From 3dbc48727eb34fc1e62d452c1cd18d5598597104 Mon Sep 17 00:00:00 2001
From: henrygd <hank@henrygd.me>
Date: Fri, 31 Oct 2025 19:09:53 -0400
Subject: [PATCH] add INTEL_GPU_DEVICE env var (#1285)

---
 agent/gpu_intel.go        |  7 ++++++-
 agent/gpu_test.go         | 41 +++++++++++++++++++++++++++++++++++++++
 supplemental/CHANGELOG.md |  4 ++++
 3 files changed, 51 insertions(+), 1 deletion(-)

diff --git a/agent/gpu_intel.go b/agent/gpu_intel.go
index 2a718a74..31c982c3 100644
--- a/agent/gpu_intel.go
+++ b/agent/gpu_intel.go
@@ -49,7 +49,12 @@ func (gm *GPUManager) updateIntelFromStats(sample *intelGpuStats) bool {
 
 // collectIntelStats executes intel_gpu_top in text mode (-l) and parses the output
 func (gm *GPUManager) collectIntelStats() (err error) {
-	cmd := exec.Command(intelGpuStatsCmd, "-s", intelGpuStatsInterval, "-l")
+	// Build command arguments, optionally selecting a device via -d
+	args := []string{"-s", intelGpuStatsInterval, "-l"}
+	if dev, ok := GetEnv("INTEL_GPU_DEVICE"); ok && dev != "" {
+		args = append(args, "-d", dev)
+	}
+	cmd := exec.Command(intelGpuStatsCmd, args...)
 	// Avoid blocking if intel_gpu_top writes to stderr
 	cmd.Stderr = io.Discard
 	stdout, err := cmd.StdoutPipe()
diff --git a/agent/gpu_test.go b/agent/gpu_test.go
index d204cf77..748037b7 100644
--- a/agent/gpu_test.go
+++ b/agent/gpu_test.go
@@ -4,8 +4,10 @@
 package agent
 
 import (
+	"fmt"
 	"os"
 	"path/filepath"
+	"strings"
 	"testing"
 	"time"
 
@@ -1624,3 +1626,42 @@ func TestParseIntelData(t *testing.T) {
 		})
 	}
 }
+
+func TestIntelCollectorDeviceEnv(t *testing.T) {
+	dir := t.TempDir()
+	t.Setenv("PATH", dir)
+
+	// Prepare a file to capture args
+	argsFile := filepath.Join(dir, "args.txt")
+
+	// Create a fake intel_gpu_top that records its arguments and prints minimal valid output
+	scriptPath := filepath.Join(dir, "intel_gpu_top")
+	script := fmt.Sprintf(`#!/bin/sh
+echo "$@" > %s
+echo "Freq MHz      IRQ RC6     Power W     IMC MiB/s             RCS             VCS"
+echo " req  act       /s   %%   gpu   pkg     rd     wr       %%  se  wa       %%  se  wa"
+echo "226  223      338  58  2.00  2.69   1820    965   0.00    0   0    0.00   0   0"
+echo "189  187      412  67  1.80  2.45   1950    823   8.50    2   1    15.00   1   0"
+`, argsFile)
+	if err := os.WriteFile(scriptPath, []byte(script), 0755); err != nil {
+		t.Fatal(err)
+	}
+
+	// Set device selector via prefixed env var
+	t.Setenv("BESZEL_AGENT_INTEL_GPU_DEVICE", "sriov")
+
+	gm := &GPUManager{GpuDataMap: make(map[string]*system.GPUData)}
+	if err := gm.collectIntelStats(); err != nil {
+		t.Fatalf("collectIntelStats error: %v", err)
+	}
+
+	// Verify that -d sriov was passed
+	data, err := os.ReadFile(argsFile)
+	if err != nil {
+		t.Fatalf("failed reading args file: %v", err)
+	}
+	argsStr := strings.TrimSpace(string(data))
+	require.Contains(t, argsStr, "-d sriov")
+	require.Contains(t, argsStr, "-s ")
+	require.Contains(t, argsStr, "-l")
+}
diff --git a/supplemental/CHANGELOG.md b/supplemental/CHANGELOG.md
index 5953cf57..25fd9215 100644
--- a/supplemental/CHANGELOG.md
+++ b/supplemental/CHANGELOG.md
@@ -1,3 +1,7 @@
+## 0.15.4
+
+- Add `INTEL_GPU_DEVICE` environment variable to select Intel GPU device. (#1285)
+
 ## 0.15.3
 
 - Improve parsing of edge case S.M.A.R.T. power on times. (#1347)