- dev - passt

[PATCH 0/5] Cleanups to tcp socket pool handling
by David Gibson 23 Jan '23

23 Jan '23

This is a bit of a diversion from what I'm notionally working on at the moment. While thinking about what we'd need to use the IP_TRANSPARENT socket option to broaden the cases where we can "splice", I noticed some inelegancies in how we handle the pool of pre-opened sockets in the TCP code, and well, here we are. This makes a number of cleanups to the handling of these pools. Most notably, tcp_splice_connect() now has simpler semantics: it now always runs in the init namespace, and is always given a pre-opened socket (which could be in the guest ns). David Gibson (5): tcp: Make a helper to refill each socket pool tcp: Split init and ns cases for tcp_sock_refill() tcp: Move socket pool declarations around tcp: Split pool lookup from creating new sockets in tcp_conn_new_sock() tcp: Improve handling of fallback if socket pool is empty on new splice tcp.c | 138 ++++++++++++++++++------------------------------- tcp.h | 2 - tcp_conn.h | 12 ++++- tcp_splice.c | 141 ++++++++++++++++++++++++++------------------------- 4 files changed, 132 insertions(+), 161 deletions(-) -- 2.39.0

2 12

[PATCH v4 0/8] Don't use additional sockets for receiving "spliced" UDP communications
by David Gibson 13 Jan '23

13 Jan '23

At present, the UDP "splice" and "tap" paths are quite separate. We have separate sockets to receive packets bound for the tap and splice paths. This leads to some code duplication, and extra open sockets. This series partially unifies the two paths, allowing us to use a single (host side) socket, bound to 0.0.0.0 or :: to receive packets for both cases. Changes since v3: * Fixed really dumb compile error, and actually ran through the tests. Oops. Changes since v2: * Don't receive multiple packets at once for pasta mode - seems to hurt throughput on balance. * Add some comments clarifying reasoning here. Changes since v1: * Renamed udp_localname[46] to udp[46]_localname * Allow handling of UDP port 0 * Fix a bug which could misidentify certain v6 packets as v4-spliceable * Some minor cosmetic fixes to code and commit messages David Gibson (8): udp: Move sending pasta tap frames to the end of udp_sock_handler() udp: Split sending to passt tap interface into separate function udp: Split receive from preparation and send in udp_sock_handler() udp: Don't handle tap receive batch size calculation within a #define udp: Pre-populate msg_names with local address udp: Unify udp_sock_handler_splice() with udp_sock_handler() udp: Decide whether to "splice" per datagram rather than per socket udp: Don't use separate sockets to listen for spliced packets udp.c | 388 ++++++++++++++++++++++++++++++--------------------------- udp.h | 2 +- util.h | 7 ++ 3 files changed, 213 insertions(+), 184 deletions(-) -- 2.39.0

2 11

[libvirt PATCH 0/9] Support libvirt-managed QEMU domain <interface> backed by a passt process
by Laine Stump 12 Jan '23

