intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2025-12-30 01:35:20 +08:00
Code Issues Packages Projects Releases Wiki Activity

feature(sysman): Support for reset event upon repair

Browse Source 
Related-To: LOCI-3720

Signed-off-by: Bellekallu Rajkiran <bellekallu.rajkiran@intel.com>
This commit is contained in:
Bellekallu Rajkiran
2023-01-16 08:29:04 +00:00
committed by Compute-Runtime-Automation
parent 31d68c42e6
commit 1c63753ebe
3 changed files with 124 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
level_zero/tools/source/sysman/events/linux/os_events_imp_prelim.cpp
View File

@@ -1,5 +1,5 @@
/*
* Copyright (C) 2022 Intel Corporation
* Copyright (C) 2022-2023 Intel Corporation
*
* SPDX-License-Identifier: MIT
*
@@ -75,8 +75,6 @@ bool LinuxEventsImp::isResetRequired(void *dev, zes_event_type_flags_t &pEvent)
return true;
}
// ZES_EVENT_TYPE_FLAG_DEVICE_RESET_REQUIRED event could also be received when, reset Reason is
// ZES_RESET_REASON_FLAG_REPAIR.
return false;
}
@@ -286,6 +284,16 @@ bool LinuxEventsImp::eventListen(zes_event_type_flags_t &pEvent, uint64_t timeou
}
}
// check if any reset required for reason field repair
if (registeredEvents & ZES_EVENT_TYPE_FLAG_DEVICE_RESET_REQUIRED) {
zes_device_state_t deviceState = {};
pLinuxSysmanImp->getSysmanDeviceImp()->pGlobalOperations->deviceGetState(&deviceState);
if (deviceState.reset & ZES_RESET_REASON_FLAG_REPAIR) {
pEvent |= ZES_EVENT_TYPE_FLAG_DEVICE_RESET_REQUIRED;
return true;
}
}
return listenSystemEvents(pEvent, timeout);
}

22
level_zero/tools/test/unit_tests/sources/sysman/events/linux/mock_events_prelim.h
View File

@@ -10,6 +10,7 @@
#include "level_zero/tools/source/sysman/events/events_imp.h"
#include "level_zero/tools/source/sysman/events/linux/os_events_imp_prelim.h"
#include "level_zero/tools/source/sysman/firmware_util/firmware_util.h"
namespace L0 {
namespace ult {
@@ -211,6 +212,27 @@ class UdevLibMock : public UdevLib {
ADDMETHOD_NOBASE_VOIDRETURN(dropDeviceReference, (void *dev));
};
struct MockEventsFwInterface : public FirmwareUtil {
bool mockIfrStatus = false;
ze_result_t fwIfrApplied(bool &ifrStatus) override {
ifrStatus = mockIfrStatus;
return ZE_RESULT_SUCCESS;
}
MockEventsFwInterface() = default;
ADDMETHOD_NOBASE(fwDeviceInit, ze_result_t, ZE_RESULT_SUCCESS, (void));
ADDMETHOD_NOBASE(getFirstDevice, ze_result_t, ZE_RESULT_SUCCESS, (igsc_device_info * info));
ADDMETHOD_NOBASE(getFwVersion, ze_result_t, ZE_RESULT_SUCCESS, (std::string fwType, std::string &firmwareVersion));
ADDMETHOD_NOBASE(flashFirmware, ze_result_t, ZE_RESULT_SUCCESS, (std::string fwType, void *pImage, uint32_t size));
ADDMETHOD_NOBASE(fwSupportedDiagTests, ze_result_t, ZE_RESULT_SUCCESS, (std::vector<std::string> & supportedDiagTests));
ADDMETHOD_NOBASE(fwRunDiagTests, ze_result_t, ZE_RESULT_SUCCESS, (std::string & osDiagType, zes_diag_result_t *pResult));
ADDMETHOD_NOBASE(fwGetMemoryErrorCount, ze_result_t, ZE_RESULT_SUCCESS, (zes_ras_error_type_t category, uint32_t subDeviceCount, uint32_t subDeviceId, uint64_t &count));
ADDMETHOD_NOBASE(fwGetEccConfig, ze_result_t, ZE_RESULT_SUCCESS, (uint8_t * currentState, uint8_t *pendingState));
ADDMETHOD_NOBASE(fwSetEccConfig, ze_result_t, ZE_RESULT_SUCCESS, (uint8_t newState, uint8_t *currentState, uint8_t *pendingState));
ADDMETHOD_NOBASE_VOIDRETURN(getDeviceSupportedFwTypes, (std::vector<std::string> & fwTypes));
ADDMETHOD_NOBASE_VOIDRETURN(fwGetMemoryHealthIndicator, (zes_mem_health_t * health));
};
class PublicLinuxEventsImp : public L0::LinuxEventsImp {
public:
PublicLinuxEventsImp(OsSysman *pOsSysman) : LinuxEventsImp(pOsSysman) {}

