Update versions for sqlite and protobuf #2347

ahsan-ca · 2023-10-18T15:32:43Z

Updated versions for sqlite and protobuf are needed to fix
security vulnerabilities.

umangyadav · 2023-10-18T15:40:01Z

requirements.txt

@@ -21,12 +21,12 @@
 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 # THE SOFTWARE.
 #####################################################################################
-google/protobuf@v3.11.0 -DCMAKE_POSITION_INDEPENDENT_CODE=On -X subdir -Dprotobuf_BUILD_TESTS=Off
+google/protobuf@v3.20.2 -DCMAKE_POSITION_INDEPENDENT_CODE=On -X subdir -Dprotobuf_BUILD_TESTS=Off


I wonder if .pb files or .onnx needs to be re-generated.
Did all tests pass ?

I think they might need to be regenerated. There seems to be an issue on the develop branch preventing it from building:

/code/AMDMIGraphX/AMDMIGraphX/src/targets/gpu/include/migraphx/gpu/ck.hpp:34:10: fatal error: 'ck/host/device_batched_gemm_softmax_gemm.hpp' file not found #include "ck/host/device_batched_gemm_softmax_gemm.hpp"

So I checked out an earlier version of develop and made the version changes for protobuf and sqlite and was able to get build/tests to pass. I put that up as PR but realized there were conflicts with the requirements.txt on develop. So I rebased it to resolve conflicts, but I was not sure about the commits ROCmSoftwarePlatform/composable_kernel@70eefcf4f263aa5c25f3c9ff0db8f6f199ef0fb9 that may need to be updated. I was planning to ask this in the afternoon meeting along with some other questions on this. In the meantime, I think its better to convert it to draft

that error is from CK.
You need to rebuild your dependencies using rbuild develop or rbuild prepare.

I tried rbuild develop but I still saw the same error. The I tried rbuild prepare with the command rbuild prepare --deps-dir ../AMDMIGraphX/. But I still see the same error. I am doing this in the build directory in a docker container. Do you know what I might be missing? :)
Edit: May be the deps-dir is incorrect?

Never mind, I think I got it fixed. Doing rbuild prepare -d depend and using -DCMAKE_PREFIX_PATH=depend seems to have done the trick.
Edit: No it did not fix it :). Lakhinder mentioned spinning a new docker container is another way. I'll try that.
Second Edit: I was able to get this resolved.

I wonder if .pb files or .onnx needs to be re-generated. Did all tests pass ?

@umangyadav All tests except the test_verify ones I mentioned in the other message pass.

codecov · 2023-10-18T17:42:23Z

Codecov Report

Merging #2347 (1d8e0bb) into develop (cbee4b7) will not change coverage.
Report is 6 commits behind head on develop.
The diff coverage is n/a.

❗ Current head 1d8e0bb differs from pull request most recent head 8f1dd36. Consider uploading reports for the commit 8f1dd36 to get more accurate results

@@           Coverage Diff            @@
##           develop    #2347   +/-   ##
========================================
  Coverage    91.36%   91.36%           
========================================
  Files          440      440           
  Lines        16530    16530           
========================================
  Hits         15101    15101           
  Misses        1429     1429

ahsan-ca · 2023-10-18T19:11:58Z

Seeing this error now:

RuntimeError: /code/AMDMIGraphX/AMDMIGraphX/src/targets/gpu/hip.cpp:67: get_available_gpu_memory: Failed getting available memory: invalid argument

Not related to this PR.

rocm versions:
Inside docker: 5.7.0-63
On bare metal it is 5.7.0-59

Machine running this on:
Linux ixt-rack-186 5.15.0-86-generic #96~20.04.1-Ubuntu SMP Thu Sep 21 13:23:37 UTC 2023 x86_64 x86_64 x86_64 GNU/Linux

