December 2025 - dev

[PATCH v3 0/6] vhost-user: Add multiqueue support
by Laurent Vivier 05 Dec '25

05 Dec '25

This series implements multiqueue support for vhost-user mode, allowing passt to utilize multiple queue pairs for improved network throughput when used with multi-CPU guest VMs. While this version uses a single thread for packet processing, it enables the guest kernel to distribute network traffic across multiple queues and vCPUs. The implementation advertises support for up to 16 queue pairs (32 virtqueues) by setting VIRTIO_NET_F_MQ and VHOST_USER_PROTOCOL_F_MQ feature flags. Packets are routed to the appropriate RX queue based on which TX queue they originated from, following the virtio specification's automatic receive steering requirements. This series adds: - Multiqueue capability advertisement (VIRTIO_NET_F_MQ and VHOST_USER_PROTOCOL_F_MQ features) - Per-queue-pair packet pools to support concurrent queue operations - Queue pair parameter throughout the network stack, propagated through all protocol handlers (TCP, UDP, ICMP, ARP, DHCP, DHCPv6, NDP) - Flow-aware queue routing that tracks the originating TX queue for each flow and routes return packets to the corresponding RX queue - Test coverage with VHOST_USER_MQ environment variable to validate multiqueue functionality across all protocols (TCP, UDP, ICMP) and services (DHCP, NDP) Current behavior: - TX queue selection is controlled by the guest kernel's networking stack - RX packets are routed to queues based on their associated flows, with the queue assignment updated on each packet from TX to maintain affinity - Host-initiated flows (e.g., from socket-side connections) currently default to queue pair 0 The changes are transparent to single-queue operation - passt/pasta modes and single-queue vhost-user configurations continue to work unchanged, always using queue pair 0. v3: - Removed --max-qpairs configuration option - multiqueue support is now always enabled up to 16 queue pairs without requiring explicit configuration - Replaced "tap: Add queue pair parameter throughout the packet processing path" with "tap: Convert packet pools to per-queue-pair arrays for multiqueue" - simplified implementation by converting global pools to arrays rather than passing pool parameters throughout - Changed qpair parameter type from int to unsigned int throughout the codebase - Simplified test infrastructure - queues parameter is always set on netdev, mq=true added to virtio-net only when VHOST_USER_MQ > 1 - Updated QEMU usage hints to always show multiqueue-capable command line v2: - New patch: "tap: Remove pool parameter from tap4_handler() and tap6_handler()" to clean up unused parameters before adding queue pair parameter - Changed to one packet pool per queue pair instead of shared pools across all queue pairs - Split "multiqueue: Add queue-aware flow management..." into two patches: - "tap: Add queue pair parameter throughout the packet processing path" - "flow: Add queue pair tracking to flow management" - Updated test infrastructure patch with refined implementation Laurent Vivier (6): tap: Remove pool parameter from tap4_handler() and tap6_handler() vhost-user: Enable multiqueue test: Add multiqueue support to vhost-user test infrastructure vhost-user: Add queue pair parameter throughout the network stack tap: Convert packet pools to per-queue-pair arrays for multiqueue flow: Add queue pair tracking to flow management arp.c | 15 +++-- arp.h | 6 +- dhcp.c | 5 +- dhcp.h | 2 +- dhcpv6.c | 12 ++-- dhcpv6.h | 2 +- flow.c | 33 +++++++++ flow.h | 17 +++++ fwd.c | 18 ++--- fwd.h | 5 +- icmp.c | 25 ++++--- icmp.h | 4 +- ndp.c | 35 ++++++---- ndp.h | 7 +- netlink.c | 2 +- tap.c | 177 ++++++++++++++++++++++++++++--------------------- tap.h | 20 +++--- tcp.c | 47 +++++++------ tcp.h | 7 +- tcp_vu.c | 8 ++- test/lib/setup | 21 +++--- test/run | 23 +++++++ udp.c | 47 +++++++------ udp.h | 6 +- udp_flow.c | 8 ++- udp_flow.h | 2 +- udp_internal.h | 4 +- udp_vu.c | 4 +- vhost_user.c | 10 +-- virtio.h | 2 +- vu_common.c | 15 +++-- vu_common.h | 3 +- 32 files changed, 374 insertions(+), 218 deletions(-) -- 2.51.1