101
level_zero/tools/test/unit_tests/sources/sysman/events/linux/test_zes_events_prelim.cpp
View File

@@ -7,7 +7,6 @@
#include "shared/test/common/os_interface/linux/sys_calls_linux_ult.h"
#include "level_zero/tools/source/sysman/global_operations/global_operations_imp.h"
#include "level_zero/tools/test/unit_tests/sources/sysman/events/linux/mock_events_prelim.h"
#include "level_zero/tools/test/unit_tests/sources/sysman/linux/mock_sysman_fixture.h"
@@ -28,8 +27,6 @@ class SysmanEventsFixture : public SysmanDeviceFixture {
FsAccess *pFsAccessOriginal = nullptr;
OsEvents *pOsEventsPrev = nullptr;
L0::EventsImp *pEventsImp;
GlobalOperations *pGlobalOperationsOriginal = nullptr;
std::unique_ptr<GlobalOperationsImp> pGlobalOperations;
std::unique_ptr<MockEventsSysfsAccess> pSysfsAccess;
SysfsAccess *pSysfsAccessOriginal = nullptr;
std::unique_ptr<MockPmuInterfaceImpForEvents> pPmuInterface;
@@ -56,12 +53,6 @@ class SysmanEventsFixture : public SysmanDeviceFixture {
pLinuxSysmanImp->pUdevLib = new UdevLibMock();
pLinuxEventsImp = new PublicLinuxEventsImp(pOsSysman);
pEventsImp->pOsEvents = pLinuxEventsImp;
pGlobalOperations = std::make_unique<GlobalOperationsImp>(pLinuxSysmanImp);
pGlobalOperationsOriginal = pSysmanDeviceImp->pGlobalOperations;
pSysmanDeviceImp->pGlobalOperations = pGlobalOperations.get();
pSysmanDeviceImp->pGlobalOperations->init();
pPmuInterface = std::make_unique<MockPmuInterfaceImpForEvents>(pLinuxSysmanImp);
pOriginalPmuInterface = pLinuxSysmanImp->pPmuInterface;
pLinuxSysmanImp->pPmuInterface = pPmuInterface.get();
@@ -94,7 +85,6 @@ class SysmanEventsFixture : public SysmanDeviceFixture {
pEventsImp = nullptr;
pLinuxSysmanImp->pSysfsAccess = pSysfsAccessOriginal;
pLinuxSysmanImp->pFsAccess = pFsAccessOriginal;
pSysmanDeviceImp->pGlobalOperations = pGlobalOperationsOriginal;
pLinuxSysmanImp->pPmuInterface = pOriginalPmuInterface;
SysmanDeviceFixture::TearDown();
}
@@ -130,6 +120,10 @@ TEST_F(SysmanEventsFixture, GivenLibUdevLibraryNotFoundWhenListeningForEventsThe
TEST_F(SysmanEventsFixture,
GivenValidDeviceHandleAndListeningForEventsWhenEventGenerationSourceDeviceIsNotDrmAndPlatformThenEventListenReturnFalse) {
VariableBackup<FirmwareUtil *> backupFwUtil(&pLinuxSysmanImp->pFwUtilInterface);
auto pMockFwInterface = new MockEventsFwInterface;
pLinuxSysmanImp->pFwUtilInterface = pMockFwInterface;
VariableBackup<decltype(SysCalls::sysCallsPoll)> mockPoll(&SysCalls::sysCallsPoll, [](struct pollfd *pollFd, unsigned long int numberOfFds, int timeout) -> int {
return 1;
});
@@ -180,6 +174,7 @@ TEST_F(SysmanEventsFixture,
pEventsImp->pOsEvents = pOsEventOriginal;
delete pUdevLibLocal;
delete pPublicLinuxEventsImp;
delete pMockFwInterface;
}
TEST_F(SysmanEventsFixture,
@@ -287,6 +282,10 @@ TEST_F(SysmanEventsFixture, GivenValidDeviceHandleAndListeningEventsWhenNullEven
TEST_F(SysmanEventsFixture,
GivenValidDeviceHandleAndListeningForEventsWhenUdevCallToAllocateDeviceFailsThenEventListenReturnFalse) {
VariableBackup<FirmwareUtil *> backupFwUtil(&pLinuxSysmanImp->pFwUtilInterface);
auto pMockFwInterface = new MockEventsFwInterface;
