From: Eric Wong <e@80x24.org>
To: "René Scharfe" <l.s.r@web.de>
Cc: git@vger.kernel.org, Jeff King <peff@peff.net>
Subject: Re: [PATCH v2 1/5] speed up alt_odb_usable() with many alternates
Date: Tue, 6 Jul 2021 23:01:15 +0000 [thread overview]
Message-ID: <20210706230115.GA8624@dcvr> (raw)
In-Reply-To: <fc342ddd-1b35-62cc-dd4b-e0462d595819@web.de>
René Scharfe <l.s.r@web.de> wrote:
> Am 29.06.21 um 22:53 schrieb Eric Wong:
> > With many alternates, the duplicate check in alt_odb_usable()
> > wastes many cycles doing repeated fspathcmp() on every existing
> > alternate. Use a khash to speed up lookups by odb->path.
> >
> > Since the kh_put_* API uses the supplied key without
> > duplicating it, we also take advantage of it to replace both
> > xstrdup() and strbuf_release() in link_alt_odb_entry() with
> > strbuf_detach() to avoid the allocation and copy.
> >
> > In a test repository with 50K alternates and each of those 50K
> > alternates having one alternate each (for a total of 100K total
> > alternates); this speeds up lookup of a non-existent blob from
> > over 16 minutes to roughly 2.7 seconds on my busy workstation.
>
> Yay for hashmaps! :)
>
> > Note: all underlying git object directories were small and
> > unpacked with only loose objects and no packs. Having to load
> > packs increases times significantly.
> >
> > Signed-off-by: Eric Wong <e@80x24.org>
> > ---
> > object-file.c | 33 ++++++++++++++++++++++-----------
> > object-store.h | 17 +++++++++++++++++
> > object.c | 2 ++
> > 3 files changed, 41 insertions(+), 11 deletions(-)
> >
> > diff --git a/object-file.c b/object-file.c
> > index f233b440b2..304af3a172 100644
> > --- a/object-file.c
> > +++ b/object-file.c
> > @@ -517,9 +517,9 @@ const char *loose_object_path(struct repository *r, struct strbuf *buf,
> > */
> > static int alt_odb_usable(struct raw_object_store *o,
> > struct strbuf *path,
> > - const char *normalized_objdir)
> > + const char *normalized_objdir, khiter_t *pos)
> > {
> > - struct object_directory *odb;
> > + int r;
> >
> > /* Detect cases where alternate disappeared */
> > if (!is_directory(path->buf)) {
> > @@ -533,14 +533,22 @@ static int alt_odb_usable(struct raw_object_store *o,
> > * Prevent the common mistake of listing the same
> > * thing twice, or object directory itself.
> > */
> > - for (odb = o->odb; odb; odb = odb->next) {
> > - if (!fspathcmp(path->buf, odb->path))
> > - return 0;
> > + if (!o->odb_by_path) {
> > + khiter_t p;
> > +
> > + o->odb_by_path = kh_init_odb_path_map();
> > + assert(!o->odb->next);
> > + p = kh_put_odb_path_map(o->odb_by_path, o->odb->path, &r);
>
> So on the first run you not just create the hashmap, but you also
> pre-populate it with the main object directory. Makes sense. The
> hashmap wouldn't even be created in repositories without alternates.
>
> > + if (r < 0) die_errno(_("kh_put_odb_path_map"));
>
> Our other callers don't handle a negative return code because it would
> indicate an allocation failure, and in our version we use ALLOC_ARRAY,
> which dies on error. So you don't need that check here, but we better
> clarify that in khash.h.
>
> > + assert(r == 1); /* never used */
> > + kh_value(o->odb_by_path, p) = o->odb;
> > }
> > if (!fspathcmp(path->buf, normalized_objdir))
> > return 0;
> > -
> > - return 1;
> > + *pos = kh_put_odb_path_map(o->odb_by_path, path->buf, &r);
> > + if (r < 0) die_errno(_("kh_put_odb_path_map"));
>
> Dito.