12 Jan '23

passt (https://passt.top) provides a method for connecting a guest to the larger network without requiring any elevated privileges. This set of patches allows libvirt/QEMU users to easily configure a QEMU domain to use passt for the backend of any emulated network interface. More details are in the individual patches, but the short explanation is that you will use: <interface type='user'> <backend type='passt'> ... to select the passt backend. (We decided to do it this way since the concept is so similar to slirp, which was the original "type='user'") The following caveats currently apply: 1) passt support requires "-netdev stream" in QEMU, which is only available starting with qemu-7.2.0. So if you want to test these patches out, you need the latest upstream release of QEMU. 2) SELinux must be set to "permissive". This is of course temporary. As I understand it, the remedy to this is a new SELinux profile for the passt binary, which is outside the control of libvirt and so not something that can be addressed in this patchset (or any other patch to libvirt). 3) Although there is a a new option for QEMU's -netdev that will tell QEMU to attempt to reconnect to a new incarnation of the same socket if passt unexpectedly exits, and a new QEMU event that will be put into QEMU to inform libvirt that the passt process has exited (so that it can start a new, identical passt process), I think this hasn't been pushed upstream yet (??), and I haven't implemented any support for it here. So, if the passt process unexpectedly exits, the guest will be without networking. However, Stefano (passt author) is emphatic that passt will never unexpectedly exit :-) passt has *many* other options that libvirt could support, but the small subset here are the things that seem most useful (and/or were specifically requested by prospective users of passt). It is always easier to add more stuff in the future than to remove "mistakes", so I tried to no go overboard. Laine Stump (9): conf: rename virDomainNetBackend* to virDomainNetDriver* conf: move anonymous backend struct from virDomainNetDef into its own struct conf: put interface <backend> parsing/formatting separate functions conf: add passt XML additions to schema conf: parse/format passt-related XML additions qemu: new capability QEMU_CAPS_NETDEV_STREAM qemu: add passtStateDir to qemu driver config qemu: hook up passt config to qemu domains specfile: require passt for the build if fedora >= 36 or rhel >= 9 docs/formatdomain.rst | 95 +++++- libvirt.spec.in | 7 + meson.build | 1 + po/POTFILES | 1 + src/conf/domain_conf.c | 303 ++++++++++++++++-- src/conf/domain_conf.h | 64 +++- src/conf/domain_validate.c | 32 +- src/conf/schemas/domaincommon.rng | 65 ++++ src/conf/virconftypes.h | 6 + src/libvirt_private.syms | 1 + src/qemu/meson.build | 2 + src/qemu/qemu_capabilities.c | 4 + src/qemu/qemu_capabilities.h | 3 + src/qemu/qemu_command.c | 11 +- src/qemu/qemu_command.h | 3 +- src/qemu/qemu_conf.c | 2 + src/qemu/qemu_conf.h | 1 + src/qemu/qemu_domain.c | 5 +- src/qemu/qemu_domain.h | 3 +- src/qemu/qemu_driver.c | 12 + src/qemu/qemu_extdevice.c | 25 +- src/qemu/qemu_hotplug.c | 26 +- src/qemu/qemu_interface.c | 8 +- src/qemu/qemu_passt.c | 284 ++++++++++++++++ src/qemu/qemu_passt.h | 38 +++ src/qemu/qemu_process.c | 1 + src/qemu/qemu_validate.c | 9 +- src/security/virt-aa-helper.c | 2 +- .../caps_7.2.0.x86_64.xml | 1 + tests/qemuxml2argvdata/net-user-passt.args | 34 ++ .../net-user-passt.x86_64-latest.args | 37 +++ tests/qemuxml2argvdata/net-user-passt.xml | 57 ++++ tests/qemuxml2argvtest.c | 2 + tests/qemuxml2xmloutdata/net-user-passt.xml | 1 + tests/qemuxml2xmltest.c | 1 + 35 files changed, 1087 insertions(+), 60 deletions(-) create mode 100644 src/qemu/qemu_passt.c create mode 100644 src/qemu/qemu_passt.h create mode 100644 tests/qemuxml2argvdata/net-user-passt.args create mode 100644 tests/qemuxml2argvdata/net-user-passt.x86_64-latest.args create mode 100644 tests/qemuxml2argvdata/net-user-passt.xml create mode 120000 tests/qemuxml2xmloutdata/net-user-passt.xml -- 2.38.1

5 27

[PATCH] Correct port ranges in --tcp-ports examples
by Laine Stump 09 Jan '23

09 Jan '23

The proper syntax is: ${start}-${end}:${dststart}-${dstend} not ${start}-${end}-${dststart}:${dstend} Signed-off-by: Laine Stump <laine(a)redhat.com> --- passt.1 | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/passt.1 b/passt.1 index 528763b..0dad878 100644 --- a/passt.1 +++ b/passt.1 @@ -349,7 +349,7 @@ Forward local ports 22 and 25 to ports 22 and 25 on the guest -t 22-80 Forward local ports 22 to 80 to corresponding ports on the guest .TP --t 22-80-32:90 +-t 22-80:32-90 Forward local ports 22 to 80 to corresponding ports on the guest plus 10 .TP -t 192.0.2.1/22 @@ -361,7 +361,7 @@ Forward local port 22, bound to 192.0.2.1 and interface eth0, to port 22 -t 2000-5000,~3000-3010 Forward local ports 2000 to 5000, but not 3000 to 3010 .TP --t 192.0.2.1/20-30,~25 +-t 192.0.2.1/20-26:24-30,~25-29 Forward local ports 20 to 24, and 26 to 30, bound to 192.0.2.1 .TP -t ~20000-20010 @@ -427,7 +427,7 @@ Forward local ports 22 and 25 to ports 22 and 25 in the target namespace -t 22-80 Forward local ports 22 to 80 to corresponding ports in the target namespace .TP --t 22-80-32:90 +-t 22-80:32-90 Forward local ports 22 to 80 to corresponding ports plus 10 in the target namespace .TP @@ -440,7 +440,7 @@ Forward local port 22, bound to 192.0.2.1 and interface eth0, to port 22 -t 2000-5000,~3000-3010 Forward local ports 2000 to 5000, but not 3000 to 3010 .TP --t 192.0.2.1/20-30,~25 +-t 192.0.2.1/20-26:24-30,~25-29 Forward local ports 20 to 24, and 26 to 30, bound to 192.0.2.1 .TP -t ~20000-20010 -- 2.38.1

