- dev - passt

passt: new version 2024_10_30.ee7d0b6 available
by Stefano Brivio 30 Oct '24

30 Oct '24

The new version with tag 2024_10_30.ee7d0b6 includes the following changes: ee7d0b6 util: Don't use errno after a successful call in __daemon() b1a607f udp: Take care of cert-int09-c clang-tidy warning for enum udp_iov_idx 099ace6 treewide: Address cert-err33-c clang-tidy warnings for clock and timer functions 59fe34e treewide: Suppress clang-tidy warning if we already use O_CLOEXEC 134b4d5 Makefile: Disable readability-math-missing-parentheses clang-tidy check 7442478 treewide: Silence cert-err33-c clang-tidy warnings for fprintf() 98efe7c treewide: Comply with CERT C rule ERR33-C for snprintf() 988a4d7 Makefile: Exclude qrap.c from clang-tidy checks ba38e67 tcp: unify l2 TCPv4 and TCPv6 queues and structures 2053c36 tcp: set ip and eth headers in l2 tap queues on the fly 5563d5f test: remove obsolete images f43f7d5 tcp: cleanup tcp_buf_data_from_sock() e7fcd0c tcp: Use runtime tests for TCP_INFO fields 8114381 tcp: Generalise probing for tcpi_snd_wnd field 13f0291 tcp: Remove compile-time dependency on struct tcp_info version 9e4615b tcp_splice: fcntl(2) returns the size of the pipe, if F_SETPIPE_SZ succeeds 149f457 tcp_splice: splice() all we have to the writing side, not what we just read 9e5df35 tcp: Use structures to construct initial TCP options b4dace8 fwd: Direct inbound spliced forwards to the guest's external address 58e6d68 test: Clarify test for spliced inbound transfers 1fa4211 passt.1: Clarify and update "Handling of local addresses" section ef8a516 passt.1: Mark --stderr as deprecated more prominently 53176ca test: Wait for DAD on DHCPv6 addresses 75b9c0f test: Explicitly wait for DAD to complete on SLAAC addresses f9d677b arp: Fix a handful of small warts 2d7f734 tcp: Send "empty" handshake ACK before first data segment 7612cb8 test: Pass TRACE from run_term() into ./run from_term b40880c test/lib/term: Always use printf for messages with escape sequences ff63ac9 conf: Add --dns-host option to configure host side nameserver 9d66df9 conf: Add command line switch to enable IP_FREEBIND socket option 151dbe0 udp: Update UDP checksum using an iovec array 3d484aa tcp: Update TCP checksum using an iovec array e6548c6 checksum: Add an offset argument in csum_iov() fd8334b pcap: Add an offset argument in pcap_iov() 72e7d30 tcp: Use tcp_payload_t rather than tcphdr def8acd test: Kernel binary can now be passed via the KERNEL environmental variable b55013b inany: Add inany_pton() helper cbde419 tcp, udp: Make {tcp,udp}_sock_init() take an inany address b8d4fac util, pif: Replace sock_l4() with pif_sock_l4() 204e77c udp: Don't attempt to get dual-stack sockets in nonsensical cases 8f8c4d2 tcp: Allow checksum to be disabled 4fe5f4e udp: Allow checksum to be disabled d836d9e util: Remove possible quadratic behaviour from write_remainder() bfc294b util: Add helper to write() all of a buffer bb41901 tcp: Make tcp_update_seqack_wnd()s force_seq parameter explicitly boolean 265b209 tcp: Simplify ifdef logic in tcp_update_seqack_wnd() 4aff6f9 tcp: Clean up tcpi_snd_wnd probing 7d8804b tcp: Make some extra functions private 5ff5d55 tcp: Avoid overlapping memcpy() in DUP_ACK handling 1f414ed tcp: Remove redundant initialisation of iov[TCP_IOV_ETH].iov_base https://passt.top/passt/log/?qt=range&q=2024_09_06.6b38f07..2024_10_30.ee7d… Packages: - Alpine Linux: https://pkgs.alpinelinux.org/packages?name=passt - Arch Linux: https://www.archlinux.org/packages/extra/x86_64/passt/ https://archlinuxarm.org/packages/aarch64/passt https://archlinuxarm.org/packages/armv7h/passt - Chimera: https://pkgs.chimera-linux.org/packages?name=passt - Clear Linux: https://github.com/clearlinux-pkgs/passt/ - Debian tracker: https://tracker.debian.org/pkg/passt - Copr (CentOS Stream, EPEL, Fedora, Mageia): https://copr.fedorainfracloud.org/coprs/sbrivio/passt/build/8189610/ permanent mirror: https://passt.top/builds/copr/0^20241030.gee7d0b6/ - Fedora updates: https://bodhi.fedoraproject.org/updates/?packages=passt - Gentoo versions: https://packages.gentoo.org/packages/net-misc/passt - GNU Guix: https://packages.guix.gnu.org/packages/passt/ - Homebrew: https://formulae.brew.sh/formula/passt - NixOS: https://github.com/NixOS/nixpkgs/tree/nixos-unstable/pkgs/by-name/pa/passt - openSUSE: https://software.opensuse.org/package/passt - PLD Linux: https://git.pld-linux.org/cgi-bin/gitweb.cgi?p=packages/passt.git - Solus: https://github.com/getsolus/packages/tree/main/packages/p/passt - Ubuntu tracker: https://launchpad.net/ubuntu/+source/passt - Void Linux: https://voidlinux.org/packages/?q=passt - Static builds: - Package for other RPM-based distributions, x86_64 only: https://passt.top/builds/latest/x86_64/passt-gee7d0b6-1.x86_64.rpm - x86_64 static binaries: https://passt.top/builds/latest/x86_64/ - Debian package, from x86_64 static build: https://passt.top/builds/latest/x86_64/passt_0ee7d0b6-1_all.deb -- Stefano