migraphx-bot · 2023-10-18T19:27:50Z

Test	Batch	Rate new 5929fe	Rate old 0a4254	Diff	Compare
torchvision-resnet50	64	2,848.23	2,848.84	-0.02%	✅
torchvision-resnet50_fp16	64	6,483.65	6,474.01	0.15%	✅
torchvision-densenet121	32	2,098.55	2,098.18	0.02%	✅
torchvision-densenet121_fp16	32	3,678.99	3,681.38	-0.06%	✅
torchvision-inceptionv3	32	1,593.17	1,593.19	-0.00%	✅
torchvision-inceptionv3_fp16	32	2,589.35	2,592.14	-0.11%	✅
cadene-inceptionv4	16	706.73	706.91	-0.03%	✅
cadene-resnext64x4	16	696.86	697.53	-0.10%	✅
slim-mobilenet	64	8,348.55	8,361.19	-0.15%	✅
slim-nasnetalarge	64	226.65	226.45	0.09%	✅
slim-resnet50v2	64	2,674.17	2,674.22	-0.00%	✅
bert-mrpc-onnx	8	824.15	824.56	-0.05%	✅
bert-mrpc-tf	1	388.89	388.58	0.08%	✅
pytorch-examples-wlang-gru	1	302.36	299.93	0.81%	✅
pytorch-examples-wlang-lstm	1	314.91	306.64	2.70%	✅
torchvision-resnet50_1	1	599.08	601.13	-0.34%	✅
torchvision-inceptionv3_1	1	337.16	336.20	0.29%	✅
cadene-dpn92_1	1	396.51	395.46	0.27%	✅
cadene-resnext101_1	1	329.63	329.48	0.05%	✅
slim-vgg16_1	1	465.09	464.93	0.03%	✅
slim-mobilenet_1	1	2,016.76	2,054.41	-1.83%	✅
slim-inceptionv4_1	1	216.94	216.37	0.27%	✅
onnx-taau-downsample	1	306.46	305.32	0.37%	✅
dlrm-criteoterabyte	1	21.65	21.68	-0.11%	✅
dlrm-criteoterabyte_fp16	1	40.71	40.71	-0.01%	✅
agentmodel	1	5,708.02	5,769.94	-1.07%	✅
unet_fp16	2	55.95	55.92	0.04%	✅
resnet50v1_fp16	1	945.49	947.58	-0.22%	✅
bert_base_cased_fp16	64	970.04	970.06	-0.00%	✅
bert_large_uncased_fp16	32	304.71	304.67	0.01%	✅
bert_large_fp16	1	166.91	167.45	-0.32%	✅
distilgpt2_fp16	16	1,277.75	1,277.04	0.06%	✅

This build is OK for merge ✅

migraphx-bot · 2023-10-18T19:27:52Z

:white_check_mark:bert-mrpc-onnx: PASSED: MIGraphX meets tolerance

:white_check_mark:bert-mrpc-tf: PASSED: MIGraphX meets tolerance

:white_check_mark:pytorch-examples-wlang-gru: PASSED: MIGraphX meets tolerance

:white_check_mark:pytorch-examples-wlang-lstm: PASSED: MIGraphX meets tolerance

:white_check_mark:torchvision-resnet50_1: PASSED: MIGraphX meets tolerance

🔴torchvision-inceptionv3_1: FAILED: MIGraphX is not within tolerance - check verbose output

:white_check_mark:cadene-dpn92_1: PASSED: MIGraphX meets tolerance

:white_check_mark:cadene-resnext101_1: PASSED: MIGraphX meets tolerance

:white_check_mark:slim-vgg16_1: PASSED: MIGraphX meets tolerance

:white_check_mark:slim-mobilenet_1: PASSED: MIGraphX meets tolerance

🔴slim-inceptionv4_1: FAILED: MIGraphX is not within tolerance - check verbose output

:white_check_mark:dlrm-criteoterabyte: PASSED: MIGraphX meets tolerance

:white_check_mark:agentmodel: PASSED: MIGraphX meets tolerance

:white_check_mark:unet: PASSED: MIGraphX meets tolerance

:white_check_mark:resnet50v1: PASSED: MIGraphX meets tolerance

🔴bert_base_cased_fp16: FAILED: MIGraphX is not within tolerance - check verbose output

🔴bert_large_uncased_fp16: FAILED: MIGraphX is not within tolerance - check verbose output

:white_check_mark:bert_large: PASSED: MIGraphX meets tolerance

🔴distilgpt2_fp16: FAILED: MIGraphX is not within tolerance - check verbose output

ahsan-ca · 2023-10-20T16:31:16Z

The get_available_gpu_memory error seem to be machine dependant. I was trying this on another machine and I do not see these errors.
However, I see the following errors. But these errors also appear on develop branch, so do not seem that this PR caused the errors.

[   RUN    ] gemm_2args_mm_5
Benchmarking gpu::mlir_op: 21 configs
Fastest solution: 64,64,32,4,4,2
FAILED: gpu
RMS Error: 0.504296
Max diff: 1.61328
Mismatch at 24: 0.121094 != -0.171875

module: "main"
2 = @param:2 -> float_type, {2, 3, 3, 4}, {36, 12, 4, 1}, target_id=0
1 = @param:1 -> float_type, {2, 1, 2, 3}, {6, 6, 3, 1}, target_id=0
@2 = multibroadcast[out_lens={2, 3, 2, 3},out_dyn_dims={}](1) -> float_type, {2, 3, 2, 3}, {6, 0, 3, 1}, target_id=0
@3 = dot(@2,2) -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0


ref:
module: "main"
2 = @param:2 -> float_type, {2, 3, 3, 4}, {36, 12, 4, 1}, target_id=0
1 = @param:1 -> float_type, {2, 1, 2, 3}, {6, 6, 3, 1}, target_id=0
@2 = multibroadcast[out_lens={2, 3, 2, 3},out_dyn_dims={}](1) -> float_type, {2, 3, 2, 3}, {6, 0, 3, 1}, target_id=0
@3 = dot(@2,2) -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0