2 1

[PATCH v2 00/18] RFC: Unify and simplify tap send path
by David Gibson 05 Jan '23

05 Jan '23

Although we have an abstraction for the "slow path" (DHCP, NDP) guest bound packets, the TCP and UDP forwarding paths write directly to the tap fd. However, it turns out how they send frames to the tap device is more similar than it originally appears. This series unifies the low-level tap send functions for TCP and UDP, and makes some clean ups along the way. This is based on my earlier outstanding series. Changes since v1: * Abstract tap header construction as well as frame send (a number of new patches) * Remove unneeded flags buf_bytes globals as well * Fix bug where we weren't correctly setting iov_len after the move to giving variable sized iovecs to send_frames(). David Gibson (18): pcap: Introduce pcap_frame() helper pcap: Replace pcapm() with pcap_multiple() tcp: Combine two parts of passt tap send path together tcp: Don't keep compute total bytes in a message until we need it tcp: Improve interface to tcp_l2_buf_flush() tcp: Combine two parts of pasta tap send path together tap, tcp: Move tap send path to tap.c util: Introduce hton*_constant() in place of #ifdefs tcp, udp: Use named field initializers in iov_init functions util: Parameterize ethernet header initializer macro tcp: Remove redundant and incorrect initialization from *_iov_init() tcp: Consolidate calculation of total frame size tap: Add "tap headers" abstraction tcp: Use abstracted tap header tap: Use different io vector bases depending on tap type udp: Use abstracted tap header udp: Use tap_send_frames() tap: Improve handling of partial frame sends dhcpv6.c | 50 +++-------- pcap.c | 78 +++++------------ pcap.h | 3 +- tap.c | 123 +++++++++++++++++++++++++++ tap.h | 54 ++++++++++++ tcp.c | 254 +++++++++++++------------------------------------------ udp.c | 213 ++++++---------------------------------------- udp.h | 2 +- util.h | 47 ++-------- 9 files changed, 303 insertions(+), 521 deletions(-) -- 2.38.1

2 29

[PATCH] tcp: Explicitly check option length field values in tcp_opt_get()
by Stefano Brivio 05 Jan '23

05 Jan '23

Reported by Coverity (CWE-606, Untrusted loop bound), and actually harmless because we'll exit the option-scanning loop if the remaining length is not enough for a new option, instead of reading past the header. In any case, it looks like a good idea to explicitly check for reasonable values of option lengths. Signed-off-by: Stefano Brivio <sbrivio(a)redhat.com> --- tcp.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/tcp.c b/tcp.c index 26037b3..1e0a338 100644 --- a/tcp.c +++ b/tcp.c @@ -1117,6 +1117,10 @@ static int tcp_opt_get(const char *opts, size_t len, uint8_t type_find, break; default: type = *(opts++); + + if (*(uint8_t *)opts < 2 || *(uint8_t *)opts > len) + return -1; + optlen = *(opts++) - 2; len -= 2; -- 2.35.1

2 1

[PATCH v3 0/8] Don't use additional sockets for receiving "spliced" UDP communications
by David Gibson 05 Jan '23