pLinuxSysmanImp->pFwUtilInterface = pMockFwInterface;
VariableBackup<decltype(SysCalls::sysCallsPoll)> mockPoll(&SysCalls::sysCallsPoll, [](struct pollfd *pollFd, unsigned long int numberOfFds, int timeout) -> int {
return 1;
});
@@ -335,9 +334,14 @@ TEST_F(SysmanEventsFixture,
pEventsImp->pOsEvents = pOsEventOriginal;
delete pUdevLibLocal;
delete pPublicLinuxEventsImp;
delete pMockFwInterface;
}
TEST_F(SysmanEventsFixture, GivenValidDeviceHandleWhenListeningForResetRequiredEventsThenEventListenAPIReturnsAfterReceivingEventWithinTimeout) {
VariableBackup<FirmwareUtil *> backupFwUtil(&pLinuxSysmanImp->pFwUtilInterface);
auto pMockFwInterface = new MockEventsFwInterface;
pLinuxSysmanImp->pFwUtilInterface = pMockFwInterface;
VariableBackup<decltype(SysCalls::sysCallsPoll)> mockPoll(&SysCalls::sysCallsPoll, [](struct pollfd *pollFd, unsigned long int numberOfFds, int timeout) -> int {
return 1;
});
@@ -389,6 +393,7 @@ TEST_F(SysmanEventsFixture, GivenValidDeviceHandleWhenListeningForResetRequiredE
pEventsImp->pOsEvents = pOsEventOriginal;
delete pUdevLibLocal;
delete pPublicLinuxEventsImp;
delete pMockFwInterface;
}
TEST_F(SysmanDeviceFixture, GivenValidDeviceHandleWhenEventRegisterIsCalledThenSuccessIsReturned) {
@@ -397,6 +402,10 @@ TEST_F(SysmanDeviceFixture, GivenValidDeviceHandleWhenEventRegisterIsCalledThenS
TEST_F(SysmanEventsFixture,
GivenValidDeviceHandleWhenListeningForResetRequiredEventsAndIfUeventReceivedWithWrongPropertyValueThenEventListenAPIReturnsWithinTimeout) {
VariableBackup<FirmwareUtil *> backupFwUtil(&pLinuxSysmanImp->pFwUtilInterface);
auto pMockFwInterface = new MockEventsFwInterface;
pLinuxSysmanImp->pFwUtilInterface = pMockFwInterface;
VariableBackup<decltype(SysCalls::sysCallsPoll)> mockPoll(&SysCalls::sysCallsPoll, [](struct pollfd *pollFd, unsigned long int numberOfFds, int timeout) -> int {
return 1;
});
@@ -447,10 +456,15 @@ TEST_F(SysmanEventsFixture,
pEventsImp->pOsEvents = pOsEventOriginal;
delete pUdevLibLocal;
delete pPublicLinuxEventsImp;
delete pMockFwInterface;
}
TEST_F(SysmanEventsFixture,
GivenValidDeviceHandleWhenListeningForResetRequiredEventsAndIfDrmEventReceivedButChangeEventNotReceivedThenEventListenAPIReturnsWithinTimeout) {
VariableBackup<FirmwareUtil *> backupFwUtil(&pLinuxSysmanImp->pFwUtilInterface);
auto pMockFwInterface = new MockEventsFwInterface;
pLinuxSysmanImp->pFwUtilInterface = pMockFwInterface;
VariableBackup<decltype(SysCalls::sysCallsPoll)> mockPoll(&SysCalls::sysCallsPoll, [](struct pollfd *pollFd, unsigned long int numberOfFds, int timeout) -> int {
return 1;
});
@@ -502,10 +516,15 @@ TEST_F(SysmanEventsFixture,
pEventsImp->pOsEvents = pOsEventOriginal;
delete pUdevLibLocal;
delete pPublicLinuxEventsImp;
delete pMockFwInterface;
}
TEST_F(SysmanEventsFixture,
GivenValidDeviceHandleWhenListeningForResetRequiredEventsAndIfUeventReceivedWithNullPropertyValueThenEventListenAPIReturnsWithinTimeout) {
VariableBackup<FirmwareUtil *> backupFwUtil(&pLinuxSysmanImp->pFwUtilInterface);
auto pMockFwInterface = new MockEventsFwInterface;
pLinuxSysmanImp->pFwUtilInterface = pMockFwInterface;