gpu:
module: "main"
@0 = check_context::migraphx::gpu::context -> float_type, {}, {}, target_id=0
output = @param:output -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0
2 = @param:2 -> float_type, {2, 3, 3, 4}, {36, 12, 4, 1}, target_id=0
1 = @param:1 -> float_type, {2, 1, 2, 3}, {6, 6, 3, 1}, target_id=0
@4 = multibroadcast[out_lens={2, 3, 2, 3},out_dyn_dims={}](1) -> float_type, {2, 3, 2, 3}, {6, 0, 3, 1}, target_id=0
@5 = gpu::code_object[code_object=3752,symbol_name=mlir_dot,global=384,local=64,](@4,2,output) -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0



void run_verify::verify(const std::string &, const migraphx::program &, const migraphx::compile_options &) const
/code/AMDMIGraphX/AMDMIGraphX/test/verify/run_verify.cpp:264:
    FAILED: passed [ 0 ]
[  FAILED  ] gemm_2args_mm_5: Test failure
[   RUN    ] gemm_2args_mm_6
Benchmarking gpu::mlir_op: 21 configs
Fastest solution: 64,64,32,8,4,2
FAILED: gpu
RMS Error: 0.292345
Max diff: 1.59375
Mismatch at 24: 0.582031 != 0

module: "main"
2 = @param:2 -> float_type, {1, 3, 3, 4}, {36, 12, 4, 1}, target_id=0
1 = @param:1 -> float_type, {2, 1, 2, 3}, {6, 6, 3, 1}, target_id=0
@2 = multibroadcast[out_lens={2, 3, 2, 3},out_dyn_dims={}](1) -> float_type, {2, 3, 2, 3}, {6, 0, 3, 1}, target_id=0
@3 = multibroadcast[out_lens={2, 3, 3, 4},out_dyn_dims={}](2) -> float_type, {2, 3, 3, 4}, {0, 12, 4, 1}, target_id=0
@4 = dot(@2,@3) -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0


ref:
module: "main"
2 = @param:2 -> float_type, {1, 3, 3, 4}, {36, 12, 4, 1}, target_id=0
1 = @param:1 -> float_type, {2, 1, 2, 3}, {6, 6, 3, 1}, target_id=0
@2 = multibroadcast[out_lens={2, 3, 2, 3},out_dyn_dims={}](1) -> float_type, {2, 3, 2, 3}, {6, 0, 3, 1}, target_id=0
@3 = multibroadcast[out_lens={2, 3, 3, 4},out_dyn_dims={}](2) -> float_type, {2, 3, 3, 4}, {0, 12, 4, 1}, target_id=0
@4 = dot(@2,@3) -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0


gpu:
module: "main"
@0 = check_context::migraphx::gpu::context -> float_type, {}, {}, target_id=0
2 = @param:2 -> float_type, {1, 3, 3, 4}, {36, 12, 4, 1}, target_id=0
1 = @param:1 -> float_type, {2, 1, 2, 3}, {6, 6, 3, 1}, target_id=0
@3 = multibroadcast[out_lens={2, 3, 3, 4},out_dyn_dims={}](2) -> float_type, {2, 3, 3, 4}, {0, 12, 4, 1}, target_id=0
@4 = multibroadcast[out_lens={2, 3, 2, 3},out_dyn_dims={}](1) -> float_type, {2, 3, 2, 3}, {6, 0, 3, 1}, target_id=0
output = @param:output -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0
@6 = gpu::code_object[code_object=3880,symbol_name=mlir_dot,global=384,local=64,](@4,@3,output) -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0



void run_verify::verify(const std::string &, const migraphx::program &, const migraphx::compile_options &) const
/code/AMDMIGraphX/AMDMIGraphX/test/verify/run_verify.cpp:264:
    FAILED: passed [ 0 ]
[  FAILED  ] gemm_2args_mm_6: Test failure

test_verify: /usr/local/cget/build/tmp-82ea508a033d44d9b28ff38af027eb7d/rocMLIR-507bb94ce7873786486d296ec81d2eadaab49003/external/llvm-project/mlir/lib/Dialect/Tensor/IR/TensorOps.cpp:1394: static RankedTensorType mlir::tensor::CollapseShapeOp::inferCollapsedType(RankedTensorType, ArrayRef<AffineMap>): Assertion `isReassociationValid(reassociation) && "invalid reassociation"' failed.
CMake Error at gdb/test_test_verify_general/run.cmake:16 (message):
  Test failed

umangyadav · 2023-10-20T21:47:42Z

However, I see the following errors. But these errors also appear on develop branch, so do not seem that this PR caused the errors.

