Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Optimize error message #6073

Merged
merged 21 commits into from
Aug 30, 2021
Merged

Optimize error message #6073

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
21 commits
Select commit Hold shift + click to select a range
112bd9e
modified error msg
luqiang-guo Aug 26, 2021
10148bf
Merge branch 'master' into dev_optimize_the_error_message
luqiang-guo Aug 26, 2021
f30bbee
fix tensor_impl.h error msg
luqiang-guo Aug 29, 2021
c60bf20
Merge branch 'dev_optimize_the_error_message' of https://github.com/O…
luqiang-guo Aug 29, 2021
907891d
fix consistent tensor device
luqiang-guo Aug 29, 2021
8deafc4
format
luqiang-guo Aug 29, 2021
416b8bb
Follow the comment to improve
luqiang-guo Aug 29, 2021
b77a6ad
fix log
luqiang-guo Aug 29, 2021
56401cc
rename RETURN_ERROR_WITH_BUG_PROMPT
luqiang-guo Aug 29, 2021
06fcfb2
fix PRINT_BUG_PROMPT_AND_ABORT()
luqiang-guo Aug 29, 2021
cb9d0a0
fix consistent test device
luqiang-guo Aug 29, 2021
717b00f
fix consistent test device
luqiang-guo Aug 29, 2021
d9ada6d
fix comment
luqiang-guo Aug 29, 2021
8655fd1
fix comment
luqiang-guo Aug 29, 2021
b136fba
fix comment error
luqiang-guo Aug 29, 2021
b42adcb
fix comment
luqiang-guo Aug 30, 2021
0fd824d
Merge branch 'master' into dev_optimize_the_error_message
luqiang-guo Aug 30, 2021
25cfda9
Merge branch 'master' into dev_optimize_the_error_message
oneflow-ci-bot Aug 30, 2021
11fdb69
Merge branch 'master' into dev_optimize_the_error_message
oneflow-ci-bot Aug 30, 2021
239b139
Merge branch 'master' into dev_optimize_the_error_message
oneflow-ci-bot Aug 30, 2021
5422ba2
Merge branch 'master' into dev_optimize_the_error_message
oneflow-ci-bot Aug 30, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions oneflow/core/common/error.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -311,4 +311,9 @@ void ThrowError(const std::shared_ptr<cfg::ErrorProto>& error) {

const std::shared_ptr<cfg::ErrorProto>& ThreadLocalError() { return *MutThreadLocalError(); }

const char* kOfBugIssueUploadPrompt =
"This is a oneflow bug, please submit issues in "
"'https://github.com/Oneflow-Inc/oneflow/issues' include the log information of the error, the "
"minimum reproduction code, and the system information.";

} // namespace oneflow
4 changes: 4 additions & 0 deletions oneflow/core/common/error.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -132,6 +132,10 @@ inline Error&& operator<<(Error&& error, const Error& other) {
return std::move(error);
}

extern const char* kOfBugIssueUploadPrompt;

} // namespace oneflow

#define PRINT_BUG_PROMPT_AND_ABORT() LOG(FATAL) << kOfBugIssueUploadPrompt

#endif // ONEFLOW_CORE_COMMON_ERROR_H_
1 change: 1 addition & 0 deletions oneflow/core/common/maybe.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -379,6 +379,7 @@ std::string GetFormatedSerializedError(const std::shared_ptr<cfg::ErrorProto>& e
#define OF_RUNTIME_ERROR() \
return Error::RuntimeError().AddStackFrame(__FILE__, __LINE__, __FUNCTION__) << "RuntimeError " \
": "
#define RETURN_ERROR_WITH_BUG_PROMPT() OF_RUNTIME_ERROR() << kOfBugIssueUploadPrompt

