- dev - passt

Re: [PATCH] Add missing includes to headers
by Stefano Brivio 25 Feb '26

25 Feb '26

On Tue, 24 Feb 2026 17:41:28 -0500 Peter Foley <pefoley(a)google.com> wrote: > On Tue, Feb 24, 2026 at 4:18 PM Stefano Brivio <sbrivio(a)redhat.com> wrote: > > > > > > I'm inclined to forget about trying to fix everything and just do the > > > minimal set of changes to make the tooling happy with parsing the > > headers. > > > > Yeah, at this point I'd totally support that. > > Does the most recent patch I sent look good to apply? > That should be the "minimal" include happiness version. Ah, right, v2 would do that... I think so, I still need to have a look in detail, test against musl and on a couple of distributions, but it looks fine so far. Give me a bit of time, I'll keep you posted. -- Stefano

2 1

[PATCH trivial] tcp_vu, udp_vu: Fix comment headers for header length functions
by Laurent Vivier 24 Feb '26

24 Feb '26

The comment headers for tcp_vu_hdrlen() and udp_vu_hdrlen() described the return value as "the size of the header in level 2 frame", but these functions also include the virtio net header, which is not part of the L2 frame. Signed-off-by: Laurent Vivier <lvivier(a)redhat.com> --- tcp_vu.c | 4 ++-- udp_vu.c | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/tcp_vu.c b/tcp_vu.c index 94169c21f700..8908d376a7dd 100644 --- a/tcp_vu.c +++ b/tcp_vu.c @@ -40,10 +40,10 @@ static struct vu_virtq_element elem[VIRTQUEUE_MAX_SIZE]; static int head[VIRTQUEUE_MAX_SIZE + 1]; /** - * tcp_vu_hdrlen() - return the size of the header in level 2 frame (TCP) + * tcp_vu_hdrlen() - Return the size of virtio net and L2 headers (TCP) * @v6: Set for IPv6 packet * - * Return: return the size of the header + * Return: total size of virtio net header, Ethernet, IP and TCP headers */ static size_t tcp_vu_hdrlen(bool v6) { diff --git a/udp_vu.c b/udp_vu.c index d190fef82d49..dd3f8e9272ff 100644 --- a/udp_vu.c +++ b/udp_vu.c @@ -37,10 +37,10 @@ static struct iovec iov_vu [VIRTQUEUE_MAX_SIZE]; static struct vu_virtq_element elem [VIRTQUEUE_MAX_SIZE]; /** - * udp_vu_hdrlen() - return the size of the header in level 2 frame (UDP) + * udp_vu_hdrlen() - Return the size of virtio net and L2 headers (UDP) * @v6: Set for IPv6 packet * - * Return: return the size of the header + * Return: total size of virtio net header, Ethernet, IP and UDP headers */ static size_t udp_vu_hdrlen(bool v6) { -- 2.53.0

3 3

[PATCH v11] conf: Support CIDR notation for -a/--address option
by Jon Maloy 24 Feb '26

