Merge branch 'main' into cmake

.buildbot/jenkins-build-llvm-project.py

@@ -58,7 +58,7 @@
 LLVM_PROJECT_SHA1_REGEX     = 'git checkout ([0-9a-f]+)'
 LLVM_PROJECT_DOCKERFILE     = 'docker/Dockerfile.llvm-project'
 LLVM_PROJECT_GITHUB_URL     = 'https://api.github.com/repos/llvm/llvm-project'
-BASE_IMAGE                  = { 'static': 'ubuntu:focal',
+BASE_IMAGE                  = { 'static': 'ubuntu:jammy',
                                 'shared': 'registry.access.redhat.com/ubi8-minimal:latest' }
 LLVM_PROJECT_IMAGE          = { 'static': docker_static_image_name,
                                 'shared': docker_shared_image_name }

.buildbot/jenkins-watch-llvm-project.py

@@ -60,7 +60,7 @@
 github_repo_access_token    = os.getenv('GITHUB_REPO_ACCESS_TOKEN')
 jenkins_rest_api_token      = os.getenv('JENKINS_REST_API_TOKEN')
 
-LLVM_PROJECT_BASE_IMAGE     = 'ubuntu:focal'
+LLVM_PROJECT_BASE_IMAGE     = 'ubuntu:jammy'
 LLVM_PROJECT_WATCH_IMAGE    = 'llvm-project-watch'
 LLVM_PROJECT_GITHUB_URL     = 'https://api.github.com/repos/llvm/llvm-project'
 LLVM_PROJECT_SHA1_REGEX     = 'git checkout ([0-9a-f]+)'

docker/Dockerfile.llvm-project

@@ -1,4 +1,4 @@
-# By default, use ubuntu:jammy
+# By default, use ubuntu:jammy, remember to change Jenkins build script as well
 ARG BASE_IMAGE="ubuntu:jammy"
 FROM ${BASE_IMAGE}
 
@@ -89,6 +89,7 @@ RUN git clone -n https://github.com/llvm/llvm-project.git \
              -DCMAKE_BUILD_TYPE=Release \
              -DLLVM_ENABLE_ASSERTIONS=ON \
              -DLLVM_ENABLE_RTTI=ON \
+             -DLLVM_ENABLE_LIBEDIT=OFF \
              -DBUILD_SHARED_LIBS=${BUILD_SHARED_LIBS} \
     && cmake --build . --parallel ${NPROC} -- ${MAKEFLAGS} \
     && (cmake --build . --parallel ${NPROC} --target check-mlir || \

docs/BuildOnLinuxOSX.md

@@ -14,7 +14,7 @@ Firstly, install MLIR (as a part of LLVM-Project):
 ``` bash
 git clone -n https://github.com/llvm/llvm-project.git
 # Check out a specific branch that is known to work with ONNX-MLIR.
-cd llvm-project && git checkout 3abae1c88416858cf2e9a7ed9417bc52033933b4 && cd ..
+cd llvm-project && git checkout 6cf7fe4a9a715bcdf3f4913753109e22dfc9940b && cd ..
 ```
 
 [same-as-file]: <> (utils/build-mlir.sh)
@@ -26,7 +26,8 @@ cmake -G Ninja ../llvm \
    -DLLVM_TARGETS_TO_BUILD="host" \
    -DCMAKE_BUILD_TYPE=Release \
    -DLLVM_ENABLE_ASSERTIONS=ON \
-   -DLLVM_ENABLE_RTTI=ON
+   -DLLVM_ENABLE_RTTI=ON \
+   -DLLVM_ENABLE_LIBEDIT=OFF
 
 cmake --build . -- ${MAKEFLAGS}
 cmake --build . --target check-mlir

docs/BuildOnWindows.md

@@ -52,7 +52,7 @@ Install MLIR (as a part of LLVM-Project):
 ```shell
 git clone -n https://github.com/llvm/llvm-project.git
 # Check out a specific branch that is known to work with ONNX-MLIR.
-cd llvm-project && git checkout 3abae1c88416858cf2e9a7ed9417bc52033933b4 && cd ..
+cd llvm-project && git checkout 6cf7fe4a9a715bcdf3f4913753109e22dfc9940b && cd ..
 ```
 
