From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.133.124]) by passt.top (Postfix) with ESMTP id 70F6A5A026F for ; Wed, 6 Dec 2023 16:00:29 +0100 (CET) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1701874828; 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: in-reply-to:in-reply-to:references:references; bh=/a9JGl2dN+MwFrxa2wd06k/8Kc0vwIkLAK6VCKXa/Lk=; b=Zi24ecHqwf5btvPGFz1kvyiuhRoD0rowAq0wkUHB9UtAk7hL1+EPZRZMibs8H4VGB3l6Ty cKZUqpBD7GiIxjVQ9KcXHG/lekk+VitbnWgQTyO6YRf4jfTgvBz3SxXnuo8gW07db/EjXa 1BvKCeSv2MY+M/2GaXyvl6phG1sxSNQ= Received: from mimecast-mx02.redhat.com (mx-ext.redhat.com [66.187.233.73]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.3, cipher=TLS_AES_256_GCM_SHA384) id us-mta-284-0Pk7SdR7Nh-9nhQQi6bkSg-1; Wed, 06 Dec 2023 10:00:23 -0500 X-MC-Unique: 0Pk7SdR7Nh-9nhQQi6bkSg-1 Received: from smtp.corp.redhat.com (int-mx08.intmail.prod.int.rdu2.redhat.com [10.11.54.8]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (2048 bits) server-digest SHA256) (No client certificate requested) by mimecast-mx02.redhat.com (Postfix) with ESMTPS id 7B3631C18CCF for ; Wed, 6 Dec 2023 15:00:23 +0000 (UTC) Received: from elisabeth (unknown [10.39.208.4]) by smtp.corp.redhat.com (Postfix) with ESMTPS id 2E022C185A5; Wed, 6 Dec 2023 15:00:21 +0000 (UTC) Date: Wed, 6 Dec 2023 15:59:40 +0100 From: Stefano Brivio To: Jon Maloy Subject: Re: tcp.c: leverage MSG_PEEK with offset kernel capability when available Message-ID: <20231206155940.51047ac1@elisabeth> In-Reply-To: <20231205233604.1491317-1-jmaloy@redhat.com> References: <20231205233604.1491317-1-jmaloy@redhat.com> Organization: Red Hat MIME-Version: 1.0 X-Scanned-By: MIMEDefang 3.4.1 on 10.11.54.8 X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Type: multipart/mixed; boundary="MP_/su5PNohzewmN8RFVGr3uH6w" Message-ID-Hash: DSYPVCPVQNUV2CZLTWYXMN7AJSYDW6DJ X-Message-ID-Hash: DSYPVCPVQNUV2CZLTWYXMN7AJSYDW6DJ 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: lvivier@redhat.com, dgibson@redhat.com, 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: --MP_/su5PNohzewmN8RFVGr3uH6w Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Disposition: inline On Tue, 5 Dec 2023 18:36:04 -0500 Jon Maloy wrote: > The kernel may support recvmsg(MSG_PEEK), starting from a given offset. This > makes it possible to avoid repeated reading of already read initial bytes of > a received message, hence saving us read cycles when forwarding TCP messages > in the host->name space direction. > > When this feature is supported, iov_sock[0].iov_base can be set to NULL. The > kernel code will interpret this as an instruction to skip reading of the first > iov_sock[0].iov_len bytes of the message. > > Since iov_sock[0].iov_base is set to point to tcp_buf_discard, we do this > by simply not allocating that buffer, letting the pointer remain NULL, when > we find that we have this kernel support. > > There is no macro or function indicating kernel support for this feature. We > therefore have to probe for it by reading from an established TCP connection. > The traffic connection cannot be used for this purpose, because it will be > broken if the probe reading fails. We therefore have to create a temporary > local connection for this task. Because of this, we also add a new function, > tcp_probe_msg_peek_offset_cap(), which creates this temporary connection > and performs the probe read on it. > > Signed-off-by: Jon Maloy > --- > tcp.c | 110 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++-- > 1 file changed, 108 insertions(+), 2 deletions(-) > > diff --git a/tcp.c b/tcp.c > index f506cfd..ab5168e 100644 > --- a/tcp.c > +++ b/tcp.c > @@ -402,6 +402,8 @@ struct tcp6_l2_head { /* For MSS6 macro: keep in sync with tcp6_l2_buf_t */ > (conn->events & (SOCK_FIN_RCVD | TAP_FIN_RCVD))) > #define CONN_HAS(conn, set) ((conn->events & (set)) == (set)) > > +static bool tcp_probe_msg_peek_offset_cap(); No need for a forward declaration, I guess. > + > static const char *tcp_event_str[] __attribute((__unused__)) = { > "SOCK_ACCEPTED", "TAP_SYN_RCVD", "ESTABLISHED", "TAP_SYN_ACK_SENT", > > @@ -506,7 +508,8 @@ static struct tcp_buf_seq_update tcp6_l2_buf_seq_update[TCP_FRAMES_MEM]; > static unsigned int tcp6_l2_buf_used; > > /* recvmsg()/sendmsg() data for tap */ > -static char tcp_buf_discard [MAX_WINDOW]; > +static char *tcp_buf_discard = NULL; > + > static struct iovec iov_sock [TCP_FRAMES_MEM + 1]; > > static struct iovec tcp4_l2_iov [TCP_FRAMES_MEM]; > @@ -573,6 +576,15 @@ static unsigned int tcp6_l2_flags_buf_used; > > #define CONN(idx) (&(FLOW(idx)->tcp)) > > + > +/** msg_peek_offset_cap() - Does the kernel support recvmsg(MSG_PEEK) with offset? > + */ > +static inline bool msg_peek_offset_cap() > +{ > + return !tcp_buf_discard; > +} > + > + > /** conn_at_idx() - Find a connection by index, if present > * @idx: Index of connection to lookup > * > @@ -2224,7 +2236,9 @@ static int tcp_data_from_sock(struct ctx *c, struct tcp_tap_conn *conn) > return 0; > } > > - sendlen = len - already_sent; > + sendlen = len; > + if (!msg_peek_offset_cap()) > + sendlen -= already_sent; > if (sendlen <= 0) { > conn_flag(c, conn, STALLED); > return 0; > @@ -3107,6 +3121,15 @@ int tcp_init(struct ctx *c) > NS_CALL(tcp_ns_socks_init, c); > } > > + /* Only allocate discard buffer if needed */ > + if (!tcp_probe_msg_peek_offset_cap()) { > + tcp_buf_discard = malloc(MAX_WINDOW); I would rather not use the heap at all, even though after commit 0515adceaa8f ("passt, pasta: Namespace-based sandboxing, defer seccomp policy application") we don't ask seccomp to terminate the process if we call a sbrk() (or similar) in this phase. The only specific issue I have in mind is rather minor, that is, at the moment we can reliably calculate our memory footprint using nm(1). But in general, having a single set of memory addresses keep things simpler and probably a bit safer. This would be the only non-static memory we use, and I don't see a strong case for it. I would rather drop this buffer after a few months (in turn, if/after the kernel change is accepted), turning the detection into a build-time step, with passt failing if we find that we don't have this buffer, and we were built for a kernel with support for MSG_PEEK with offset. > + if (!tcp_buf_discard) { > + perror("failed to allocate discard buffer\n"); > + exit(EXIT_FAILURE); > + } > + debug("allocated discard buffer of size %i\n", MAX_WINDOW); > + } > return 0; > } > > @@ -3213,3 +3236,86 @@ void tcp_timer(struct ctx *c, const struct timespec *ts) > if (c->mode == MODE_PASTA) > tcp_splice_refill(c); > } > + > +/** tcp_probe_msg_peek_offset_cap() - Probe kernel for MSG_PEEK with offset support > + */ > +static bool tcp_probe_msg_peek_offset_cap() For consistency: (void). I have two main criticisms to this approach: 1. it uses fork() (and that's another usage of heap memory) but we don't actually need connect() and send() to be synchronous for this test, and 2. we bind an actual TCP port where we run. I attached a sketch (pkt_selfie.c) of a slightly different approach that solves 1. by using a non-blocking socket on the client side, and solves 2. by creating the pair of sockets in a detached network namespace, which is essentially invisible and goes away once we're done with the probing. For some reason, the negative case works: $ gcc -o pkt_selfie pkt_selfie.c; strace -f ./pkt_selfie [...] sendto(5, "ab", 2, 0, NULL, 0) = 2 recvmsg(6, {msg_name=0x7ffd2ba5d130, msg_namelen=2 => 0, msg_iov=NULL, msg_iovlen=0, msg_controllen=0, msg_flags=0}, MSG_PEEK) = 