Bugfixes for test failure (#1106)

- Bugfixes - Update test instructions using docker
2025-10-01 01:18:41 +05:30
@@ -30,28 +30,31 @@ Users may checkout `amd-staging` to preview upcoming features.

 ## Testing

-Populate the empty variables in `Dockerfile.customrocmtest` based on latest CI build information.
+Populate the <usename> variable in `docker/docker-compose.customrocmtest.yml`.
+Populate the <rocm_build_image> variable in `docker/Dockerfile.customrocmtest` based on latest ROCm CI build information.

 To quickly get the environment (bash shell) for building and testing, run the following commands:
 * `cd docker`
-* `docker compose -f docker-compose.customrocmtest.yml up --force-recreate -d && docker attach docker-customrocmtest-1`
+* If the docker image is not available on the machine, then build the image, otherwise skip this step: `docker compose -f docker-compose.customrocmtest.yml build`
+* Launch the container, and check the name of the container: `docker compose -f docker-compose.customrocmtest.yml up --force-recreate -d `
+* Run bash shell on the launched container: `docker exec -it <container_name> bash`
+* If testing is done, kill the container: `docker container kill <container_name>`

-Inside the docker container, clean, build and install the project with tests enabled:
+Inside the docker container, clean, build, then install the project with tests enabled:
 ```
 rm -rf build install && cmake -B build -D CMAKE_INSTALL_PREFIX=install -D ENABLE_TESTS=ON -D INSTALL_TESTS=ON -DENABLE_COVERAGE=ON -S . && cmake --build build --target install --parallel 8
 ```

 Note that per the above command, build assets will be stored under `build` directory and installed assets will be stored under `install` directory.

-Then, to run the automated test suite, run the following command:
+Then, to run the automated test suite, run the following commands:
 ```
+mkdir build
 ctest
 ```

 For manual testing, you can find the executable at `install/bin/rocprof-compute`

-NOTE: This Dockerfile uses `ubuntu 22.04` as the base operating system image
-
 ## Standalone binary

 To create a standalone binary, run the following commands:
@@ -1,5 +1,5 @@
 # Use a base image
-FROM ubuntu:22.04
+FROM <rocm_build_image>

 # Set the working directory
 WORKDIR /app
@@ -18,33 +18,15 @@ RUN locale-gen en_US.UTF-8

 # Install Python 3.10 and pip
 RUN apt-get install -y python3.10 python3.10-venv python3.10-dev python3-pip libsqlite3-dev
-
-# Update pip
-RUN apt remove -y python3-wheel
-RUN curl https://bootstrap.pypa.io/get-pip.py -o get-pip.py
-RUN python3.10 get-pip.py
-RUN python3.10 -m pip install --upgrade pip setuptools wheel
-
-# Set Python 3.10 as the default python3
-RUN update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.10 1
-
-# Remove blinker python package
-RUN apt-get remove -y python3-blinker
-
-# Install rocm
-# Define custom version
-ARG DEB_FILE=""
-ARG AMDGPU_BUILD=""
-ARG ROCM_BUILD=""
-RUN curl -O "https://artifactory-cdn.amd.com/artifactory/list/amdgpu-deb/${DEB_FILE}"
-RUN apt-get install -y "./${DEB_FILE}"
-RUN amdgpu-repo --amdgpu-build="${AMDGPU_BUILD}" --rocm-build="compute-rocm-dkms-no-npi-hipclang/${ROCM_BUILD}"
-RUN DEBIAN_FRONTEND=noninteractive TZ="America/Toronto" amdgpu-install --yes --usecase=rocm
+RUN python3.10 -m venv venv
+ENV PATH="venv/bin:$PATH"
+RUN python -m pip install --upgrade pip

 # Install any dependencies specified in requirements.txt
+WORKDIR /app/projects/rocprofiler-compute
+COPY projects/rocprofiler-compute/requirements.txt /app/projects/rocprofiler-compute/requirements.txt
+COPY projects/rocprofiler-compute/requirements-test.txt /app/projects/rocprofiler-compute/requirements-test.txt
+RUN python -m pip install -r requirements.txt -r requirements-test.txt
+
 # Run interactive bash shell
-CMD ["/bin/bash", "-c", "\
-    cd /app/projects/rocprofiler-compute \
-    && python3.10 -m pip install -r requirements.txt -r requirements-test.txt \
-    && exec /bin/bash \
-"]
+CMD ["/bin/bash"]
@@ -1,8 +1,12 @@
 services:
-  customrocmtest: # service name
+  rocprofiler-compute-<username>: # service name
    build:
      context: ../../../
      dockerfile: projects/rocprofiler-compute/docker/Dockerfile.customrocmtest
+    image: rocprofiler-compute-<username>
+    # Forward some env vars
+    environment:
+      ROCR_VISIBLE_DEVICES: ${ROCR_VISIBLE_DEVICES}
    devices:
      - /dev/kfd
      - /dev/dri
@@ -372,12 +372,12 @@ class RocProfCompute:
            if not hasattr(self.__args, "name") or not self.__args.name:
                console_error("-n/--name is required")
            self.__args.path = str(Path(self.__args.path) / self.__args.name)
-        # Add node name to workload path
-        if self.__args.subpath == "node_name":
-            self.__args.path = str(Path(self.__args.path) / socket.gethostname())
-        # Add gpu model name to workload path
-        elif self.__args.subpath == "gpu_model":
-            self.__args.path = str(Path(self.__args.path) / self.__mspec.gpu_model)
+            # Add node name to workload path
+            if self.__args.subpath == "node_name":
+                self.__args.path = str(Path(self.__args.path) / socket.gethostname())
+            # OR, Add gpu model name to workload path
+            elif self.__args.subpath == "gpu_model":
+                self.__args.path = str(Path(self.__args.path) / self.__mspec.gpu_model)

        # Create workload directory if it does not exist
        p = Path(self.__args.path)
@@ -32,7 +32,6 @@ from abc import abstractmethod
 from pathlib import Path
 from typing import Any, Optional, Union

-import numpy as np
 import pandas as pd
 import yaml

@@ -466,7 +465,7 @@ class RocProfCompute_Base:
                    console_debug(output)

            console_log("profiling", f"Current input file: {fname}")
-            options = self.get_profiler_options(fname, self._soc)
+            options = self.get_profiler_options(str(fname), self._soc)
            start_time = time.time()
            if self.__profiler == "rocprofv3" or self.__profiler == "rocprofiler-sdk":
                # Only 1-run case is permitted for attach/detach
@@ -542,5 +541,5 @@ class RocProfCompute_Base:
        self._soc.post_profiling()


-def test_df_column_equality(df: pd.DataFrame) -> np.bool:
+def test_df_column_equality(df: pd.DataFrame) -> bool:
    return df.eq(df.iloc[:, 0], axis=0).all(1).all()
@@ -70,7 +70,7 @@ class rocprofiler_sdk_profiler(RocProfCompute_Base):
            "ROCPROFILER_LIBRARY_CTOR": "1",
            "LD_PRELOAD": ":".join(ld_preload),
            "ROCP_TOOL_LIBRARIES": rocprofiler_sdk_tool_path,
-            "LD_LIBRARY_PATH": rocm_libdir,
+            "LD_LIBRARY_PATH": str(rocm_libdir),
            "ROCPROF_KERNEL_TRACE": "1",
            "ROCPROF_OUTPUT_FORMAT": args.format_rocprof_output,
            "ROCPROF_OUTPUT_PATH": f"{args.path}/out/pmc_1",