1 0

[PATCH v5 0/8] Take care of clang-tidy warnings with LLVM >= 16
by Stefano Brivio 30 Oct '24

30 Oct '24

So I started hitting some clang-tidy warnings with LLVM 16, some looked bogus, so I upgraded to LLVM 19, and... I got even more. This series takes care of them in different ways. v5: - ouch, it looks like I didn't test with 2/8 while rebasing this thing back and forth... replace snprintf() with vsnprintf() there, otherwise nothing works v4: - drop 5/9 and keep O_APPEND for the log file, turned off around log rotation, so that a hypothetical log file with multiple writers would still be somewhat consistent v3: - split 5/8 into 5/9 and 6/9: in the first, drop O_APPEND so that we can have a helper to open any output file we need, and in the second one, always use O_CLOEXEC for pcap file (and use the new helper, now that we can) v2: - make snprintf_check() return and set errno on failure, in 2/8 - add missing err_perror() calls on clock_gettime() failures in 6/8 - drop all explicit integer assignments in enum udp_iov_idx in 7/8 Stefano Brivio (8): Makefile: Exclude qrap.c from clang-tidy checks treewide: Comply with CERT C rule ERR33-C for snprintf() treewide: Silence cert-err33-c clang-tidy warnings for fprintf() Makefile: Disable readability-math-missing-parentheses clang-tidy check treewide: Suppress clang-tidy warning if we already use O_CLOEXEC treewide: Address cert-err33-c clang-tidy warnings for clock and timer functions udp: Take care of cert-int09-c clang-tidy warning for enum udp_iov_idx util: Don't use errno after a successful call in __daemon() Makefile | 13 +++++++--- arch.c | 6 ++++- conf.c | 64 +++++++++++++++++++++++++++---------------------- log.c | 9 ++++--- passt.c | 9 ++++--- pasta.c | 11 ++++++--- pcap.c | 24 ++++++++++--------- tap.c | 5 ++-- tcp.c | 12 +++++++--- udp.c | 10 ++++---- util.c | 72 ++++++++++++++++++++++++++++++++++++-------------------- util.h | 7 +++++- 12 files changed, 151 insertions(+), 91 deletions(-) -- 2.43.0