They don't seem to be failing on CI with MLIR. Can you check if your MLIR dependency is up-to-date ?
it may actually have to do with SQLite3 version.
CC: @manupak @krzysz00

krzysz00 · 2023-10-23T14:37:15Z

(Ping noted, no idea what I can do to help here)

manupak · 2023-10-23T15:05:39Z

We dont have a (meaningful) sqlite dependency ( tagging @jerryyin to confirm as jerry worked on it last).
I checked on 7604ecf in develop and I dont see these issues either.

manupak · 2023-10-23T15:07:00Z

I suspect its dev environment issue ? as 1) I cant reproduce it and 2) CI cant reproduce it.. So might worth checking the MLIR commit hash being used in the dev env.

ahsan-ca · 2023-10-23T15:11:48Z

I suspect its dev environment issue ? as 1) I cant reproduce it and 2) CI cant reproduce it.. So might worth checking the MLIR commit hash being used in the dev env.

Thanks for checking it out. I'll investigate it more at my end. I also suspect this to be an env issue at my end.

ahsan-ca · 2023-10-25T18:37:55Z

The get_available_gpu_memory error seem to be machine dependant. I was trying this on another machine and I do not see these errors. However, I see the following errors. But these errors also appear on develop branch, so do not seem that this PR caused the errors.

[   RUN    ] gemm_2args_mm_5
Benchmarking gpu::mlir_op: 21 configs
Fastest solution: 64,64,32,4,4,2
FAILED: gpu
RMS Error: 0.504296
Max diff: 1.61328
Mismatch at 24: 0.121094 != -0.171875

module: "main"
2 = @param:2 -> float_type, {2, 3, 3, 4}, {36, 12, 4, 1}, target_id=0
1 = @param:1 -> float_type, {2, 1, 2, 3}, {6, 6, 3, 1}, target_id=0
@2 = multibroadcast[out_lens={2, 3, 2, 3},out_dyn_dims={}](1) -> float_type, {2, 3, 2, 3}, {6, 0, 3, 1}, target_id=0
@3 = dot(@2,2) -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0


ref:
module: "main"
2 = @param:2 -> float_type, {2, 3, 3, 4}, {36, 12, 4, 1}, target_id=0
1 = @param:1 -> float_type, {2, 1, 2, 3}, {6, 6, 3, 1}, target_id=0
@2 = multibroadcast[out_lens={2, 3, 2, 3},out_dyn_dims={}](1) -> float_type, {2, 3, 2, 3}, {6, 0, 3, 1}, target_id=0
@3 = dot(@2,2) -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0


gpu:
module: "main"
@0 = check_context::migraphx::gpu::context -> float_type, {}, {}, target_id=0
output = @param:output -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0
2 = @param:2 -> float_type, {2, 3, 3, 4}, {36, 12, 4, 1}, target_id=0
1 = @param:1 -> float_type, {2, 1, 2, 3}, {6, 6, 3, 1}, target_id=0
@4 = multibroadcast[out_lens={2, 3, 2, 3},out_dyn_dims={}](1) -> float_type, {2, 3, 2, 3}, {6, 0, 3, 1}, target_id=0
@5 = gpu::code_object[code_object=3752,symbol_name=mlir_dot,global=384,local=64,](@4,2,output) -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0



void run_verify::verify(const std::string &, const migraphx::program &, const migraphx::compile_options &) const
/code/AMDMIGraphX/AMDMIGraphX/test/verify/run_verify.cpp:264:
    FAILED: passed [ 0 ]
[  FAILED  ] gemm_2args_mm_5: Test failure
[   RUN    ] gemm_2args_mm_6
Benchmarking gpu::mlir_op: 21 configs
Fastest solution: 64,64,32,8,4,2
FAILED: gpu
RMS Error: 0.292345
Max diff: 1.59375
Mismatch at 24: 0.582031 != 0

module: "main"
2 = @param:2 -> float_type, {1, 3, 3, 4}, {36, 12, 4, 1}, target_id=0
1 = @param:1 -> float_type, {2, 1, 2, 3}, {6, 6, 3, 1}, target_id=0
@2 = multibroadcast[out_lens={2, 3, 2, 3},out_dyn_dims={}](1) -> float_type, {2, 3, 2, 3}, {6, 0, 3, 1}, target_id=0
@3 = multibroadcast[out_lens={2, 3, 3, 4},out_dyn_dims={}](2) -> float_type, {2, 3, 3, 4}, {0, 12, 4, 1}, target_id=0
@4 = dot(@2,@3) -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0


ref:
module: "main"
2 = @param:2 -> float_type, {1, 3, 3, 4}, {36, 12, 4, 1}, target_id=0
1 = @param:1 -> float_type, {2, 1, 2, 3}, {6, 6, 3, 1}, target_id=0
@2 = multibroadcast[out_lens={2, 3, 2, 3},out_dyn_dims={}](1) -> float_type, {2, 3, 2, 3}, {6, 0, 3, 1}, target_id=0
@3 = multibroadcast[out_lens={2, 3, 3, 4},out_dyn_dims={}](2) -> float_type, {2, 3, 3, 4}, {0, 12, 4, 1}, target_id=0
@4 = dot(@2,@3) -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0


