Date: Mon, 24 Jan 2022 10:47:00 +0100 From: Wojciech Macek <wma@semihalf.com> To: Gleb Smirnoff <glebius@freebsd.org> Cc: Kristof Provost <kp@freebsd.org>, Wojciech Macek <wma@freebsd.org>, src-committers <src-committers@freebsd.org>, "<dev-commits-src-all@freebsd.org>" <dev-commits-src-all@freebsd.org>, dev-commits-src-main@freebsd.org Subject: Re: git: 9ce46cbc95d7 - main - ip_mroute: move ip_mrouter_done outside lock Message-ID: <CANsEV8croaGGMJq9VHSQ70eO4%2BmfCAsEAHhEb-ZGPyJ=kLEEzA@mail.gmail.com> In-Reply-To: <Ye4t7c%2BqWUlyWkbC@FreeBSD.org> References: <202201210519.20L5J1mJ029323@gitrepo.freebsd.org> <7A991003-0F24-4C8C-81FD-84F39FC78D8A@FreeBSD.org> <C25EE57B-BE5A-4BC8-AD00-C8D0D582D073@FreeBSD.org> <CANsEV8cmZcQKh3U8w2S37cg9heFd=icuEmF7dVwrE486OM3Yug@mail.gmail.com> <Ye2FOvi32Xu23qIB@FreeBSD.org> <CANsEV8cEWc9GsTZk%2BVm-_%2B9rsU-=yXLjRkZZz--dg7-ZJRgOnw@mail.gmail.com> <Ye4t7c%2BqWUlyWkbC@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--00000000000046aa1f05d650da99 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable I must have copied it from some line above, now I see you recently refactored that part. Indeed, epoch_assert is called inside ip_input so I see no reason why not to remove RLOCK/RUNLOCK there. Will test it and update once ready. Thanks, Wojtek pon., 24 sty 2022 o 05:41 Gleb Smirnoff <glebius@freebsd.org> napisa=C5=82(= a): > On Sun, Jan 23, 2022 at 07:24:36PM +0100, Wojciech Macek wrote: > W> This is an orginal fix. The issue was visible due to race between ip > W> input/output and mrouter uninit. > W> https://reviews.freebsd.org/D29946 > > Why did you add another net epoch tracker under different name? Just addi= ng > NET_EPOCH_WAIT() to X_ip_mrouter_done() would provide the same effect > without > any modifications to the rest of the stack. > > -- > Gleb Smirnoff > --00000000000046aa1f05d650da99 Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable <div dir=3D"ltr">I must have copied it from some line above, now I see you = recently refactored that part. Indeed, epoch_assert is called inside ip_inp= ut so I see no reason why not to remove RLOCK/RUNLOCK=C2=A0there. Will test= =C2=A0it and update once ready.<div><br></div><div>Thanks,</div><div>Wojtek= </div></div><br><div class=3D"gmail_quote"><div dir=3D"ltr" class=3D"gmail_= attr">pon., 24 sty 2022 o 05:41=C2=A0Gleb Smirnoff <<a href=3D"mailto:gl= ebius@freebsd.org">glebius@freebsd.org</a>> napisa=C5=82(a):<br></div><b= lockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-le= ft:1px solid rgb(204,204,204);padding-left:1ex">On Sun, Jan 23, 2022 at 07:= 24:36PM +0100, Wojciech Macek wrote:<br> W> This is an orginal fix. The issue was visible due to race between ip<= br> W> input/output and mrouter uninit.<br> W> <a href=3D"https://reviews.freebsd.org/D29946" rel=3D"noreferrer" tar= get=3D"_blank">https://reviews.freebsd.org/D29946</a><br> <br> Why did you add another net epoch tracker under different name? Just adding= <br> NET_EPOCH_WAIT() to X_ip_mrouter_done() would provide the same effect witho= ut<br> any modifications to the rest of the stack.<br> <br> -- <br> Gleb Smirnoff<br> </blockquote></div> --00000000000046aa1f05d650da99--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CANsEV8croaGGMJq9VHSQ70eO4%2BmfCAsEAHhEb-ZGPyJ=kLEEzA>