24 Feb '26

We extend the -a/--address option to accept addresses in CIDR notation (e.g., 192.168.1.1/24 or 2001:db8::1/64) as an alternative to using separate -a and -n options. We add a new inany_prefix_pton() helper function that: - Parses address strings with a compulsory /prefix_len suffix - Validates prefix length based on address family (0-32 for IPv4, 0-128 for IPv6), including handling of IPv4-to-IPv6 mapping case. For IPv4, the prefix length is stored in ip4.prefix_len when provided. For IPv6, the given prefix length is still overridden by the default value 64 Mixing -n and CIDR notation results in an error to catch likely user mistakes. Also fix a bug in conf_ip4_prefix() that was incorrectly using the global 'optarg' instead of its 'arg' parameter. Signed-off-by: Jon Maloy <jmaloy(a)redhat.com> --- v3: Fixes after feedback from Laurent, David and Stefano Notably, updated man page for the -a option v4: Fixes based on feedback from David G: - Handling prefix length adjustment when IPv4-to-IPv6 mapping - Removed redundant !IN6_IS_ADDR_V4MAPPED(&addr.a6) test - Simplified tests of acceptable address types - Merged documentation and code commits - Some documentation text clarifications v5: - Moved address/prefix parsing into a refactored inany_prefix_pton() function. - inany_prefix_pton() now only caluclates IPv6 style prefix lengths - Stricter distinction between error causes. - Some refactoring of the 'case a:' branch in conf() - Some small fixes in passt.1 v6: - Refactored inany_prefix_pton() and conf()::'case -a' code after input from David Gibson. v7: - More refactoring after input from David Gibson. - I kept the return values 1 and 0. This is consistent with the return values of inet_pton() and inany_pton(). v8: - Changed condition for updating ipv4 prefix length v9: - Made char *src and char *pstr in inany_prefix_pton() const - Updated logics in conf.c:: case 'a' and case 'n' to be clearer (I think) v10: -More (final?) updates after feedback from David and Stefano v11: - Updated man pages to document that prefix length attached to IPv6 address are still ignored. - Re-added zero-initialization of stack array char astr[], to guarantee that we get a zero-terminated string after the strncpy() call. --- conf.c | 72 +++++++++++++++++++++++++++++++++++++-------------------- inany.c | 50 +++++++++++++++++++++++++++++++++++++++ inany.h | 15 ++++++++++++ ip.c | 21 +++++++++++++++++ ip.h | 2 ++ passt.1 | 19 +++++++++++---- 6 files changed, 149 insertions(+), 30 deletions(-) diff --git a/conf.c b/conf.c index 2942c8c..db94f92 100644 --- a/conf.c +++ b/conf.c @@ -682,7 +682,7 @@ static int conf_ip4_prefix(const char *arg) return -1; } else { errno = 0; - len = strtoul(optarg, NULL, 0); + len = strtoul(arg, NULL, 0); if (len > 32 || errno) return -1; } @@ -896,7 +896,7 @@ static void usage(const char *name, FILE *f, int status) " a zero value disables assignment\n" " default: 65520: maximum 802.3 MTU minus 802.3 header\n" " length, rounded to 32 bits (IPv4 words)\n" - " -a, --address ADDR Assign IPv4 or IPv6 address ADDR\n" + " -a, --address ADDR Assign IPv4 or IPv6 address ADDR[/PREFIXLEN]\n" " can be specified zero to two times (for IPv4 and IPv6)\n" " default: use addresses from interface with default route\n" " -n, --netmask MASK Assign IPv4 MASK, dot-decimal or bits\n" @@ -1504,6 +1504,8 @@ void conf(struct ctx *c, int argc, char **argv) unsigned dns4_idx = 0, dns6_idx = 0; unsigned long max_mtu = IP_MAX_MTU; struct fqdn *dnss = c->dns_search; + bool addr_has_prefix_len = false; + uint8_t prefix_len_from_opt = 0; unsigned int ifi4 = 0, ifi6 = 0; const char *logfile = NULL; size_t logsize = 0; @@ -1808,36 +1810,56 @@ void conf(struct ctx *c, int argc, char **argv) c->mtu = mtu; break; } - case 'a': - if (inet_pton(AF_INET6, optarg, &c->ip6.addr) && - !IN6_IS_ADDR_UNSPECIFIED(&c->ip6.addr) && - !IN6_IS_ADDR_LOOPBACK(&c->ip6.addr) && - !IN6_IS_ADDR_V4MAPPED(&c->ip6.addr) && - !IN6_IS_ADDR_V4COMPAT(&c->ip6.addr) && - !IN6_IS_ADDR_MULTICAST(&c->ip6.addr)) { - if (c->mode == MODE_PASTA) - c->ip6.no_copy_addrs = true; - break; - } + case 'a': { + union inany_addr addr; + uint8_t prefix_len; + + addr_has_prefix_len = inany_prefix_pton(optarg, &addr, + &prefix_len); + + if (addr_has_prefix_len && prefix_len_from_opt) + die("Redundant prefix length specification"); + + if (!addr_has_prefix_len && !inany_pton(optarg, &addr)) + die("Invalid address: %s", optarg); + + if (prefix_len_from_opt && inany_v4(&addr)) + prefix_len = prefix_len_from_opt; + else if (!addr_has_prefix_len) + prefix_len = inany_default_prefix_len(&addr); + + if (inany_is_unspecified(&addr) || + inany_is_multicast(&addr) || + inany_is_loopback(&addr) || + IN6_IS_ADDR_V4COMPAT(&addr.a6)) + die("Invalid address: %s", optarg); - if (inet_pton(AF_INET, optarg, &c->ip4.addr) && - !IN4_IS_ADDR_UNSPECIFIED(&c->ip4.addr) && - !IN4_IS_ADDR_BROADCAST(&c->ip4.addr) && - !IN4_IS_ADDR_LOOPBACK(&c->ip4.addr) && - !IN4_IS_ADDR_MULTICAST(&c->ip4.addr)) { + if (inany_v4(&addr)) { + c->ip4.addr = *inany_v4(&addr); + c->ip4.prefix_len = prefix_len - 96; if (c->mode == MODE_PASTA) c->ip4.no_copy_addrs = true; - break; + } else { + c->ip6.addr = addr.a6; + if (c->mode == MODE_PASTA) + c->ip6.no_copy_addrs = true; } - - die("Invalid address: %s", optarg); break; - case 'n': - c->ip4.prefix_len = conf_ip4_prefix(optarg); - if (c->ip4.prefix_len < 0) - die("Invalid netmask: %s", optarg); + } + case 'n': { + int plen; + + if (addr_has_prefix_len) + die("Redundant prefix length specification"); + plen = conf_ip4_prefix(optarg); + if (plen < 0) + die("Invalid prefix length: %s", optarg); + + prefix_len_from_opt = plen + 96; + c->ip4.prefix_len = plen; break; + } case 'M': parse_mac(c->our_tap_mac, optarg); break; diff --git a/inany.c b/inany.c index 7680439..aa5e6c7 100644 --- a/inany.c +++ b/inany.c @@ -11,6 +11,7 @@ #include <assert.h> #include <netinet/in.h> #include <arpa/inet.h> +#include <errno.h> #include "util.h" #include "ip.h" @@ -57,3 +58,52 @@ int inany_pton(const char *src, union inany_addr *dst) return 0; } + +/** + * inany_prefix_pton() - Parse an IPv[46] address with prefix length + * @src: IPv[46] address and prefix length string in CIDR format + * @dst: Output buffer, filled with parsed address + * @prefix_len: Prefix length, to be filled in IPv6 format + * + * Return: 1 on success, 0 if no parseable address or prefix is found + */ +int inany_prefix_pton(const char *src, union inany_addr *dst, + uint8_t *prefix_len) +{ + char astr[INANY_ADDRSTRLEN] = { 0 }; + size_t alen = strcspn(src, "/"); + const char *pstr = &src[alen + 1]; + unsigned long plen; + char *end; + + if (alen >= INANY_ADDRSTRLEN) + return 0; + + if (src[alen] != '/') + return 0; + + strncpy(astr, src, alen); + + /* Read prefix length */ + errno = 0; + plen = strtoul(pstr, &end, 10); + if (errno || *end || plen > 128) + return 0; + + /* Read address */ + if (inet_pton(AF_INET6, astr, dst)) { + if (inany_v4(dst) && plen < 96) + return 0; + *prefix_len = plen; + return 1; + } + + if (inany_pton(astr, dst)) { + if (plen > 32) + return 0; + *prefix_len = plen + 96; + return 1; + } + + return 0; +} diff --git a/inany.h b/inany.h index 61b36fb..c7d6027 100644 --- a/inany.h +++ b/inany.h @@ -96,6 +96,19 @@ static inline struct in_addr *inany_v4(const union inany_addr *addr) return (struct in_addr *)&addr->v4mapped.a4; } +/** inany_default_prefix_len() - Get default prefix length for address + * @addr: IPv4 or iPv6 address + * + * Return: Class-based prefix length for IPv4 (in IPv6 format: 104-128), + * or 64 for IPv6 + */ +static inline int inany_default_prefix_len(const union inany_addr *addr) +{ + const struct in_addr *v4 = inany_v4(addr); + + return v4 ? ip4_class_prefix_len(v4) + 96 : 64; +} + /** inany_equals - Compare two IPv[46] addresses * @a, @b: IPv[46] addresses * @@ -295,5 +308,7 @@ static inline void inany_siphash_feed(struct siphash_state *state, const char *inany_ntop(const union inany_addr *src, char *dst, socklen_t size); int inany_pton(const char *src, union inany_addr *dst); +int inany_prefix_pton(const char *src, union inany_addr *dst, + uint8_t *prefix_len); #endif /* INANY_H */ diff --git a/ip.c b/ip.c index 9a7f4c5..40dc24e 100644 --- a/ip.c +++ b/ip.c @@ -13,6 +13,8 @@ */ #include <stddef.h> +#include <netinet/in.h> + #include "util.h" #include "ip.h" @@ -67,3 +69,22 @@ found: *proto = nh; return true; } + +/** + * ip4_class_prefix_len() - Get class based prefix length for IPv4 address + * @addr: IPv4 address + * + * Return: prefix length based on address class, or 32 for other + */ +int ip4_class_prefix_len(const struct in_addr *addr) +{ + in_addr_t a = ntohl(addr->s_addr); + + if (IN_CLASSA(a)) + return 32 - IN_CLASSA_NSHIFT; + if (IN_CLASSB(a)) + return 32 - IN_CLASSB_NSHIFT; + if (IN_CLASSC(a)) + return 32 - IN_CLASSC_NSHIFT; + return 32; +} diff --git a/ip.h b/ip.h index 5830b92..bd28640 100644 --- a/ip.h +++ b/ip.h @@ -135,4 +135,6 @@ static const struct in_addr in4addr_broadcast = { 0xffffffff }; #define IPV6_MIN_MTU 1280 #endif +int ip4_class_prefix_len(const struct in_addr *addr); + #endif /* IP_H */ diff --git a/passt.1 b/passt.1 index db0d662..70552be 100644 --- a/passt.1 +++ b/passt.1 @@ -156,10 +156,16 @@ By default, the advertised MTU is 65520 bytes, that is, the maximum 802.3 MTU minus the length of a 802.3 header, rounded to 32 bits (IPv4 words). .TP -.BR \-a ", " \-\-address " " \fIaddr +.BR \-a ", " \-\-address " " \fIaddr\fR[/\fIprefix_len\fR] Assign IPv4 \fIaddr\fR via DHCP (\fByiaddr\fR), or \fIaddr\fR via DHCPv6 (option 5) and an \fIaddr\fR-based prefix via NDP Router Advertisement (option type 3) for an IPv6 \fIaddr\fR. +An optional /\fIprefix_len\fR (0-32 for IPv4, 0-128 for IPv6) can be +appended in CIDR notation (e.g. 192.0.2.1/24). This is an alternative to +using the \fB-n\fR, \fB--netmask\fR option. Mixing CIDR notation with +\fB-n\fR results in an error. +If a prefix length is assigned to an IPv6 address using this method, it will +in the current code version be overridden by the default value of 64. This option can be specified zero (for defaults) to two times (once for IPv4, once for IPv6). By default, assigned IPv4 and IPv6 addresses are taken from the host interfaces @@ -172,10 +178,13 @@ is assigned for IPv4, and no additional address will be assigned for IPv6. .TP .BR \-n ", " \-\-netmask " " \fImask Assign IPv4 netmask \fImask\fR, expressed as dot-decimal or number of bits, via -DHCP (option 1). -By default, the netmask associated to the host address matching the assigned one -is used. If there's no matching address on the host, the netmask is determined -according to the CIDR block of the assigned address (RFC 4632). +DHCP (option 1). Alternatively, the prefix length can be specified using CIDR +notation with the \fB-a\fR, \fB--address\fR option (e.g. \fB-a\fR 192.0.2.1/24). +Mixing \fB-n\fR with CIDR notation results in an error. +If no address is indicated, the netmask associated with the adopted host address, +if any, is used. If an address is indicated, but without a prefix length, the +netmask is determined based on the corresponding network class. In all other +cases, the netmask is determined by using the indicated prefix length. .TP .BR \-M ", " \-\-mac-addr " " \fIaddr -- 2.52.0