2 9

[PATCH v6 0/8] Take care of clang-tidy warnings with LLVM >= 16
by Stefano Brivio 30 Oct '24

30 Oct '24

So I started hitting some clang-tidy warnings with LLVM 16, some looked bogus, so I upgraded to LLVM 19, and... I got even more. This series takes care of them in different ways. v6: - ...and I didn't test 5/8 either :( the log file needs O_RDWR, whereas O_WRONLY suffices for pcap and PID file v5: - ouch, it looks like I didn't test with 2/8 while rebasing this thing back and forth... replace snprintf() with vsnprintf() there, otherwise nothing works v4: - drop 5/9 and keep O_APPEND for the log file, turned off around log rotation, so that a hypothetical log file with multiple writers would still be somewhat consistent v3: - split 5/8 into 5/9 and 6/9: in the first, drop O_APPEND so that we can have a helper to open any output file we need, and in the second one, always use O_CLOEXEC for pcap file (and use the new helper, now that we can) v2: - make snprintf_check() return and set errno on failure, in 2/8 - add missing err_perror() calls on clock_gettime() failures in 6/8 - drop all explicit integer assignments in enum udp_iov_idx in 7/8 Stefano Brivio (8): Makefile: Exclude qrap.c from clang-tidy checks treewide: Comply with CERT C rule ERR33-C for snprintf() treewide: Silence cert-err33-c clang-tidy warnings for fprintf() Makefile: Disable readability-math-missing-parentheses clang-tidy check treewide: Suppress clang-tidy warning if we already use O_CLOEXEC treewide: Address cert-err33-c clang-tidy warnings for clock and timer functions udp: Take care of cert-int09-c clang-tidy warning for enum udp_iov_idx util: Don't use errno after a successful call in __daemon() Makefile | 13 +++++++--- arch.c | 6 ++++- conf.c | 65 ++++++++++++++++++++++++++++---------------------- log.c | 9 ++++--- passt.c | 9 ++++--- pasta.c | 11 ++++++--- pcap.c | 24 ++++++++++--------- tap.c | 5 ++-- tcp.c | 12 +++++++--- udp.c | 10 ++++---- util.c | 72 ++++++++++++++++++++++++++++++++++++-------------------- util.h | 7 +++++- 12 files changed, 152 insertions(+), 91 deletions(-) -- 2.43.0

1 8

[PATCH v4 0/8] Take care of clang-tidy warnings with LLVM >= 16
by Stefano Brivio 30 Oct '24

30 Oct '24

So I started hitting some clang-tidy warnings with LLVM 16, some looked bogus, so I upgraded to LLVM 19, and... I got even more. This series takes care of them in different ways. v4: - drop 5/9 and keep O_APPEND for the log file, turned off around log rotation, so that a hypothetical log file with multiple writers would still be somewhat consistent v3: - split 5/8 into 5/9 and 6/9: in the first, drop O_APPEND so that we can have a helper to open any output file we need, and in the second one, always use O_CLOEXEC for pcap file (and use the new helper, now that we can) v2: - make snprintf_check() return and set errno on failure, in 2/8 - add missing err_perror() calls on clock_gettime() failures in 6/8 - drop all explicit integer assignments in enum udp_iov_idx in 7/8 Stefano Brivio (8): Makefile: Exclude qrap.c from clang-tidy checks treewide: Comply with CERT C rule ERR33-C for snprintf() treewide: Silence cert-err33-c clang-tidy warnings for fprintf() Makefile: Disable readability-math-missing-parentheses clang-tidy check treewide: Suppress clang-tidy warning if we already use O_CLOEXEC or if we can't treewide: Address cert-err33-c clang-tidy warnings for clock and timer functions udp: Take care of cert-int09-c clang-tidy warning for enum udp_iov_idx util: Don't use errno after a successful call in __daemon() Makefile | 13 +++++++--- arch.c | 6 ++++- conf.c | 64 +++++++++++++++++++++++++++---------------------- log.c | 9 ++++--- passt.c | 9 ++++--- pasta.c | 11 ++++++--- pcap.c | 24 ++++++++++--------- tap.c | 5 ++-- tcp.c | 12 +++++++--- udp.c | 10 ++++---- util.c | 72 ++++++++++++++++++++++++++++++++++++-------------------- util.h | 7 +++++- 12 files changed, 151 insertions(+), 91 deletions(-) -- 2.43.0

