From: Derrick Stolee <stolee@gmail.com>
To: "René Scharfe" <l.s.r@web.de>,
"SZEDER Gábor via GitGitGadget" <gitgitgadget@gmail.com>,
git@vger.kernel.org
Cc: me@ttaylorr.com, szeder.dev@gmail.com,
Derrick Stolee <dstolee@microsoft.com>
Subject: Re: [PATCH v2 10/11] commit-graph: check all leading directories in changed path Bloom filters
Date: Thu, 25 Jun 2020 11:05:04 -0400 [thread overview]
Message-ID: <0fe96c75-2946-8160-2ced-3d9781dca8c0@gmail.com> (raw)
In-Reply-To: <8d8b1763-bbf3-682e-0649-f5a0885399e6@web.de>
On 6/25/2020 3:25 AM, René Scharfe wrote:
> Am 23.06.20 um 19:47 schrieb SZEDER Gábor via GitGitGadget:
>> From: =?UTF-8?q?SZEDER=20G=C3=A1bor?= <szeder.dev@gmail.com>
>>
>> The file 'dir/subdir/file' can only be modified if its leading
>> directories 'dir' and 'dir/subdir' are modified as well.
>>
>> So when checking modified path Bloom filters looking for commits
>> modifying a path with multiple path components, then check not only
>> the full path in the Bloom filters, but all its leading directories as
>> well. Take care to check these paths in "deepest first" order,
>> because it's the full path that is least likely to be modified, and
>> the Bloom filter queries can short circuit sooner.
>>
>> This can significantly reduce the average false positive rate, by
>> about an order of magnitude or three(!), and can further speed up
>> pathspec-limited revision walks. The table below compares the average
>> false positive rate and runtime of
>>
>> git rev-list HEAD -- "$path"
>>
>> before and after this change for 5000+ randomly* selected paths from
>> each repository:
>>
>> Average false Average Average
>> positive rate runtime runtime
>> before after before after difference
>> ------------------------------------------------------------------
>> git 3.220% 0.7853% 0.0558s 0.0387s -30.6%
>> linux 2.453% 0.0296% 0.1046s 0.0766s -26.8%
>> tensorflow 2.536% 0.6977% 0.0594s 0.0420s -29.2%
>>
>> *Path selection was done with the following pipeline:
>>
>> git ls-tree -r --name-only HEAD | sort -R | head -n 5000
>>
>> The improvements in runtime are much smaller than the improvements in
>> average false positive rate, as we are clearly reaching diminishing
>> returns here. However, all these timings depend on that accessing
>> tree objects is reasonably fast (warm caches). If we had a partial
>> clone and the tree objects had to be fetched from a promisor remote,
>> e.g.:
>>
>> $ git clone --filter=tree:0 --bare file://.../webkit.git webkit.notrees.git
>> $ git -C webkit.git -c core.modifiedPathBloomFilters=1 \
>> commit-graph write --reachable
>> $ cp webkit.git/objects/info/commit-graph webkit.notrees.git/objects/info/
>> $ git -C webkit.notrees.git -c core.modifiedPathBloomFilters=1 \
>> rev-list HEAD -- "$path"
>>
>> then checking all leading path component can reduce the runtime from
>> over an hour to a few seconds (and this is with the clone and the
>> promisor on the same machine).
>>
>> This adjusts the tracing values in t4216-log-bloom.sh, which provides a
>> concrete way to notice the improvement.
>>
>> Helped-by: Taylor Blau <me@ttaylorr.com>
>> Helped-by: René Scharfe <l.s.r@web.de>
>> Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com>
>> Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
>> ---
>> revision.c | 41 ++++++++++++++++++++++++++++++++---------
>> revision.h | 6 ++++--
>> t/t4216-log-bloom.sh | 2 +-
>> 3 files changed, 37 insertions(+), 12 deletions(-)
>>
>> diff --git a/revision.c b/revision.c
>> index b53377cd52..077888ee51 100644
>> --- a/revision.c
>> +++ b/revision.c
>> @@ -670,9 +670,10 @@ static void prepare_to_use_bloom_filter(struct rev_info *revs)
>> {
>> struct pathspec_item *pi;
>> char *path_alloc = NULL;
>> - const char *path;
>> + const char *path, *p;
>> int last_index;
>> - int len;
>> + size_t len;
>> + int path_component_nr = 1;
>>
>> if (!revs->commits)
>> return;
>> @@ -709,8 +710,28 @@ static void prepare_to_use_bloom_filter(struct rev_info *revs)
>> return;
>> }
>>
>> - revs->bloom_key = xmalloc(sizeof(struct bloom_key));
>> - fill_bloom_key(path, len, revs->bloom_key, revs->bloom_filter_settings);
>> + p = path;
>> + while (*p) {
>> + if (is_dir_sep(*p))
>> + path_component_nr++;
>> + p++;
>> + }
>> +
>> + revs->bloom_keys_nr = path_component_nr;
>> + ALLOC_ARRAY(revs->bloom_keys, revs->bloom_keys_nr);
>> +
>> + fill_bloom_key(path, len, &revs->bloom_keys[0],
>> + revs->bloom_filter_settings);
>> + path_component_nr = 1;
>> +
>> + p = path + len - 1;
>
> len cannot be 0 at this point, as patch 9 made sure, so this is safe.
