped when running git fetch repeatedly while
remote has no update by quickfetch. So I disabled it to see the
performance of fetch-pack. In chromium repository, master branch
is updated several times in an hour, so git fetch invokes fetch-pack
in such frequency.
> In any case, do_fetch_pack() tries to see if all of the tip commits
> we are going to fetch exist locally, so when you are trying a fetch
> that grabs huge number of refs (by the way, it means that the first
> sentence of the proposed log message is not quite true---it is "When
> fetching a large number of refs", as it does not matter how many
> refs _we_ have, no?), everything_local() ends up making repeated
> calls to has_object_file_with_flags() to all of the refs.
>
I fixed description by changing 'refs' to 'remote refs'. In my understanding,
git tries to check existence of remote refs even if we won't fetch such refs.
> I like the idea---this turns "for each of these many things, check
> if it exists with lstat(2)" into "enumerate what exists with
> lstat(2), and then use that for the existence test"; if you need to
> try N objects for existence, and you only have M objects loose where
> N is vastly larger than M, it will be a huge win. If you have very
> many loose objects and checking only a handful of objects for
> existence check, you would lose big, though, no?
>
Yes. But I think the default limit for the number of loose objects, 7000,
gives us small overhead when we do enumeration of all objects.
>> diff --git a/fetch-pack.c b/fetch-pack.c
>> index d97461296..1658487f7 100644
>> --- a/fetch-pack.c
>> +++ b/fetch-pack.c
>> @@ -711,6 +711,15 @@ static void mark_alternate_complete(struct object *obj)
>> mark_complete(>oid);
>> }
>>
>> +static int add_loose_objects_to_set(const struct object_id *oid,
>> + const char *path,
>> + void *data)
>> +{
>> + struct oidset* set = (struct oidset*)(data);
>
> Style: in our codebase, asterisk does not stick to the type.
>
> struct oidset *set = (struct oidset *)(data);
>
>> @@ -719,16 +728,21 @@ static int everything_local(struct fetch_pack_args
>> *args,
>> int retval;
>> int old_save_commit_buffer = save_commit_buffer;
>> timestamp_t cutoff = 0;
>> + struct oidset loose_oid_set = OIDSET_INIT;
>> +
>> + for_each_loose_object(add_loose_objects_to_set, _oid_set, 0);
>
> OK, so this is the "enumerate all loose objects" phase.
>
>> save_commit_buffer = 0;
>>
>> for (ref = *refs; ref; ref = ref->next) {
>> struct object *o;
>> + unsigned int flag = OBJECT_INFO_QUICK;
>
> Hmm, OBJECT_INFO_QUICK optimization was added in dfdd4afc
> ("sha1_file: teach sha1_object_info_extended more flags",
> 2017-06-21), but since 8b4c0103 ("sha1_file: support lazily fetching
> missing objects", 2017-12-08) it appears that passing
> OBJECT_INFO_QUICK down the codepath does not do anything
> interesting. Jonathan (cc'ed), are all remaining hits from "git
> grep OBJECT_INFO_QUICK" all dead no-ops these days?
>
Yes the flag is no-op now, but let me untouched the flag in this patch.
>> diff --git a/sha1_file.c b/sha1_file.c
>> index 1b94f39c4..c903cbcec 100644
>> --- a/sha1_file.c
>> +++ b/sha1_file.c
>> @@ -1262,6 +1262,9 @@ int sha1_object_info_extended(const unsigned char
>> *sha1, struct object_info *oi,
>> if (find_pack_entry(real, ))
>> break;
>>
>> + if (flags & OBJECT_INFO_SKIP_LOOSE)
>> + return -1;
>> +
>
> I cannot quite convince myself that this is done at the right layer;
> it smells to be at a bit too low a layer. This change makes sense
> only to a caller that is interested in the existence test. If the
> flag is named after what it does, i.e. "ignore loose object", then
> it does sort-of make sense, though.
>
Couldn't come up with good alternative for this, I followed your
flag name suggestion in patch v3.