>
> > + /* r: 0 = exists, 1 = never used, 2 = deleted */
> > + return r == 0 ? 0 : 1;
>
> The comment indicates that khash would be nicer to use if it had an
> enum for the kh_put return values. Perhaps, but that should be done in
> another series.
Agreed for another series. I've also found myself wishing khash
used enums. But I'm also not sure how much changing of 3rd
party code we should be doing...
> I like the solution in oidset.c to make this more readable, though: Call
> the return value "added" instead of "r" and then a "return !added;"
> makes sense without additional comments.
>
> > }
> >
> > /*
> > diff --git a/object-store.h b/object-store.h
> > index ec32c23dcb..20c1cedb75 100644
> > --- a/object-store.h
> > +++ b/object-store.h
> > @@ -7,6 +7,8 @@
> > #include "oid-array.h"
> > #include "strbuf.h"
> > #include "thread-utils.h"
> > +#include "khash.h"
> > +#include "dir.h"
> >
> > struct object_directory {
> > struct object_directory *next;
> > @@ -30,6 +32,19 @@ struct object_directory {
> > char *path;
> > };
> >
> > +static inline int odb_path_eq(const char *a, const char *b)
> > +{
> > + return !fspathcmp(a, b);
> > +}
>
> This is not specific to the object store. It could be called fspatheq
> and live in dir.h. Or dir.c -- a surprising amount of code seems to
> necessary for that negation (https://godbolt.org/z/MY7Wda3a7). Anyway,
> it's just an idea for another series.
No JS here for godbolt, but there's also a bunch of "!fspathcmp"
here that could probably be changed to fspatheq.
> > +
> > +static inline int odb_path_hash(const char *str)
> > +{
> > + return ignore_case ? strihash(str) : __ac_X31_hash_string(str);
> > +}
>
> The internal Attractive Chaos (__ac_*) macros should be left confined
> to khash.h, I think. Its alias kh_str_hash_func would be better
> suited here.
>
> Do we want to use the K&R hash function here at all, though? If we
> use FNV-1 when ignoring case, why not also use it (i.e. strhash) when
> respecting it? At least that's done in builtin/sparse-checkout.c,
> dir.c and merge-recursive.c. This is just handwaving and yammering
> about lack of symmetry, but I do wonder how your performance numbers
> look with strhash. If it's fine then we could package this up as
> fspathhash..
Yeah, I think fspathhash should be path_hash in merge-recursive.c
(and path_hash eliminated).
I don't have performance numbers, and I doubt hash function
performance is much overhead, here. I used X31 since it was
local to khash.
I would prefer we only have one non-cryptographic hash
implementation to reduce cognitive overhead, so maybe we can
drop X31 entirely for FNV-1. I'd also prefer we only have khash
or hashmap, not both.
> And I also wonder how it looks if you use strihash unconditionally.
> I guess case collisions are usually rare and branching based on a
> global variable may be more expensive than case folding.
*shrug* I'll let somebody with more appropriate systems do
benchmarks, there. But it could be an easy switch once
fspathhash is in place.
