Add support for ras l3 fabric errors

Related-To: LOCI-3966

Signed-off-by: Mayank Raghuwanshi <mayank.raghuwanshi@intel.com>
This commit is contained in:
Mayank Raghuwanshi
2023-03-18 11:42:33 +00:00
committed by Compute-Runtime-Automation
parent 3f6443edfb
commit 065232eac7
3 changed files with 36 additions and 29 deletions

View File

@@ -20,20 +20,20 @@ static const std::map<zes_ras_error_cat_t, std::vector<std::string>> categoryToL
{ZES_RAS_ERROR_CAT_PROGRAMMING_ERRORS,
{"eu-attention"}},
{ZES_RAS_ERROR_CAT_NON_COMPUTE_ERRORS,
{"soc-fatal-mdfi-east", "soc-fatal-mdfi-south", "soc-fatal-mdfi-west",
"soc-fatal-psf-0", "soc-fatal-psf-1", "soc-fatal-psf-2", "soc-fatal-psf-csc-0",
{"soc-fatal-psf-0", "soc-fatal-psf-1", "soc-fatal-psf-2", "soc-fatal-psf-csc-0",
"soc-fatal-psf-csc-1", "soc-fatal-psf-csc-2", "soc-fatal-punit",
"sgunit-fatal", "soc-nonfatal-punit", "sgunit-fatal", "sgunit-nonfatal", "gsc-nonfatal-mia-shutdown",
"gsc-nonfatal-aon-parity", "gsc-nonfatal-rom-parity", "gsc-nonfatal-fuse-crc-check",
"gsc-nonfatal-selfmbist", "gsc-nonfatal-fuse-pull", "gsc-nonfatal-sram-ecc", "gsc-nonfatal-glitch-det",
"gsc-nonfatal-ucode-parity", "gsc-nonfatal-mia-int", "gsc-nonfatal-wdg-timeout"}},
{ZES_RAS_ERROR_CAT_COMPUTE_ERRORS,
{"fatal-fpu", "fatal-l3-fabric", "fatal-eu-grf", "fatal-sampler", "fatal-slm",
{"fatal-fpu", "fatal-eu-grf", "fatal-sampler", "fatal-slm",
"fatal-guc", "fatal-eu-ic", "fatal-subslice"}},
{ZES_RAS_ERROR_CAT_DRIVER_ERRORS,
{"driver-object-migration", "driver-engine-other", "driver-ggtt",
"driver-gt-interrupt", "driver-gt-other", "driver-guc-communication",
"driver-rps"}}};
"driver-rps"}},
{ZES_RAS_ERROR_CAT_L3FABRIC_ERRORS, {"soc-fatal-mdfi-east", "soc-fatal-mdfi-south", "soc-fatal-mdfi-west", "fatal-l3-fabric", "soc-fatal-cd0-mdfi"}}};
static const std::map<zes_ras_error_cat_t, std::vector<std::string>> categoryToListOfEventsCorrectable = {
{ZES_RAS_ERROR_CAT_CACHE_ERRORS,