Takuto
Re: [PATCH] fetch-pack.c: use oidset to check existence of loose object
2018-03-09 2:19 GMT+09:00 René Scharfe <l@web.de>:
> Am 08.03.2018 um 13:06 schrieb Takuto Ikuta:
>> +static int add_loose_objects_to_set(const struct object_id *oid,
>> + const char *path,
>> + void *data)
>> +{
>> + struct oidset* set = (struct oidset*)(data);
>
> This cast is not needed (unlike in C++). And the asterisk should be stuck
> to the variable, not the type (see Documentation/CodingGuidelines).
>
>> + oidset_insert(set, oid);
>
> In fact, you could just put "data" in here instead of "set" (without a
> cast), with no loss in readability or safety.
>
Thank you for review, changed to use data directly in v2.
>> + return 0;
>> +}
>> +
>> static int everything_local(struct fetch_pack_args *args,
>> struct ref **refs,
>> struct ref **sought, int nr_sought)
>> @@ -719,16 +728,21 @@ static int everything_local(struct fetch_pack_args
>> *args,
>> int retval;
>> int old_save_commit_buffer = save_commit_buffer;
>> timestamp_t cutoff = 0;
>> + struct oidset loose_oid_set = OIDSET_INIT;
>> +
>> + for_each_loose_object(add_loose_objects_to_set, _oid_set, 0);
>>
>> save_commit_buffer = 0;
>>
>> for (ref = *refs; ref; ref = ref->next) {
>> struct object *o;
>> + unsigned int flag = OBJECT_INFO_QUICK;
>>
>> - if (!has_object_file_with_flags(>old_oid,
>> - OBJECT_INFO_QUICK))
>> - continue;
>> + if (!oidset_contains(_oid_set, >old_oid))
>> + flag |= OBJECT_INFO_SKIP_LOOSE;
>>
>> + if (!has_object_file_with_flags(>old_oid, flag))
>> + continue;
>> o = parse_object(>old_oid);
>> if (!o)
>> continue;
>> @@ -744,6 +758,8 @@ static int everything_local(struct fetch_pack_args *args,
>> }
>> }
>>
>> + oidset_clear(_oid_set);
>> +
>
> This part looks fine to me. (Except perhaps call the variable "flags"
> because you sometimes have two?)
>
Changed flag names.
> Why not include packed objects as well? Probably because packs have
> indexes which can queried quickly to determine object existence, and
> because there are only few loose objects in typical repositories,
> right?
>
Correct. In my target repository, chromium, fetch-pack's slowness
comes from many lstat
to non-existing loose objects for remote refs. I focus on to remove such lstat.
> A similar cache was introduced by cc817ca3ef (sha1_name: cache
> readdir(3) results in find_short_object_filename()) to speed up
> finding unambiguous shorts object hashes. I wonder if it could be
> used here as well, but I don't see an easy way.
>
>> if (!args->no_dependents) {
>> if (!args->deepen) {
>> for_each_ref(mark_complete_oid, NULL);
>> diff --git a/sha1_file.c b/sha1_file.c
>> index 1b94f39c4..c903cbcec 100644
>> --- a/sha1_file.c
>> +++ b/sha1_file.c
>> @@ -1262,6 +1262,9 @@ int sha1_object_info_extended(const unsigned char
>> *sha1, struct object_info *oi,
>> if (find_pack_entry(real, ))
>> break;
>>
>> + if (flags & OBJECT_INFO_SKIP_LOOSE)
>> + return -1;
>> +
>> /* Most likely it's a loose object. */
>> if (!sha1_loose_object_info(real, oi, flags))
>> return 0;
>>
>
> This early return doesn't just skip checking loose objects. It
> also skips reloading packs and fetching missing objects for
> partial clones. That may not be a problem for fetch-pack, but
> it means the flag has a misleading name. Do you get the same
> performance improvement if you make it only skip that
> sha1_loose_object_info() call?
>
I changed the flag name following Junio's suggestion. If we skip
sha1_loose_object_info call,
reprepare_packed_git(...) runs for every non-existing refs and git
fetch becomes slower.
Takuto
[PATCH v3] fetch-pack.c: use oidset to check existence of loose object
In repository having large number of remote refs, because to check
existence of each refs in local repository, 'git fetch' ends up doing a
lot of lstat(2) calls to see if it exists in loose form, which makes it
slow.
This patch enumerates loose objects in hashmap beforehand and uses it to
check existence instead of using lstat(2) to improve performance of
fetch-pack for repositories having large number of remote refs compared
to the number of loose objects.