2 2

[PATCH] Fix build when HAS_GETRANDOM is undefined
by Peter Foley 24 Feb '26

24 Feb '26

e.g. util.c:1163:14: error: use of undeclared identifier 'dev_random'; did you mean 'raw_random'? 1163 | ret = read(dev_random, (char *)buf + random_read, Signed-off-by: Peter Foley <pefoley(a)google.com> --- util.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/util.c b/util.c index db27431..a4f2be4 100644 --- a/util.c +++ b/util.c @@ -1160,7 +1160,7 @@ void raw_random(void *buf, size_t buflen) ret = getrandom((char *)buf + random_read, buflen - random_read, GRND_RANDOM); #else - ret = read(dev_random, (char *)buf + random_read, + ret = read(fd, (char *)buf + random_read, buflen - random_read); #endif @@ -1177,7 +1177,7 @@ void raw_random(void *buf, size_t buflen) } #ifndef HAS_GETRANDOM - close(dev_random); + close(fd); #endif if (random_read < buflen) -- 2.53.0.371.g1d285c8824-goog

4 3

[PATCH v2 0/2] Fix minimum frame size checks in vhost-user paths
by Laurent Vivier 24 Feb '26

24 Feb '26

In the vhost-user code paths, buffers from the virtio queue include a virtio net header prepended to the Ethernet frame. The minimum frame size checks (ETH_ZLEN, i.e. 60 bytes per IEEE 802.3) must account for this extra header, otherwise the size requested from vu_collect() and validated in ASSERT() is too small. v2: - PATCH 1/2 is merged, rebase the series on top of master - Add a patch to remove VNET_HLEN from l2len in vu_pad() callers - Remove assert on ETH_ZLEN in tcp_vu_prepare() as the size is trimmed to the actual data size Laurent Vivier (2): tcp_vu: vu_pad() expects l2 length tcp_vu, udp_vu: Account for virtio net header in minimum frame size tcp_vu.c | 17 ++++++++++------- udp_vu.c | 2 +- 2 files changed, 11 insertions(+), 8 deletions(-) -- 2.53.0

2 4

[PATCH 0/2] Fix minimum frame size checks in vhost-user paths
by Laurent Vivier 23 Feb '26

23 Feb '26

In the vhost-user code paths, buffers from the virtio queue include a virtio net header prepended to the Ethernet frame. The minimum frame size checks (ETH_ZLEN, i.e. 60 bytes per IEEE 802.3) must account for this extra header, otherwise the size requested from vu_collect() and validated in ASSERT() is too small. This two-patch series: 1. Introduce a VNET_HLEN macro in virtio.h to replace all open-coded sizeof(struct virtio_net_hdr_mrg_rxbuf) throughout the vhost-user code, improving readability and consistency. 2. Fix the minimum frame size calculations in tcp_vu.c and udp_vu.c to use ETH_ZLEN + VNET_HLEN instead of bare ETH_ZLEN, correctly accounting for the virtio net header in the buffer size. Laurent Vivier (2): virtio: Introduce VNET_HLEN macro for virtio net header length tcp_vu, udp_vu: Account for virtio net header in minimum frame size tcp_vu.c | 29 +++++++++++------------------ udp_vu.c | 14 +++++--------- virtio.h | 2 ++ vu_common.c | 13 +++++-------- vu_common.h | 2 +- 5 files changed, 24 insertions(+), 36 deletions(-) -- 2.52.0