View File

@@ -58,8 +58,9 @@ constexpr uint64_t driverEngineOther = 3u;
constexpr uint64_t initialUncorrectableCacheErrors = 2u;
constexpr uint64_t initialEngineReset = 2u;
constexpr uint64_t initialProgrammingErrors = 7u;
constexpr uint64_t initialUncorrectableNonComputeErrors = 8u;
constexpr uint64_t initialUncorrectableComputeErrors = 10u;
constexpr uint64_t initialUncorrectableNonComputeErrors = 3u;
constexpr uint64_t initialUncorrectableFabricErrors = 8u;
constexpr uint64_t initialUncorrectableComputeErrors = 7u;
constexpr uint64_t initialCorrectableComputeErrors = 6u;
constexpr uint64_t initialUncorrectableDriverErrors = 5u;
@@ -67,16 +68,18 @@ constexpr uint64_t initialUncorrectableCacheErrorsTile0 = 2u;
constexpr uint64_t initialCorrectableCacheErrorTile0 = 2u;
constexpr uint64_t initialEngineResetTile0 = 2u;
constexpr uint64_t initialProgrammingErrorsTile0 = 7u;
constexpr uint64_t initialUncorrectableNonComputeErrorsTile0 = 15u;
constexpr uint64_t initialUncorrectableNonComputeErrorsTile0 = 10u;
constexpr uint64_t initialCorrectableNonComputeErrorsTile0 = 2u;
constexpr uint64_t initialUncorrectableComputeErrorsTile0 = 11u;
constexpr uint64_t initialUncorrectableComputeErrorsTile0 = 8u;
constexpr uint64_t initialUncorrectableFabricErrorsTile0 = 8u;
constexpr uint64_t initialCorrectableComputeErrorsTile0 = 6u;
constexpr uint64_t initialUncorrectableDriverErrorsTile0 = 5u;
constexpr uint64_t initialUncorrectableCacheErrorsTile1 = 1u;
constexpr uint64_t initialEngineResetTile1 = 4u;
constexpr uint64_t initialProgrammingErrorsTile1 = 5u;
constexpr uint64_t initialCorrectableComputeErrorsTile1 = 7u;
constexpr uint64_t initialUncorrectableNonComputeErrorsTile1 = 5u;
constexpr uint64_t initialUncorrectableNonComputeErrorsTile1 = 3u;
constexpr uint64_t initialUncorrectableFabricErrorsTile1 = 2u;
constexpr uint64_t initialUncorrectableComputeErrorsTile1 = 6u;
constexpr uint64_t initialUncorrectableDriverErrorsTile1 = 4u;
constexpr uint64_t timeStamp = 1000u;
@@ -132,11 +135,11 @@ struct MockRasPmuInterfaceImp : public PmuInterfaceImp {
data[5] = driverGgtt;
data[6] = driverRps;
data[7] = 0;
data[8] = 0;
data[9] = fatalEuErrorCount;
data[10] = socFatalMdfiEastCount;
data[11] = socFatalPsfCsc0Count;
data[12] = fatalTlb;
data[8] = fatalEuErrorCount;
data[9] = socFatalPsfCsc0Count;
data[10] = fatalTlb;
data[11] = 0;
data[12] = socFatalMdfiEastCount;
return 0;
}
@@ -159,14 +162,14 @@ struct MockRasPmuInterfaceImp : public PmuInterfaceImp {
data[5] = driverGgtt;
data[6] = driverRps;
data[7] = 0;
data[8] = 0;
data[9] = fatalSubslice;
data[10] = fatalEuErrorCount;
data[11] = socFatalMdfiEastCount;
data[12] = socFatalPsfCsc0Count;
data[13] = nonFatalGscAonParity;
data[14] = nonFataGscSelfmBist;
data[15] = fatalTlb;
data[8] = fatalSubslice;
data[9] = fatalEuErrorCount;
data[10] = socFatalPsfCsc0Count;
data[11] = nonFatalGscAonParity;
data[12] = nonFataGscSelfmBist;
data[13] = fatalTlb;
data[14] = 0;
data[15] = socFatalMdfiEastCount;
return 0;
}
@@ -187,10 +190,10 @@ struct MockRasPmuInterfaceImp : public PmuInterfaceImp {
data[4] = driverMigration;
data[5] = driverEngineOther;
data[6] = fatalGucErrorCountTile1;
data[7] = socFatalMdfiWestCountTile1;
data[8] = socFatalPunitTile1;
data[9] = fatalIdiParityErrorCountTile1;
data[10] = fatalL3BankTile1;
data[7] = socFatalPunitTile1;
data[8] = fatalIdiParityErrorCountTile1;
data[9] = fatalL3BankTile1;
data[10] = socFatalMdfiWestCountTile1;
return 0;
}