With this patch, the number of lstat(2) calls in `git fetch` is reduced
from 411412 to 13794 for chromium repository, it has more than 48
remote refs.
I took time stat of `git fetch` disabling quickfetch, so that fetch-pack
runs, for chromium repository 3 time on linux with SSD.
* with this patch
8.105s
8.309s
7.640s
avg: 8.018s
* master
12.287s
11.175s
12.227s
avg: 11.896s
On my MacBook Air which has slower lstat(2).
* with this patch
14.501s
* master
1m16.027s
`git fetch` on slow disk will be improved largely.
Signed-off-by: Takuto Ikuta <tik...@chromium.org>
---
cache.h | 2 ++
fetch-pack.c | 26 +++---
sha1_file.c | 3 +++
3 files changed, 28 insertions(+), 3 deletions(-)
diff --git a/cache.h b/cache.h
index d06932ed0..6a72f54d7 100644
--- a/cache.h
+++ b/cache.h
@@ -1773,6 +1773,8 @@ struct object_info {
#define OBJECT_INFO_SKIP_CACHED 4
/* Do not retry packed storage after checking packed and loose storage */
#define OBJECT_INFO_QUICK 8
+/* Do not check loose object */
+#define OBJECT_INFO_IGNORE_LOOSE 16
extern int sha1_object_info_extended(const unsigned char *, struct object_info
*, unsigned flags);
/*
diff --git a/fetch-pack.c b/fetch-pack.c
index d97461296..aed4aa213 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c
@@ -711,6 +711,14 @@ static void mark_alternate_complete(struct object *obj)
mark_complete(>oid);
}
+static int add_loose_objects_to_set(const struct object_id *oid,
+ const char *path,
+ void *data)
+{
+ oidset_insert(data, oid);
+ return 0;
+}
+
static int everything_local(struct fetch_pack_args *args,
struct ref **refs,
struct ref **sought, int nr_sought)
@@ -719,16 +727,26 @@ static int everything_local(struct fetch_pack_args *args,
int retval;
int old_save_commit_buffer = save_commit_buffer;
timestamp_t cutoff = 0;
+ struct oidset loose_oid_set = OIDSET_INIT;
+
+ /* Enumerate all loose objects. */
+ for_each_loose_object(add_loose_objects_to_set, _oid_set, 0);
save_commit_buffer = 0;
for (ref = *refs; ref; ref = ref->next) {
struct object *o;
+ unsigned int flags = OBJECT_INFO_QUICK;
- if (!has_object_file_with_flags(>old_oid,
- OBJECT_INFO_QUICK))
- continue;
+ if (!oidset_contains(_oid_set, >old_oid)) {
+ /* I know this does not exist in the loose form,
+* so check if it exists in a non-loose form.
+*/
+ flags |= OBJECT_INFO_IGNORE_LOOSE;
+ }
+ if (!has_object_file_with_flags(>old_oid, flags))
+ continue;
o = parse_object(>old_oid);
if (!o)
continue;
@@ -744,6 +762,8 @@ static int everything_local(struct fetch_pack_args *args,
}
}
+ oidset_clear(_oid_set);
+
if (!args->no_dependents) {
if (!args->deepen) {
for_each_ref(mark_complete_oid, NULL);
diff --git a/sha1_file.c b/sha1_file.c
index 1b94f39c4..c0a197947 100644
--- a/sha1_file.c
+++ b/sha1_file.c
@@ -1262,6 +1262,9 @@ int sha1_object_info_extended(const unsigned char *sha1,
struct object_info *oi,
if (find_pack_entry(real, ))
break;
+ if (flags & OBJECT_INFO_IGNORE_LOOSE)
+ return -1;
+
/* Most likely it's a loose object. */
if (!sha1_loose_object_info(real, oi, flags))
return 0;
--
2.16.2
[PATCH v2 1/1] fetch-pack.c: use oidset to check existence of loose object
In repository having large number of remote refs, because to check
existence of each refs in local repository, 'git fetch' ends up doing a
lot of lstat(2) calls to see if it exists in loose form, which makes it
slow.
This patch enumerates loose objects in hashmap beforehand and uses it to
check existence instead of using lstat(2) to improve performance of
fetch-pack for repositories having large number of remote refs compared
to the number of loose objects.