2 7

[RFC PATCH 0/5] Pad all inbound frames to 802.3 minimum size if needed
by Stefano Brivio 05 Dec '25

05 Dec '25

Patch 1/5 handles the easy non-batched case with a copy to a padded buffer (only if needed). Patches 2/5 and 3/5 clean coding style up before further changes. Patch 4/5 deals with TCP and UDP batched frames in non-vhost-user modes. Patch 5/5 is for batched frames in vhost-user mode instead, and this is the part I'm the least confident about, in particular with regard to the size assumptions on vhost-user provided buffers. Stefano Brivio (5): tap: Pad non-batched frames to 802.3 minimum (60 bytes) if needed tcp: Fix coding style for comment to enum tcp_iov_parts udp: Fix coding style for comment to enum udp_iov_idx tcp, udp: Pad batched frames to 60 bytes (802.3 minimum) in non-vhost-user modes tcp, udp: Pad batched frames for vhost-user modes to 60 bytes (802.3 minimum) tap.c | 7 +++++++ tcp.c | 2 -- tcp_buf.c | 23 +++++++++++++++++++++++ tcp_internal.h | 5 ++++- tcp_vu.c | 27 +++++++++++++++++++++++++++ udp.c | 31 ++++++++++++++++++++++++++----- udp_vu.c | 11 ++++++++++- util.c | 3 +++ util.h | 3 +++ 9 files changed, 103 insertions(+), 9 deletions(-) -- 2.43.0

3 22

[PATCH] vu_common: Clarify prototype of vu_collect()
by Stefano Brivio 05 Dec '25

05 Dec '25

The last argument of vu_collect() can't exceed @size, but this isn't documented and it wasn't obvious to me, to the point that I actually proposed a change to "fix" callers so that they wouldn't overrun input buffers. Rename @frame_size to @collected, and clarify that it can only be up to @size bytes. Signed-off-by: Stefano Brivio <sbrivio(a)redhat.com> --- vu_common.c | 8 ++++---- vu_common.h | 2 +- 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/vu_common.c b/vu_common.c index b13b7c3..ce61fa6 100644 --- a/vu_common.c +++ b/vu_common.c @@ -76,13 +76,13 @@ void vu_init_elem(struct vu_virtq_element *elem, struct iovec *iov, int elem_cnt * in the in_sg array. * @max_elem: Number of virtqueue elements in the array * @size: Maximum size of the data in the frame - * @frame_size: The total size of the buffers (output) + * @collected: Collected buffer length, up to @size, set on return * * Return: number of elements used to contain the frame */ int vu_collect(const struct vu_dev *vdev, struct vu_virtq *vq, struct vu_virtq_element *elem, int max_elem, - size_t size, size_t *frame_size) + size_t size, size_t *collected) { size_t current_size = 0; int elem_cnt = 0; @@ -113,8 +113,8 @@ int vu_collect(const struct vu_dev *vdev, struct vu_virtq *vq, break; } - if (frame_size) - *frame_size = current_size; + if (collected) + *collected = current_size; return elem_cnt; } diff --git a/vu_common.h b/vu_common.h index f538f23..c0883b2 100644 --- a/vu_common.h +++ b/vu_common.h @@ -48,7 +48,7 @@ void vu_init_elem(struct vu_virtq_element *elem, struct iovec *iov, int elem_cnt); int vu_collect(const struct vu_dev *vdev, struct vu_virtq *vq, struct vu_virtq_element *elem, int max_elem, size_t size, - size_t *frame_size); + size_t *collected); void vu_set_vnethdr(const struct vu_dev *vdev, struct virtio_net_hdr_mrg_rxbuf *vnethdr, int num_buffers); -- 2.43.0

2 1

[PATCH] test: Expand tmux right status bar to fit pass/fail/skipped counter and time
by Stefano Brivio 05 Dec '25