05 Jan '23

At present, the UDP "splice" and "tap" paths are quite separate. We have separate sockets to receive packets bound for the tap and splice paths. This leads to some code duplication, and extra open sockets. This series partially unifies the two paths, allowing us to use a single (host side) socket, bound to 0.0.0.0 or :: to receive packets for both cases. Changes since v2: * Don't receive multiple packets at once for pasta mode - seems to hurt throughput on balance. * Add some comments clarifying reasoning here. Changes since v1: * Renamed udp_localname[46] to udp[46]_localname * Allow handling of UDP port 0 * Fix a bug which could misidentify certain v6 packets as v4-spliceable * Some minor cosmetic fixes to code and commit messages David Gibson (8): udp: Move sending pasta tap frames to the end of udp_sock_handler() udp: Split sending to passt tap interface into separate function udp: Split receive from preparation and send in udp_sock_handler() udp: Don't handle tap receive batch size calculation within a #define udp: Pre-populate msg_names with local address udp: Unify udp_sock_handler_splice() with udp_sock_handler() udp: Decide whether to "splice" per datagram rather than per socket udp: Don't use separate sockets to listen for spliced packets udp.c | 389 ++++++++++++++++++++++++++++++--------------------------- udp.h | 2 +- util.h | 7 ++ 3 files changed, 214 insertions(+), 184 deletions(-) -- 2.39.0

2 10

[PATCH] test/perf/pasta_tcp: Add host to namespace cases for traffic via tap
by Stefano Brivio 05 Jan '23

05 Jan '23

Similarly to UDP cases, these were missing as it wasn't clear, when the other tests were introduced, if using the global address of a namespace, from the host, should have resulted in connections being routed via the tap interface. Signed-off-by: Stefano Brivio <sbrivio(a)redhat.com> --- test/perf/pasta_tcp | 57 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 57 insertions(+) diff --git a/test/perf/pasta_tcp b/test/perf/pasta_tcp index 1847c83..8ebb25c 100644 --- a/test/perf/pasta_tcp +++ b/test/perf/pasta_tcp @@ -240,4 +240,61 @@ nsout LAT tcp_crr --nolog -P 10003 -C 10013 -4 -c -H __GW__ | sed -n 's/^through hostw lat __LAT__ 1500 500 + +tr TCP throughput over IPv6: host to ns +nsout IFNAME ip -j link show | jq -rM '.[] | select(.link_type == "ether").ifname' +nsout ADDR6 ip -j -6 addr show|jq -rM '.[] | select(.ifname == "__IFNAME__").addr_info[] | select(.scope == "global" and .prefixlen == 64).local' +bw - +bw - +bw - +iperf3 BW host ns __ADDR6__ 100${i}2 __THREADS__ __TIME__ __OPTS__ +bw __BW__ 8.0 10.0 + +tl TCP RR latency over IPv6: host to ns +lat - +lat - +lat - +nsb tcp_rr --nolog -P 10002 -C 10012 -6 +hout LAT tcp_rr --nolog -P 10002 -C 10012 -6 -c -H __ADDR6__ | sed -n 's/^throughput=$.*$/\1/p' +nsw +lat __LAT__ 150 100 + +tl TCP CRR latency over IPv6: host to ns +lat - +lat - +lat - +sleep 1 +nsb tcp_crr --nolog -P 10002 -C 10012 -6 +hout LAT tcp_crr --nolog -P 10002 -C 10012 -6 -c -H __ADDR6__ | sed -n 's/^throughput=$.*$/\1/p' +nsw +lat __LAT__ 5000 10000 + + +tr TCP throughput over IPv4: host to ns +nsout ADDR ip -j -4 addr show|jq -rM '.[] | select(.ifname == "__IFNAME__").addr_info[0].local' +bw - +bw - +bw - +iperf3 BW host ns __ADDR__ 100${i}2 __THREADS__ __TIME__ __OPTS__ +bw __BW__ 8.0 10.0 + +tl TCP RR latency over IPv4: host to ns +lat - +lat - +lat - +nsb tcp_rr --nolog -P 10002 -C 10012 -4 +hout LAT tcp_rr --nolog -P 10002 -C 10012 -4 -c -H __ADDR__ | sed -n 's/^throughput=$.*$/\1/p' +nsw +lat __LAT__ 150 100 + +tl TCP CRR latency over IPv4: host to ns +lat - +lat - +lat - +sleep 1 +nsb tcp_crr --nolog -P 10002 -C 10012 -4 +hout LAT tcp_crr --nolog -P 10002 -C 10012 -4 -c -H __ADDR__ | sed -n 's/^throughput=$.*$/\1/p' +nsw +lat __LAT__ 5000 10000 + te -- 2.35.1