With this patch, the number of lstat(2) calls in `git fetch` is reduced
from 411412 to 13794 for chromium repository, it has more than 48
remote refs.
I took time stat of `git fetch` disabling quickfetch, so that fetch-pack
runs, for chromium repository 3 time on linux with SSD.
* with this patch
8.105s
8.309s
7.640s
avg: 8.018s
* master
12.287s
11.175s
12.227s
avg: 11.896s
On my MacBook Air which has slower lstat(2).
* with this patch
14.501s
* master
1m16.027s
`git fetch` on slow disk will be improved largely.
Signed-off-by: Takuto Ikuta <tik...@chromium.org>
---
cache.h | 2 ++
fetch-pack.c | 26 +++---
sha1_file.c | 3 +++
3 files changed, 28 insertions(+), 3 deletions(-)
diff --git a/cache.h b/cache.h
index d06932ed0..6a72f54d7 100644
--- a/cache.h
+++ b/cache.h
@@ -1773,6 +1773,8 @@ struct object_info {
#define OBJECT_INFO_SKIP_CACHED 4
/* Do not retry packed storage after checking packed and loose storage */
#define OBJECT_INFO_QUICK 8
+/* Do not check loose object */
+#define OBJECT_INFO_IGNORE_LOOSE 16
extern int sha1_object_info_extended(const unsigned char *, struct object_info
*, unsigned flags);
/*
diff --git a/fetch-pack.c b/fetch-pack.c
index d97461296..ef8b93424 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c
@@ -711,6 +711,14 @@ static void mark_alternate_complete(struct object *obj)
mark_complete(>oid);
}
+static int add_loose_objects_to_set(const struct object_id *oid,
+ const char *path,
+ void *data)
+{
+ oidset_insert(data, oid);
+ return 0;
+}
+
static int everything_local(struct fetch_pack_args *args,
struct ref **refs,
struct ref **sought, int nr_sought)
@@ -719,16 +727,26 @@ static int everything_local(struct fetch_pack_args *args,
int retval;
int old_save_commit_buffer = save_commit_buffer;
timestamp_t cutoff = 0;
+ struct oidset loose_oid_set = OIDSET_INIT;
+
+ /* Enumerate all loose objects. */
+ for_each_loose_object(add_loose_objects_to_set, _oid_set, 0);
save_commit_buffer = 0;
for (ref = *refs; ref; ref = ref->next) {
struct object *o;
+ unsigned int flag = OBJECT_INFO_QUICK;
- if (!has_object_file_with_flags(>old_oid,
- OBJECT_INFO_QUICK))
- continue;
+ if (!oidset_contains(_oid_set, >old_oid)) {
+ /* I know this does not exist in the loose form,
+* so check if it exists in a non-loose form.
+*/
+ flag |= OBJECT_INFO_IGNORE_LOOSE;
+ }
+ if (!has_object_file_with_flags(>old_oid, flag))
+ continue;
o = parse_object(>old_oid);
if (!o)
continue;
@@ -744,6 +762,8 @@ static int everything_local(struct fetch_pack_args *args,
}
}
+ oidset_clear(_oid_set);
+
if (!args->no_dependents) {
if (!args->deepen) {
for_each_ref(mark_complete_oid, NULL);
diff --git a/sha1_file.c b/sha1_file.c
index 1b94f39c4..c0a197947 100644
--- a/sha1_file.c
+++ b/sha1_file.c
@@ -1262,6 +1262,9 @@ int sha1_object_info_extended(const unsigned char *sha1,
struct object_info *oi,
if (find_pack_entry(real, ))
break;
+ if (flags & OBJECT_INFO_IGNORE_LOOSE)
+ return -1;
+
/* Most likely it's a loose object. */
if (!sha1_loose_object_info(real, oi, flags))
return 0;
--
2.16.2
[PATCH v2 0/1] fetch-pack.c: use oidset to check existence of loose object
I removed unnecessary type cast, added comment in everthing_local(...), changed flag name and updated description. Takuto Ikuta (1): fetch-pack.c: use oidset to check existence of loose object cache.h | 2 ++ fetch-pack.c | 26 +++--- sha1_file.c | 3 +++ 3 files changed, 28 insertions(+), 3 deletions(-) -- 2.16.2
[PATCH] fetch-pack.c: use oidset to check existence of loose object
In repository having large number of refs, lstat for non-existing loose
objects makes `git fetch` slow.
