From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:49547) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ft7iO-00009J-CB for qemu-devel@nongnu.org; Fri, 24 Aug 2018 04:45:09 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ft7iK-0004KN-N3 for qemu-devel@nongnu.org; Fri, 24 Aug 2018 04:45:08 -0400 Received: from mx3-rdu2.redhat.com ([66.187.233.73]:50520 helo=mx1.redhat.com) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1ft7iI-0004Hi-Nh for qemu-devel@nongnu.org; Fri, 24 Aug 2018 04:45:02 -0400 Received: from smtp.corp.redhat.com (int-mx06.intmail.prod.int.rdu2.redhat.com [10.11.54.6]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id E01A9402315B for ; Fri, 24 Aug 2018 08:45:01 +0000 (UTC) Date: Fri, 24 Aug 2018 16:44:54 +0800 From: Peter Xu Message-ID: <20180824084454.GA3020@xz-x1> References: <20180823143125.16767-1-marcandre.lureau@redhat.com> <20180824034517.GF24044@xz-x1> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [PATCH v2 0/5] Fix socket chardev regression List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: =?utf-8?Q?Marc-Andr=C3=A9?= Lureau Cc: qemu-devel , Paolo Bonzini , "P. Berrange, Daniel" On Fri, Aug 24, 2018 at 10:32:58AM +0200, Marc-Andr=C3=A9 Lureau wrote: > Hi >=20 > On Fri, Aug 24, 2018 at 5:45 AM, Peter Xu wrote: > > On Thu, Aug 23, 2018 at 04:31:20PM +0200, Marc-Andr=C3=A9 Lureau wrot= e: > >> Hi, > >> > >> In commit 25679e5d58e "chardev: tcp: postpone async connection setup= " > >> (and its follow up 99f2f54174a59), Peter moved chardev socket > >> connection to machine_done event. However, chardev created later wil= l > >> no longer attempt to connect, and chardev created in tests do not ha= ve > >> machine_done event (breaking some of vhost-user-test). > >> > >> The goal was to move the "connect" source to the chardev frontend > >> context (the monitor thread context in his case). chr->gcontext is s= et > >> with qemu_chr_fe_set_handlers(). But there is no guarantee that the > >> function will be called in general, so we can't delay connection unt= il > >> then: the chardev should still attempt to connect during open(), usi= ng > >> the main context. > >> > >> An alternative would be to specify the iothread during chardev > >> creation. Setting up monitor OOB would be quite different too, it > >> would take the same iothread as argument. > >> > >> 99f2f54174a595e is also a bit problematic, since it will behave > >> differently before and after machine_done (the first case gives a > >> chance to use a different context reliably, the second looks racy) > >> > >> In the end, I am not sure this is all necessary, as chardev callback= s > >> are called after qemu_chr_fe_set_handlers(), at which point the > >> context of sources are updated. In "char-socket: update all ioc > >> handlers when changing context", I moved also the hup handler to the > >> updated context. So unless the main thread is already stuck, we can > >> setup a different context for the chardev at that time. Or not? > >> > >> v2: > >> - fix a random socket chardev test failure > > > > I have no problem on patch 3-5, but aren't patch 1-2 still breaking > > the context switch of chardev? Or did I misunderstood? >=20 > Switching context is broken/racy regardless. My current plan is to > suspend the context to switch to, and update all sources before > thawing the target context. >=20 > If you prefer we can delay the revert, but I would rather have them > because the regression seems much worse to me. Have you started working on above? I started to code up some RFC patch to let the chardev maintain the iothreads (now it'll only have a monitor iothread) then we don't need to switch context for chardev any more. I plan to post it today for a quick look. With that, this series will not break anything then. >=20 > > > > Regards, > > > > -- > > Peter Xu Regards, --=20 Peter Xu