3 7

[PATCH v4 00/12] Introduce multiple addresses
by Jon Maloy 18 Feb '26

18 Feb '26

This version contains what I perceive as the least controversial parts of my previous RFC series. It basically makes address handling behave like before, but now allowing multiple addresses both at the host side and the guest side. v2: - Added the earlier standalone CIDR commit to the head of the series. - Replaced the guest namespace interface subscriptions with just an address observation feature, so that it works with both PASTA and PASST. - Unified 'no_copy_addrs' and 'copy_addrs' code paths, as suggested by David G. - Multiple other changes, also based on feedback from David. - Removed the host interface subscription patches, -for now. I intend to re-add them once this series is applied. - Outstanding question: When do we add an IPv4 link local address to the guest? Only in local/opaque mode? Only when explicitly requested? Always? v3: - Unified the IPv4 and IPv6 arrays into one array - Changed prefix_len to always be in IPv6/IpV4 mapped format - Updated migration protocol to v3, handling multiple addresses - Many other smaller changes, based on feedback from the PASST team v4: - Numerous changes based on feedback - Added several new commits, mostly broken out of the pre-existing ones. Jon Maloy (12): ip: Introduce unified multi-address data structures ip: Introduce for_each_addr() macro for address iteration fwd: Unify guest accessibility checks with unified address array arp: Check all configured addresses in ARP filtering pasta: Extract pasta_ns_conf_ip4/6() to reduce nesting netlink: Return prefix length for IPv6 addresses in nl_addr_get() conf: Allow multiple -a/--address options per address family ip: Track observed guest IPv4 addresses in unified address array ip: Track observed guest IPv6 addresses in unified address array fwd: Unify fwd_set_observed_ip4() and fwd_set_observed_ip6() migrate: Rename v1 address functions to v2 for clarity migrate: Update protocol to v3 for multi-address support arp.c | 15 +++- conf.c | 147 +++++++++++++++++++++--------------- conf.h | 8 ++ dhcp.c | 13 +++- dhcpv6.c | 11 ++- dhcpv6.h | 2 +- fwd.c | 221 ++++++++++++++++++++++++++++++++++++++---------------- fwd.h | 3 + inany.h | 3 + ip.h | 5 ++ migrate.c | 185 ++++++++++++++++++++++++++++++++++++++++----- ndp.c | 17 ++++- netlink.c | 4 +- passt.h | 104 +++++++++++++++++++++---- pasta.c | 200 ++++++++++++++++++++++++++++-------------------- tap.c | 72 +++++++++++++----- 16 files changed, 736 insertions(+), 274 deletions(-) -- 2.52.0