gpu:
module: "main"
@0 = check_context::migraphx::gpu::context -> float_type, {}, {}, target_id=0
2 = @param:2 -> float_type, {1, 3, 3, 4}, {36, 12, 4, 1}, target_id=0
1 = @param:1 -> float_type, {2, 1, 2, 3}, {6, 6, 3, 1}, target_id=0
@3 = multibroadcast[out_lens={2, 3, 3, 4},out_dyn_dims={}](2) -> float_type, {2, 3, 3, 4}, {0, 12, 4, 1}, target_id=0
@4 = multibroadcast[out_lens={2, 3, 2, 3},out_dyn_dims={}](1) -> float_type, {2, 3, 2, 3}, {6, 0, 3, 1}, target_id=0
output = @param:output -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0
@6 = gpu::code_object[code_object=3880,symbol_name=mlir_dot,global=384,local=64,](@4,@3,output) -> float_type, {2, 3, 2, 4}, {24, 8, 4, 1}, target_id=0



void run_verify::verify(const std::string &, const migraphx::program &, const migraphx::compile_options &) const
/code/AMDMIGraphX/AMDMIGraphX/test/verify/run_verify.cpp:264:
    FAILED: passed [ 0 ]
[  FAILED  ] gemm_2args_mm_6: Test failure

test_verify: /usr/local/cget/build/tmp-82ea508a033d44d9b28ff38af027eb7d/rocMLIR-507bb94ce7873786486d296ec81d2eadaab49003/external/llvm-project/mlir/lib/Dialect/Tensor/IR/TensorOps.cpp:1394: static RankedTensorType mlir::tensor::CollapseShapeOp::inferCollapsedType(RankedTensorType, ArrayRef<AffineMap>): Assertion `isReassociationValid(reassociation) && "invalid reassociation"' failed.
CMake Error at gdb/test_test_verify_general/run.cmake:16 (message):
  Test failed

The failures seem to be happening on Navi32, the tests run successfully on an MI100 system. I have opened an issue for this with the details: #2365

FYI @manupak @umangyadav

TedThemistokleous · 2023-10-25T21:20:04Z

@ahsan-ca once CI passes set this to ready-to-merge unless anyone has any other unresolved comments

ahsan-ca · 2023-10-26T14:48:45Z

@ahsan-ca once CI passes set this to ready-to-merge unless anyone has any other unresolved comments

@TedThemistokleous There are a few failures, but I am unable to sign in to Jenkins to see the failures. How does one get access to Jenkins to see the failures?

causten · 2023-10-26T14:55:02Z

/opt/rocm/llvm/bin/clang++ -Werror -g -O2 -fno-omit-frame-pointer -fsanitize=undefined,address -fno-sanitize-recover=undefined,address CMakeFiles/test_tf.dir/tf/tf_test.cpp.o -o ../bin/test_tf -Wl,-rpath,/var/jenkins/workspace/AMDMIGraphX_PR-2347/build/lib ../lib/libmigraphx_tf.so.2008000.0 ../lib/libmigraphx.so.2008000.0 -lpthread
ld.lld: error: undefined reference due to --no-allow-shlib-undefined: google::protobuf::internal::InternalMetadata::~InternalMetadata()

referenced by ../lib/libmigraphx_tf.so.2008000.0
clang++: error: linker command failed with exit code 1 (use -v to see invocation)
make[2]: *** [test/CMakeFiles/test_tf.dir/build.make:99: bin/test_tf] Error 1
make[2]: Leaving directory '/var/jenkins/workspace/AMDMIGraphX_PR-2347/build'
make[1]: *** [CMakeFiles/Makefile2:20470: test/CMakeFiles/test_tf.dir/all] E

ahsan-ca · 2023-10-26T17:46:15Z

/opt/rocm/llvm/bin/clang++ -Werror -g -O2 -fno-omit-frame-pointer -fsanitize=undefined,address -fno-sanitize-recover=undefined,address CMakeFiles/test_tf.dir/tf/tf_test.cpp.o -o ../bin/test_tf -Wl,-rpath,/var/jenkins/workspace/AMDMIGraphX_PR-2347/build/lib ../lib/libmigraphx_tf.so.2008000.0 ../lib/libmigraphx.so.2008000.0 -lpthread ld.lld: error: undefined reference due to --no-allow-shlib-undefined: google::protobuf::internal::InternalMetadata::~InternalMetadata()

referenced by ../lib/libmigraphx_tf.so.2008000.0
clang++: error: linker command failed with exit code 1 (use -v to see invocation)
make[2]: *** [test/CMakeFiles/test_tf.dir/build.make:99: bin/test_tf] Error 1
make[2]: Leaving directory '/var/jenkins/workspace/AMDMIGraphX_PR-2347/build'
make[1]: *** [CMakeFiles/Makefile2:20470: test/CMakeFiles/test_tf.dir/all] E

