bee/audit/internal/collector/nic_mellanox.go

package collector

import (
	"bee/audit/internal/schema"
	"context"
	"log/slog"
	"os"
	"os/exec"
	"path/filepath"
	"strings"
	"time"
)

const nicProbeTimeout = 2 * time.Second

var (
	mstflintQuery = func(bdf string) (string, error) {
		out, err := commandOutputWithTimeout(nicProbeTimeout, "mstflint", "-d", bdf, "q")
		if err != nil {
			return "", err
		}
		return string(out), nil
	}

	ethtoolInfoQuery = func(iface string) (string, error) {
		out, err := commandOutputWithTimeout(nicProbeTimeout, "ethtool", "-i", iface)
		if err != nil {
			return "", err
		}
		return string(out), nil
	}

	netIfacesByBDF = listNetIfacesByBDF
	readNetCarrierFile = func(iface string) (string, error) {
		path := filepath.Join("/sys/class/net", iface, "carrier")
		raw, err := os.ReadFile(path)
		if err != nil {
			return "", err
		}
		return strings.TrimSpace(string(raw)), nil
	}
)

// enrichPCIeWithMellanox enriches Mellanox/NVIDIA Networking devices with
// firmware/serial information from mstflint, with ethtool fallback for firmware.
func enrichPCIeWithMellanox(devs []schema.HardwarePCIeDevice) []schema.HardwarePCIeDevice {
	enriched := 0
	for i := range devs {
		if !isMellanoxDevice(devs[i]) {
			continue
		}

		bdf := ""
		if devs[i].BDF != nil {
			bdf = normalizePCIeBDF(*devs[i].BDF)
		}
		if bdf == "" {
			continue
		}

		fw, serial := queryMellanoxFromMstflint(bdf)
		if fw == "" {
			fw = queryFirmwareFromEthtool(bdf)
		}

		if fw != "" {
			devs[i].Firmware = &fw
		}
		if serial != "" {
			devs[i].SerialNumber = &serial
		}
		if fw != "" || serial != "" {
			enriched++
		}
	}

	slog.Info("mellanox: enriched", "count", enriched)
	return devs
}

func isMellanoxDevice(dev schema.HardwarePCIeDevice) bool {
	return dev.VendorID != nil && *dev.VendorID == MellanoxVendorID
}

func queryMellanoxFromMstflint(bdf string) (firmware, serial string) {
	out, err := mstflintQuery(bdf)
	if err != nil {
		return "", ""
	}
	return parseMstflintQuery(out)
}

func parseMstflintQuery(raw string) (firmware, serial string) {
	for _, line := range strings.Split(raw, "\n") {
		line = strings.TrimSpace(line)
		if line == "" {
			continue
		}
		idx := strings.Index(line, ":")
		if idx < 0 {
			continue
		}
		key := strings.ToLower(strings.TrimSpace(line[:idx]))
		val := strings.TrimSpace(line[idx+1:])
		switch key {
		case "fw version":
			if val != "" {
				firmware = val
			}
		case "board serial number":
			if val != "" {
				serial = val
			}
		}
	}
	return firmware, serial
}

func queryFirmwareFromEthtool(bdf string) string {
	for _, iface := range netIfacesByBDF(bdf) {
		out, err := ethtoolInfoQuery(iface)
		if err != nil {
			continue
		}
		if fw := parseEthtoolFirmwareInfo(out); fw != "" {
			return fw
		}
	}
	return ""
}

func parseEthtoolFirmwareInfo(raw string) string {
	for _, line := range strings.Split(raw, "\n") {
		line = strings.TrimSpace(line)
		if line == "" {
			continue
		}
		idx := strings.Index(line, ":")
		if idx < 0 {
			continue
		}
		key := strings.ToLower(strings.TrimSpace(line[:idx]))
		val := strings.TrimSpace(line[idx+1:])
		if key == "firmware-version" && val != "" {
			return val
		}
	}
	return ""
}

func listNetIfacesByBDF(bdf string) []string {
	path := filepath.Join("/sys/bus/pci/devices", bdf, "net")
	entries, err := os.ReadDir(path)
	if err != nil {
		return nil
	}
	ifaces := make([]string, 0, len(entries))
	for _, e := range entries {
		if e.Name() == "" {
			continue
		}
		ifaces = append(ifaces, e.Name())
	}
	return ifaces
}

func commandOutputWithTimeout(timeout time.Duration, name string, args ...string) ([]byte, error) {
	ctx, cancel := context.WithTimeout(context.Background(), timeout)
	defer cancel()
	return exec.CommandContext(ctx, name, args...).Output()
}