1 13

[PATCH v3] udp: Split activity timeouts for UDP flows
by Yumei Huang 15 Feb '26

15 Feb '26

Frequent DNS queries over UDP from a container or guest can result in many sockets shown in ss(8), typically one per flow. This is expected and harmless, but it can make the output of ss(8) look noisy and potentially concern users. This patch splits UDP flow timeouts into two, mirroring the Linux kernel, and sources the values from kernel parameters. The shorter timeout is applied to unidirectional flows and minimal bidirectional exchanges (single datagram and reply), while the longer timeout is used for bidirectional flows with multiple datagrams on either side. Link: https://bugs.passt.top/show_bug.cgi?id=197 Suggested-by: Stefano Brivio <sbrivio(a)redhat.com> Signed-off-by: Yumei Huang <yuhuang(a)redhat.com> --- contrib/apparmor/abstractions/passt | 4 ++++ udp.c | 34 +++++++++++++++++++++++++++-- udp.h | 4 ++++ udp_flow.c | 30 ++++++++++++++++++++----- udp_flow.h | 4 ++++ 5 files changed, 69 insertions(+), 7 deletions(-) diff --git a/contrib/apparmor/abstractions/passt b/contrib/apparmor/abstractions/passt index 0ffadaf..85bd1ee 100644 --- a/contrib/apparmor/abstractions/passt +++ b/contrib/apparmor/abstractions/passt @@ -41,6 +41,10 @@ @{PROC}/sys/net/ipv4/tcp_syn_linear_timeouts r, @{PROC}/sys/net/ipv4/tcp_rto_max_ms r, + # udp_get_timeout_params(), udp.c + @{PROC}/sys/net/netfilter/nf_conntrack_udp_timeout r, + @{PROC}/sys/net/netfilter/nf_conntrack_udp_timeout_stream r, + network netlink raw, # nl_sock_init_do(), netlink.c network inet stream, # tcp.c diff --git a/udp.c b/udp.c index b2383e2..19adda0 100644 --- a/udp.c +++ b/udp.c @@ -26,7 +26,10 @@ * * We track pseudo-connections of this type as flow table entries of type * FLOW_UDP. We store the time of the last traffic on the flow in uflow->ts, - * and let the flow expire if there is no traffic for UDP_CONN_TIMEOUT seconds. + * and let the flow expire if there is no traffic for UDP_TIMEOUT seconds for + * unidirectional flows and flows with only one datagram and one reply, or + * UDP_TIMEOUT_STREAM seconds for bidirectional flows with more than one + * datagram on either side. * * NOTE: This won't handle multicast protocols, or some protocols with different * port usage. We'll need specific logic if we want to handle those. @@ -118,6 +121,13 @@ #define UDP_MAX_FRAMES 32 /* max # of frames to receive at once */ +#define UDP_TIMEOUT "/proc/sys/net/netfilter/nf_conntrack_udp_timeout" +#define UDP_TIMEOUT_STREAM \ + "/proc/sys/net/netfilter/nf_conntrack_udp_timeout_stream" + +#define UDP_TIMEOUT_DEFAULT 30 /* s */ +#define UDP_TIMEOUT_STREAM_DEFAULT 120 /* s */ + /* Maximum UDP data to be returned in ICMP messages */ #define ICMP4_MAX_DLEN 8 #define ICMP6_MAX_DLEN (IPV6_MIN_MTU \ @@ -953,7 +963,7 @@ void udp_sock_handler(const struct ctx *c, union epoll_ref ref, int s = ref.fd; flow_trace(uflow, "Received data on reply socket"); - uflow->ts = now->tv_sec; + udp_flow_activity(uflow, !tosidx.sidei, now); if (pif_is_socket(topif)) { udp_sock_to_sock(c, ref.fd, n, tosidx); @@ -1179,6 +1189,24 @@ static void udp_splice_iov_init(void) } } +/** + * udp_get_timeout_params() - Get host kernel UDP timeout parameters + * @c: Execution context + */ +static void udp_get_timeout_params(struct ctx *c) +{ + intmax_t v; + + v = read_file_integer(UDP_TIMEOUT, UDP_TIMEOUT_DEFAULT); + c->udp.timeout = v; + + v = read_file_integer(UDP_TIMEOUT_STREAM, UDP_TIMEOUT_STREAM_DEFAULT); + c->udp.stream_timeout = v; + + debug("Using UDP timeout parameters, timeout: %d, stream_timeout: %d", + c->udp.timeout, c->udp.stream_timeout); +} + /** * udp_init() - Initialise per-socket data, and sockets in namespace * @c: Execution context @@ -1189,6 +1217,8 @@ int udp_init(struct ctx *c) { ASSERT(!c->no_udp); + udp_get_timeout_params(c); + udp_iov_init(c); if (fwd_listen_sync(c, &c->udp.fwd_in, PIF_HOST, IPPROTO_UDP) < 0) diff --git a/udp.h b/udp.h index 2b91d72..da9c2df 100644 --- a/udp.h +++ b/udp.h @@ -24,11 +24,15 @@ void udp_update_l2_buf(const unsigned char *eth_d); * @fwd_in: Port forwarding configuration for inbound packets * @fwd_out: Port forwarding configuration for outbound packets * @timer_run: Timestamp of most recent timer run + * @timeout: Timeout for unidirectional flows (in s) + * @stream_timeout: Timeout for stream-like flows (in s) */ struct udp_ctx { struct fwd_ports fwd_in; struct fwd_ports fwd_out; struct timespec timer_run; + int timeout; + int stream_timeout; }; #endif /* UDP_H */ diff --git a/udp_flow.c b/udp_flow.c index 1f5e84e..0023837 100644 --- a/udp_flow.c +++ b/udp_flow.c @@ -17,8 +17,6 @@ #include "udp_internal.h" #include "epoll_ctl.h" -#define UDP_CONN_TIMEOUT 180 /* s, timeout for ephemeral or local bind */ - /** * udp_at_sidx() - Get UDP specific flow at given sidx * @sidx: Flow and side to retrieve @@ -152,6 +150,8 @@ static flow_sidx_t udp_flow_new(const struct ctx *c, union flow *flow, uflow->ts = now->tv_sec; uflow->s[INISIDE] = uflow->s[TGTSIDE] = -1; uflow->ttl[INISIDE] = uflow->ttl[TGTSIDE] = 0; + uflow->activity[INISIDE] = 1; + uflow->activity[TGTSIDE] = 0; flow_foreach_sidei(sidei) { if (pif_is_socket(uflow->f.pif[sidei])) @@ -228,7 +228,7 @@ flow_sidx_t udp_flow_from_sock(const struct ctx *c, uint8_t pif, sidx = flow_lookup_sa(c, IPPROTO_UDP, pif, s_in, dst, port); if ((uflow = udp_at_sidx(sidx))) { - uflow->ts = now->tv_sec; + udp_flow_activity(uflow, sidx.sidei, now); return flow_sidx_opposite(sidx); } @@ -285,7 +285,7 @@ flow_sidx_t udp_flow_from_tap(const struct ctx *c, sidx = flow_lookup_af(c, IPPROTO_UDP, pif, af, saddr, daddr, srcport, dstport); if ((uflow = udp_at_sidx(sidx))) { - uflow->ts = now->tv_sec; + udp_flow_activity(uflow, sidx.sidei, now); return flow_sidx_opposite(sidx); } @@ -362,9 +362,29 @@ bool udp_flow_defer(const struct ctx *c, struct udp_flow *uflow, bool udp_flow_timer(const struct ctx *c, struct udp_flow *uflow, const struct timespec *now) { - if (now->tv_sec - uflow->ts <= UDP_CONN_TIMEOUT) + int timeout = c->udp.timeout; + + if (uflow->activity[TGTSIDE] && + (uflow->activity[INISIDE] > 1 || uflow->activity[TGTSIDE] > 1)) + timeout = c->udp.stream_timeout; + + if (now->tv_sec - uflow->ts <= timeout) return false; udp_flow_close(c, uflow); return true; } + +/** + * udp_flow_activity() - Track activity of a UDP flow + * @uflow: UDP flow + * @sidei: Side index of the flow (INISIDE or TGTSIDE) + * @now: Current timestamp + */ +void udp_flow_activity(struct udp_flow *uflow, unsigned int sidei, + const struct timespec *now) +{ + uflow->ts = now->tv_sec; + if (uflow->activity[sidei] < UINT8_MAX) + uflow->activity[sidei]++; +} diff --git a/udp_flow.h b/udp_flow.h index 14e0f92..3ca49c5 100644 --- a/udp_flow.h +++ b/udp_flow.h @@ -16,6 +16,7 @@ * @flush1: @s[1] may have datagrams queued for other flows * @ts: Activity timestamp * @s: Socket fd (or -1) for each side of the flow + * @activity: Packets seen from each side of the flow, up to UINT8_MAX */ struct udp_flow { /* Must be first element */ @@ -29,6 +30,7 @@ struct udp_flow { time_t ts; int s[SIDES]; + uint8_t activity[SIDES]; }; struct udp_flow *udp_at_sidx(flow_sidx_t sidx); @@ -46,5 +48,7 @@ bool udp_flow_defer(const struct ctx *c, struct udp_flow *uflow, const struct timespec *now); bool udp_flow_timer(const struct ctx *c, struct udp_flow *uflow, const struct timespec *now); +void udp_flow_activity(struct udp_flow *uflow, unsigned int sidei, + const struct timespec *now); #endif /* UDP_FLOW_H */ -- 2.52.0