0 but on a kernel with your patch, I get ENOTCONN on recvmsg(). If I replace that by a simple recv(): sendto(5, "ab", 2, 0, NULL, 0) = 2 recvfrom(6, "ab", 10, 0, NULL, NULL) = 2 ...so I don't think it's a fundamental issue with this approach, rather something with your patch, but I'm not yet sure what. :) Most of pkt_selfie.c is copied and pasted (with minimal adaptations) from existing passt's codebase, the actual implementation starts at line 107. Of course it's missing all the error checks etc. -- Stefano --MP_/su5PNohzewmN8RFVGr3uH6w Content-Type: text/x-c++src Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename=pkt_selfie.c #define _GNU_SOURCE #include #include #include #include #include #include #include #include #include #include #include #include #include /* ===> from passt's Makefile and code... */ #define RLIMIT_STACK_VAL 8192 #define NS_FN_STACK_SIZE (RLIMIT_STACK_VAL * 1024 / 8) int do_clone(int (*fn)(void *), char *stack_area, size_t stack_size, int flags, void *arg) { #ifdef __ia64__ return __clone2(fn, stack_area + stack_size / 2, stack_size / 2, flags, arg); #else return clone(fn, stack_area + stack_size / 2, flags, arg); #endif } static int nl_sock; static int nl_sock_init_do(void *arg) { struct sockaddr_nl addr = { .nl_family = AF_NETLINK, }; int *s = &nl_sock; #ifdef NETLINK_GET_STRICT_CHK int y = 1; #endif *s = socket(AF_NETLINK, SOCK_RAW | SOCK_CLOEXEC, NETLINK_ROUTE); if (*s < 0 || bind(*s, (struct sockaddr *)&addr, sizeof(addr))) { *s = -1; return 0; } return 0; } /** * nl_send() - Prepare and send netlink request * @s: Netlink socket * @req: Request (will fill netlink header) * @type: Request type * @flags: Extra request flags (NLM_F_REQUEST and NLM_F_ACK assumed) * @len: Request length * * Return: sequence number of request on success, terminates on error */ static uint32_t nl_send(int s, void *req, uint16_t type, uint16_t flags, ssize_t len) { struct nlmsghdr *nh; ssize_t n; nh = (struct nlmsghdr *)req; nh->nlmsg_type = type; nh->nlmsg_flags = NLM_F_REQUEST | NLM_F_ACK | flags; nh->nlmsg_len = len; nh->nlmsg_seq = 1; nh->nlmsg_pid = 0; n = send(s, req, len, 0); return nh->nlmsg_seq; } int nl_link_up(int s, unsigned int ifi, int mtu) { struct req_t { struct nlmsghdr nlh; struct ifinfomsg ifm; struct rtattr rta; unsigned int mtu; } req = { .ifm.ifi_family = AF_UNSPEC, .ifm.ifi_index = ifi, .ifm.ifi_flags = IFF_UP, .ifm.ifi_change = IFF_UP, .rta.rta_type = IFLA_MTU, .rta.rta_len = RTA_LENGTH(sizeof(unsigned int)), .mtu = mtu, }; ssize_t len = sizeof(req); if (!mtu) /* Shorten request to drop MTU attribute */ len = offsetof(struct req_t, rta); return nl_send(s, &req, RTM_NEWLINK, 0, len); /* was nl_do() */ } /* <=== ...until here */ static int tcp_probe_sockets(void *arg) { int *s = (int *)arg; nl_sock_init_do(NULL); nl_link_up(nl_sock, 1 /* lo */, 0); s[0] = socket(AF_INET, SOCK_STREAM, IPPROTO_TCP); s[1] = socket(AF_INET, SOCK_STREAM | SOCK_NONBLOCK, IPPROTO_TCP); return 0; } int main(int argc, char **argv) { char ns_fn_stack[NS_FN_STACK_SIZE], b; struct iovec iov[2] = { { NULL, 1 }, { &b, 1 }, }; struct sockaddr a = { AF_INET, htons(6666), }; struct msghdr msg = { iov, 2 }; int s[2], s_nl, s_recv; ssize_t len; char buf[10]; do_clone(tcp_probe_sockets, ns_fn_stack, sizeof(ns_fn_stack), CLONE_NEWNET | CLONE_NEWUSER | CLONE_VM | CLONE_VFORK | CLONE_FILES | SIGCHLD, (void *)s); bind(s[0], &a, sizeof(a)); getsockname(s[0], &a, &((int){ sizeof(a) })); listen(s[0], 0); connect(s[1], &a, sizeof(a)); s_recv = accept(s[0], NULL, NULL); send(s[1], (char *)("ab"), 2, 0); len = recvmsg(s_recv, &msg, MSG_PEEK); printf("MSG_PEEK with offset %ssupported\n", len == 1 ? "" : "not "); close(s_recv); close(s[1]); close(s[0]); return 0; } --MP_/su5PNohzewmN8RFVGr3uH6w--