05 Dec '25

I recently realised I couldn't see seconds elapsing in asciinema test recordings. That's because now that we display a SKIPPED counter, we need some extra room for characters, 64 in total: PASS: xxx | FAIL: xxx | SKIPPED: xxx | YYYY-MM-DDThh:mm:ss+00:00 Fixes: 4234ace84cdf ("test: Display count of skipped tests in status and summary") Signed-off-by: Stefano Brivio <sbrivio(a)redhat.com> --- test/lib/term | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/test/lib/term b/test/lib/term index 79aa2b7..f596364 100755 --- a/test/lib/term +++ b/test/lib/term @@ -707,7 +707,7 @@ term() { tmux set window-status-current-style 'bg=colour1 fg=colour233 bold' tmux set status-right '#(TZ="UTC" date -Iseconds)' - tmux set status-right-length 50 + tmux set status-right-length 64 tmux set status-right-style 'bg=colour1 fg=colour233 bold' tmux set history-limit 500000 -- 2.43.0

2 1

[PATCH] tcp: Enable SO_KEEPALIVE if we see keep-alive segments from container / guest
by Stefano Brivio 05 Dec '25

05 Dec '25

This is an approximation, as enabling SO_KEEPALIVE doesn't mean that a keep-alive segment will be sent right away, rather that keep-alive segments will start being sent if the connection is idle. On the other hand, we don't have direct control over the host-side TCP, so this is probably the best approximation we can get. By default, namespaces inherit keep-alive parameters from their parent namespace, so we can assume that, in case of a container, we'll wait for the same interval it took for the container to start sending us keep-alives, effectively doubling that interval. To keep this simple, set SO_KEEPALIVE whenever we see a keep-alive segment, instead of tracking its state. Keep-alive segments are relatively infrequent, so we don't expect any substantial cost from doing that. Reported-by: Dominic Kohls <Dominic.Wenzler(a)rewe-group.com> Signed-off-by: Stefano Brivio <sbrivio(a)redhat.com> --- tcp.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/tcp.c b/tcp.c index 8357c0e..c5486bc 100644 --- a/tcp.c +++ b/tcp.c @@ -1838,6 +1838,9 @@ static int tcp_data_from_tap(const struct ctx *c, struct tcp_tap_conn *conn, tcp_send_flag(c, conn, ACK); tcp_timer_ctl(c, conn); + setsockopt(conn->sock, SOL_SOCKET, SO_KEEPALIVE, + &((int){ 1 }), sizeof(int)); + if (p->count == 1) { tcp_tap_window_update(c, conn, ntohs(th->window)); -- 2.43.0

2 1

[PATCH] seccomp: Fix build and operation on 32-bit musl targets
by Stefano Brivio 04 Dec '25