 [same-as-file]: <> (utils/build-mlir.cmd)
@@ -68,7 +68,8 @@ call cmake %root_dir%\llvm-project\llvm -G "Ninja" ^
    -DLLVM_ENABLE_ASSERTIONS=ON ^
    -DLLVM_ENABLE_RTTI=ON ^
    -DLLVM_ENABLE_ZLIB=OFF ^
-   -DLLVM_INSTALL_UTILS=ON
+   -DLLVM_INSTALL_UTILS=ON ^
+   -DLLVM_ENABLE_LIBEDIT=OFF
 
 call cmake --build . --config Release
 call cmake --build . --config Release --target install

docs/Dialects/krnl.md

@@ -147,21 +147,32 @@ means to block the for loop referred to by %i using a tile size of 4.
 
 call operation
 
-The call operation provides a generic way to call an external function
-at Krnl level.  The `funcName` determines which function to call.
-The `result` is the Value to store the function return. Currently only
-one output is supported. The `result` has to be resulted memref.
-Since resolution of the output MemRef involves shape inference on ONNX Op,
-resolution should be done at lowering ONNX Op, not within krnl.Call.
-Another reason is that Krnl.call need to be defined with AllocationOp
-interface if `result` is allocated inside this Op.
-The parameters can be of any type: MemRef, NoneType or any llvm type.
-Different types of parameters will be converted, if needed, when KrnlCallOp
-is lowered. Attributes will be converted to parameters too (To be Added).
-The function signature will be determined with the types of parameters.
-An LLVM::CallOp to either a runtime library or a llvm intrinsic function
-will be generated.
-The krnl.call op will be lowered to llvm at krnl-to-llvm conversion.
+The call operation provides a generic way to replace an ONNX Op with a call
+to an external function at Krnl level. 
+`funcName` attributes determines which function to call. 
+`parameters` is the inputs to Krnl.Call. It includes the outputs and inputs 
+of the ONNX Op. The outputs and inputs are already lowered to MemRefs.
+The external function is assumed NOT to allocate or free any memory.
+'numOfOutput` attribute to tell how manu outputs Memref in parameters.
+mlir::OpTrait::AttrSizedOperandSegments is not used to put outputs and
+inputs into separate variadic parameters because I am thinking of mixing
+the inputs and outpus as required by external library.
+
+The attributes of the ONNX Op will be copied to KrnlCallOp under the control
+of the user.
+In Krnl To llvm lowering, the parameters and attributes will be lowered to
+parameters of the llvm function call.
+
+Several builder is defined to help translating an ONNX Op to Krnl.Call.
+User can provides the allocated MemRefs for outputs and the inputs
+separately. The inputs are usually the operands of the ONNX Op.
+The attributes of ONNX Op can be copied or not copied based on a bool
+parameter in the builder. Builder also provide a mechanism for user
+to selectively copy some attributes.
+
+The krnl.call op will be lowered to llvm at krnl-to-llvm conversion in which
+OMTensor is used as a container for MemRef arguments. Other representation
+of parameters, such as data pointer only, will be supported in future.
 
 Interfaces: MemoryEffectOpInterface
 
@@ -170,12 +181,12 @@ Interfaces: MemoryEffectOpInterface
 | Attribute | MLIR Type | Description |
 | :-------: | :-------: | ----------- |
 | `funcName` | ::mlir::StringAttr | string attribute
+| `numOfOutput` | ::mlir::IntegerAttr | 64-bit signed integer attribute
 
 #### Operands:
 
 | Operand | Description |
 | :-----: | ----------- |
-| `result` | any type
 | `parameters` | any type
 
 ### `krnl.copy_from_tile_buffer` (::mlir::KrnlCopyFromBufferOp)