2 9

[PATCH v5 0/2] tcp: unify IPv4 and IPv6 tap queues
by Jon Maloy 29 Oct '24

29 Oct '24

This should save us some memory and code. --- v2: - Setting pointers to pre-set IP and MAC headers on the fly instead of copying them. - Merged patch #2 and #3 from v1 v3: - Changes based on feedback from team v4: - Rebased v5: - Changes based on feedback from team and rebased Jon Maloy (2): tcp: set ip and eth headers in l2 tap queues on the fly tcp: unify l2 TCPv4 and TCPv6 queues and structures tcp.c | 6 +- tcp_buf.c | 272 ++++++++++++++++++------------------------------------ tcp_buf.h | 3 +- 3 files changed, 91 insertions(+), 190 deletions(-) -- 2.45.2

3 5

[PATCH 0/7] Rework some IOV handling in TCP code
by David Gibson 29 Oct '24

29 Oct '24

These reworks are largely aimed at making the vhost-user integration easier, and with luck allowing more logic to be shared between it and the existing "buffer" paths. Of course, in the short term, these will probably conflict with the patches... I hope it ends up as a net positive, Laurent, let me know. I think a number of similar changes should be possible for UDP, but I haven't tackled that yet. David Gibson (7): tcp: Pass TCP header and payload separately to tcp_update_check_tcp[46]() tcp: Move tcp_l2_buf_fill_headers() to tcp_buf.c tcp: Rework tcp_l2_buf_fill_headers() into tcp_buf_make_frame() tcp: Don't use return value from tcp_fill_headers[46] to adjust iov_len tcp: Pass TCP header and payload separately to tcp_fill_headers[46]() tcp: Merge tcp_update_check_tcp[46]() tcp: Fold tcp_update_csum() into tcp_fill_header() tcp.c | 232 +++++++++++-------------------------------------- tcp_buf.c | 48 +++++++--- tcp_internal.h | 15 +++- 3 files changed, 100 insertions(+), 195 deletions(-) -- 2.47.0

2 13

[PATCH v2 0/8] Take care of clang-tidy warnings with LLVM >= 16
by Stefano Brivio 28 Oct '24

28 Oct '24

So I started hitting some clang-tidy warnings with LLVM 16, some looked bogus, so I upgraded to LLVM 19, and... I got even more. This series takes care of them in different ways. v2: - make snprintf_check() return and set errno on failure, in 2/8 - add missing err_perror() calls on clock_gettime() failures in 6/8 - drop all explicit integer assignments in enum udp_iov_idx in 7/8 Stefano Brivio (8): Makefile: Exclude qrap.c from clang-tidy checks treewide: Comply with CERT C rule ERR33-C for snprintf() treewide: Silence cert-err33-c clang-tidy warnings for fprintf() Makefile: Disable readability-math-missing-parentheses clang-tidy check treewide: Suppress clang-tidy warning if we already use O_CLOEXEC or if we can't treewide: Address cert-err33-c clang-tidy warnings for clock and timer functions udp: Take care of cert-int09-c clang-tidy warning for enum udp_iov_idx util: Don't use errno after a successful call in __daemon() Makefile | 13 ++++++++++--- arch.c | 6 +++++- conf.c | 59 ++++++++++++++++++++++++++++++-------------------------- log.c | 10 +++++++--- passt.c | 9 ++++++--- pasta.c | 11 ++++++++--- pcap.c | 18 +++++++++++------ tap.c | 5 +++-- tcp.c | 12 +++++++++--- udp.c | 10 +++++----- util.c | 49 ++++++++++++++++++++++++++++++++++++---------- util.h | 5 +++++ 12 files changed, 141 insertions(+), 66 deletions(-) -- 2.43.0

