From: David Gibson <david@gibson.dropbear.id.au>
To: Michael Roth <mdroth@linux.vnet.ibm.com>
Cc: aik@ozlabs.ru, qemu-devel@nongnu.org, agraf@suse.de,
qemu-ppc@nongnu.org, bharata@linux.vnet.ibm.com,
pbonzini@redhat.com
Subject: Re: [Qemu-devel] [RFCv2 2/2] spapr: Don't use QOM [*] syntax for DR connectors.
Date: Wed, 16 Sep 2015 13:18:04 +1000 [thread overview]
Message-ID: <20150916031804.GZ2547@voom.fritz.box> (raw)
In-Reply-To: <20150915040300.19285.85127@loki>
[-- Attachment #1: Type: text/plain, Size: 4043 bytes --]
On Mon, Sep 14, 2015 at 11:03:00PM -0500, Michael Roth wrote:
> Quoting David Gibson (2015-09-13 20:41:53)
> > The dynamic reconfiguration (hotplug) code for the pseries machine type
> > uses a "DR connector" QOM object for each resource it will be possible
> > to hotplug. Each of these is added to its owner using
> > object_property_add_child(owner, "dr-connector[*], ...);
> >
> > That works ok, mostly, but it means that the property indices are
> > arbitrary, depending on the order in which the connectors are constructed.
> > When we have both memory and cpu hotplug, the connectors will be under the
> > same parent (at least in the current drafts), meaning the indices don't
> > correspond to any meaningful ID.
> >
> > It gets worse when large amounts of hotpluggable RAM is configured. For
> > RAM, there's a DR connector object for every 256MB of potential memory. So
> > if maxmem=2T, for example, there are 8192 objects under the same parent.
> >
> > The QOM interfaces aren't really designed for this. In particular
> > object_property_add() with [*] has O(n^2) time complexity (in the number of
> > existing children): first it has a linear search through array indices to
> > find a free slot, each of which is attempted to a recursive call to
> > object_property_add() with a specific [N]. Those calls are O(n) because
> > there's a linear search through all properties to check for duplicates.
> >
> > By using a meaningful index value, which we already know is unique we can
> > avoid the [*] special behaviour. That lets us reduce the total time for
> > creating the DR objects from O(n^3) to O(n^2).
> >
> > O(n^2) is still kind of crappy, but it's enough to reduce the startup time
> > of qemu with maxmem=2T from ~20 minutes to ~4 seconds.
> >
> > Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
> > Cc: Bharata B Rao <bharata@linux.vnet.ibm.com>
>
> This is the second patch I've seen that drops use of "[*]" for
> performance reasons, but looking at the code I don't really see
> any reason that logic can't be implemented in object_property_add()
> in O(n) time. For instance I think it can be achieved by
> storing/hashing the base string to an array of auto-incremented
> indicies that we update whenever a child with the corresponding
> <base_string>[n] format is added.
Oh, there's no question that could be done. The question is: is
thousands of QOM objects under a single parent a use case that we
actually want to build QOM for?
> I wouldn't hold this real fix up for that though, and in fact the
> use of DRC indexes make for much easier debugging anyway so I'd
> probably still prefer this approach anyway.
>
> Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
>
> > ---
> > hw/ppc/spapr_drc.c | 5 ++++-
> > 1 file changed, 4 insertions(+), 1 deletion(-)
> >
> > diff --git a/hw/ppc/spapr_drc.c b/hw/ppc/spapr_drc.c
> > index 68e0c3e..2f95259 100644
> > --- a/hw/ppc/spapr_drc.c
> > +++ b/hw/ppc/spapr_drc.c
> > @@ -451,13 +451,16 @@ sPAPRDRConnector *spapr_dr_connector_new(Object *owner,
> > {
> > sPAPRDRConnector *drc =
> > SPAPR_DR_CONNECTOR(object_new(TYPE_SPAPR_DR_CONNECTOR));
> > + char *prop_name;
> >
> > g_assert(type);
> >
> > drc->type = type;
> > drc->id = id;
> > - object_property_add_child(owner, "dr-connector[*]", OBJECT(drc), NULL);
> > + prop_name = g_strdup_printf("dr-connector[%"PRIu32"]", get_index(drc));
> > + object_property_add_child(owner, prop_name, OBJECT(drc), NULL);
> > object_property_set_bool(OBJECT(drc), true, "realized", NULL);
> > + g_free(prop_name);
> >
> > /* human-readable name for a DRC to encode into the DT
> > * description. this is mainly only used within a guest in place
>
--
David Gibson | I'll have my music baroque, and my code
david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_
| _way_ _around_!
http://www.ozlabs.org/~dgibson
[-- Attachment #2: Type: application/pgp-signature, Size: 819 bytes --]
prev parent reply other threads:[~2015-09-16 3:25 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-09-14 1:41 [Qemu-devel] [RFCv2 0/2] spapr: Cleanups to dynamic reconfiguration mechanism David Gibson
2015-09-14 1:41 ` [Qemu-devel] [RFCv2 1/2] spapr: Remove unnecessary owner field from sPAPRDRConnector David Gibson
2015-09-14 6:23 ` Bharata B Rao
2015-09-14 8:22 ` Alexey Kardashevskiy
2015-09-14 11:45 ` David Gibson
2015-09-14 12:11 ` Paolo Bonzini
2015-09-14 14:06 ` Alexey Kardashevskiy
2015-09-14 14:24 ` Paolo Bonzini
2015-09-16 3:16 ` David Gibson
2015-09-17 15:50 ` Michael Roth
2015-09-17 15:53 ` Paolo Bonzini
2015-09-17 23:03 ` Michael Roth
2015-09-17 15:01 ` Michael Roth
2015-09-15 0:31 ` David Gibson
2015-09-15 0:30 ` David Gibson
2015-09-15 11:05 ` Paolo Bonzini
2015-09-14 1:41 ` [Qemu-devel] [RFCv2 2/2] spapr: Don't use QOM [*] syntax for DR connectors David Gibson
2015-09-14 4:07 ` Bharata B Rao
2015-09-14 4:14 ` David Gibson
2015-09-14 4:41 ` Bharata B Rao
2015-09-14 5:20 ` David Gibson
2015-09-14 8:13 ` Alexey Kardashevskiy
2015-09-15 4:03 ` Michael Roth
2015-09-15 4:21 ` Michael Roth
2015-09-16 3:18 ` David Gibson [this message]
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=20150916031804.GZ2547@voom.fritz.box \
--to=david@gibson.dropbear.id.au \
--cc=agraf@suse.de \
--cc=aik@ozlabs.ru \
--cc=bharata@linux.vnet.ibm.com \
--cc=mdroth@linux.vnet.ibm.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=qemu-ppc@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.