> Good.
>
>> + while (p > path) {
>> + if (is_dir_sep(*p))
>> + fill_bloom_key(path, p - path,
>> + &revs->bloom_keys[path_component_nr++],
>> + revs->bloom_filter_settings);
>> + p--;
>> + }
>
> This walks the directory hierarchy upwards and adds bloom filters for
> shorter and shorter paths, ("deepest first"). Good.
>
> And it supports all directory separators. On Windows that would be
> slash (/) and backslash (\). I assume paths are normalized to use
> only slashes when bloom filters are written, correct? Then the lookup
> side needs to normalize a given path to only use slashes as well,
> otherwise paths with backslashes cannot be found. This part seems to
> be missing.
Yes, that's a good point. We _require_ the paths to be normalized
here to be Unix-style paths or else the Bloom filter keys are
incorrect. Thankfully, they are. Let's make that clear in-code by
using '/' instead of is_dir_sep().
Thanks,
-Stolee
next prev parent reply other threads:[~2020-06-25 15:05 UTC|newest]
Thread overview: 76+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-06-15 20:14 [PATCH 0/8] More commit-graph/Bloom filter improvements Derrick Stolee via GitGitGadget
2020-06-15 20:14 ` [PATCH 1/8] commit-graph: place bloom_settings in context Derrick Stolee via GitGitGadget
2020-06-18 20:30 ` René Scharfe
2020-06-19 12:58 ` Derrick Stolee
2020-06-15 20:14 ` [PATCH 2/8] commit-graph: unify the signatures of all write_graph_chunk_*() functions SZEDER Gábor via GitGitGadget
2020-06-18 20:30 ` René Scharfe
2020-06-15 20:14 ` [PATCH 3/8] commit-graph: simplify chunk writes into loop SZEDER Gábor via GitGitGadget
2020-06-18 20:30 ` René Scharfe
2020-06-15 20:14 ` [PATCH 4/8] commit-graph: check chunk sizes after writing SZEDER Gábor via GitGitGadget
2020-06-15 20:14 ` [PATCH 5/8] commit-graph: check all leading directories in changed path Bloom filters SZEDER Gábor via GitGitGadget
2020-06-18 20:31 ` René Scharfe
2020-06-19 9:14 ` René Scharfe
2020-06-19 17:17 ` Taylor Blau
2020-06-19 17:19 ` Taylor Blau
2020-06-23 13:47 ` Derrick Stolee
2020-06-15 20:14 ` [PATCH 6/8] bloom: enforce a minimum size of 8 bytes Derrick Stolee via GitGitGadget
2020-06-15 20:14 ` [PATCH 7/8] commit-graph: change test to die on parse, not load Derrick Stolee via GitGitGadget
2020-06-15 20:14 ` [PATCH 8/8] commit-graph: persist existence of changed-paths Derrick Stolee via GitGitGadget
2020-06-17 21:21 ` [PATCH 0/8] More commit-graph/Bloom filter improvements Junio C Hamano
2020-06-18 1:46 ` Derrick Stolee
2020-06-23 17:46 ` [PATCH v2 00/11] " Derrick Stolee via GitGitGadget
2020-06-23 17:47 ` [PATCH v2 01/11] commit-graph: place bloom_settings in context Derrick Stolee via GitGitGadget
2020-06-23 17:47 ` [PATCH v2 02/11] commit-graph: change test to die on parse, not load Derrick Stolee via GitGitGadget
2020-06-23 17:47 ` [PATCH v2 03/11] bloom: get_bloom_filter() cleanups Derrick Stolee via GitGitGadget
2020-06-25 7:24 ` René Scharfe
2020-06-23 17:47 ` [PATCH v2 04/11] commit-graph: persist existence of changed-paths Derrick Stolee via GitGitGadget
2020-06-23 17:47 ` [PATCH v2 05/11] commit-graph: unify the signatures of all write_graph_chunk_*() functions SZEDER Gábor via GitGitGadget
2020-06-25 7:25 ` René Scharfe
2020-06-23 17:47 ` [PATCH v2 06/11] commit-graph: simplify chunk writes into loop SZEDER Gábor via GitGitGadget
2020-06-25 7:25 ` René Scharfe
2020-06-25 14:59 ` Derrick Stolee
2020-06-23 17:47 ` [PATCH v2 07/11] commit-graph: check chunk sizes after writing SZEDER Gábor via GitGitGadget
2020-06-25 7:25 ` René Scharfe
2020-06-25 15:02 ` Derrick Stolee
2020-06-23 17:47 ` [PATCH v2 08/11] revision.c: fix whitespace Derrick Stolee via GitGitGadget
2020-06-23 17:47 ` [PATCH v2 09/11] revision: empty pathspecs should not use Bloom filters Taylor Blau via GitGitGadget