04 Dec '25

On 32-bit musl targets (for example, Alpine i386 / i586), we need to: - use the set of system calls already defined for i686. While Alpine's kernel supports i586 as well (Debian for example doesn't), it's the same architecture, so change i386/i486/i586 machine strings to i686 in seccomp.sh - allow clock_gettime64() as an alternative to clock_gettime() (not just added on top, rather replacing it), because clock_gettime() isn't available as a system call name at all in musl headers (while glibc aliases it to the kernel's clock_gettime64()) - similarly, allow timerfd_gettime64() as a name for timerfd_gettime() - allow timerfd_settime32() as a name for timerfd_settime(), even though there's no such system call declared in Linux kernel headers, as musl uses that name to distinguish it from timerfd_settime64() - unconditionally allow timerfd_settime64(), because musl uses it whenever the 'old' argument is non-NULL and sizeof(time_t) > 4, which happens to be the case in tcp_timer_handler() Reported-by: John D. Rowell <me(a)jdrowell.com> Link: https://bugs.passt.top/show_bug.cgi?id=177 Link: https://gitlab.alpinelinux.org/alpine/aports/-/issues/17686 Signed-off-by: Stefano Brivio <sbrivio(a)redhat.com> --- passt.c | 3 ++- seccomp.sh | 1 + tcp.c | 7 ++++--- 3 files changed, 7 insertions(+), 4 deletions(-) diff --git a/passt.c b/passt.c index 0b84ac6..5ed88d0 100644 --- a/passt.c +++ b/passt.c @@ -329,7 +329,8 @@ static void passt_worker(void *opaque, int nfds, struct epoll_event *events) * #syscalls bind connect recvfrom sendto shutdown * #syscalls arm:recv ppc64le:recv arm:send ppc64le:send * #syscalls accept4 accept listen epoll_ctl epoll_wait|epoll_pwait epoll_pwait - * #syscalls clock_gettime arm:clock_gettime64 i686:clock_gettime64 + * #syscalls clock_gettime|clock_gettime64 + * #syscalls arm:clock_gettime64 i686:clock_gettime64 */ int main(int argc, char **argv) { diff --git a/seccomp.sh b/seccomp.sh index ba92b29..60ebe84 100755 --- a/seccomp.sh +++ b/seccomp.sh @@ -21,6 +21,7 @@ IN="$@" [ -z "${ARCH}" ] && ARCH="$(uname -m)" [ -z "${CC}" ] && CC="cc" +case "${ARCH}" in i[345]86) ARCH=i686 ;; esac AUDIT_ARCH="AUDIT_ARCH_$(echo ${ARCH} | tr '[a-z]' '[A-Z]' \ | sed 's/^ARM.*/ARM/' \ diff --git a/tcp.c b/tcp.c index fc986a2..8357c0e 100644 --- a/tcp.c +++ b/tcp.c @@ -566,8 +566,7 @@ static int tcp_epoll_ctl(const struct ctx *c, struct tcp_tap_conn *conn) * tcp_timer_ctl() - Set timerfd based on flags/events, create timerfd if needed * @c: Execution context * @conn: Connection pointer - * - * #syscalls timerfd_create timerfd_settime + * #syscalls timerfd_create timerfd_settime|timerfd_settime32 */ static void tcp_timer_ctl(const struct ctx *c, struct tcp_tap_conn *conn) { @@ -2461,7 +2460,9 @@ cancel: * @c: Execution context * @ref: epoll reference of timer (not connection) * - * #syscalls timerfd_gettime arm:timerfd_gettime64 i686:timerfd_gettime64 + * #syscalls timerfd_gettime|timerfd_gettime64 + * #syscalls arm:timerfd_gettime64 i686:timerfd_gettime64 + * #syscalls arm:timerfd_settime64 i686:timerfd_settime64 */ void tcp_timer_handler(const struct ctx *c, union epoll_ref ref) { -- 2.43.0

1 0

[PATCH v5 00/15] Reduce differences between inbound and outbound socket binding
by David Gibson 03 Dec '25

03 Dec '25

This series is based on my series fixing bug 176 (regression in auto forwarding). The fact that outbound forwarding sockets are bound to the loopback address, whereas inbound forwarding sockets are (by default) bound to the unspecified address leads to some unexpected differences between the paths setting up each of them. An idea for tackling bug 100 suggested a different approach which will also reduce some of those differences and allow more code to be shared between the two paths. I've since discovered that this approach doesn't help for bug 100, but I think it's still worthwhile for other reasons. v5: - Combine with SO_BINDTODEVICE and bug 113 patch series - Add fallback handling for kernels without SO_BINDTODEVICE - Add missing struct field documentation for no_bindtodevice v4: - Add cleanup patch removing unused structure field - Rebase, fixing conflicts with Laurent's changes - A bunch of spelling and other cosmetic fixes - Clarify relation to bug 100 and bug 113 v3: - A number of additional fixes covering the handling of IPV6_V6ONLY sockopt - Assorted trivial changes v2: - Some rearrangements and rewordings for clarity David Gibson (15): util: Correct error message on SO_BINDTODEVICE failure util: Extend sock_probe_mem() to sock_probe_features() conf: More useful errors for kernels without SO_BINDTODEVICE flow: Remove bogus @path field from flowside_sock_args inany: Let length of sockaddr_inany be implicit from the family util, flow, pif: Simplify sock_l4_sa() interface tcp: Merge tcp_ns_sock_init[46]() into tcp_sock_init_one() udp: Unify some more inbound/outbound parts of udp_sock_init() udp: Move udp_sock_init() special case to its caller util: Fix setting of IPV6_V6ONLY socket option tcp, udp: Remove fallback if creating dual stack socket fails tcp, udp: Bind outbound listening sockets by interface instead of address util: Rename sock_l4_dualstack() to sock_l4_dualstack_any() tcp: Always populate oaddr field for socket initiated flows fwd: Preserve non-standard loopback address when splice forwarding conf.c | 16 ++++- flow.c | 22 +++---- fwd.c | 4 +- icmp.c | 3 +- inany.h | 17 ++++++ passt.c | 2 +- passt.h | 2 + pif.c | 27 ++------- pif.h | 2 +- tcp.c | 162 +++++++++++++++++++++------------------------------ tcp.h | 5 +- tcp_splice.c | 5 +- udp.c | 122 +++++++++++++++++++++----------------- udp.h | 5 +- util.c | 102 +++++++++++++++++++++++++++----- util.h | 10 ++-- 16 files changed, 289 insertions(+), 217 deletions(-) -- 2.52.0

