Fix generating localIDs in zebin

This commit fixes generating localIDs in zebin.
With thix fix, Emit Local in compute walker will be set accordingly with
the size of local_id argtype (currently, Emit Local is set to Emit None,
which prevents generating local IDs).

Related-To: NEO-6089
Signed-off-by: Kacper Nowak <[email protected]>

Author: km-nowak

shared/source/device_binary_format/zebin_decoder.cpp

@@ -687,6 +687,9 @@ NEO::DecodeError populateArgDescriptor(const NEO::Elf::ZebinKernelMetadata::Type
             dst.kernelAttributes.numLocalIdChannels = static_cast<uint8_t>(tupleSize);
             break;
         }
+        dst.kernelAttributes.localId[0] = tupleSize > 0;
+        dst.kernelAttributes.localId[1] = tupleSize > 1;
+        dst.kernelAttributes.localId[2] = tupleSize > 2;
         dst.kernelAttributes.perThreadDataSize = dst.kernelAttributes.simdSize;
         dst.kernelAttributes.perThreadDataSize *= sizeof(LocalIdT);
         dst.kernelAttributes.perThreadDataSize = alignUp(dst.kernelAttributes.perThreadDataSize, grfSize);
@@ -711,6 +714,9 @@ NEO::DecodeError populateArgDescriptor(const NEO::Elf::ZebinKernelMetadata::Type
             dst.kernelAttributes.numLocalIdChannels = static_cast<uint8_t>(tupleSize);
             break;
         }
+        dst.kernelAttributes.localId[0] = tupleSize > 0;
+        dst.kernelAttributes.localId[1] = tupleSize > 1;
+        dst.kernelAttributes.localId[2] = tupleSize > 2;
         dst.kernelAttributes.simdSize = 1;
         dst.kernelAttributes.perThreadDataSize = dst.kernelAttributes.simdSize;
         dst.kernelAttributes.perThreadDataSize *= dst.kernelAttributes.numLocalIdChannels;

shared/test/unit_test/device_binary_format/zebin_decoder_tests.cpp

@@ -3808,7 +3808,7 @@ TEST(PopulateArgDescriptorPerThreadPayload, GivenArgTypeLocalIdWhenSizeIsInvalid
     EXPECT_TRUE(warnings.empty()) << warnings;
 }
 
-TEST(PopulateArgDescriptorPerThreadPayload, GivenArgTypeLocalIdWhenSizeIsValidThenCalculateNumChannelAccordingly) {
+TEST(PopulateArgDescriptorPerThreadPayload, GivenArgTypeLocalIdWhenSizeIsValidThenCalculateNumChannelAndSetEmitLocalIdAccordingly) {
     uint32_t simdSizes[] = {8, 16, 32};
     uint32_t numChannelsOpts[] = {1, 2, 3};
 
@@ -3849,6 +3849,11 @@ TEST(PopulateArgDescriptorPerThreadPayload, GivenArgTypeLocalIdWhenSizeIsValidTh
             ASSERT_EQ(1U, programInfo.kernelInfos.size());
             EXPECT_EQ(numChannels, programInfo.kernelInfos[0]->kernelDescriptor.kernelAttributes.numLocalIdChannels) << warnings << "simd : " << simdSize << ", num channels : " << numChannels;
             EXPECT_EQ(simdSize, programInfo.kernelInfos[0]->kernelDescriptor.kernelAttributes.simdSize) << warnings << "simd : " << simdSize << ", num channels : " << numChannels;
+
+            const auto &emitLocalId = programInfo.kernelInfos[0]->kernelDescriptor.kernelAttributes.localId;
+            EXPECT_EQ(static_cast<uint8_t>(numChannels > 0), emitLocalId[0]);
+            EXPECT_EQ(static_cast<uint8_t>(numChannels > 1), emitLocalId[1]);
+            EXPECT_EQ(static_cast<uint8_t>(numChannels > 2), emitLocalId[2]);
         }
     }
 }
@@ -3919,7 +3924,7 @@ TEST(PopulateArgDescriptorPerThreadPayload, GivenArgTypePackedLocalIdWhenSizeIsI
     EXPECT_TRUE(warnings.empty()) << warnings;
 }
 
-TEST(PopulateArgDescriptorPerThreadPayload, GivenArgTypePackedLocalIdWhenSizeIsValidThenCalculateNumChannelAccordingly) {
+TEST(PopulateArgDescriptorPerThreadPayload, GivenArgTypePackedLocalIdWhenSizeIsValidThenCalculateNumChannelAndSetEmitLocalIdAccordingly) {
     uint32_t simdSizes[] = {1};
     uint32_t numChannelsOpts[] = {1, 2, 3};
 
@@ -3960,6 +3965,11 @@ TEST(PopulateArgDescriptorPerThreadPayload, GivenArgTypePackedLocalIdWhenSizeIsV
             ASSERT_EQ(1U, programInfo.kernelInfos.size());
             EXPECT_EQ(numChannels, programInfo.kernelInfos[0]->kernelDescriptor.kernelAttributes.numLocalIdChannels) << warnings << "simd : " << simdSize << ", num channels : " << numChannels;
             EXPECT_EQ(simdSize, programInfo.kernelInfos[0]->kernelDescriptor.kernelAttributes.simdSize) << warnings << "simd : " << simdSize << ", num channels : " << numChannels;
+
+            const auto &emitLocalId = programInfo.kernelInfos[0]->kernelDescriptor.kernelAttributes.localId;
+            EXPECT_EQ(static_cast<uint8_t>(numChannels > 0), emitLocalId[0]);
+            EXPECT_EQ(static_cast<uint8_t>(numChannels > 1), emitLocalId[1]);
+            EXPECT_EQ(static_cast<uint8_t>(numChannels > 2), emitLocalId[2]);
         }
     }
 }
@@ -5099,4 +5109,4 @@ TEST_F(IntelGTNotesFixture, WhenValidatingTargetDeviceGivenValidTargetDeviceAndI
     ASSERT_TRUE(outErrReason.empty());
 
     EXPECT_FALSE(validateTargetDevice(elf, targetDevice));
-}
+}