2 1

[PATCH 0/8] Don't use additional sockets for receiving "spliced" UDP communications
by David Gibson 04 Jan '23

04 Jan '23

At present, the UDP "splice" and "tap" paths are quite separate. We have separate sockets to receive packets bound for the tap and splice paths. This leads to some code duplication, and extra open sockets. This series partially unifies the two paths, allowing us to use a single (host side) socket, bound to 0.0.0.0 or :: to receive packets for both cases. This is based on my earlier series with some fixes for the tap path. David Gibson (8): udp: Move sending pasta tap frames to the end of udp_sock_handler() udp: Split sending to passt tap interface into separate function udp: Split receive from preparation and send in udp_sock_handler() udp: Receive multiple datagrams at once on the pasta sock->tap path udp: Pre-populate msg_names with local address udp: Unify udp_sock_handler_splice() with udp_sock_handler() udp: Decide whether to "splice" per datagram rather than per socket udp: Don't use separate sockets to listen for spliced packets udp.c | 382 ++++++++++++++++++++++++++++++--------------------------- udp.h | 2 +- util.h | 7 ++ 3 files changed, 207 insertions(+), 184 deletions(-) -- 2.38.1

2 27

[PATCH] test/perf/pasta_udp: Add host to namespace cases for traffic via tap
by Stefano Brivio 04 Jan '23

04 Jan '23

These were missing as it wasn't clear, when the other tests were introduced, if using the global address of a namespace, from the host, should have resulted in traffic being routed via the tap interface (as opposed to the loopback interface). We now clarified that's actually the case. Use same values and thresholds as the tests for loopback traffic, as throughput figures currently indicate there isn't much difference. Signed-off-by: Stefano Brivio <sbrivio(a)redhat.com> --- Contrary to what I stated a while ago, there's actually no need to adapt the "code" writing the JavaScript snippets for performance reports -- the trick is simply to keep those ordered like the previous ones, and we'll get the right output. test/perf/pasta_udp | 37 +++++++++++++++++++++++++++++++++++++ 1 file changed, 37 insertions(+) diff --git a/test/perf/pasta_udp b/test/perf/pasta_udp index 27ea724..112d70d 100644 --- a/test/perf/pasta_udp +++ b/test/perf/pasta_udp @@ -179,4 +179,41 @@ nsout LAT udp_rr --nolog -P 10003 -C 10013 -4 -c -H __GW__ | sed -n 's/^throughp hostw lat __LAT__ 200 150 + +tr UDP throughput over IPv6: host to ns +nsout IFNAME ip -j link show | jq -rM '.[] | select(.link_type == "ether").ifname' +nsout ADDR6 ip -j -6 addr show|jq -rM '.[] | select(.ifname == "__IFNAME__").addr_info[] | select(.scope == "global" and .prefixlen == 64).local' +bw - +bw - +bw - +iperf3 BW host ns __ADDR6__ 100${i}2 __THREADS__ __TIME__ __OPTS__ -b 15G +bw __BW__ 7.0 9.0 + +tl UDP RR latency over IPv6: host to ns +lat - +lat - +lat - +nsb udp_rr --nolog -P 10002 -C 10012 -6 +hout LAT udp_rr --nolog -P 10002 -C 10012 -6 -c -H __ADDR6__ | sed -n 's/^throughput=$.*$/\1/p' +nsw +lat __LAT__ 200 150 + + +tr UDP throughput over IPv4: host to ns +nsout ADDR ip -j -4 addr show|jq -rM '.[] | select(.ifname == "__IFNAME__").addr_info[0].local' +bw - +bw - +bw - +iperf3 BW host ns __ADDR__ 100${i}2 __THREADS__ __TIME__ __OPTS__ -b 15G +bw __BW__ 7.0 9.0 + +tl UDP RR latency over IPv4: host to ns +lat - +lat - +lat - +nsb udp_rr --nolog -P 10002 -C 10012 -4 +hout LAT udp_rr --nolog -P 10002 -C 10012 -4 -c -H __ADDR__ | sed -n 's/^throughput=$.*$/\1/p' +nsw +lat __LAT__ 200 150 + te -- 2.35.1

2 1