From: Junio C Hamano <gitster@pobox.com>
To: Stefan Beller <sbeller@google.com>
Cc: "git\@vger.kernel.org" <git@vger.kernel.org>,
Brandon Williams <bmwill@google.com>
Subject: Re: [PATCHv2] attr: convert to new threadsafe API
Date: Wed, 12 Oct 2016 15:57:31 -0700 [thread overview]
Message-ID: <xmqqlgxt18h0.fsf@gitster.mtv.corp.google.com> (raw)
In-Reply-To: <CAGZ79kaBhXHLDEK0XMLjm3QofmtaGZspA3EEx5x4-qCYY--wZA@mail.gmail.com> (Stefan Beller's message of "Wed, 12 Oct 2016 13:02:40 -0700")
Stefan Beller <sbeller@google.com> writes:
>> But many callers do not follow that; rather they do
>>
>> loop to iterate over paths {
>> call a helper func to learn attr X for path
>> use the value of attr X
>> }
>>
>> using a callchain that embeds a helper function deep inside, and
>> "check" is kept in the helper, check-attr function is called from
>> there, and "result" is not passed from the caller to the helper
>> (obviously, because it does not exist in the current API). See the
>> callchain that leads down to convert.c::convert_attrs() for a
>> typical example. When converted to the new API, it needs to have a
>> new "result" structure every time it is called, and cannot reuse the
>> one that was used in its previous call.
>
> Why would that be? i.e. I do not understand the reasoning/motivation
> as well as what you propose to change here.
The leaf function may look like
check_eol(const char *path)
{
static struct git_attr_check *check;
initl(&check, "eol", NULL);
git_check_attr(&check, path, result);
return nth_element_in_result(result, 0);
}
and we want "result" to be thread-ready.
A naive and undesired way to put it on stack is like so:
const char *check_eol(const char *path)
{
static struct git_attr_check *check;
struct git_attr_result result = RESULT_INIT;
const char *eol;
initl(&check, "eol", NULL);
init_result(&check, &result);
git_check_attr(&check, path, &result);
eol = nth_element_in_result(&result, 0);
clear_result(&result);
return eol;
}
where your "struct git_attr_result" has a fixed size, and the actual
result array is allocated via ALLOC_GROW() etc. That's overhead
that we do not want. Instead can't we do this?
const char *check_eol(const char *path)
{
static struct git_attr_check *check;
/* we know we only want one */
struct git_attr_result result[1];
const char *eol;
initl(&check, "eol", NULL);
git_check_attr(&check, path, result);
return result[0];
}
That way, we won't be doing ALLOC_GROW() in init_result() or free in
clear_result().
If you use a structure that has pointer to an array (i.e. the "naive
and undesired way" above), you cannot amortize the alloc/free by
making result "static" if you want to be thread-ready.
next prev parent reply other threads:[~2016-10-12 23:10 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-10-11 23:59 [PATCHv2] attr: convert to new threadsafe API Stefan Beller
2016-10-12 6:12 ` Junio C Hamano
2016-10-12 6:51 ` Stefan Beller
2016-10-12 16:20 ` Junio C Hamano
2016-10-12 20:02 ` Stefan Beller
2016-10-12 20:59 ` Junio C Hamano
2016-10-12 22:57 ` Junio C Hamano [this message]
2016-10-12 20:07 ` Johannes Sixt
2016-10-12 21:38 ` Junio C Hamano
2016-10-12 21:42 ` Stefan Beller
2016-10-12 21:50 ` Junio C Hamano
2016-10-12 21:45 ` Jacob Keller
2016-10-12 21:47 ` Stefan Beller
2016-10-12 21:28 ` Junio C Hamano
2016-10-12 21:39 ` Stefan Beller
2016-10-12 21:44 ` 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=xmqqlgxt18h0.fsf@gitster.mtv.corp.google.com \
--to=gitster@pobox.com \
--cc=bmwill@google.com \
--cc=git@vger.kernel.org \
--cc=sbeller@google.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).