2 1

checksum: add VSX fast path for POWER8/POWER9
by jfiusdq 15 Feb '26

15 Feb '26

Tested with podman on Debian 13 for a while and works ok. It's difficult to run all the tests on POWER but 505-networking-pasta.bats test suite passes.

4 6

[PATCH] tcp: Move tap header update out of tcp_fill_headers()
by Laurent Vivier 15 Feb '26

15 Feb '26

tcp_fill_headers() currently calls tap_hdr_update() to set the frame length in the tap-specific header. This is backend-specific: the tap backend needs this for its frame length header, but the vhost-user backend passes NULL for the tap header and doesn't use it at all. Remove the tap_hdr parameter from tcp_fill_headers() and instead return the computed L2 frame length. The tap backend caller, tcp_l2_buf_fill_headers(), now calls tap_hdr_update() itself with the returned length. The vhost-user callers, tcp_vu_send_flag() and tcp_vu_prepare(), no longer need to pass a NULL tap header. Signed-off-by: Laurent Vivier <lvivier(a)redhat.com> --- tap.h | 3 +-- tcp.c | 16 +++++++++------- tcp_buf.c | 6 ++++-- tcp_internal.h | 11 ++++++----- tcp_vu.c | 4 ++-- 5 files changed, 22 insertions(+), 18 deletions(-) diff --git a/tap.h b/tap.h index ee22a9d78c44..cc780d168941 100644 --- a/tap.h +++ b/tap.h @@ -65,8 +65,7 @@ static inline struct iovec tap_hdr_iov(const struct ctx *c, */ static inline void tap_hdr_update(struct tap_hdr *thdr, size_t l2len) { - if (thdr) - thdr->vnet_len = htonl(l2len); + thdr->vnet_len = htonl(l2len); } unsigned long tap_l2_max_len(const struct ctx *c); diff --git a/tcp.c b/tcp.c index 0a64892afbaf..048da2ab59bc 100644 --- a/tcp.c +++ b/tcp.c @@ -924,7 +924,6 @@ static void tcp_fill_header(struct tcphdr *th, * tcp_fill_headers() - Fill 802.3, IP, TCP headers * @c: Execution context * @conn: Connection pointer - * @taph: tap backend specific header * @eh: Pointer to Ethernet header * @ip4h: Pointer to IPv4 header, or NULL * @ip6h: Pointer to IPv6 header, or NULL @@ -933,12 +932,15 @@ static void tcp_fill_header(struct tcphdr *th, * @ip4_check: IPv4 checksum, if already known * @seq: Sequence number for this segment * @no_tcp_csum: Do not set TCP checksum + * + * Return: frame length (including L2 headers) */ -void tcp_fill_headers(const struct ctx *c, struct tcp_tap_conn *conn, - struct tap_hdr *taph, struct ethhdr *eh, - struct iphdr *ip4h, struct ipv6hdr *ip6h, - struct tcphdr *th, struct iov_tail *payload, - const uint16_t *ip4_check, uint32_t seq, bool no_tcp_csum) +size_t tcp_fill_headers(const struct ctx *c, struct tcp_tap_conn *conn, + struct ethhdr *eh, + struct iphdr *ip4h, struct ipv6hdr *ip6h, + struct tcphdr *th, struct iov_tail *payload, + const uint16_t *ip4_check, uint32_t seq, + bool no_tcp_csum) { const struct flowside *tapside = TAPFLOW(conn); size_t l4len = iov_tail_size(payload) + sizeof(*th); @@ -1004,7 +1006,7 @@ void tcp_fill_headers(const struct ctx *c, struct tcp_tap_conn *conn, else tcp_update_csum(psum, th, payload); - tap_hdr_update(taph, MAX(l3len + sizeof(struct ethhdr), ETH_ZLEN)); + return MAX(l3len + sizeof(struct ethhdr), ETH_ZLEN); } /** diff --git a/tcp_buf.c b/tcp_buf.c index d29254103364..675ec138c460 100644 --- a/tcp_buf.c +++ b/tcp_buf.c @@ -183,14 +183,16 @@ static void tcp_l2_buf_fill_headers(const struct ctx *c, struct ethhdr *eh = iov[TCP_IOV_ETH].iov_base; struct ipv6hdr *ip6h = NULL; struct iphdr *ip4h = NULL; + size_t l2len; if (a4) ip4h = iov[TCP_IOV_IP].iov_base; else ip6h = iov[TCP_IOV_IP].iov_base; - tcp_fill_headers(c, conn, taph, eh, ip4h, ip6h, th, &tail, - check, seq, no_tcp_csum); + l2len = tcp_fill_headers(c, conn, eh, ip4h, ip6h, th, &tail, check, seq, + no_tcp_csum); + tap_hdr_update(taph, l2len); } /** diff --git a/tcp_internal.h b/tcp_internal.h index 5f8fb3584355..518913b5bc03 100644 --- a/tcp_internal.h +++ b/tcp_internal.h @@ -176,11 +176,12 @@ void tcp_rst_do(const struct ctx *c, struct tcp_tap_conn *conn); struct tcp_info_linux; -void tcp_fill_headers(const struct ctx *c, struct tcp_tap_conn *conn, - struct tap_hdr *taph, struct ethhdr *eh, - struct iphdr *ip4h, struct ipv6hdr *ip6h, - struct tcphdr *th, struct iov_tail *payload, - const uint16_t *ip4_check, uint32_t seq, bool no_tcp_csum); +size_t tcp_fill_headers(const struct ctx *c, struct tcp_tap_conn *conn, + struct ethhdr *eh, + struct iphdr *ip4h, struct ipv6hdr *ip6h, + struct tcphdr *th, struct iov_tail *payload, + const uint16_t *ip4_check, uint32_t seq, + bool no_tcp_csum); int tcp_update_seqack_wnd(const struct ctx *c, struct tcp_tap_conn *conn, bool force_seq, struct tcp_info_linux *tinfo); diff --git a/tcp_vu.c b/tcp_vu.c index b9e9b55ed3d3..f654f3193bd7 100644 --- a/tcp_vu.c +++ b/tcp_vu.c @@ -135,7 +135,7 @@ int tcp_vu_send_flag(const struct ctx *c, struct tcp_tap_conn *conn, int flags) flags_elem[0].in_sg[0].iov_len = hdrlen + optlen; payload = IOV_TAIL(flags_elem[0].in_sg, 1, hdrlen); - tcp_fill_headers(c, conn, NULL, eh, ip4h, ip6h, th, &payload, + tcp_fill_headers(c, conn, eh, ip4h, ip6h, th, &payload, NULL, seq, !*c->pcap); vu_pad(&flags_elem[0].in_sg[0], hdrlen + optlen); @@ -335,7 +335,7 @@ static void tcp_vu_prepare(const struct ctx *c, struct tcp_tap_conn *conn, th->ack = 1; th->psh = push; - tcp_fill_headers(c, conn, NULL, eh, ip4h, ip6h, th, &payload, + tcp_fill_headers(c, conn, eh, ip4h, ip6h, th, &payload, *check, conn->seq_to_tap, no_tcp_csum); if (ip4h) *check = &ip4h->check; -- 2.52.0

2 1