haproxy/src/xprt_quic.c
Amaury Denoyelle a8968701c0 BUG/MAJOR: quic: complete thread migration before tcp-rules
A quic_conn is instantiated and tied on the first thread which has
received the first INITIAL packet. After handshake completion,
listener_accept() is called. For each quic_conn, a new thread is
selected among the least loaded ones Note that this occurs earlier if
handling 0-RTT data.

This thread connection migration is done in two steps :
* inside listener_accept(), on the origin thread, quic_conn
  tasks/tasklet are killed. After this, no quic_conn related processing
  will occur on this thread. The connection is flagged with
  QUIC_FL_CONN_AFFINITY_CHANGED.
* as soon as the first quic_conn related processing occurs on the new
  thread, the migration is finalized. This allows to allocate the new
  tasks/tasklet directly on the destination thread.

This last step on the new thread must be done prior to other quic_conn
access. There is two events which may trigger it :
* a packet is received on the new thread. In this case,
  qc_finalize_affinity_rebind() is called from quic_dgram_parse().
* the recently accepted connection is popped from accept_queue_ring via
  accept_queue_process(). This will called session_accept_fd() as
  listener.bind_conf.accept callback. This instantiates a new session
  and start connection stack via conn_xprt_start(), which itself calls
  qc_xprt_start() where qc_finalize_affinity_rebind() is used.

A condition was recently found which could cause a closing to be used
with qc_finalize_affinity_rebind() which is forbidden with a BUG_ON().

This lat step was not compatible with layer 4 rule such as "tcp-request
connection reject" which closes the connection early. In this case, most
of the body of session_accept_fd() is skipped, including
qc_xprt_start(), so thread migration is not finalized. At the end of the
function, conn_xprt_close() is then called which flags the connection as
CLOSING.

If a datagram is received for this connection before it is released,
this will call qc_finalize_affinity_rebind() which triggers its BUG_ON()
to prevent thread migration for CLOSING quic_conn.

FATAL: bug condition "qc->flags & ((1U << 29)|(1U << 30))" matched at src/quic_conn.c:2036
Thread 3 "haproxy" received signal SIGILL, Illegal instruction.
[Switching to Thread 0x7ffff794f700 (LWP 2973030)]
0x00005555556221f3 in qc_finalize_affinity_rebind (qc=0x7ffff002d060) at src/quic_conn.c:2036
2036            BUG_ON(qc->flags & (QUIC_FL_CONN_CLOSING|QUIC_FL_CONN_DRAINING));
(gdb) bt
 #0  0x00005555556221f3 in qc_finalize_affinity_rebind (qc=0x7ffff002d060) at src/quic_conn.c:2036
 #1  0x0000555555682463 in quic_dgram_parse (dgram=0x7fff5003ef10, from_qc=0x0, li=0x555555f38670) at src/quic_rx.c:2602
 #2  0x0000555555651aae in quic_lstnr_dghdlr (t=0x555555fc4440, ctx=0x555555fc3f78, state=32832) at src/quic_sock.c:189
 #3  0x00005555558c9393 in run_tasks_from_lists (budgets=0x7ffff7944c90) at src/task.c:596
 #4  0x00005555558c9e8e in process_runnable_tasks () at src/task.c:876
 #5  0x000055555586b7b2 in run_poll_loop () at src/haproxy.c:2966
 #6  0x000055555586be87 in run_thread_poll_loop (data=0x555555d3d340 <ha_thread_info+64>) at src/haproxy.c:3165
 #7  0x00007ffff7b59609 in start_thread () from /lib/x86_64-linux-gnu/libpthread.so.0
 #8  0x00007ffff7a7e133 in clone () from /lib/x86_64-linux-gnu/libc.so.6

To fix this issue, ensure quic_conn migration is completed earlier
inside session_accept_fd(), before any tcp rules processing. This is
done by moving qc_finalize_affinity_rebind() invocation from
qc_xprt_start() to qc_conn_init().