next prev parent reply other threads:[~2021-07-06 23:01 UTC|newest]
Thread overview: 99+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-06-24 0:58 [PATCH] speed up alt_odb_usable() with many alternates Eric Wong
2021-06-27 2:47 ` [PATCH 0/5] optimizations for many odb alternates Eric Wong
2021-06-27 2:47 ` [PATCH 1/5] speed up alt_odb_usable() with many alternates Eric Wong
2021-06-27 2:47 ` [PATCH 2/5] avoid strlen via strbuf_addstr in link_alt_odb_entry Eric Wong
2021-06-27 2:47 ` [PATCH 3/5] make object_directory.loose_objects_subdir_seen a bitmap Eric Wong
2021-06-27 10:23 ` René Scharfe
2021-06-28 23:09 ` Eric Wong
2021-06-27 2:47 ` [PATCH 4/5] oidcpy_with_padding: constify `src' arg Eric Wong
2021-06-27 2:47 ` [PATCH 5/5] oidtree: a crit-bit tree for odb_loose_cache Eric Wong
2021-06-29 14:42 ` Junio C Hamano
2021-06-29 20:17 ` Eric Wong
2021-06-29 20:53 ` [PATCH v2 0/5] optimizations for many alternates Eric Wong
2021-07-07 23:10 ` [PATCH v3 " Eric Wong
2021-07-07 23:10 ` [PATCH v3 1/5] speed up alt_odb_usable() with " Eric Wong
2021-07-08 0:20 ` Junio C Hamano
2021-07-08 1:14 ` Eric Wong
2021-07-08 4:30 ` Junio C Hamano
2021-07-07 23:10 ` [PATCH v3 2/5] avoid strlen via strbuf_addstr in link_alt_odb_entry Eric Wong
2021-07-08 4:57 ` Junio C Hamano
2021-07-07 23:10 ` [PATCH v3 3/5] make object_directory.loose_objects_subdir_seen a bitmap Eric Wong
2021-07-07 23:10 ` [PATCH v3 4/5] oidcpy_with_padding: constify `src' arg Eric Wong
2021-07-07 23:10 ` [PATCH v3 5/5] oidtree: a crit-bit tree for odb_loose_cache Eric Wong
2021-06-29 20:53 ` [PATCH v2 1/5] speed up alt_odb_usable() with many alternates Eric Wong
2021-07-03 10:05 ` René Scharfe
2021-07-04 9:02 ` René Scharfe
2021-07-06 23:01 ` Eric Wong [this message]
2021-06-29 20:53 ` [PATCH v2 2/5] avoid strlen via strbuf_addstr in link_alt_odb_entry Eric Wong
2021-06-29 20:53 ` [PATCH v2 3/5] make object_directory.loose_objects_subdir_seen a bitmap Eric Wong
2021-06-29 20:53 ` [PATCH v2 4/5] oidcpy_with_padding: constify `src' arg Eric Wong
2021-06-29 20:53 ` [PATCH v2 5/5] oidtree: a crit-bit tree for odb_loose_cache Eric Wong
2021-07-04 9:02 ` René Scharfe
2021-07-06 23:21 ` Eric Wong
2021-07-04 9:32 ` Ævar Arnfjörð Bjarmason
2021-07-07 23:12 ` Eric Wong
2021-08-06 15:31 ` Andrzej Hunt
2021-08-06 17:53 ` René Scharfe
2021-08-07 22:49 ` Eric Wong
2021-08-09 1:35 ` Carlo Arenas
2021-08-09 1:38 ` [PATCH/RFC 0/3] pedantic errors in next Carlo Marcelo Arenas Belón
2021-08-09 1:38 ` [PATCH/RFC 1/3] oidtree: avoid nested struct oidtree_node Carlo Marcelo Arenas Belón
2021-08-09 1:38 ` [PATCH/RFC 2/3] object-store: avoid extra ';' from KHASH_INIT Carlo Marcelo Arenas Belón
2021-08-09 15:53 ` Junio C Hamano
2021-08-09 1:38 ` [PATCH/RFC 3/3] ci: run a pedantic build as part of the GitHub workflow Carlo Marcelo Arenas Belón
2021-08-09 10:50 ` Bagas Sanjaya
2021-08-09 22:03 ` Carlo Arenas
2021-08-09 14:56 ` Phillip Wood
2021-08-09 22:48 ` Carlo Arenas
2021-08-10 15:24 ` Phillip Wood
2021-08-10 18:25 ` Junio C Hamano
2021-08-30 11:36 ` Ævar Arnfjörð Bjarmason
2021-08-31 20:28 ` Carlo Arenas
2021-08-31 20:51 ` Ævar Arnfjörð Bjarmason
