From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.129.124]) by passt.top (Postfix) with ESMTP id 90C0A5A0272 for ; Thu, 10 Aug 2023 09:51:02 +0200 (CEST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1691653861; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=f8tReVvkmATI+EOnTlovBSpuaIXTw3KhYJpaJBfRw6o=; b=PSogFkbosZjwmRQrFO/adydvYX3nWt5El3p17tZfT+8aaKBzDW1S8/R7bKeL4NXTRWZQ0I aU+C5WBVFcjawOke8ePlVGPgbFJU5dWMxZ6mFem50noY2NoVASL90LqrickFUYbZsCdQz3 a4GRZGdyenX3HkjlT0HVwdznG3aAs3U= Received: from mimecast-mx02.redhat.com (66.187.233.73 [66.187.233.73]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id us-mta-619-RV_zXncrN4GupmVs4JHMsA-1; Thu, 10 Aug 2023 03:50:58 -0400 X-MC-Unique: RV_zXncrN4GupmVs4JHMsA-1 Received: from smtp.corp.redhat.com (int-mx07.intmail.prod.int.rdu2.redhat.com [10.11.54.7]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mimecast-mx02.redhat.com (Postfix) with ESMTPS id EBAB11C01B38; Thu, 10 Aug 2023 07:50:57 +0000 (UTC) Received: from elisabeth (unknown [10.39.208.5]) by smtp.corp.redhat.com (Postfix) with ESMTPS id 50D12140E914; Thu, 10 Aug 2023 07:50:56 +0000 (UTC) Date: Thu, 10 Aug 2023 09:50:55 +0200 From: Stefano Brivio To: David Gibson Subject: Re: [PATCH 8/9] epoll: Split listening Unix domain socket into its own type Message-ID: <20230810095055.421e2636@elisabeth> In-Reply-To: References: <20230807134631.1400119-1-david@gibson.dropbear.id.au> <20230807134631.1400119-9-david@gibson.dropbear.id.au> <20230809215927.006be951@elisabeth> Organization: Red Hat MIME-Version: 1.0 X-Scanned-By: MIMEDefang 3.1 on 10.11.54.7 X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Message-ID-Hash: M2KT3ZUVBTZUJOPQ55NTPE5CCFABWIVR X-Message-ID-Hash: M2KT3ZUVBTZUJOPQ55NTPE5CCFABWIVR X-MailFrom: sbrivio@redhat.com X-Mailman-Rule-Misses: dmarc-mitigation; no-senders; approved; emergency; loop; banned-address; member-moderation; nonmember-moderation; administrivia; implicit-dest; max-recipients; max-size; news-moderation; no-subject; digests; suspicious-header CC: passt-dev@passt.top X-Mailman-Version: 3.3.8 Precedence: list List-Id: Development discussion and patches for passt Archived-At: Archived-At: List-Archive: List-Archive: List-Help: List-Owner: List-Post: List-Subscribe: List-Unsubscribe: On Thu, 10 Aug 2023 11:08:19 +1000 David Gibson wrote: > On Wed, Aug 09, 2023 at 09:59:27PM +0200, Stefano Brivio wrote: > > On Mon, 7 Aug 2023 23:46:30 +1000 > > David Gibson wrote: > > > > > tap_handler() actually handles events on three different types of object: > > > the /dev/tap character device (pasta), a connected Unix domain socket > > > (passt) or a listening Unix domain socket (passt). > > > > > > The last, in particular, really has no handling in common with the others, > > > so split it into its own epoll type and directly dispatch to the relevant > > > handler from the top level. > > > > > > Signed-off-by: David Gibson > > > --- > > > passt.c | 6 +++++- > > > passt.h | 6 ++++-- > > > tap.c | 20 ++++++++------------ > > > tap.h | 4 ++-- > > > 4 files changed, 19 insertions(+), 17 deletions(-) > > > > > > diff --git a/passt.c b/passt.c > > > index c32981d..c60f346 100644 > > > --- a/passt.c > > > +++ b/passt.c > > > @@ -64,6 +64,7 @@ char *epoll_type_str[EPOLL_TYPE_MAX+1] = { > > > [EPOLL_TYPE_ICMPV6] = "ICMPv6 socket", > > > [EPOLL_TYPE_NSQUIT] = "namespace inotify", > > > [EPOLL_TYPE_TAP] = "tap device", > > > + [EPOLL_TYPE_TAP_LISTEN] = "listening qemu socket", > > > }; > > > > > > /** > > > @@ -317,7 +318,10 @@ loop: > > > > > > switch (ref.type) { > > > case EPOLL_TYPE_TAP: > > > - tap_handler(&c, ref.fd, events[i].events, &now); > > > + tap_handler(&c, events[i].events, &now); > > > + break; > > > + case EPOLL_TYPE_TAP_LISTEN: > > > + tap_listen_handler(&c, eventmask); > > > break; > > > case EPOLL_TYPE_NSQUIT: > > > pasta_netns_quit_handler(&c, quit_fd); > > > diff --git a/passt.h b/passt.h > > > index 176bc85..7dae08b 100644 > > > --- a/passt.h > > > +++ b/passt.h > > > @@ -61,10 +61,12 @@ enum epoll_type { > > > EPOLL_TYPE_ICMPV6, > > > /* inotify fd watching for end of netns (pasta) */ > > > EPOLL_TYPE_NSQUIT, > > > - /* tap char device, or qemu socket fd */ > > > + /* tap char device, or connected qemu socket fd */ > > > EPOLL_TYPE_TAP, > > > + /* socket listening for qemu socket connections */ > > > + EPOLL_TYPE_TAP_LISTEN, > > > > > > - EPOLL_TYPE_MAX = EPOLL_TYPE_TAP, > > > + EPOLL_TYPE_MAX = EPOLL_TYPE_TAP_LISTEN, > > > }; > > > > > > /** > > > diff --git a/tap.c b/tap.c > > > index ad0decf..8468f86 100644 > > > --- a/tap.c > > > +++ b/tap.c > > > @@ -1076,7 +1076,7 @@ restart: > > > static void tap_sock_unix_init(struct ctx *c) > > > { > > > int fd = socket(AF_UNIX, SOCK_STREAM, 0); > > > - union epoll_ref ref = { .type = EPOLL_TYPE_TAP }; > > > + union epoll_ref ref = { .type = EPOLL_TYPE_TAP_LISTEN }; > > > struct epoll_event ev = { 0 }; > > > struct sockaddr_un addr = { > > > .sun_family = AF_UNIX, > > > @@ -1142,10 +1142,11 @@ static void tap_sock_unix_init(struct ctx *c) > > > } > > > > > > /** > > > - * tap_sock_unix_new() - Handle new connection on listening socket > > > + * tap_listen_handler() - Handle new connection on listening socket > > > * @c: Execution context > > > + * @events: epoll events on the socket > > > */ > > > -static void tap_sock_unix_new(struct ctx *c) > > > +void tap_listen_handler(struct ctx *c, uint32_t events) > > > { > > > union epoll_ref ref = { .type = EPOLL_TYPE_TAP }; > > > struct epoll_event ev = { 0 }; > > > @@ -1153,6 +1154,9 @@ static void tap_sock_unix_new(struct ctx *c) > > > struct ucred ucred; > > > socklen_t len; > > > > > > + if (events != EPOLLIN) > > > + return; > > > + > > > > This comment actually belongs to 2/4 of the tap reset clean-up series, > > but posting it here for clarity: ...wouldn't it be appropriate to > > handle errors while at it? Otherwise I guess we'll just spin on > > EPOLLERR or EPOLLRDHUP. > > So, I don't think we'll spin, because we set EPOLLET (edge > triggered). That said, EPOLLRDHUP makes no sense on a listening > socket, and we're not subscribed to EPOLLERR There's no need to subscribe to EPOLLERR (and EPOLLHUP): both types of events are always reported. I guess it's also fine to indicate it explicitly in 2/13 v2 as you did, to remark that we handle it, but if you look around we never add EPOLLERR or EPOLLHUP (except for a stray occurrence in udp_splice_new()) to the set of events. -- Stefano