I suspect this to be caused by an issue in Protobuf : ABI may depend on NDEBUG. This issue has been fixed in Protobuf v21.3. I have updated the Protobuf version to v21.3 to see if it fixes the error and how this version interacts with other packages. Need to wait for Jenkins tests to see how this plays out.

ahsan-ca · 2023-10-26T18:20:32Z

/opt/rocm/llvm/bin/clang++ -Werror -g -O2 -fno-omit-frame-pointer -fsanitize=undefined,address -fno-sanitize-recover=undefined,address CMakeFiles/test_tf.dir/tf/tf_test.cpp.o -o ../bin/test_tf -Wl,-rpath,/var/jenkins/workspace/AMDMIGraphX_PR-2347/build/lib ../lib/libmigraphx_tf.so.2008000.0 ../lib/libmigraphx.so.2008000.0 -lpthread ld.lld: error: undefined reference due to --no-allow-shlib-undefined: google::protobuf::internal::InternalMetadata::~InternalMetadata()

referenced by ../lib/libmigraphx_tf.so.2008000.0
clang++: error: linker command failed with exit code 1 (use -v to see invocation)
make[2]: *** [test/CMakeFiles/test_tf.dir/build.make:99: bin/test_tf] Error 1
make[2]: Leaving directory '/var/jenkins/workspace/AMDMIGraphX_PR-2347/build'
make[1]: *** [CMakeFiles/Makefile2:20470: test/CMakeFiles/test_tf.dir/all] E

I suspect this to be caused by an issue in Protobuf : ABI may depend on NDEBUG. This issue has been fixed in Protobuf v21.3. I have updated the Protobuf version to v21.3 to see if it fixes the error and how this version interacts with other packages. Need to wait for Jenkins tests to see how this plays out.

Ah, this did not work out. Looking into it.

ahsan-ca · 2023-10-26T23:43:41Z

/opt/rocm/llvm/bin/clang++ -Werror -g -O2 -fno-omit-frame-pointer -fsanitize=undefined,address -fno-sanitize-recover=undefined,address CMakeFiles/test_tf.dir/tf/tf_test.cpp.o -o ../bin/test_tf -Wl,-rpath,/var/jenkins/workspace/AMDMIGraphX_PR-2347/build/lib ../lib/libmigraphx_tf.so.2008000.0 ../lib/libmigraphx.so.2008000.0 -lpthread ld.lld: error: undefined reference due to --no-allow-shlib-undefined: google::protobuf::internal::InternalMetadata::~InternalMetadata()

referenced by ../lib/libmigraphx_tf.so.2008000.0
clang++: error: linker command failed with exit code 1 (use -v to see invocation)
make[2]: *** [test/CMakeFiles/test_tf.dir/build.make:99: bin/test_tf] Error 1
make[2]: Leaving directory '/var/jenkins/workspace/AMDMIGraphX_PR-2347/build'
make[1]: *** [CMakeFiles/Makefile2:20470: test/CMakeFiles/test_tf.dir/all] E

I suspect this to be caused by an issue in Protobuf : ABI may depend on NDEBUG. This issue has been fixed in Protobuf v21.3. I have updated the Protobuf version to v21.3 to see if it fixes the error and how this version interacts with other packages. Need to wait for Jenkins tests to see how this plays out.

Ah, this did not work out. Looking into it.

The install dependencies step failed with Protobuf v21.3. I also tried Protobuf v24.3 but it still failed the install dependencies step.

ahsan-ca · 2023-10-27T00:18:59Z

The original version of Protobuf 3.11.0 had the security vulnerability CVE-2021-22570. Protobuf version 3.15.0 or greater fixes this vulnerability (Source).
I am now trying Protobuf version 3.19.6 (as Protobuf <= 3.19, most likely, does not have the undefined reference issue) to see if it does not break our CI. If it does not, then we should be good as it satisfies the requirement of fixing security vulnerability CVE-2021-22570

ahsan-ca · 2023-10-27T14:12:21Z

The original version of Protobuf 3.11.0 had the security vulnerability CVE-2021-22570. Protobuf version 3.15.0 or greater fixes this vulnerability. I am now trying Protobuf version 3.19.6 (as Protobuf <= 3.19, most likely, does not have the undefined reference issue) to see if it does not break our CI. If it does not, then we should be good as it satisfies the requirement of fixing security vulnerability CVE-2021-22570

All the CI tests pass so far, just waiting on Jenkins tests to pass before this is ready to be merged.