This must be backported up to 2.7.
2023-11-20 16:11:26 +01:00

176 lines
4.7 KiB
C

/*
* QUIC xprt layer. Act as an abstraction between quic_conn and MUX layers.
*
* Copyright 2020 HAProxy Technologies, Frederic Lecaille <flecaille@haproxy.com>
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*
*/
#include <haproxy/api.h>
#include <haproxy/connection.h>
#include <haproxy/quic_conn.h>
#include <haproxy/ssl_sock.h>
#include <haproxy/quic_trace.h>
#include <haproxy/trace.h>
static void quic_close(struct connection *conn, void *xprt_ctx)
{
struct ssl_sock_ctx *conn_ctx = xprt_ctx;
struct quic_conn *qc = conn_ctx->qc;
TRACE_ENTER(QUIC_EV_CONN_CLOSE, qc);
/* Next application data can be dropped. */
qc->mux_state = QC_MUX_RELEASED;
/* If the quic-conn timer has already expired or if already in "connection close"
* state, free the quic-conn.
*/
if (qc->flags & (QUIC_FL_CONN_EXP_TIMER|QUIC_FL_CONN_CLOSING)) {
quic_conn_release(qc);
qc = NULL;
goto leave;
}
/* Schedule a CONNECTION_CLOSE emission. If process stopping is in
* progress, quic-conn idle-timer will be scheduled immediately after
* its emission to ensure an immediate connection closing.
*/
qc_check_close_on_released_mux(qc);
leave:
TRACE_LEAVE(QUIC_EV_CONN_CLOSE, qc);
}
/* Called from the upper layer, to subscribe <es> to events <event_type>. The
* event subscriber <es> is not allowed to change from a previous call as long
* as at least one event is still subscribed. The <event_type> must only be a
* combination of SUB_RETRY_RECV and SUB_RETRY_SEND. It always returns 0.
*/
static int quic_conn_subscribe(struct connection *conn, void *xprt_ctx, int event_type, struct wait_event *es)
{
struct quic_conn *qc = conn->handle.qc;
TRACE_ENTER(QUIC_EV_CONN_SUB, qc);
BUG_ON(event_type & ~(SUB_RETRY_SEND|SUB_RETRY_RECV));
BUG_ON(qc->subs && qc->subs != es);
es->events |= event_type;
qc->subs = es;
/* TODO implement a check_events to detect if subscriber should be
* woken up immediately ?
*/
if (event_type & SUB_RETRY_RECV)
TRACE_DEVEL("subscribe(recv)", QUIC_EV_CONN_XPRTRECV, qc);
if (event_type & SUB_RETRY_SEND)
TRACE_DEVEL("subscribe(send)", QUIC_EV_CONN_XPRTSEND, qc);
TRACE_LEAVE(QUIC_EV_CONN_SUB, qc);
return 0;
}
/* Called from the upper layer, to unsubscribe <es> from events <event_type>.
* The <es> pointer is not allowed to differ from the one passed to the
* subscribe() call. It always returns zero.
*/
static int quic_conn_unsubscribe(struct connection *conn, void *xprt_ctx, int event_type, struct wait_event *es)
{
struct quic_conn *qc = conn->handle.qc;
TRACE_ENTER(QUIC_EV_CONN_SUB, qc);
if (event_type & SUB_RETRY_RECV)
TRACE_DEVEL("unsubscribe(recv)", QUIC_EV_CONN_XPRTRECV, qc);
if (event_type & SUB_RETRY_SEND)
TRACE_DEVEL("unsubscribe(send)", QUIC_EV_CONN_XPRTSEND, qc);
es->events &= ~event_type;
if (!es->events)
qc->subs = NULL;
/* TODO implement ignore_events similar to conn_unsubscribe() ? */
TRACE_LEAVE(QUIC_EV_CONN_SUB, qc);
return 0;
}
/* Store in <xprt_ctx> the context attached to <conn>.
* Returns always 0.
*/
static int qc_conn_init(struct connection *conn, void **xprt_ctx)
{
struct quic_conn *qc = conn->handle.qc;
TRACE_ENTER(QUIC_EV_CONN_NEW, qc);
/* Ensure thread connection migration is finalized ASAP. */
if (qc->flags & QUIC_FL_CONN_AFFINITY_CHANGED)
qc_finalize_affinity_rebind(qc);
/* do not store the context if already set */
if (*xprt_ctx)
goto out;
*xprt_ctx = qc->xprt_ctx;
out:
TRACE_LEAVE(QUIC_EV_CONN_NEW, qc);
return 0;
}
/* Start the QUIC transport layer */
static int qc_xprt_start(struct connection *conn, void *ctx)
{
int ret = 0;
struct quic_conn *qc;
qc = conn->handle.qc;
TRACE_ENTER(QUIC_EV_CONN_NEW, qc);
/* mux-quic can now be considered ready. */
qc->mux_state = QC_MUX_READY;
ret = 1;
out:
TRACE_LEAVE(QUIC_EV_CONN_NEW, qc);
return ret;
}
static struct ssl_sock_ctx *qc_get_ssl_sock_ctx(struct connection *conn)
{
if (!conn || conn->xprt != xprt_get(XPRT_QUIC) || !conn->handle.qc || !conn->xprt_ctx)
return NULL;
return conn->handle.qc->xprt_ctx;
}
/* transport-layer operations for QUIC connections. */
static struct xprt_ops ssl_quic = {
.close = quic_close,
.subscribe = quic_conn_subscribe,
.unsubscribe = quic_conn_unsubscribe,
.init = qc_conn_init,
.start = qc_xprt_start,
.prepare_bind_conf = ssl_sock_prepare_bind_conf,
.destroy_bind_conf = ssl_sock_destroy_bind_conf,
.get_alpn = ssl_sock_get_alpn,
.get_ssl_sock_ctx = qc_get_ssl_sock_ctx,
.name = "QUIC",
};
static void __quic_conn_init(void)
{
xprt_register(XPRT_QUIC, &ssl_quic);
}
INITCALL0(STG_REGISTER, __quic_conn_init);