This patch stores existing loose objects in hashmap beforehand and use
it to check existence instead of using lstat.
With this patch, the number of lstat calls in `git fetch` is reduced
from 411412 to 13794 for chromium repository.
I took time stat of `git fetch` disabling quickfetch for chromium
repository 3 time on linux with SSD.
* with this patch
8.105s
8.309s
7.640s
avg: 8.018s
* master
12.287s
11.175s
12.227s
avg: 11.896s
On my MacBook Air which has slower lstat.
* with this patch
14.501s
* master
1m16.027s
`git fetch` on slow disk will be improved largely.
Signed-off-by: Takuto Ikuta <tik...@chromium.org>
---
cache.h | 2 ++
fetch-pack.c | 22 +++---
sha1_file.c | 3 +++
3 files changed, 24 insertions(+), 3 deletions(-)
diff --git a/cache.h b/cache.h
index d06932ed0..db38db40e 100644
--- a/cache.h
+++ b/cache.h
@@ -1773,6 +1773,8 @@ struct object_info {
#define OBJECT_INFO_SKIP_CACHED 4
/* Do not retry packed storage after checking packed and loose storage */
#define OBJECT_INFO_QUICK 8
+/* Do not check loose object */
+#define OBJECT_INFO_SKIP_LOOSE 16
extern int sha1_object_info_extended(const unsigned char *, struct object_info
*, unsigned flags);
/*
diff --git a/fetch-pack.c b/fetch-pack.c
index d97461296..1658487f7 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c
@@ -711,6 +711,15 @@ static void mark_alternate_complete(struct object *obj)
mark_complete(>oid);
}
+static int add_loose_objects_to_set(const struct object_id *oid,
+ const char *path,
+ void *data)
+{
+ struct oidset* set = (struct oidset*)(data);
+ oidset_insert(set, oid);
+ return 0;
+}
+
static int everything_local(struct fetch_pack_args *args,
struct ref **refs,
struct ref **sought, int nr_sought)
@@ -719,16 +728,21 @@ static int everything_local(struct fetch_pack_args *args,
int retval;
int old_save_commit_buffer = save_commit_buffer;
timestamp_t cutoff = 0;
+ struct oidset loose_oid_set = OIDSET_INIT;
+
+ for_each_loose_object(add_loose_objects_to_set, _oid_set, 0);
save_commit_buffer = 0;
for (ref = *refs; ref; ref = ref->next) {
struct object *o;
+ unsigned int flag = OBJECT_INFO_QUICK;
- if (!has_object_file_with_flags(>old_oid,
- OBJECT_INFO_QUICK))
- continue;
+ if (!oidset_contains(_oid_set, >old_oid))
+ flag |= OBJECT_INFO_SKIP_LOOSE;
+ if (!has_object_file_with_flags(>old_oid, flag))
+ continue;
o = parse_object(>old_oid);
if (!o)
continue;
@@ -744,6 +758,8 @@ static int everything_local(struct fetch_pack_args *args,
}
}
+ oidset_clear(_oid_set);
+
if (!args->no_dependents) {
if (!args->deepen) {
for_each_ref(mark_complete_oid, NULL);
diff --git a/sha1_file.c b/sha1_file.c
index 1b94f39c4..c903cbcec 100644
--- a/sha1_file.c
+++ b/sha1_file.c
@@ -1262,6 +1262,9 @@ int sha1_object_info_extended(const unsigned char *sha1,
struct object_info *oi,
if (find_pack_entry(real, ))
break;
+ if (flags & OBJECT_INFO_SKIP_LOOSE)
+ return -1;
+
/* Most likely it's a loose object. */
if (!sha1_loose_object_info(real, oi, flags))
return 0;
--
2.16.2
Re: [PATCH] Use OBJECT_INFO_QUICK to speedup git fetch-pack
Hi, Thank you for merging! Takuto 2017-11-28 20:27 GMT+09:00 Johannes Schindelin <johannes.schinde...@gmx.de>: > Hi, > > On Tue, 28 Nov 2017, Takuto Ikuta wrote: > >> As long as this PR is included in next Git for Windows release, I >> won't suffer from slow git fetch. >> https://github.com/git-for-windows/git/pull/1372 >> >> But I sent you 2 PRs to follow right way. >> https://github.com/git-for-windows/git/pull/1379 >> https://github.com/git-for-windows/git/pull/1380 >> Feel free to merge these PRs. > > I amended them slightly, and merged them. > > Thank you, > Dscho -- Takuto Ikuta Software Engineer in Tokyo Chrome Infrastructure (goma team)
