Commit Graph
2 Commits
Author SHA1 Message Date
Brad Fitzpatrick 80c71a6b06 kernel/linux, tstest: find the tswg module in this repo
Now that the tswg kernel module is under kernel/linux, the tests and
scripts that build it default to that tree rather than a separate
~/src/tslinux checkout, and the scripts find the tailscale.com
checkout they're in. TSWG_DIR and TAILSCALE_DIR still override them.

Also rename the module's COPYING to LICENSE, give the ML-KEM test the
repo's license header (and skip it without a C compiler, now that it
runs with the rest of the module's tests), and update DESIGN.md for
the new layout, which no longer has a pristine upstream/ copy.

Updates #426

Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: Ifcd99dd6d2197550d9c4abd13d4d4d80ed26dc60
2026-11-09 15:47:37 +00:00
Brad Fitzpatrick 7648a28d11 kernel/linux/{scripts,DESIGN.md}: add leanbench.sh and record lean benchmark results
leanbench.sh runs tailscale.com's new lean data plane benchmark, which
compares wireguard-go, tswg, the kernel's WireGuard and no tunnel between
two QEMU VMs on a host bridge with vhost-net, without natlab, over IPv4
and IPv6. With -smoke it runs the quick correctness test instead.

The results overturn the natlab ones, which natlab's userspace network
had capped: kernel WireGuard (and tswg, identically) gets about a third
of wireguard-go's bulk TCP throughput, at 1.5x to 2.5x the CPU per byte,
without being CPU-bound. So outer UDP GSO/GRO and finding where the
kernel serializes are now the critical performance work.

Updates #426

Signed-off-by: Brad Fitzpatrick <[email protected]>
Change-Id: I2a24de8d82370fc8dfb6a9af0971da20831ea307
2026-11-07 06:06:16 +00:00