DPDK patches and discussions
 help / color / mirror / Atom feed
From: Frode Nordahl <frode.nordahl@canonical.com>
To: David Marchand <david.marchand@redhat.com>
Cc: dev@dpdk.org, thomas@monjalon.net, mattias.ronnblom@ericsson.com,
	 anatoly.burakov@intel.com
Subject: Re: [PATCH v2 0/5] Defer lcore variables allocation
Date: Thu, 19 Dec 2024 17:19:30 +0100	[thread overview]
Message-ID: <CAKpbOAQsvUwEFt6J=VVakf1Zsuwjc5mQv6f8qZbAvbS9ajozhQ@mail.gmail.com> (raw)
In-Reply-To: <20241217085954.3310414-1-david.marchand@redhat.com>

On Tue, Dec 17, 2024 at 10:00 AM David Marchand
<david.marchand@redhat.com> wrote:
>
> As I had reported in rc2, the lcore variables allocation have a
> noticeable impact on applications consuming DPDK, even when such
> applications does not use DPDK, or use features associated to
> some lcore variables.
>
> While the amount has been reduced in a rush before rc2,
> there are still cases when the increased memory footprint is noticed
> like in scaling tests.
> See https://bugs.launchpad.net/ubuntu/+source/dpdk/+bug/2090931
>
>
> lcore variable allocations in constructor is a bad idea, as the
> application consuming DPDK has no control over such allocation:
> linking some code does not mean that all of it will be used at runtime.
>
> The general question on whether lcore variables in constructor should
> be forbidden, is left to a later discussion.
>
> For now, this series only focus on fixing subsystems using lcore
> variables so that those allocations are deferred either in rte_eal_init()
> or in the path that does require such lcore variables.

While I have not done code review of this series, I can confirm that
building Open vSwitch against DPDK built with this set of patches
resolves the issue outlined in
https://bugs.launchpad.net/ubuntu/+source/dpdk/+bug/2090931 as in just
linking against DPDK no longer make a program increase its resident
size.

Acked-by: Frode Nordahl <frode.nordahl@canonical.com>

Thanks alot for addressing the issue!

-- 
Frode Nordahl

> --
> David Marchand
>
> Changes since v1:
> - added a check on lcore handle,
> - fixed lcore variable size in lib/power,
> - renamed/introduced allocation helpers,
>
>
> David Marchand (5):
>   eal: check lcore variable handle
>   random: defer seeding to EAL init
>   power: defer lcore variable allocation
>   power: reduce memory footprint of per-lcore state
>   eal/x86: defer power intrinsics variable allocation
>
>  lib/eal/common/eal_private.h       |  6 ++++++
>  lib/eal/common/rte_random.c        |  7 +++++--
>  lib/eal/freebsd/eal.c              |  2 ++
>  lib/eal/include/rte_lcore_var.h    |  2 ++
>  lib/eal/linux/eal.c                |  2 ++
>  lib/eal/windows/eal.c              |  2 ++
>  lib/eal/x86/rte_power_intrinsics.c | 15 +++++++++++++--
>  lib/power/rte_power_pmd_mgmt.c     | 29 ++++++++++++++++++++---------
>  8 files changed, 52 insertions(+), 13 deletions(-)
>
> --
> 2.47.0
>

      parent reply	other threads:[~2024-12-23 10:36 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-12-05 17:57 [PATCH 0/3] " David Marchand
2024-12-05 17:57 ` [PATCH 1/3] random: defer seeding to EAL init David Marchand
2024-12-06 11:09   ` Mattias Rönnblom
2024-12-16  9:38   ` Burakov, Anatoly
2024-12-05 17:57 ` [PATCH 2/3] power: defer lcore variable allocation David Marchand
2024-12-06 11:29   ` Mattias Rönnblom
2024-12-12  7:57     ` David Marchand
2024-12-13  6:58       ` Mattias Rönnblom
2024-12-16 10:02         ` David Marchand
2024-12-05 17:57 ` [PATCH 3/3] eal/x86: defer power intrinsics " David Marchand
2024-12-06 11:32   ` Mattias Rönnblom
2024-12-06 11:01 ` [PATCH 0/3] Defer lcore variables allocation Mattias Rönnblom
2024-12-06 15:55   ` Thomas Monjalon
2024-12-10 17:09     ` Stephen Hemminger
2024-12-09 11:03   ` David Marchand
2024-12-09 15:39     ` Mattias Rönnblom
2024-12-09 17:40       ` David Marchand
2024-12-10  9:41         ` Mattias Rönnblom
2024-12-16 10:01           ` Burakov, Anatoly
2024-12-16  9:42 ` Burakov, Anatoly
2024-12-16  9:49   ` David Marchand
2024-12-17  9:06     ` David Marchand
2024-12-18 20:10     ` Mattias Rönnblom
2024-12-17  8:59 ` [PATCH v2 0/5] " David Marchand
2024-12-17  8:59   ` [PATCH v2 1/5] eal: check lcore variable handle David Marchand
2024-12-18 11:18     ` Burakov, Anatoly
2024-12-17  8:59   ` [PATCH v2 2/5] random: defer seeding to EAL init David Marchand
2024-12-18 16:35     ` Stephen Hemminger
2024-12-18 17:03       ` Mattias Rönnblom
2024-12-17  8:59   ` [PATCH v2 3/5] power: defer lcore variable allocation David Marchand
2024-12-18 11:17     ` Burakov, Anatoly
2024-12-17  8:59   ` [PATCH v2 4/5] power: reduce memory footprint of per-lcore state David Marchand
2024-12-18 11:17     ` Burakov, Anatoly
2024-12-17  8:59   ` [PATCH v2 5/5] eal/x86: defer power intrinsics variable allocation David Marchand
2024-12-18 11:17     ` Burakov, Anatoly
2024-12-19 16:19   ` Frode Nordahl [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='CAKpbOAQsvUwEFt6J=VVakf1Zsuwjc5mQv6f8qZbAvbS9ajozhQ@mail.gmail.com' \
    --to=frode.nordahl@canonical.com \
    --cc=anatoly.burakov@intel.com \
    --cc=david.marchand@redhat.com \
    --cc=dev@dpdk.org \
    --cc=mattias.ronnblom@ericsson.com \
    --cc=thomas@monjalon.net \
    /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).