2 19

[PATCH] test: Update Makefile to avoid failing on missing images
by Yumei Huang 03 Dec '25

03 Dec '25

1 0

[PATCH v10 0/5] Retry SYNs for inbound connections
by Yumei Huang 03 Dec '25

03 Dec '25

When a client connects, SYN would be sent to guest only once. If the guest is not connected or ready at that time, the connection will be reset in 10s. These patches introduce the SYN retry mechanism using the similar backoff timeout as linux kernel. Also update the data retransmission timeout using the backoff timeout. v10: - Update read_file() to check buf_size first - Improve debug message in read_file_integer() - Replace RTO_MAX_MS_DEFAULT with RTO_MAX_DEFAULT - Wording fixes v9: - Squash the patch of adding parameter c to tcp_timer_ctl() into the later one - Fix the cppcheck error - Update the return values of read_file(), remove the error message of it and add check for buf_size - Rename RTO_INIT_ACK to RTO_INIT_AFTER_SYN_RETRIES - Round up rto_max, fix the conversion in subtraction - Some other minor fixes related to wording v8: - Remove the TCP_/tcp_ suffix from certain macro and struct member names - Add parameter struct ctx *c to tcp_timer_ctl() - Modify rto_max type from size_t to int - Clamp syn_retries and syn_linear_timeouts with MAX_SYNCNT(127) - Some other minor fixes related to wording and formatting v7: - Update read_file() and read_file_integer() - Rename tcp_syn_params_init() to tcp_get_rto_params() - Modify the implementation of the timeout assignment - Add a patch to clamp the retry timeout Yumei Huang (5): tcp: Rename "retrans" to "retries" util: Introduce read_file() and read_file_integer() function tcp: Resend SYN for inbound connections tcp: Update data retransmission timeout tcp: Clamp the retry timeout tcp.c | 107 +++++++++++++++++++++++++++++++++++++++-------------- tcp.h | 6 +++ tcp_conn.h | 13 ++++--- util.c | 91 +++++++++++++++++++++++++++++++++++++++++++++ util.h | 1 + 5 files changed, 185 insertions(+), 33 deletions(-) -- 2.51.1

3 9

[PATCH v5 01/15] util: Correct error message on SO_BINDTODEVICE failure
by David Gibson 02 Dec '25

02 Dec '25

The warning message we print of SO_BINDTODEVICE fails is incorrect: we include EPOLL_TYPE_STR(proto), but the proto variable is not an epoll type so this will generate misleading garbage. This has been wrong ever since 340164445 but the message is rare enough that we never noticed. Correct that, and reword the message a bit for clarity while we're there. Fixes: 340164445341 ("epoll: Generalize epoll_ref to cover things other than sockets") Signed-off-by: David Gibson <david(a)gibson.dropbear.id.au> --- util.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/util.c b/util.c index ab23463b..347f34f5 100644 --- a/util.c +++ b/util.c @@ -126,9 +126,10 @@ int sock_l4_sa(const struct ctx *c, enum epoll_type type, char str[SOCKADDR_STRLEN]; ret = -errno; - warn("Can't bind %s socket for %s to %s, closing", - EPOLL_TYPE_STR(proto), - sockaddr_ntop(sa, str, sizeof(str)), ifname); + warn("SO_BINDTODEVICE %s failed for %s on %s: %s", + ifname, EPOLL_TYPE_STR(type), + sockaddr_ntop(sa, str, sizeof(str)), + strerror_(-ret)); close(fd); return ret; } -- 2.52.0

1 0