2 13

[PATCH] check: remove obsolete images
by Laurent Vivier 25 Oct '24

25 Oct '24

Remove debian-9-nocloud-amd64-daily-20200210-166.qcow2 and openSUSE-Tumbleweed-JeOS.x86_64-kvm-and-xen.qcow2 as they cannot be downloaded anymore Signed-off-by: Laurent Vivier <lvivier(a)redhat.com> --- test/Makefile | 10 +--------- 1 file changed, 1 insertion(+), 9 deletions(-) diff --git a/test/Makefile b/test/Makefile index 35a3b559d4f2..5e4904750653 100644 --- a/test/Makefile +++ b/test/Makefile @@ -8,7 +8,6 @@ WGET = wget -c DEBIAN_IMGS = debian-8.11.0-openstack-amd64.qcow2 \ - debian-9-nocloud-amd64-daily-20200210-166.qcow2 \ debian-10-nocloud-amd64.qcow2 \ debian-10-generic-arm64.qcow2 \ debian-10-generic-ppc64el-20220911-1135.qcow2 \ @@ -42,8 +41,7 @@ OPENSUSE_IMGS = openSUSE-Leap-15.1-JeOS.x86_64-kvm-and-xen.qcow2 \ openSUSE-Leap-15.2-JeOS.x86_64-kvm-and-xen.qcow2 \ openSUSE-Leap-15.3-JeOS.x86_64-kvm-and-xen.qcow2 \ openSUSE-Tumbleweed-ARM-JeOS-efi.aarch64.raw.xz \ - openSUSE-Tumbleweed-ARM-JeOS-efi.armv7l.raw.xz \ - openSUSE-Tumbleweed-JeOS.x86_64-kvm-and-xen.qcow2 + openSUSE-Tumbleweed-ARM-JeOS-efi.armv7l.raw.xz UBUNTU_OLD_IMGS = trusty-server-cloudimg-amd64-disk1.img \ trusty-server-cloudimg-i386-disk1.img \ @@ -135,9 +133,6 @@ realclean: clean debian-8.11.0-openstack-%.qcow2: $(WGET) -O $@ https://cloud.debian.org/images/cloud/OpenStack/archive/8.11.0/debian-8.11.… -debian-9-nocloud-%-daily-20200210-166.qcow2: - $(WGET) -O $@ https://cloud.debian.org/images/cloud/stretch/daily/20200210-166/debian-9-n… - debian-10-nocloud-%.qcow2: $(WGET) -O $@ https://cloud.debian.org/images/cloud/buster/latest/debian-10-nocloud-$*.qc… @@ -203,9 +198,6 @@ openSUSE-Tumbleweed-ARM-JeOS-efi.aarch64.raw.xz: openSUSE-Tumbleweed-ARM-JeOS-efi.armv7l.raw.xz: $(WGET) -O $@ http://download.opensuse.org/ports/armv7hl/tumbleweed/appliances/openSUSE-T… -openSUSE-Tumbleweed-JeOS.x86_64-kvm-and-xen.qcow2: - $(WGET) -O $@ https://download.opensuse.org/tumbleweed/appliances/openSUSE-Tumbleweed-JeO… - # Ubuntu downloads trusty-server-cloudimg-%-disk1.img: $(WGET) -O $@ https://cloud-images.ubuntu.com/trusty/current/trusty-server-cloudimg-$*-di… -- 2.47.0

