WithBindCapability assumed the kernel's privileged port range ends at 1024, but it is set by net.ipv4.ip_unprivileged_port_start, which can be anywhere from 0 (Docker's default) to 65535. With a value above 1024, a non-root bind to e.g. :1500 failed even though the capability was permitted. Now listen is called normally first, and only if it fails with EACCES on a TCP/UDP port, and CAP_NET_BIND_SERVICE is permitted but not effective, the capability is raised and the bind retried. This also skips raising it when the range starts lower than the port. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|---|---|---|
| .. | ||
| filesystems | ||
| metrics | ||
| testmocks | ||
| bindcap_linux.go | ||
| bindcap_linux_test.go | ||
| bindcap_other.go | ||
| logbuffer.go | ||
| logmarshalers.go | ||
| logs.go | ||
| logs_test.go | ||
| ranges.go | ||
| sockets.go | ||