#define OF_COMPLIE_OPTION_ERROR() \
return Error::CompileOptionWrongError().AddStackFrame(__FILE__, __LINE__, __FUNCTION__) \
Expand Down
144 changes: 56 additions & 88 deletions oneflow/core/framework/tensor.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -121,146 +121,108 @@ class StaticZerosTensor final : public Tensor {
// Getters
const std::shared_ptr<const Shape>& shape() const { return shape_; }
Symbol<DType> dtype() const { return CHECK_JUST(DType::Get(dtype_)); }
Maybe<TransportToken> transport_token() const {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no transport_token property";
}
Maybe<Symbol<cfg::NdSbp>> nd_sbp() const {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no nd_sbp property";
}
Maybe<Symbol<ParallelDesc>> parallel_desc() const {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no parallel_desc property";
}
Maybe<TransportToken> transport_token() const { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<Symbol<cfg::NdSbp>> nd_sbp() const { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<Symbol<ParallelDesc>> parallel_desc() const { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<Symbol<Device>> device() const { return device_; }
Maybe<Symbol<Device>*> mut_device() {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no mut_device property";
}
Maybe<Symbol<Device>*> mut_device() { RETURN_ERROR_WITH_BUG_PROMPT(); }
bool is_cuda() const {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have is_cuda property";
PRINT_BUG_PROMPT_AND_ABORT();
return false;
}
bool is_consistent() const { return false; }
bool is_local() const { return !is_consistent(); }
bool is_lazy() const {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have is_lazy property";
PRINT_BUG_PROMPT_AND_ABORT();
return false;
}
bool is_eager() const { return !is_lazy(); }
const TensorMeta& tensor_meta() const {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have tensor_meta property";
PRINT_BUG_PROMPT_AND_ABORT();
return *(TensorMeta*)nullptr;
}
Maybe<Symbol<ConsistentTensorMeta>> consistent_tensor_meta() const {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no consistent_tensor_meta property";
RETURN_ERROR_WITH_BUG_PROMPT();
}

// Getters valid only for EagerMirroredTensor
Maybe<EagerMirroredTensorImpl*> mut_eager_mirrored_tensor_impl() {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no mut_eager_mirrored_tensor_impl property";
}
Maybe<vm::EagerBlobObject> eager_blob_object() const {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no eager_blob_object property";
}
Maybe<LocalDepObject*> compute_local_dep_object() const {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no compute_local_dep_object property";
}
Maybe<bool> has_eager_blob_object() const {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no has_eager_blob_object property";
}
Maybe<TensorStorage> tensor_storage() const {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no tensor_storage property";
}
Maybe<const Stride> stride() const {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no stride property";
}
Maybe<int64_t> storage_offset() const {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no storage_offset property";
RETURN_ERROR_WITH_BUG_PROMPT();
}
Maybe<vm::EagerBlobObject> eager_blob_object() const { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<LocalDepObject*> compute_local_dep_object() const { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<bool> has_eager_blob_object() const { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<TensorStorage> tensor_storage() const { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<const Stride> stride() const { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<int64_t> storage_offset() const { RETURN_ERROR_WITH_BUG_PROMPT(); }

// Getters/Setters valid only for EagerConsistentTensor
Maybe<const Optional<Symbol<cfg::NdSbp>>&> consumer_nd_sbp_constraint() const {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no consumer_nd_sbp_constraint property";
}
Maybe<MirroredTensor> cur_rank_phy_tensor() const {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no cur_rank_phy_tensor property";
RETURN_ERROR_WITH_BUG_PROMPT();
}
Maybe<MirroredTensor> cur_rank_phy_tensor() const { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<void> set_consumer_nd_sbp_constraint(Symbol<cfg::NdSbp> val) {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no set_consumer_nd_sbp_constraint property";
RETURN_ERROR_WITH_BUG_PROMPT();
}

