fix RDMA char device accumulation across multiple devices

[gauravkghildiyal]

What type of PR is this?

/kind bug
/kind cleanup

What this PR does / why we need it:

Previously, charDevices was declared outside the device loop in prepareResourceClaim, causing character device paths to accumulate across iterations when a single ResourceClaim allocated multiple RDMA devices. This change ensures each DeviceConfig stored in podConfigStore only contains its own character devices

Which issue(s) this PR is related to:

Special notes for your reviewer:

Does this PR introduce a user-facing change?

NONE

[netlify]

✅ Deploy Preview for dranet canceled.

Name	Link
🔨 Latest commit	aa168c8
🔍 Latest deploy log	https://app.netlify.com/projects/dranet/deploys/6a7599f5d128f7000801c945

[aojea]

good catch, reproduced with

--- a/pkg/driver/dra_hooks_test.go
+++ b/pkg/driver/dra_hooks_test.go
@@ -32,6 +32,7 @@
     k8sresource "k8s.io/apimachinery/pkg/api/resource"
     metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
     "k8s.io/apimachinery/pkg/types"
+	"k8s.io/apimachinery/pkg/util/sets"
     "k8s.io/client-go/tools/record"
     "k8s.io/dynamic-resource-allocation/kubeletplugin"
     "k8s.io/utils/ptr"
@@ -954,3 +955,24 @@
         })
     }
 }
+
+func TestBuildRDMAConfigCharDeviceAccumulation(t *testing.T) {
+	// Regression test for the RDMA char device accumulation bug: charDevices is
+	// declared once outside the device loop in prepareResourceClaim and passed by
+	// reference into buildRDMAConfig for every device in the claim, so character
+	// device paths inserted while processing one device leak into the RDMAConfig
+	// computed for the next device.
+	//
+	// "/dev/null" stands in for a char device path belonging to a device
+	// processed earlier in the loop; it's a stock character device present on
+	// any Linux system, so the test needs no real RDMA hardware.
+	charDevices := sets.New[string]("/dev/null")
+
+	cfg := buildRDMAConfig("mlx5_1", charDevices)
+
+	for _, dc := range cfg.DevChars {
+		if dc.Path == "/dev/null" {
+			t.Errorf("RDMAConfig for mlx5_1 unexpectedly contains a char device left over from a different device: %v", cfg.DevChars)
+		}
+	}
+} 

please add an unit test as regression test

[gauravkghildiyal]

Yep done! Thanks

[aojea]

/lgtm
/approve

I think that we can make GetRdmaDeviceForInterface(ifName string) (string, error) mockable in the db and improve the cogerage,

[kubernetes-prow]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: aojea, gauravkghildiyal

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

• OWNERS [aojea,gauravkghildiyal]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[gauravkghildiyal]

/lgtm /approve

I think that we can make GetRdmaDeviceForInterface(ifName string) (string, error) mockable in the db and improve the cogerage,

+100. I saw that opportunity but plan on addressing this alongside other test improvements separately.