causten · 2023-10-27T14:16:27Z

[ 22%] Linking CXX executable ../bin/test_tf
cd /var/jenkins/workspace/AMDMIGraphX_PR-2347/build/test && /usr/local/lib/python3.8/dist-packages/cmake/data/bin/cmake -E cmake_link_script CMakeFiles/test_tf.dir/link.txt --verbose=1
/opt/rocm/llvm/bin/clang++ -Werror -g -O2 -fsanitize=undefined -fno-sanitize=vptr,function -fno-sanitize-recover=undefined CMakeFiles/test_tf.dir/tf/tf_test.cpp.o -o ../bin/test_tf  -Wl,-rpath,/var/jenkins/workspace/AMDMIGraphX_PR-2347/build/lib ../lib/libmigraphx_tf.so.2008000.0 ../lib/libmigraphx.so.2008000.0 -lpthread 
ld.lld: error: undefined reference due to --no-allow-shlib-undefined: google::protobuf::internal::InternalMetadata::~InternalMetadata()
>>> referenced by ../lib/libmigraphx_tf.so.2008000.0
clang++: error: linker command failed with exit code 1 (use -v to see invocation)
make[2]: *** [test/CMakeFiles/test_tf.dir/build.make:99: bin/test_tf] Error 1
make[2]: Leaving directory '/var/jenkins/workspace/AMDMIGraphX_PR-2347/build'
make[1]: *** [CMakeFiles/Makefile2:25531: test/CMakeFiles/test_tf.dir/all] Error 2
make[1]: *** Waiting for unfinished jobs....

ahsan-ca · 2023-10-27T14:19:58Z

[ 22%] Linking CXX executable ../bin/test_tf
cd /var/jenkins/workspace/AMDMIGraphX_PR-2347/build/test && /usr/local/lib/python3.8/dist-packages/cmake/data/bin/cmake -E cmake_link_script CMakeFiles/test_tf.dir/link.txt --verbose=1
/opt/rocm/llvm/bin/clang++ -Werror -g -O2 -fsanitize=undefined -fno-sanitize=vptr,function -fno-sanitize-recover=undefined CMakeFiles/test_tf.dir/tf/tf_test.cpp.o -o ../bin/test_tf  -Wl,-rpath,/var/jenkins/workspace/AMDMIGraphX_PR-2347/build/lib ../lib/libmigraphx_tf.so.2008000.0 ../lib/libmigraphx.so.2008000.0 -lpthread 
ld.lld: error: undefined reference due to --no-allow-shlib-undefined: google::protobuf::internal::InternalMetadata::~InternalMetadata()
>>> referenced by ../lib/libmigraphx_tf.so.2008000.0
clang++: error: linker command failed with exit code 1 (use -v to see invocation)
make[2]: *** [test/CMakeFiles/test_tf.dir/build.make:99: bin/test_tf] Error 1
make[2]: Leaving directory '/var/jenkins/workspace/AMDMIGraphX_PR-2347/build'
make[1]: *** [CMakeFiles/Makefile2:25531: test/CMakeFiles/test_tf.dir/all] Error 2
make[1]: *** Waiting for unfinished jobs....

Ah, the same error persists.

ahsan-ca · 2023-10-27T19:58:37Z

Seems like 3.18.0 worked, I am now trying 3.19.0 in a bid to get the most recent version of Protobuf that does not have the error we saw with later versions of Protobuf.

ahsan-ca · 2023-12-15T18:10:24Z

Seems like 3.18.0 worked, I am now trying 3.19.0 in a bid to get the most recent version of Protobuf that does not have the error we saw with later versions of Protobuf.

For Posterity; 3.19.0 worked so updated Protobuf to it.

ahsan-ca requested review from pfultz2, causten, CharlieL7 and umangyadav October 18, 2023 15:32

ahsan-ca self-assigned this Oct 18, 2023

ahsan-ca linked an issue Oct 18, 2023 that may be closed by this pull request

Bump protobuf and sqlite3 to resolve CVE issues #2341

Closed

ahsan-ca force-pushed the update-python-packages branch from b48134a to 87c1ddd Compare October 18, 2023 15:36

umangyadav reviewed Oct 18, 2023

View reviewed changes

ahsan-ca marked this pull request as draft October 18, 2023 15:47

ahsan-ca force-pushed the update-python-packages branch from 87c1ddd to 3e1ca8e Compare October 19, 2023 22:16

ahsan-ca marked this pull request as ready for review October 20, 2023 16:35

ahsan-ca requested a review from umangyadav October 20, 2023 20:34

pfultz2 approved these changes Oct 25, 2023

View reviewed changes

ahsan-ca force-pushed the update-python-packages branch from 3e1ca8e to 0166274 Compare October 25, 2023 20:04

ahsan-ca requested a review from TedThemistokleous October 25, 2023 20:54

umangyadav approved these changes Oct 25, 2023

View reviewed changes

TedThemistokleous approved these changes Oct 25, 2023

View reviewed changes

ahsan-ca force-pushed the update-python-packages branch from 8bb716d to 6c502ac Compare October 26, 2023 17:53

ahsan-ca force-pushed the update-python-packages branch 6 times, most recently from e544df2 to 59516c0 Compare October 26, 2023 23:02

ahsan-ca force-pushed the update-python-packages branch from 59516c0 to 3313614 Compare October 26, 2023 23:51

