From: Han Xin <chiyutianyi@gmail.com>
To: "Ævar Arnfjörð Bjarmason" <avarab@gmail.com>
Cc: Junio C Hamano <gitster@pobox.com>,
Git List <git@vger.kernel.org>,
Neeraj Singh <neerajsi@microsoft.com>,
Han Xin <hanxin.hx@alibaba-inc.com>,
Johannes Schindelin <Johannes.Schindelin@gmx.de>
Subject: Re: ab/squelch-empty-fsync-traces & hx/unpack-streaming bug (was: What's cooking in git.git (Jul 2022, #04; Wed, 13))
Date: Sat, 16 Jul 2022 20:23:23 +0800 [thread overview]
Message-ID: <CAO0brD3fdjQNfQaUBJRAHxDc24K00zpBUa62zST0=cZ5uz3vGA@mail.gmail.com> (raw)
In-Reply-To: <220715.86bktqzdb8.gmgdl@evledraar.gmail.com>
CC: Johannes Schindelin
On Fri, Jul 15, 2022 at 10:18 PM Ævar Arnfjörð Bjarmason
<avarab@gmail.com> wrote:
>
> I hadn't had time to look at this until now. There's some interesting
> behavior here.
>
> The code to check the hardware flush was added in aaf81223f48
> (unpack-objects: use stream_loose_object() to unpack large objects,
> 2022-06-11) (that series is now on master).
>
> But as my ab/squelch-empty-fsync-traces notes we always add this to the
> event, so the:
>
> grep fsync/hardware-flush trace2.txt &&
>
> Is equivalent to:
>
> true &&
>
> I.e. it's not testing worthwhile at all. The reason you're seeing a
> failure is deu to 412e4caee38 (tests: disable fsync everywhere,
> 2021-10-29), i.e. our tests disable fsync(). What you have queued will
> pass as:
>
> GIT_TEST_FSYNC=true ./t5351-unpack-large-objects.sh
>
> But I think that would be meaningless, since we'll write out that on
> FSYNC_HARDWARE_FLUSH whether we actually support "bulk" or not. AFAICT
> the way to detect if we support "bulk" at all is to check for
> fsync/writeout-only.
>
> *Except* that we we unconditionally increment the "writeout only"
> counter, even if we don't actually support that "bulk" mode. We're just
> doing a regular fsync().
>
Agree with you.
In fact, since stream_loose_object() only works with objects of type
*blob*, the rest objects of *commit* and *tree* will still use
write_loose_object(), so "grep fsync/hardware-flush trace2.txt" did
not check for the changes in stream_loose_object() at all.
I haven't found any reference cases in the existing tests.
Perhaps, we need more efficient "fsync" test cases?
Thanks.
-Han Xin
next prev parent reply other threads:[~2022-07-16 12:26 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-07-14 1:32 What's cooking in git.git (Jul 2022, #04; Wed, 13) Junio C Hamano
2022-07-14 12:11 ` ds/rebase-update-ref (was Re: What's cooking in git.git (Jul 2022, #04; Wed, 13)) Derrick Stolee
2022-07-14 16:35 ` Junio C Hamano
2022-07-15 13:40 ` ab/squelch-empty-fsync-traces & hx/unpack-streaming bug (was: " Ævar Arnfjörð Bjarmason
2022-07-16 12:23 ` Han Xin [this message]
2022-07-17 3:46 ` en/merge-restore-to-pristine (Was: " Elijah Newren
2022-07-17 16:58 ` ZheNing Hu
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='CAO0brD3fdjQNfQaUBJRAHxDc24K00zpBUa62zST0=cZ5uz3vGA@mail.gmail.com' \
--to=chiyutianyi@gmail.com \
--cc=Johannes.Schindelin@gmx.de \
--cc=avarab@gmail.com \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=hanxin.hx@alibaba-inc.com \
--cc=neerajsi@microsoft.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).