From 5afee1620376d273eff1e359633a023dd1c7825d Mon Sep 17 00:00:00 2001
From: William Cohen <wcohen@redhat.com>
Date: Wed, 6 May 2026 15:45:22 -0400
Subject: [PATCH] pmdaamdgpu: Prevent out-of-bound references in amdgpu_fetch
 function

The valid range of indices for the pmidlist array is 0 to numpmid-1.
The loop in amdgpu_fetch was incorrectly using a "i <= numpmid" rather
than "i < numpmid" to terminate the loop.

Resolves: https://github.com/performancecopilot/pcp/issues/2584
---
 src/pmdas/amdgpu/amdgpu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/pmdas/amdgpu/amdgpu.c b/src/pmdas/amdgpu/amdgpu.c
index 49885e1adf3..d26be4b9d24 100644
--- a/src/pmdas/amdgpu/amdgpu.c
+++ b/src/pmdas/amdgpu/amdgpu.c
@@ -353,7 +353,7 @@ static int amdgpu_fetch(int numpmid, pmID pmidlist[], pmdaResult **resp,
 {
   uint32_t i = 0;
 
-  for (i = 0; i <= numpmid; i++) {
+  for (i = 0; i < numpmid; i++) {
       unsigned int cluster = pmID_cluster(pmidlist[i]);
       unsigned int item = pmID_item(pmidlist[i]);