Update versions for sqlite and protobuf

92defb8

ahsan-ca force-pushed the update-python-packages branch from 3313614 to 92defb8 Compare October 27, 2023 15:01

causten and others added 2 commits October 27, 2023 12:29

Merge branch 'develop' into update-python-packages

a61ce76

CHeck version 3.19.0 to see if it works

5929fe6

Merge branch 'develop' into update-python-packages

8f1dd36

causten merged commit d1305d0 into develop Oct 30, 2023
8 of 10 checks passed

causten deleted the update-python-packages branch October 30, 2023 02:26

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Update versions for sqlite and protobuf #2347

Update versions for sqlite and protobuf #2347

ahsan-ca commented Oct 18, 2023

umangyadav Oct 18, 2023

ahsan-ca Oct 18, 2023

umangyadav Oct 18, 2023

ahsan-ca Oct 18, 2023 •

edited

Loading

ahsan-ca Oct 18, 2023 •

edited

Loading

ahsan-ca Oct 20, 2023

codecov bot commented Oct 18, 2023 •

edited

Loading

ahsan-ca commented Oct 18, 2023 •

edited

Loading

migraphx-bot commented Oct 18, 2023 •

edited

Loading

migraphx-bot commented Oct 18, 2023

ahsan-ca commented Oct 20, 2023 •

edited

Loading

umangyadav commented Oct 20, 2023 •

edited

Loading

krzysz00 commented Oct 23, 2023

manupak commented Oct 23, 2023 •

edited

Loading

manupak commented Oct 23, 2023 •

edited

Loading

ahsan-ca commented Oct 23, 2023

ahsan-ca commented Oct 25, 2023

TedThemistokleous commented Oct 25, 2023

ahsan-ca commented Oct 26, 2023

causten commented Oct 26, 2023

ahsan-ca commented Oct 26, 2023 •

edited

Loading

ahsan-ca commented Oct 26, 2023

ahsan-ca commented Oct 26, 2023

ahsan-ca commented Oct 27, 2023 •

edited

Loading

ahsan-ca commented Oct 27, 2023

causten commented Oct 27, 2023 •

edited

Loading

ahsan-ca commented Oct 27, 2023

ahsan-ca commented Oct 27, 2023

ahsan-ca commented Dec 15, 2023

Update versions for sqlite and protobuf #2347

Update versions for sqlite and protobuf #2347

Conversation

ahsan-ca commented Oct 18, 2023

umangyadav Oct 18, 2023

Choose a reason for hiding this comment

ahsan-ca Oct 18, 2023

Choose a reason for hiding this comment

umangyadav Oct 18, 2023

Choose a reason for hiding this comment

ahsan-ca Oct 18, 2023 • edited Loading

Choose a reason for hiding this comment

ahsan-ca Oct 18, 2023 • edited Loading

Choose a reason for hiding this comment

ahsan-ca Oct 20, 2023

Choose a reason for hiding this comment

codecov bot commented Oct 18, 2023 • edited Loading

Codecov Report

ahsan-ca commented Oct 18, 2023 • edited Loading

migraphx-bot commented Oct 18, 2023 • edited Loading

migraphx-bot commented Oct 18, 2023

ahsan-ca commented Oct 20, 2023 • edited Loading

umangyadav commented Oct 20, 2023 • edited Loading

krzysz00 commented Oct 23, 2023

manupak commented Oct 23, 2023 • edited Loading

manupak commented Oct 23, 2023 • edited Loading

ahsan-ca commented Oct 23, 2023

ahsan-ca commented Oct 25, 2023

TedThemistokleous commented Oct 25, 2023

ahsan-ca commented Oct 26, 2023

causten commented Oct 26, 2023

ahsan-ca commented Oct 26, 2023 • edited Loading

ahsan-ca commented Oct 26, 2023

ahsan-ca commented Oct 26, 2023

ahsan-ca commented Oct 27, 2023 • edited Loading

ahsan-ca commented Oct 27, 2023

causten commented Oct 27, 2023 • edited Loading

ahsan-ca commented Oct 27, 2023

ahsan-ca commented Oct 27, 2023

ahsan-ca commented Dec 15, 2023

ahsan-ca Oct 18, 2023 •

edited

Loading

ahsan-ca Oct 18, 2023 •

edited

Loading

codecov bot commented Oct 18, 2023 •

edited

Loading

ahsan-ca commented Oct 18, 2023 •

edited

Loading

migraphx-bot commented Oct 18, 2023 •

edited

Loading

ahsan-ca commented Oct 20, 2023 •

edited

Loading

umangyadav commented Oct 20, 2023 •

edited

Loading

manupak commented Oct 23, 2023 •

edited

Loading

manupak commented Oct 23, 2023 •

edited

Loading

ahsan-ca commented Oct 26, 2023 •

edited

Loading

ahsan-ca commented Oct 27, 2023 •

edited

Loading

causten commented Oct 27, 2023 •

edited

Loading