[clang] [llvm] [Clang] Add env var for nvptx-arch/amdgpu-arch timeout (PR #102521)
via cfe-commits
cfe-commits at lists.llvm.org
Thu Aug 8 12:11:52 PDT 2024
llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-clang
Author: Joel E. Denny (jdenny-ornl)
<details>
<summary>Changes</summary>
When working on very busy systems, check-offload frequently fails many tests with this diagnostic:
```
clang: error: cannot determine amdgcn architecture: /tmp/llvm/build/bin/amdgpu-arch: Child timed out: ; consider passing it via '-march'
```
The timeout is 10 seconds. This patch accepts the environment variable `CLANG_TOOL_CHAIN_PROGRAM_WAIT` to increase it.
It should be documented somewhere. Any suggestions on where?
---
Full diff: https://github.com/llvm/llvm-project/pull/102521.diff
4 Files Affected:
- (modified) clang/lib/Driver/ToolChain.cpp (+9-1)
- (modified) clang/lib/Driver/ToolChains/AMDGPU.cpp (+2-1)
- (modified) clang/lib/Driver/ToolChains/Cuda.cpp (+2-1)
- (modified) llvm/utils/lit/lit/TestingConfig.py (+1)
``````````diff
diff --git a/clang/lib/Driver/ToolChain.cpp b/clang/lib/Driver/ToolChain.cpp
index 2d50c2cbbc881c..04b281e1bb10cd 100644
--- a/clang/lib/Driver/ToolChain.cpp
+++ b/clang/lib/Driver/ToolChain.cpp
@@ -40,6 +40,7 @@
#include "llvm/Support/FileSystem.h"
#include "llvm/Support/FileUtilities.h"
#include "llvm/Support/Path.h"
+#include "llvm/Support/Process.h"
#include "llvm/Support/VersionTuple.h"
#include "llvm/Support/VirtualFileSystem.h"
#include "llvm/TargetParser/AArch64TargetParser.h"
@@ -105,7 +106,7 @@ ToolChain::ToolChain(const Driver &D, const llvm::Triple &T,
llvm::Expected<std::unique_ptr<llvm::MemoryBuffer>>
ToolChain::executeToolChainProgram(StringRef Executable,
- unsigned SecondsToWait) const {
+ unsigned DefaultSecondsToWait) const {
llvm::SmallString<64> OutputFile;
llvm::sys::fs::createTemporaryFile("toolchain-program", "txt", OutputFile);
llvm::FileRemover OutputRemover(OutputFile.c_str());
@@ -116,6 +117,13 @@ ToolChain::executeToolChainProgram(StringRef Executable,
};
std::string ErrorMessage;
+ int SecondsToWait = DefaultSecondsToWait;
+ if (std::optional<std::string> Str =
+ llvm::sys::Process::GetEnv("CLANG_TOOL_CHAIN_PROGRAM_WAIT")) {
+ int Val = std::atoi(Str->c_str());
+ if (Val > 0)
+ SecondsToWait = Val;
+ }
if (llvm::sys::ExecuteAndWait(Executable, {}, {}, Redirects, SecondsToWait,
/*MemoryLimit=*/0, &ErrorMessage))
return llvm::createStringError(std::error_code(),
diff --git a/clang/lib/Driver/ToolChains/AMDGPU.cpp b/clang/lib/Driver/ToolChains/AMDGPU.cpp
index aa8f9197cfabc3..4ed366d21f5c43 100644
--- a/clang/lib/Driver/ToolChains/AMDGPU.cpp
+++ b/clang/lib/Driver/ToolChains/AMDGPU.cpp
@@ -899,7 +899,8 @@ AMDGPUToolChain::getSystemGPUArchs(const ArgList &Args) const {
else
Program = GetProgramPath("amdgpu-arch");
- auto StdoutOrErr = executeToolChainProgram(Program, /*SecondsToWait=*/10);
+ auto StdoutOrErr = executeToolChainProgram(Program,
+ /*DefaultSecondsToWait=*/10);
if (!StdoutOrErr)
return StdoutOrErr.takeError();
diff --git a/clang/lib/Driver/ToolChains/Cuda.cpp b/clang/lib/Driver/ToolChains/Cuda.cpp
index 17c952c808f725..104217eaf5d849 100644
--- a/clang/lib/Driver/ToolChains/Cuda.cpp
+++ b/clang/lib/Driver/ToolChains/Cuda.cpp
@@ -804,7 +804,8 @@ NVPTXToolChain::getSystemGPUArchs(const ArgList &Args) const {
else
Program = GetProgramPath("nvptx-arch");
- auto StdoutOrErr = executeToolChainProgram(Program, /*SecondsToWait=*/10);
+ auto StdoutOrErr = executeToolChainProgram(Program,
+ /*DefaultSecondsToWait=*/10);
if (!StdoutOrErr)
return StdoutOrErr.takeError();
diff --git a/llvm/utils/lit/lit/TestingConfig.py b/llvm/utils/lit/lit/TestingConfig.py
index eb9f8de2a7f960..06713429d06b4b 100644
--- a/llvm/utils/lit/lit/TestingConfig.py
+++ b/llvm/utils/lit/lit/TestingConfig.py
@@ -26,6 +26,7 @@ def fromdefaults(litConfig):
"SYSTEMROOT",
"TERM",
"CLANG",
+ "CLANG_TOOL_CHAIN_PROGRAM_WAIT",
"LLDB",
"LD_PRELOAD",
"LLVM_SYMBOLIZER_PATH",
``````````
</details>
https://github.com/llvm/llvm-project/pull/102521
More information about the cfe-commits
mailing list