fix: Update handles upon device reset

Outdated info that is present in several modules
because of reset result in crash when module info
is retrieved after reset.

Release and re-create module handles during reset
to avoid crash.

Related-To: NEO-9100

Signed-off-by: Bellekallu Rajkiran <bellekallu.rajkiran@intel.com>
This commit is contained in:
Bellekallu Rajkiran
2023-10-13 06:54:01 +00:00
committed by Compute-Runtime-Automation
parent d3d15542fb
commit 84cdf47abf
24 changed files with 281 additions and 7 deletions

View File

@@ -13,7 +13,13 @@
namespace L0 {
namespace Sysman {
TemperatureHandleContext::~TemperatureHandleContext() {}
TemperatureHandleContext::~TemperatureHandleContext() {
releaseTemperatureHandles();
}
void TemperatureHandleContext::releaseTemperatureHandles() {
handleList.clear();
}
void TemperatureHandleContext::createHandle(bool onSubdevice, uint32_t subDeviceId, zes_temp_sensors_t type) {
std::unique_ptr<Temperature> pTemperature = std::make_unique<TemperatureImp>(pOsSysman, onSubdevice, subDeviceId, type);
@@ -42,6 +48,7 @@ ze_result_t TemperatureHandleContext::init(uint32_t subDeviceCount) {
ze_result_t TemperatureHandleContext::temperatureGet(uint32_t *pCount, zes_temp_handle_t *phTemperature) {
std::call_once(initTemperatureOnce, [this]() {
this->init(pOsSysman->getSubDeviceCount());
this->tempInitDone = true;
});
uint32_t handleListSize = static_cast<uint32_t>(handleList.size());
uint32_t numToCopy = std::min(*pCount, handleListSize);

View File

@@ -39,13 +39,19 @@ struct TemperatureHandleContext {
ze_result_t init(uint32_t subDeviceCount);
ze_result_t temperatureGet(uint32_t *pCount, zes_temp_handle_t *phTemperature);
void releaseTemperatureHandles();
OsSysman *pOsSysman = nullptr;
std::vector<std::unique_ptr<Temperature>> handleList = {};
bool isTempInitDone() {
return tempInitDone;
}
private:
void createHandle(bool onSubdevice, uint32_t subDeviceId, zes_temp_sensors_t type);
std::once_flag initTemperatureOnce;
bool tempInitDone = false;
};
} // namespace Sysman