2 2

[PATCH] tcp: cleanup tcp_buf_data_from_sock()
by Laurent Vivier 25 Oct '24

25 Oct '24

Remove the err label as there is only one caller, and move code to the caller position. ret is not needed here anymore as it is always 0. Remove sendlen as we can user directly len. Signed-off-by: Laurent Vivier <lvivier(a)redhat.com> --- tcp_buf.c | 34 ++++++++++++++++------------------ 1 file changed, 16 insertions(+), 18 deletions(-) diff --git a/tcp_buf.c b/tcp_buf.c index 44df0e492c7b..cb6742ca1a39 100644 --- a/tcp_buf.c +++ b/tcp_buf.c @@ -382,8 +382,8 @@ int tcp_buf_data_from_sock(const struct ctx *c, struct tcp_tap_conn *conn) { uint32_t wnd_scaled = conn->wnd_from_tap << conn->ws_from_tap; int fill_bufs, send_bufs = 0, last_len, iov_rem = 0; - int sendlen, len, dlen, v4 = CONN_V4(conn); - int s = conn->sock, i, ret = 0; + int len, dlen, v4 = CONN_V4(conn); + int s = conn->sock, i; struct msghdr mh_sock = { 0 }; uint16_t mss = MSS_GET(conn); uint32_t already_sent, seq; @@ -453,12 +453,19 @@ int tcp_buf_data_from_sock(const struct ctx *c, struct tcp_tap_conn *conn) len = recvmsg(s, &mh_sock, MSG_PEEK); while (len < 0 && errno == EINTR); - if (len < 0) - goto err; + if (len < 0) { + if (errno != EAGAIN && errno != EWOULDBLOCK) { + tcp_rst(c, conn); + return -errno; + } + + return 0; + } if (!len) { if ((conn->events & (SOCK_FIN_RCVD | TAP_FIN_SENT)) == SOCK_FIN_RCVD) { - if ((ret = tcp_buf_send_flag(c, conn, FIN | ACK))) { + int ret = tcp_buf_send_flag(c, conn, FIN | ACK); + if (ret) { tcp_rst(c, conn); return ret; } @@ -469,19 +476,18 @@ int tcp_buf_data_from_sock(const struct ctx *c, struct tcp_tap_conn *conn) return 0; } - sendlen = len; if (!peek_offset_cap) - sendlen -= already_sent; + len -= already_sent; - if (sendlen <= 0) { + if (len <= 0) { conn_flag(c, conn, STALLED); return 0; } conn_flag(c, conn, ~STALLED); - send_bufs = DIV_ROUND_UP(sendlen, mss); - last_len = sendlen - (send_bufs - 1) * mss; + send_bufs = DIV_ROUND_UP(len, mss); + last_len = len - (send_bufs - 1) * mss; /* Likely, some new data was acked too. */ tcp_update_seqack_wnd(c, conn, false, NULL); @@ -502,12 +508,4 @@ int tcp_buf_data_from_sock(const struct ctx *c, struct tcp_tap_conn *conn) conn_flag(c, conn, ACK_FROM_TAP_DUE); return 0; - -err: - if (errno != EAGAIN && errno != EWOULDBLOCK) { - ret = -errno; - tcp_rst(c, conn); - } - - return ret; } -- 2.47.0

3 2

[PATCH v2 0/3] tcp: Runtime checks for availability of TCP_INFO fields
by David Gibson 25 Oct '24

25 Oct '24

We have a number of Makefile tests for the presence of various fields in tcp_info. However that only tells us if they're present in the kernel headers against which we're compiling, not whether they're available in the kernel we're running on. For anything outside the earliest forms of TCP_INFO (as described in netinet/tcp.h and shared with BSD), we should verify at runtime if the kernel is returning that information. This was supposed to be leading into improved handling of lazily calling TCP_INFO, but I hit additional complications there, so I've postponed it. Might as well get these bits merged, though. v2: * Previous version removed some #ifdefs that were still needed. Simply putting them back leads to awkward duplication between the "compile time missing" and "run time missing" cases, so take a different approach which allows us to remove all of them. * Some wording changes David Gibson (3): tcp: Remove compile-time dependency on struct tcp_info version tcp: Generalise probing for tcpi_snd_wnd field tcp: Use runtime tests for TCP_INFO fields Makefile | 15 ------- tcp.c | 114 ++++++++++++++++++++++------------------------ tcp_info.h | 120 +++++++++++++++++++++++++++++++++++++++++++++++++ tcp_internal.h | 4 +- 4 files changed, 176 insertions(+), 77 deletions(-) create mode 100644 tcp_info.h -- 2.47.0

2 4