From: Eric Blake <eblake@redhat.com>
To: Paolo Bonzini <pbonzini@redhat.com>, qemu-devel@nongnu.org
Cc: armbru@redhat.com
Subject: Re: [Qemu-devel] [PATCH 4/4] qjson: surprise, allocating 6 QObjects per token is expensive
Date: Fri, 20 Nov 2015 11:48:55 -0700 [thread overview]
Message-ID: <564F6B17.5040201@redhat.com> (raw)
In-Reply-To: <1448010269-21694-5-git-send-email-pbonzini@redhat.com>
[-- Attachment #1: Type: text/plain, Size: 1950 bytes --]
On 11/20/2015 02:04 AM, Paolo Bonzini wrote:
> Replace the contents of the tokens GQueue with a simple struct. This cuts
> the amount of memory allocated by tests/check-qjson from ~500MB to ~20MB,
> and the execution time from 600ms to 80ms on my laptop. Still a lot (some
> could be saved by using an intrusive list, such as QSIMPLEQ, instead of
> the GQueue), but the savings are already massive and the right thing to
> do would probably be to get rid of json-streamer completely.
>
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
> include/qapi/qmp/json-streamer.h | 7 ++++
> qobject/json-parser.c | 81 +++++++++++++++++++---------------------
> qobject/json-streamer.c | 19 ++++------
> 3 files changed, 53 insertions(+), 54 deletions(-)
>
> @@ -54,15 +50,16 @@ static void json_message_process_token(JSONLexer *lexer, GString *input, JSONTok
> }
> }
>
> - dict = qdict_new();
> - qdict_put(dict, "type", qint_from_int(type));
> - qdict_put(dict, "token", qstring_from_str(input->str));
> - qdict_put(dict, "x", qint_from_int(x));
> - qdict_put(dict, "y", qint_from_int(y));
> + token = g_malloc(sizeof(JSONToken) + input->len + 1);
> + token->type = type;
> + memcpy(token->str, input->str, input->len + 1);
> + token->str[input->len] = 0;
Looks like you are writing the last byte twice. Either the +1 in the
memcpy() always copies a NUL byte and we don't need the second
assignment, or you should drop the +1.
Otherwise, looks like a sane replacement that saves a lot of memory.
Reviewed-by: Eric Blake <eblake@redhat.com>
Are we hoping to get this in 2.5 because it fixes the memory hog bug, or
are we considering that it is not a regression from 2.4 and therefore
something that should wait for 2.6?
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library http://libvirt.org
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 604 bytes --]
next prev parent reply other threads:[~2015-11-20 18:49 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-11-20 9:04 [Qemu-devel] [PATCH 0/4] qjson: save a lot of memory Paolo Bonzini
2015-11-20 9:04 ` [Qemu-devel] [PATCH 1/4] qjson: replace QString in JSONLexer with GString Paolo Bonzini
2015-11-20 18:23 ` Eric Blake
2015-11-20 20:16 ` Paolo Bonzini
2015-11-20 9:04 ` [Qemu-devel] [PATCH 2/4] qjson: do not save/restore contexts Paolo Bonzini
2015-11-20 18:28 ` Eric Blake
2015-11-20 20:17 ` Paolo Bonzini
2015-11-20 9:04 ` [Qemu-devel] [PATCH 3/4] qjson: store tokens in a GQueue Paolo Bonzini
2015-11-20 18:40 ` Eric Blake
2015-11-20 9:04 ` [Qemu-devel] [PATCH 4/4] qjson: surprise, allocating 6 QObjects per token is expensive Paolo Bonzini
2015-11-20 18:48 ` Eric Blake [this message]
2015-11-20 20:19 ` Paolo Bonzini
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
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=564F6B17.5040201@redhat.com \
--to=eblake@redhat.com \
--cc=armbru@redhat.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
/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.
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 NNTP newsgroup(s).