Re: [PATCH] Use OBJECT_INFO_QUICK to speedup git fetch-pack
Hi Johannes, As long as this PR is included in next Git for Windows release, I won't suffer from slow git fetch. https://github.com/git-for-windows/git/pull/1372 But I sent you 2 PRs to follow right way. https://github.com/git-for-windows/git/pull/1379 https://github.com/git-for-windows/git/pull/1380 Feel free to merge these PRs. Thanks. 2017-11-28 9:33 GMT+09:00 Johannes Schindelin <johannes.schinde...@gmx.de>: > Hi Junio, > > On Tue, 28 Nov 2017, Junio C Hamano wrote: > >> Johannes Schindelin <johannes.schinde...@gmx.de> writes: >> >> > My current plan is to release a new Git for Windows version on Wednesday, >> > to allow for a new cURL version to be bundled. >> >> Is this an updated 2.15.0 or are you planning to package 2.15.1? > > If there is a 2.15.1 to pick up for me, I'll take it. Otherwise it'll be > Git for Windows v2.15.0(2). > > Ciao, > Dscho -- Takuto Ikuta Software Engineer in Tokyo Chrome Infrastructure (goma team)
Re: [PATCH] Use OBJECT_INFO_QUICK to speedup git fetch-pack
2017-11-27 13:53 GMT+09:00 Junio C Hamano <gits...@pobox.com>: > Jeff King <p...@peff.net> writes: > >>> cf. >>> https://public-inbox.org/git/20171120202920.7ppcwmzkxifyw...@sigill.intra.peff.net/ >> >> It's funny that we'd get two patches so close together. AFAIK the >> slowness here has been with us for years, and I just happened to >> investigate it recently. Yes, thank you for let me know. Please ignore my patch, sorry. >> >>> The 5-patch series that contains the same change as this one is >>> cooking and will hopefully be in the released version before the end >>> of the year. >> >> I'd be curious if the 5th patch there provides an additional speedup for >> Takuto's case. > > Indeed, it is a very good point. > > IIUC, the 5th one is about fetching tons of refs that you have never > seen, right? If a repository that has trouble with everything-local > is suffering because it right now has 300k remote-tracking branches, > I'd imagine that these remote-tracking branches are being added at a > considerable rate, so I'd not be surprised if these "new" refs > benefits from that patch. And it would be nice to know how much a > real life scenario actually does improve. > > Thanks. In chromium repository, your 5th patch does not improve performance, took more than 5 minutes to run fetch on windows. 4th patch is very important for the repository in daily fetch. I hope your 4th patch will be merged. Thanks. -- Takuto Ikuta Software Engineer in Tokyo Chrome Infrastructure (goma team)
[PATCH] Use OBJECT_INFO_QUICK to speedup git fetch-pack
Do not call prepare_packed_git for every refs.
In fetch-pack, git list ups entries in .git/objects/pack
directory for each refs.
Such behavior makes git fetch-pack slow for the repository having the
large number of refs, especially for windows.
For chromium repository, having more than 30 remote refs, git fetch
takes more than 3 minutes if fetch-pack runs on my windows workstation.
This patch improves the time to around 17 seconds in the same machine.
Signed-off-by: Takuto Ikuta <tik...@google.com>
---
fetch-pack.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fetch-pack.c b/fetch-pack.c
index 008b25d3db087..0184584e80599 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c
@@ -716,7 +716,7 @@ static int everything_local(struct fetch_pack_args *args,
for (ref = *refs; ref; ref = ref->next) {
struct object *o;
- if (!has_object_file(>old_oid))
+ if (!has_object_file_with_flags(>old_oid,
OBJECT_INFO_QUICK))
continue;
o = parse_object(>old_oid);
--
https://github.com/git/git/pull/437
