Use the same mutex as rocm-smi

Share the same mutex as rocm-smi implementation. Handle the crash
when a user is not in render group.

Change-Id: I486b26569f9b523b41bbdaf95d51f4a730978cfd

include/amd_smi/impl/amd_smi_gpu_device.h

@@ -63,7 +63,6 @@ class AMDSmiGPUDevice: public AMDSmiProcessor {
                 if (check_if_drm_is_supported()) this->get_drm_data();
             }
     ~AMDSmiGPUDevice() {
-        if (check_if_drm_is_supported()) shared_mutex_close(mutex_);
     }
 
     amdsmi_status_t get_drm_data();
@@ -91,7 +90,6 @@ class AMDSmiGPUDevice: public AMDSmiProcessor {
     amdsmi_bdf_t bdf_;
     uint32_t vendor_id_;
     AMDSmiDrm& drm_;
-    shared_mutex_t mutex_;
 };
 
 

src/amd_smi/amd_smi_drm.cc

@@ -131,6 +131,7 @@ amdsmi_status_t AMDSmiDrm::init() {
     amd::smi::RocmSMI& smi = amd::smi::RocmSMI::getInstance();
     auto devices = smi.devices();
 
+    bool has_valid_fds = false;
     for (uint32_t i=0; i < devices.size(); i++) {
         auto rocm_smi_device = devices[i];
         std::string render_file_name;
@@ -171,6 +172,7 @@ amdsmi_status_t AMDSmiDrm::init() {
             continue;
         }
 
+        has_valid_fds = true;
         bdf.fields.function_number = device->businfo.pci->func;
         bdf.fields.device_number = device->businfo.pci->dev;
         bdf.fields.bus_number = device->businfo.pci->bus;
@@ -182,6 +184,12 @@ amdsmi_status_t AMDSmiDrm::init() {
         drm_free_device(&device);
     }
 
+    // cannot find any valid fds.
+    if (!has_valid_fds) {
+        drm_bdfs_.clear();
+        return AMDSMI_STATUS_INIT_ERROR;
+    }
+
     return AMDSMI_STATUS_SUCCESS;
 }
 
@@ -315,7 +323,7 @@ std::vector<std::string>& AMDSmiDrm::get_drm_paths() {
 }
 
 bool AMDSmiDrm::check_if_drm_is_supported() {
-    return drm_cmd_write_ != NULL ? true : false;
+    return (drm_cmd_write_ != NULL && drm_bdfs_.size() >0) ? true : false;
 }
 
 std::vector<amdsmi_bdf_t> AMDSmiDrm::get_bdfs() {

src/amd_smi/amd_smi_gpu_device.cc

@@ -43,6 +43,7 @@
 
 #include <functional>
 #include "amd_smi/impl/amd_smi_gpu_device.h"
+#include "rocm_smi/rocm_smi_utils.h"
 
 
 namespace amd {
@@ -80,11 +81,6 @@ amdsmi_status_t AMDSmiGPUDevice::get_drm_data() {
     ret = drm_.get_bdf_by_index(gpu_id_, &bdf);
     if (ret != AMDSMI_STATUS_SUCCESS) return AMDSMI_STATUS_NOT_SUPPORTED;
 
-    mutex_ = shared_mutex_init(path.c_str(), 0777);
-    if (mutex_.ptr == nullptr) {
-        printf("Failed to create shared mem. mutex.");
-        return AMDSMI_STATUS_INIT_ERROR;
-    }
     bdf_ = bdf, path_ = path, fd_ = fd;
     vendor_id_ = drm_.get_vendor_id();
 
@@ -92,7 +88,7 @@ amdsmi_status_t AMDSmiGPUDevice::get_drm_data() {
 }
 
 pthread_mutex_t* AMDSmiGPUDevice::get_mutex() {
-    return mutex_.ptr;
+    return amd::smi::GetMutex(gpu_id_);
 }
 
 amdsmi_status_t AMDSmiGPUDevice::amdgpu_query_info(unsigned info_id,