Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 24 Nov 2022 11:06:14 +0100
From:   Pavel Machek <pavel@ucw.cz>
To:     Yu Kuai <yukuai1@huaweicloud.com>
Cc:     Eric Blake <eblake@redhat.com>,
        kernel list <linux-kernel@vger.kernel.org>,
        josef@toxicpanda.com, linux-block@vger.kernel.org,
        nbd@other.debian.org, "yukuai (C)" <yukuai3@huawei.com>
Subject: Re: nbd: please don't spawn 16 threads when nbd is not even in use
Message-ID: <Y39CFnRwNGBGIsCH@duo.ucw.cz>
References: <Y3y4+QqOlF00X9ET@duo.ucw.cz>
 <20221123200845.cuct5euvikqksojm@redhat.com>
 <Y36YHNVmbozzSdxH@duo.ucw.cz>
 <ccee0ff5-309c-b430-09c7-8d2081c9981d@huaweicloud.com>
MIME-Version: 1.0
Content-Type: multipart/signed; micalg=pgp-sha1;
        protocol="application/pgp-signature"; boundary="HZjYMFTE2RCR7pyM"
Content-Disposition: inline
In-Reply-To: <ccee0ff5-309c-b430-09c7-8d2081c9981d@huaweicloud.com>
Precedence: bulk


--HZjYMFTE2RCR7pyM
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Thu 2022-11-24 09:17:51, Yu Kuai wrote:
> Hi,
>=20
> =E5=9C=A8 2022/11/24 6:01, Pavel Machek =E5=86=99=E9=81=93:
> > Hi!
> >=20
> > > > I see this... and it looks like there are 16 workqueues before nbd =
is
> > > > even used. Surely there are better ways to do that?
> > >=20
> > > Yes, it would be nice to create a pool of workers that only spawns up
> > > threads when actual parallel requests are made.  Are you willing to
> > > help write the patch?
> >=20
> > I was thinking more "only spawn a workqueue when nbd is opened" or so.
> >=20
> > I have 16 of them, and I'm not using nbd. Workqueue per open device is
> > okay, current situation... not so much.
>=20
> You can take a look at this commit:
>=20
> e2daec488c57 ("nbd: Fix hungtask when nbd_config_put")
>=20
> Allocation of recv_workq is moved from start device to alloc device to
> fix hungtask. You might need to be careful if you want to move this.

Can we get that reverted?

That is rather obscure bug (how many GFP_KERNEL failures do you
normally see?) and it costs, dunno, 100KB? of unswappable memory.

Best regards,
								Pavel
--=20
People of Russia, stop Putin before his war on Ukraine escalates.

--HZjYMFTE2RCR7pyM
Content-Type: application/pgp-signature; name="signature.asc"

-----BEGIN PGP SIGNATURE-----

iF0EABECAB0WIQRPfPO7r0eAhk010v0w5/Bqldv68gUCY39CFgAKCRAw5/Bqldv6
8vWYAJ9yQ6drFnNpEutVeHvEesUcP4FckgCgvgMVBq1JANQNm02eRF+IlRzW0uc=
=IyCl
-----END PGP SIGNATURE-----

--HZjYMFTE2RCR7pyM--