2020-06-23 17:47 ` [PATCH v2 10/11] commit-graph: check all leading directories in changed path " SZEDER Gábor via GitGitGadget
2020-06-25 7:25 ` René Scharfe
2020-06-25 15:05 ` Derrick Stolee [this message]
2020-06-26 6:34 ` SZEDER Gábor
2020-06-26 14:42 ` Derrick Stolee
2020-06-23 17:47 ` [PATCH v2 11/11] bloom: enforce a minimum size of 8 bytes Derrick Stolee via GitGitGadget
2020-06-24 23:11 ` [PATCH v2 00/11] More commit-graph/Bloom filter improvements Junio C Hamano
2020-06-24 23:32 ` Derrick Stolee
2020-06-25 0:38 ` Junio C Hamano
2020-06-25 13:38 ` Derrick Stolee
2020-06-25 16:34 ` Junio C Hamano
2020-06-26 12:30 ` [PATCH v3 00/10] " Derrick Stolee via GitGitGadget
2020-06-26 12:30 ` [PATCH v3 01/10] commit-graph: place bloom_settings in context Derrick Stolee via GitGitGadget
2020-06-26 12:30 ` [PATCH v3 02/10] commit-graph: change test to die on parse, not load Derrick Stolee via GitGitGadget
2020-06-26 12:30 ` [PATCH v3 03/10] bloom: fix logic in get_bloom_filter() Derrick Stolee via GitGitGadget
2020-06-27 16:33 ` SZEDER Gábor
2020-06-29 13:02 ` Derrick Stolee
2020-06-26 12:30 ` [PATCH v3 04/10] commit-graph: persist existence of changed-paths Derrick Stolee via GitGitGadget
2020-06-26 12:30 ` [PATCH v3 05/10] commit-graph: unify the signatures of all write_graph_chunk_*() functions SZEDER Gábor via GitGitGadget
2020-06-26 12:30 ` [PATCH v3 06/10] commit-graph: simplify chunk writes into loop SZEDER Gábor via GitGitGadget
2020-06-26 12:30 ` [PATCH v3 07/10] commit-graph: check chunk sizes after writing SZEDER Gábor via GitGitGadget
2020-06-26 12:30 ` [PATCH v3 08/10] revision.c: fix whitespace Derrick Stolee via GitGitGadget
2020-06-26 12:30 ` [PATCH v3 09/10] revision: empty pathspecs should not use Bloom filters Taylor Blau via GitGitGadget
2020-06-26 12:30 ` [PATCH v3 10/10] commit-graph: check all leading directories in changed path " SZEDER Gábor via GitGitGadget
2020-07-01 13:27 ` [PATCH v4 00/10] More commit-graph/Bloom filter improvements Derrick Stolee via GitGitGadget
2020-07-01 13:27 ` [PATCH v4 01/10] commit-graph: place bloom_settings in context Derrick Stolee via GitGitGadget
2020-07-01 13:27 ` [PATCH v4 02/10] commit-graph: change test to die on parse, not load Derrick Stolee via GitGitGadget
2020-07-01 13:27 ` [PATCH v4 03/10] bloom: fix logic in get_bloom_filter() Derrick Stolee via GitGitGadget
2020-07-01 13:27 ` [PATCH v4 04/10] commit-graph: persist existence of changed-paths Derrick Stolee via GitGitGadget
2020-10-15 13:21 ` SZEDER Gábor
2020-10-15 21:41 ` Taylor Blau
2020-10-16 2:18 ` Derrick Stolee
2020-10-16 3:18 ` Taylor Blau
2020-10-16 13:52 ` Derrick Stolee
2020-07-01 13:27 ` [PATCH v4 05/10] commit-graph: unify the signatures of all write_graph_chunk_*() functions SZEDER Gábor via GitGitGadget
2020-07-01 13:27 ` [PATCH v4 06/10] commit-graph: simplify chunk writes into loop SZEDER Gábor via GitGitGadget
2020-07-01 13:27 ` [PATCH v4 07/10] commit-graph: check chunk sizes after writing SZEDER Gábor via GitGitGadget
2020-07-01 13:27 ` [PATCH v4 08/10] revision.c: fix whitespace Derrick Stolee via GitGitGadget
2020-07-01 13:27 ` [PATCH v4 09/10] revision: empty pathspecs should not use Bloom filters Taylor Blau via GitGitGadget
2020-07-01 13:27 ` [PATCH v4 10/10] commit-graph: check all leading directories in changed path " SZEDER Gábor via GitGitGadget
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=0fe96c75-2946-8160-2ced-3d9781dca8c0@gmail.com \
--to=stolee@gmail.com \
--cc=dstolee@microsoft.com \
--cc=git@vger.kernel.org \
--cc=gitgitgadget@gmail.com \
--cc=l.s.r@web.de \
--cc=me@ttaylorr.com \
--cc=szeder.dev@gmail.com \
/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).