April 2025 - dev

[PATCH 0/3] Add vhost-net kernel support
by Eugenio Pérez 03 Apr '25

03 Apr '25

vhost-net is a kernel device that allows to read packets from a tap device using virtio queues instead of regular read(2) and write(2). This enables a more eficient packet processing, as the memory can be written directly by the kernel to the userspace and back, instead of wasting bandwith on copies, and it enables to batch many packets in a single notification (through eventfds) both tx and rx. --- RFC: At this moment only to receive packets from the vhost kernel to the pasta binary is supported. It does not even support to refill more rx descriptors so the kernel can keep receiving packets! Just for early review to make sure we're on the same page. Eugenio Pérez (3): tap: specify the packet pool tap: implement vhost_call_cb tap: add die() on vhost error epoll_type.h | 4 + passt.c | 8 ++ passt.h | 13 +- tap.c | 366 +++++++++++++++++++++++++++++++++++++++++++++++++-- tap.h | 6 +- vu_common.c | 7 +- 6 files changed, 387 insertions(+), 17 deletions(-) -- 2.49.0

4 12

[PATCH 0/3] udp: Further preliminaries towards using connected sockets on both sides
by David Gibson 03 Apr '25

03 Apr '25

Some additional small cleanups David Gibson (3): udp: Remove redundant udp_at_sidx() call in udp_tap_handler() udp: Simplify updates to UDP flow timestamp udp: Correct some seccomp filter annotations udp.c | 23 +++++++++-------------- udp_vu.c | 9 +-------- udp_vu.h | 3 +-- 3 files changed, 11 insertions(+), 24 deletions(-) -- 2.49.0

2 6

[PATCH] passt-repair: Correct off-by-one error verifying name
by David Gibson 02 Apr '25

02 Apr '25

passt-repair will generate an error if the name it gets from the kernel is too long or not NUL terminated. Downstream testing has reported occasionally seeing this error in practice. In turns out there is a trivial off-by-one error in the check: ev->len is the length of the name, including terminating \0 characters, so to check for a \0 at the end of the buffer we need to check ev->name[len - 1] not ev->name[len]. Fixes: 42a854a52 ("pasta, passt-repair: Support multiple events per...") Signed-off-by: David Gibson <david(a)gibson.dropbear.id.au> --- passt-repair.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/passt-repair.c b/passt-repair.c index 86f02934..440c77ae 100644 --- a/passt-repair.c +++ b/passt-repair.c @@ -157,7 +157,7 @@ int main(int argc, char **argv) } } while (!found); - if (ev->len > NAME_MAX + 1 || ev->name[ev->len] != '\0') { + if (ev->len > NAME_MAX + 1 || ev->name[ev->len - 1] != '\0') { fprintf(stderr, "Invalid filename from inotify\n"); _exit(1); } -- 2.49.0

2 1

[PATCH 0/4] RFC: bind() migrated connections in repair mode
by David Gibson 02 Apr '25

02 Apr '25

Downstream testing recently discovered that inbound connections can't be migrated properly, because the new socket gets an address conflict with its corresponding listening socket. It turns out this can be avoided by delaying bind() until after we're already in repair mode. Patch 4/4 is the actual fix here. Patch 3/4 adds a test program checking the behaviour to doc/platform-requirements. Patches 1 & 2 fix minor problems I spotted in doc/platform-requirements writing 3/4. Only 4/4 will need to be backported. David Gibson (4): platform requirements: Fix clang-tidy warning platform requirements: Add attributes to die() function platform requirements: Add test for address conflicts with TCP_REPAIR migrate, tcp: bind() migrated sockets in repair mode doc/platform-requirements/.gitignore | 1 + doc/platform-requirements/Makefile | 4 +- doc/platform-requirements/common.h | 1 + doc/platform-requirements/listen-vs-repair.c | 128 ++++++++++++++++++ .../reuseaddr-priority.c | 6 +- tcp.c | 38 ++++-- 6 files changed, 162 insertions(+), 16 deletions(-) create mode 100644 doc/platform-requirements/listen-vs-repair.c -- 2.49.0

2 5