// Getters for autograd
bool requires_grad() const {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have requires_grad property";
PRINT_BUG_PROMPT_AND_ABORT();
return false;
}
bool is_leaf() const {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have is_leaf property";
PRINT_BUG_PROMPT_AND_ABORT();
return false;
}
bool retain_grad() const {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have retain_grad property";
PRINT_BUG_PROMPT_AND_ABORT();
return false;
}
std::shared_ptr<const FunctionNode> grad_fn_node() const {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have grad_fn_node property";
PRINT_BUG_PROMPT_AND_ABORT();
return nullptr;
}
Maybe<Tensor> acc_grad() const {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no acc_grad property";
}
Maybe<TensorArg> current_grad() const {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no current_grad property";
}
Maybe<Tensor> detach() const { OF_RUNTIME_ERROR() << "StaticZerosTensor has no detach property"; }
Maybe<Tensor> clone() const { OF_RUNTIME_ERROR() << "StaticZerosTensor has no clone property"; }
Maybe<Tensor> acc_grad() const { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<TensorArg> current_grad() const { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<Tensor> detach() const { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<Tensor> clone() const { RETURN_ERROR_WITH_BUG_PROMPT(); }
std::shared_ptr<Tensor> data() const {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have data property";
PRINT_BUG_PROMPT_AND_ABORT();
return nullptr;
}

// Setters for autograd
void set_requires_grad(bool requires_grad) {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have set_requires_grad property";
}
Maybe<void> set_retain_grad(bool retain_grad) {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no retain_grad property";
}
void set_requires_grad(bool requires_grad) { PRINT_BUG_PROMPT_AND_ABORT(); }
Maybe<void> set_retain_grad(bool retain_grad) { RETURN_ERROR_WITH_BUG_PROMPT(); }
void set_grad_fn_node(const std::shared_ptr<FunctionNode>& grad_fn_node) {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have set_grad_fn_node property";
PRINT_BUG_PROMPT_AND_ABORT();
}
const std::shared_ptr<FunctionNode>& mut_grad_fn_node() {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have mut_grad_fn_node property";
PRINT_BUG_PROMPT_AND_ABORT();
return *(std::shared_ptr<FunctionNode>*)nullptr;
}
Maybe<void> set_acc_grad(const std::shared_ptr<Tensor>& grad) {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no set_acc_grad property";
}
Maybe<Tensor> mut_acc_grad() {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no mut_acc_grad property";
}
void set_is_leaf(bool is_leaf) {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have set_is_leaf property";
}
Maybe<void> set_acc_grad(const std::shared_ptr<Tensor>& grad) { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<Tensor> mut_acc_grad() { RETURN_ERROR_WITH_BUG_PROMPT(); }
void set_is_leaf(bool is_leaf) { PRINT_BUG_PROMPT_AND_ABORT(); }
std::shared_ptr<AutogradMeta> mut_autograd_meta() {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have mut_autograd_meta property";
PRINT_BUG_PROMPT_AND_ABORT();
return nullptr;
}
bool has_autograd_meta() const {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have has_autograd_meta property";
PRINT_BUG_PROMPT_AND_ABORT();
return false;
}
void set_autograd_meta(const std::shared_ptr<AutogradMeta>& autograd_meta) {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have set_autograd_meta property";
PRINT_BUG_PROMPT_AND_ABORT();
}

user_op::TensorDesc* mut_tensor_meta() {
LOG(FATAL) << "RuntimeError: StaticZerosTensor don't have mut_tensor_meta property";
PRINT_BUG_PROMPT_AND_ABORT();
return nullptr;
}

Maybe<MirroredTensor> AsMirroredTensor();
Maybe<ConsistentTensor> AsConsistentTensor() {
OF_RUNTIME_ERROR() << "StaticZerosTensor has no AsConsistentTensor property";
}
Maybe<ConsistentTensor> AsConsistentTensor() { RETURN_ERROR_WITH_BUG_PROMPT(); }

private:
StaticZerosTensor(const std::shared_ptr<const Shape>& shape, DataType dtype,
Expand Down Expand Up @@ -386,7 +348,7 @@ class Parameter final : public TensorIf<Parameter> {
if (const auto& consistent_tensor = std::dynamic_pointer_cast<ConsistentTensor>(tensor_)) {
return consistent_tensor;
}
OF_RUNTIME_ERROR() << "Parameter Tensor has no AsConsistentTensor property";
RETURN_ERROR_WITH_BUG_PROMPT();
}