VariableBackup<decltype(SysCalls::sysCallsPoll)> mockPoll(&SysCalls::sysCallsPoll, [](struct pollfd *pollFd, unsigned long int numberOfFds, int timeout) -> int {
return 1;
});
@@ -556,6 +575,68 @@ TEST_F(SysmanEventsFixture,
pEventsImp->pOsEvents = pOsEventOriginal;
delete pUdevLibLocal;
delete pPublicLinuxEventsImp;
delete pMockFwInterface;
}
HWTEST2_F(SysmanEventsFixture,
GivenValidDeviceHandleWhenListeningForResetRequiredEventsAfterInFieldRepairAndIfUeventReceivedWithNullPropertyValueThenEventListenAPIReturnsResetEvent, IsPVC) {
VariableBackup<FirmwareUtil *> backupFwUtil(&pLinuxSysmanImp->pFwUtilInterface);
auto pMockFwInterface = new MockEventsFwInterface;
pLinuxSysmanImp->pFwUtilInterface = pMockFwInterface;
VariableBackup<decltype(SysCalls::sysCallsPoll)> mockPoll(&SysCalls::sysCallsPoll, [](struct pollfd *pollFd, unsigned long int numberOfFds, int timeout) -> int {
return 1;
});
VariableBackup<decltype(SysCalls::sysCallsOpen)> mockOpen(&SysCalls::sysCallsOpen, [](const char *pathname, int flags) -> int {
if (strcmp(pathname, "i915.iaf.31/iaf_fabric_id") == 0) {
return fabricDeviceFd;
} else {
return drmDeviceFd;
}
});
VariableBackup<decltype(NEO::SysCalls::sysCallsFstat)> mockFstat(&NEO::SysCalls::sysCallsFstat, [](int fd, struct stat *buf) -> int {
if (fd == fabricDeviceFd) {
buf->st_rdev = 10;
} else {
buf->st_rdev = 0;
}
return 0;
});
// Step 1: Initialize a mocked udev lib object for this test case
auto pUdevLibLocal = new UdevLibMock();
int a = 0;
void *ptr = &a; // Initialize a void pointer with dummy data
pUdevLibLocal->allocateDeviceToReceiveDataResult = ptr;
pUdevLibLocal->getEventGenerationSourceDeviceResult = 0;
pUdevLibLocal->getEventPropertyValueResult = nullptr;
// Step 2: Create a new PublicLinuxEventsImp, where we will attach the above created Udev Lib object
auto pPublicLinuxEventsImp = new PublicLinuxEventsImp(pOsSysman);
pPublicLinuxEventsImp->pUdevLib = pUdevLibLocal;
// Step 3: Backup original pOsEvent created during set up
auto pOsEventOriginal = pEventsImp->pOsEvents;
pEventsImp->pOsEvents = static_cast<OsEvents *>(pPublicLinuxEventsImp);
// Step 4: Call APIs for validation
EXPECT_EQ(ZE_RESULT_SUCCESS, zesDeviceEventRegister(device->toHandle(), ZES_EVENT_TYPE_FLAG_DEVICE_RESET_REQUIRED));
zes_device_handle_t *phDevices = new zes_device_handle_t[1];
phDevices[0] = device->toHandle();
uint32_t numDeviceEvents = 0;
pMockFwInterface->mockIfrStatus = true;
zes_event_type_flags_t *pDeviceEvents = new zes_event_type_flags_t[1];
EXPECT_EQ(ZE_RESULT_SUCCESS, zesDriverEventListen(driverHandle->toHandle(), 1u, 1u, phDevices, &numDeviceEvents, pDeviceEvents));
EXPECT_EQ(1u, numDeviceEvents);
EXPECT_EQ(ZES_EVENT_TYPE_FLAG_DEVICE_RESET_REQUIRED, pDeviceEvents[0]);
// Step 5: Cleanup
delete[] phDevices;
delete[] pDeviceEvents;
pEventsImp->pOsEvents = pOsEventOriginal;
delete pUdevLibLocal;
delete pPublicLinuxEventsImp;
delete pMockFwInterface;
}
TEST_F(SysmanEventsFixture, GivenValidDeviceHandleWhenListeningForCurrentlyUnsupportedEventsThenEventListenAPIWaitForTimeoutIfEventNotReceived) {