View File

@@ -192,9 +192,10 @@ TEST_F(SysmanRasFixture, GivenValidRasHandleWhenCallingzesRasGeStateForGtThenSuc
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_RESET], fatalEngineResetCount + initialEngineReset);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_PROGRAMMING_ERRORS], euAttention + initialProgrammingErrors);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_COMPUTE_ERRORS], fatalEuErrorCount + initialUncorrectableComputeErrors);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_NON_COMPUTE_ERRORS], socFatalMdfiEastCount + socFatalPsfCsc0Count + initialUncorrectableNonComputeErrors);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_NON_COMPUTE_ERRORS], socFatalPsfCsc0Count + initialUncorrectableNonComputeErrors);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_DISPLAY_ERRORS], 0u);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_DRIVER_ERRORS], driverMigration + driverGgtt + driverRps + initialUncorrectableDriverErrors);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_L3FABRIC_ERRORS], socFatalMdfiEastCount + initialUncorrectableFabricErrors);
}
}
}
@@ -229,9 +230,10 @@ TEST_F(SysmanRasFixture, GivenValidRasHandleWhenCallingzesRasGeStateForGtAfterCl
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_RESET], fatalEngineResetCount + initialEngineReset);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_PROGRAMMING_ERRORS], euAttention + initialProgrammingErrors);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_COMPUTE_ERRORS], fatalEuErrorCount + initialUncorrectableComputeErrors);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_NON_COMPUTE_ERRORS], socFatalMdfiEastCount + socFatalPsfCsc0Count + initialUncorrectableNonComputeErrors);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_NON_COMPUTE_ERRORS], socFatalPsfCsc0Count + initialUncorrectableNonComputeErrors);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_DISPLAY_ERRORS], 0u);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_DRIVER_ERRORS], driverMigration + driverGgtt + driverRps + initialUncorrectableDriverErrors);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_L3FABRIC_ERRORS], socFatalMdfiEastCount + initialUncorrectableFabricErrors);
}
}
correctable = true;
@@ -655,9 +657,10 @@ TEST_F(SysmanRasMultiDeviceFixture, GivenValidRasHandleWhenCallingzesRasGeStateF
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_RESET], fatalEngineResetCount + initialEngineResetTile0);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_PROGRAMMING_ERRORS], euAttention + initialProgrammingErrorsTile0);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_COMPUTE_ERRORS], fatalSubslice + fatalEuErrorCount + initialUncorrectableComputeErrorsTile0);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_NON_COMPUTE_ERRORS], socFatalMdfiEastCount + socFatalPsfCsc0Count + nonFatalGscAonParity + nonFataGscSelfmBist + initialUncorrectableNonComputeErrorsTile0);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_NON_COMPUTE_ERRORS], socFatalPsfCsc0Count + nonFatalGscAonParity + nonFataGscSelfmBist + initialUncorrectableNonComputeErrorsTile0);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_DISPLAY_ERRORS], 0u);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_DRIVER_ERRORS], driverMigration + driverGgtt + driverRps + initialUncorrectableDriverErrorsTile0);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_L3FABRIC_ERRORS], socFatalMdfiEastCount + initialUncorrectableFabricErrorsTile0);
} else if (handleIndex == 2u) {
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_CACHE_ERRORS], 0u); // No. of correctable error type for subdevice 1
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_RESET], 0u);
@@ -674,6 +677,7 @@ TEST_F(SysmanRasMultiDeviceFixture, GivenValidRasHandleWhenCallingzesRasGeStateF
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_NON_COMPUTE_ERRORS], socFatalMdfiWestCountTile1 + socFatalPunitTile1 + initialUncorrectableNonComputeErrorsTile1);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_DISPLAY_ERRORS], 0u);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_DRIVER_ERRORS], driverMigration + driverEngineOther + initialUncorrectableDriverErrorsTile1);
EXPECT_EQ(state.category[ZES_RAS_ERROR_CAT_L3FABRIC_ERRORS], socFatalMdfiWestCountTile1 + initialUncorrectableFabricErrorsTile1);
}
handleIndex++;
}