2021-08-31 23:54 ` Carlo Arenas
2021-09-01 1:52 ` Jeff King
2021-09-01 17:55 ` Junio C Hamano
2021-08-30 11:40 ` Ævar Arnfjörð Bjarmason
2021-09-01 9:19 ` [RFC PATCH v2 0/4] developer: support pedantic Carlo Marcelo Arenas Belón
2021-09-01 9:19 ` [RFC PATCH v2 1/4] developer: retire USE_PARENS_AROUND_GETTEXT_N support Carlo Marcelo Arenas Belón
2021-09-01 9:19 ` [RFC PATCH v2 2/4] developer: enable pedantic by default Carlo Marcelo Arenas Belón
2021-09-01 9:19 ` [RFC PATCH v2 3/4] developer: add an alternative script for detecting broken N_() Carlo Marcelo Arenas Belón
2021-09-01 9:19 ` [RFC PATCH v2 4/4] developer: move detect-compiler out of the main directory Carlo Marcelo Arenas Belón
2021-09-01 10:10 ` [RFC PATCH v2 0/4] developer: support pedantic Jeff King
2021-09-01 11:25 ` [PATCH] gettext: remove optional non-standard parens in N_() definition Ævar Arnfjörð Bjarmason
2021-09-01 17:31 ` Eric Sunshine
2021-09-02 9:13 ` Jeff King
2021-09-02 19:19 ` Junio C Hamano
2021-09-01 11:27 ` [RFC PATCH v2 0/4] developer: support pedantic Ævar Arnfjörð Bjarmason
2021-09-01 18:03 ` Carlo Arenas
2021-09-03 17:02 ` [PATCH v3 0/3] support pedantic in developer mode Carlo Marcelo Arenas Belón
2021-09-03 17:02 ` [PATCH v3 1/3] gettext: remove optional non-standard parens in N_() definition Carlo Marcelo Arenas Belón
2021-09-10 15:39 ` Ævar Arnfjörð Bjarmason
2021-09-03 17:02 ` [PATCH v3 2/3] win32: allow building with pedantic mode enabled Carlo Marcelo Arenas Belón
2021-09-03 18:47 ` René Scharfe
2021-09-03 20:13 ` Carlo Marcelo Arenas Belón
2021-09-03 20:32 ` Junio C Hamano
2021-09-03 20:38 ` René Scharfe
2021-09-04 9:37 ` René Scharfe
2021-09-04 14:42 ` Carlo Arenas
2021-09-27 23:04 ` Jonathan Tan
2021-09-28 0:30 ` Carlo Arenas
2021-09-28 16:50 ` Jonathan Tan
2021-09-28 17:37 ` Junio C Hamano
2021-09-28 20:16 ` Jonathan Tan
2021-09-29 1:00 ` Carlo Arenas
2021-09-29 15:55 ` Junio C Hamano
2021-09-03 17:02 ` [PATCH v3 3/3] developer: enable pedantic by default Carlo Marcelo Arenas Belón
2021-09-05 7:54 ` [PATCH v3 0/3] support pedantic in developer mode Ævar Arnfjörð Bjarmason
2021-08-09 16:44 ` [PATCH/RFC 0/3] pedantic errors in next Junio C Hamano
2021-08-09 20:10 ` Eric Wong
2021-08-10 6:16 ` Carlo Marcelo Arenas Belón
2021-08-10 19:30 ` René Scharfe
2021-08-10 23:49 ` Carlo Arenas
2021-08-11 0:57 ` Carlo Arenas
2021-08-11 14:57 ` René Scharfe
2021-08-11 17:20 ` Junio C Hamano
2021-08-10 18:59 ` [PATCH v2 5/5] oidtree: a crit-bit tree for odb_loose_cache René Scharfe
2021-08-10 19:40 ` René Scharfe
2021-08-14 20:00 ` [PATCH] oidtree: avoid unaligned access to crit-bit tree René Scharfe
2021-08-16 19:11 ` Junio C Hamano
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
List information: http://vger.kernel.org/majordomo-info.html
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20210706230115.GA8624@dcvr \
--to=e@80x24.org \
--cc=git@vger.kernel.org \
--cc=l.s.r@web.de \
--cc=peff@peff.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
Code repositories for project(s) associated with this public inbox
https://80x24.org/mirrors/git.git
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).