private:
Expand All @@ -405,13 +367,22 @@ class MirroredTensor final : public TensorIf<MirroredTensor>,
const std::shared_ptr<const Shape>& shape() const override { return impl_->shape(); }
Symbol<DType> dtype() const override { return CHECK_JUST(DType::Get(impl_->dtype())); }
Maybe<TransportToken> transport_token() const override {
OF_RUNTIME_ERROR() << "MirroredTensor Tensor has no transport_token property";
OF_RUNTIME_ERROR() << "Only consistent tensors have 'consistent_id', Consistent id is used to "
"synchronize rank";
}
Maybe<Symbol<cfg::NdSbp>> nd_sbp() const override {
OF_RUNTIME_ERROR() << "MirroredTensor has no nd_sbp property";
OF_RUNTIME_ERROR()
<< "Local tensor has no sbp property. "
"sbp is the description in the oneflow distributed case, you can refer to "
"https://docs.oneflow.org/master/basics_topics/essentials_of_oneflow.html; "
"For example, create a consistent tensor like this : 'x = flow.tensor((2,3, "
"placement=flow.placement(\"cuda\", {0: 0}), sbp=flow.sbp.broadcast))', then 'x.sbp' is "
"'flow.sbp.broadcast'";
}
Maybe<Symbol<ParallelDesc>> parallel_desc() const override {
OF_RUNTIME_ERROR() << "MirroredTensor has no parallel_desc property";
OF_RUNTIME_ERROR() << "Only consistent tensors have 'placement'. Placement is used to describe "
"the distribution of consistent tensor in multiple GPUs. Please use "
"'.device' for local tensors.";
}
Maybe<Symbol<Device>> device() const override { return impl_->device(); }
Maybe<Symbol<Device>*> mut_device() override { return impl_->mut_device(); }
Expand Down Expand Up @@ -470,9 +441,7 @@ class MirroredTensor final : public TensorIf<MirroredTensor>,
user_op::TensorDesc* mut_tensor_meta() override { return impl_->mut_tensor_meta(); }

Maybe<MirroredTensor> AsMirroredTensor() override { return shared_from_this(); }
Maybe<ConsistentTensor> AsConsistentTensor() override {
OF_RUNTIME_ERROR() << "MirroredTensor has no AsConsistentTensor property";
}
Maybe<ConsistentTensor> AsConsistentTensor() override { RETURN_ERROR_WITH_BUG_PROMPT(); }

private:
std::shared_ptr<MirroredTensorImpl> impl_;
Expand All @@ -493,7 +462,8 @@ class ConsistentTensor final : public TensorIf<ConsistentTensor>,
Maybe<Symbol<cfg::NdSbp>> nd_sbp() const override { return impl_->nd_sbp(); }
Maybe<Symbol<ParallelDesc>> parallel_desc() const override { return impl_->parallel_desc(); }
Maybe<Symbol<Device>> device() const override {
OF_RUNTIME_ERROR() << "ConsistentTensor has no device property";
OF_RUNTIME_ERROR() << "Only local tensors have 'device'. Please use "
"'.placement' for consistent tensors.";
}
Maybe<Symbol<Device>*> mut_device() override {
OF_RUNTIME_ERROR() << "ConsistentTensor has no mut_device property";
Expand Down Expand Up @@ -566,9 +536,7 @@ class ConsistentTensor final : public TensorIf<ConsistentTensor>,

user_op::TensorDesc* mut_tensor_meta() override { return impl_->mut_tensor_meta(); }

