From 1815fc808d35cfde8e58f37d738f41c7c35ad00a Mon Sep 17 00:00:00 2001
From: Satyanvesh Dittakavi <Satyanvesh.Dittakavi@amd.com>
Date: Thu, 6 Jun 2024 08:34:09 +0000
Subject: [PATCH] SWDEV-464927 - Update the Get by PCI BusId logic and Hop
 count

- Update the intra socket weight for partitions within single socket as
it is changed to 13 by the driver.
- Use the PCIe function to distinguish the partitions of the same device
such as TPX mode in gfx942.

Change-Id: I8e64023d44e37c2dbb105cbb343441a48021ba7b
---
 hipamd/src/hip_device_runtime.cpp | 10 +++++++---
 rocclr/device/rocm/rocdevice.cpp  |  2 +-
 2 files changed, 8 insertions(+), 4 deletions(-)

diff --git a/hipamd/src/hip_device_runtime.cpp b/hipamd/src/hip_device_runtime.cpp
index a735d2039a..c453820f66 100644
--- a/hipamd/src/hip_device_runtime.cpp
+++ b/hipamd/src/hip_device_runtime.cpp
@@ -462,10 +462,12 @@ hipError_t hipDeviceGetByPCIBusId(int* device, const char* pciBusIdstr) {
   int pciBusID = -1;
   int pciDeviceID = -1;
   int pciDomainID = -1;
+  int pciFunction = -1;
   bool found = false;
-  if (sscanf(pciBusIdstr, "%04x:%02x:%02x", reinterpret_cast<unsigned int*>(&pciDomainID),
+  if (sscanf(pciBusIdstr, "%04x:%02x:%02x.%01x", reinterpret_cast<unsigned int*>(&pciDomainID),
              reinterpret_cast<unsigned int*>(&pciBusID),
-             reinterpret_cast<unsigned int*>(&pciDeviceID)) == 0x3) {
+             reinterpret_cast<unsigned int*>(&pciDeviceID),
+             reinterpret_cast<unsigned int*>(&pciFunction)) == 0x4) {
     int count = 0;
     HIP_RETURN_ONFAIL(ihipDeviceGetCount(&count));
     for (cl_int i = 0; i < count; i++) {
@@ -473,9 +475,11 @@ hipError_t hipDeviceGetByPCIBusId(int* device, const char* pciBusIdstr) {
       hipDeviceProp_tR0600 prop;
       HIP_RETURN_ONFAIL(ihipDeviceGet(&dev, i));
       HIP_RETURN_ONFAIL(ihipGetDeviceProperties(&prop, dev));
+      auto* deviceHandle = g_devices[dev]->devices()[0];
 
       if ((pciBusID == prop.pciBusID) && (pciDomainID == prop.pciDomainID) &&
-          (pciDeviceID == prop.pciDeviceID)) {
+          (pciDeviceID == prop.pciDeviceID) &&
+          (pciFunction == deviceHandle->info().deviceTopology_.pcie.function)) {
         *device = i;
         found = true;
         break;
diff --git a/rocclr/device/rocm/rocdevice.cpp b/rocclr/device/rocm/rocdevice.cpp
index 05e7ff169d..57438e6f1f 100644
--- a/rocclr/device/rocm/rocdevice.cpp
+++ b/rocclr/device/rocm/rocdevice.cpp
@@ -3356,7 +3356,7 @@ bool Device::findLinkInfo(const hsa_amd_memory_pool_t& pool,
           distance += link_info[hop_idx].numa_distance;
         }
         uint32_t oneHopDistance
-          = (link_info[0].link_type == HSA_AMD_LINK_INFO_TYPE_XGMI) ? 15 : 20;
+          = (link_info[0].link_type == HSA_AMD_LINK_INFO_TYPE_XGMI) ? 13 : 20;
         link_attr.second = static_cast<int32_t>(distance/oneHopDistance);
         break;
       }