Date: Thu, 23 Mar 2017 17:02:07 +0300 From: "Andrey V. Elsukov" <bu7cher@yandex.ru> To: Mike Karels <mike@karels.net> Cc: Eugene Grosbein <eugen@freebsd.org>, freebsd-net@FreeBSD.org, George Neville-Neil <gnn@freebsd.org>, "Alexander V. Chernikov" <melifaro@freebsd.org>, karels@FreeBSD.org, =?UTF-8?Q?Olivier_Cochard-Labb=c3=a9?= <olivier@cochard.me> Subject: Re: LLE reference leak in the L2 cache Message-ID: <18768a71-3169-469a-f3c3-b9c9e544ff6b@yandex.ru> In-Reply-To: <FC62E82D-4019-4FF8-8EAD-87CC4A369755@karels.net> References: <201703140840.v2E8ecH2040827@mail.karels.net> <3a4c5d87-d42e-5615-5d2b-2a8801376600@yandex.ru> <70D2287B-664C-48E4-9E8B-68B574BE6CE6@karels.net> <FC62E82D-4019-4FF8-8EAD-87CC4A369755@karels.net>
next in thread | previous in thread | raw e-mail | index | archive | help
This is an OpenPGP/MIME signed message (RFC 4880 and 3156) --DCn7JwDqN4diIFwVbARFL3I5WvTEpTkx8 Content-Type: multipart/mixed; boundary="G7SoA0K46AlP3qxWoAJSncMEGO5PC5SqI"; protected-headers="v1" From: "Andrey V. Elsukov" <bu7cher@yandex.ru> To: Mike Karels <mike@karels.net> Cc: Eugene Grosbein <eugen@freebsd.org>, freebsd-net@FreeBSD.org, George Neville-Neil <gnn@freebsd.org>, "Alexander V. Chernikov" <melifaro@freebsd.org>, karels@FreeBSD.org, =?UTF-8?Q?Olivier_Cochard-Labb=c3=a9?= <olivier@cochard.me> Message-ID: <18768a71-3169-469a-f3c3-b9c9e544ff6b@yandex.ru> Subject: Re: LLE reference leak in the L2 cache References: <201703140840.v2E8ecH2040827@mail.karels.net> <3a4c5d87-d42e-5615-5d2b-2a8801376600@yandex.ru> <70D2287B-664C-48E4-9E8B-68B574BE6CE6@karels.net> <FC62E82D-4019-4FF8-8EAD-87CC4A369755@karels.net> In-Reply-To: <FC62E82D-4019-4FF8-8EAD-87CC4A369755@karels.net> --G7SoA0K46AlP3qxWoAJSncMEGO5PC5SqI Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable On 20.03.2017 03:46, Mike Karels wrote: > The context has gotten messy here, so I=E2=80=99m going to break down a= nd top-post. >=20 > I started review https://reviews.freebsd.org/D10059 with a fix for the > reference-count leak. > It changes the semantics so only routes within an in_pcb automatically > do L2 caching. >=20 > I=E2=80=99ll put the tcp_output change for V6 in a separate review when= this one > is done. >=20 > Andrey, could you try your iperf test again? Thanks, Hi Mike, The test with IPv6 works without reference leak now, as supposed, because it doesn't use LLE cache :) For IPv4 forwarding the problem seems also fixed, but I did only basic test. --=20 WBR, Andrey V. Elsukov --G7SoA0K46AlP3qxWoAJSncMEGO5PC5SqI-- --DCn7JwDqN4diIFwVbARFL3I5WvTEpTkx8 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQEzBAEBCAAdFiEE5lkeG0HaFRbwybwAAcXqBBDIoXoFAljT1V8ACgkQAcXqBBDI oXpiRwgAiNCkhLpZ3UmqraduJ4AmlCUD4+AWzlVWf5BnYahI7hdhZwKsLvODqykm IHznT5XB9H2IFQr5dQsYoAuv7c1f5DceioLNhUMrsoBCiC8IO0DEiYpjNGdCrcjz uOa8z52U3ed1z2YobTzy5Y/uVag4ygoHj8fwMjGweg32aqHvcnWBLHo33oIcZpGL l4KbMiFTng1CNr2YtaCyuQcmKIuremBF7UNXUhkCyb3LiUIUVAy3JsysmdU8ky8J p0XN7bylo9eTwD8k4Cen7Af1Lb6+5H+aZUVhxxpradw1VByg8C+2heelbNMZDZxS SmJDx/fWitolP74k0cQJH0wgdleJBg== =M+Rl -----END PGP SIGNATURE----- --DCn7JwDqN4diIFwVbARFL3I5WvTEpTkx8--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?18768a71-3169-469a-f3c3-b9c9e544ff6b>