Maybe<MirroredTensor> AsMirroredTensor() override {
OF_RUNTIME_ERROR() << "ConsistentTensor has no AsMirroredTensor property";
}
Maybe<MirroredTensor> AsMirroredTensor() override { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<ConsistentTensor> AsConsistentTensor() override { return shared_from_this(); }

private:
Expand Down
36 changes: 11 additions & 25 deletions oneflow/core/framework/tensor_impl.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -112,12 +112,10 @@ class MirroredTensorImpl : public TensorImpl {
}
Maybe<Symbol<Device>*> mut_device() { return mut_tensor_meta()->mut_device(); }
virtual Maybe<EagerMirroredTensorImpl*> mut_eager_mirrored_tensor_impl() {
OF_RUNTIME_ERROR() << "MirroredTensorImpl has no mut_eager_mirrored_tensor_impl property";
RETURN_ERROR_WITH_BUG_PROMPT();
}

virtual Maybe<MirroredTensorImpl> detach() const {
OF_RUNTIME_ERROR() << "MirroredTensorImpl has no detach property";
}
virtual Maybe<MirroredTensorImpl> detach() const { RETURN_ERROR_WITH_BUG_PROMPT(); }

protected:
MirroredTensorImpl(const std::shared_ptr<const MirroredTensorMeta>& tensor_meta,
Expand All @@ -141,27 +139,21 @@ class ConsistentTensorImpl : public TensorImpl {
const Optional<Symbol<cfg::NdSbp>>& consumer_nd_sbp_constraint() const {
return consumer_nd_sbp_constraint_;
}
virtual Maybe<MirroredTensor> cur_rank_phy_tensor() const {
OF_RUNTIME_ERROR() << "ConsistentTensorImpl has no cur_rank_phy_tensor property";
}
virtual Maybe<MirroredTensor> cur_rank_phy_tensor() const { RETURN_ERROR_WITH_BUG_PROMPT(); }
Symbol<ConsistentTensorMeta> tensor_meta() const { return tensor_meta_; }

// Getters valid only for EagerMirroredTensorImpl
Maybe<vm::EagerBlobObject> eager_blob_object() const override {
OF_RUNTIME_ERROR() << "ConsistentTensorImpl has no eager_blob_object property";
}
Maybe<vm::EagerBlobObject> eager_blob_object() const override { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<LocalDepObject*> compute_local_dep_object() const override {
OF_RUNTIME_ERROR() << "ConsistentTensorImpl has no compute_local_dep_object property";
}
Maybe<bool> has_eager_blob_object() const override {
OF_RUNTIME_ERROR() << "ConsistentTensorImpl has no has_eager_blob_object property";
RETURN_ERROR_WITH_BUG_PROMPT();
}
Maybe<bool> has_eager_blob_object() const override { RETURN_ERROR_WITH_BUG_PROMPT(); }

// Setters
void set_consumer_nd_sbp_constraint(Symbol<cfg::NdSbp> val) { consumer_nd_sbp_constraint_ = val; }

ConsistentTensorMeta* mut_tensor_meta() {
LOG(FATAL) << "RuntimeError: ConsistentTensorImpl don't have mut_tensor_meta property";
PRINT_BUG_PROMPT_AND_ABORT();
return nullptr;
}

Expand Down Expand Up @@ -198,18 +190,12 @@ class LazyMirroredTensorImpl final : public MirroredTensorImpl {
bool is_lazy() const override { return true; }

// Getters valid only for EagerMirroredTensorImpl
Maybe<vm::EagerBlobObject> eager_blob_object() const override {
OF_RUNTIME_ERROR() << "LazyMirroredTensorImpl has no eager_blob_object property";
}
Maybe<vm::EagerBlobObject> eager_blob_object() const override { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<LocalDepObject*> compute_local_dep_object() const override {
OF_RUNTIME_ERROR() << "LazyMirroredTensorImpl has no compute_local_dep_object property";
}
Maybe<TensorStorage> tensor_storage() const override {
OF_RUNTIME_ERROR() << "LazyMirroredTensorImpl has no tensor_storage property";
}
Maybe<bool> has_eager_blob_object() const override {
OF_RUNTIME_ERROR() << "LazyMirroredTensorImpl has no has_eager_blob_object property";
RETURN_ERROR_WITH_BUG_PROMPT();
}
Maybe<TensorStorage> tensor_storage() const override { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<bool> has_eager_blob_object() const override { RETURN_ERROR_WITH_BUG_PROMPT(); }
Maybe<MirroredTensorImpl> detach() const override;
};

Expand Down