[llvm-branch-commits] [llvm] 8ffda23 - MCContext::reportError: don't call report_fatal_error

[Fangrui Song via llvm-branch-commits]

Author: Fangrui Song
Date: 2020-12-20T23:23:12-08:00
New Revision: 8ffda237a6699780bc657a97ae178a3ea078957c

URL: https://github.com/llvm/llvm-project/commit/8ffda237a6699780bc657a97ae178a3ea078957c
DIFF: https://github.com/llvm/llvm-project/commit/8ffda237a6699780bc657a97ae178a3ea078957c.diff

LOG: MCContext::reportError: don't call report_fatal_error

Errors from MCAssembler, MCObjectStreamer and *ObjectWriter typically cause a crash:

```
% cat c.c
int bar;
extern int foo __attribute__((alias("bar")));
% clang -c -fcommon c.c
fatal error: error in backend: Common symbol 'bar' cannot be used in assignment expr
PLEASE submit a bug report to ...
Stack dump:
...
```

`LLVMTargetMachine::addPassesToEmitFile` constructs `MachineModuleInfoWrapperPass`
which creates a MCContext without SourceMgr. `MCContext::reportError` calls
`report_fatal_error` which gets captured by Clang `LLVMErrorHandler` and gets translated
to the output above.

Since `MCContext::reportError` errors indicate user errors, such a crashing style error
is inappropriate. So this patch changes `report_fatal_error` to `SourceMgr().PrintMessage`.
```
% clang -c -fcommon c.c
<unknown>:0: error: Common symbol 'bar' cannot be used in assignment expr
```

Ideally we should at least recover the original filename (the line information
is generally lost).  That requires general improvement to MC diagnostics,
because currently in many cases SMLoc information is lost.

Added: 
    

Modified: 
    llvm/lib/MC/MCContext.cpp
    llvm/test/CodeGen/AMDGPU/GlobalISel/lds-zero-initializer.ll
    llvm/test/CodeGen/AMDGPU/lds-initializer.ll
    llvm/test/CodeGen/AMDGPU/lds-zero-initializer.ll
    llvm/test/CodeGen/XCore/section-name.ll

Removed: 
    


################################################################################
diff  --git a/llvm/lib/MC/MCContext.cpp b/llvm/lib/MC/MCContext.cpp
index fe84283a9112..d054c93e37f4 100644
--- a/llvm/lib/MC/MCContext.cpp
+++ b/llvm/lib/MC/MCContext.cpp
@@ -825,13 +825,13 @@ void MCContext::reportError(SMLoc Loc, const Twine &Msg) {
 
   // If we have a source manager use it. Otherwise, try using the inline source
   // manager.
-  // If that fails, use the generic report_fatal_error().
+  // If that fails, construct a temporary SourceMgr.
   if (SrcMgr)
     SrcMgr->PrintMessage(Loc, SourceMgr::DK_Error, Msg);
   else if (InlineSrcMgr)
     InlineSrcMgr->PrintMessage(Loc, SourceMgr::DK_Error, Msg);
   else
-    report_fatal_error(Msg, false);
+    SourceMgr().PrintMessage(Loc, SourceMgr::DK_Error, Msg);
 }
 
 void MCContext::reportWarning(SMLoc Loc, const Twine &Msg) {

diff  --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/lds-zero-initializer.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/lds-zero-initializer.ll
index 8c84c1c5e306..f4b8561a4343 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/lds-zero-initializer.ll
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/lds-zero-initializer.ll
@@ -1,3 +1,3 @@
-; RUN: not --crash llc -global-isel -march=amdgcn -mcpu=tonga < %S/../lds-zero-initializer.ll 2>&1 | FileCheck %s
+; RUN: not llc -global-isel -march=amdgcn -mcpu=tonga < %S/../lds-zero-initializer.ll 2>&1 | FileCheck %s
 
-; CHECK: error: <unknown>:0:0: in function load_zeroinit_lds_global void (i32 addrspace(1)*, i1): unsupported initializer for address space
+; CHECK: <unknown>:0: error: lds: unsupported initializer for address space

diff  --git a/llvm/test/CodeGen/AMDGPU/lds-initializer.ll b/llvm/test/CodeGen/AMDGPU/lds-initializer.ll
index 42a9782ad0b7..2c2ccd0582ac 100644
--- a/llvm/test/CodeGen/AMDGPU/lds-initializer.ll
+++ b/llvm/test/CodeGen/AMDGPU/lds-initializer.ll
@@ -1,5 +1,5 @@
-; RUN: not --crash llc -march=amdgcn -mcpu=tahiti < %s 2>&1 | FileCheck %s
-; RUN: not --crash llc -march=amdgcn -mcpu=tonga < %s 2>&1 | FileCheck %s
+; RUN: llc -march=amdgcn -mcpu=tahiti < %s -o /dev/null 2>&1 | FileCheck %s
+; RUN: llc -march=amdgcn -mcpu=tonga < %s -o /dev/null 2>&1 | FileCheck %s
 
 ; CHECK: lds: unsupported initializer for address space
 

diff  --git a/llvm/test/CodeGen/AMDGPU/lds-zero-initializer.ll b/llvm/test/CodeGen/AMDGPU/lds-zero-initializer.ll
index 70bf4e951b99..8b3eeb125900 100644
--- a/llvm/test/CodeGen/AMDGPU/lds-zero-initializer.ll
+++ b/llvm/test/CodeGen/AMDGPU/lds-zero-initializer.ll
@@ -1,5 +1,5 @@
-; RUN: not --crash llc -march=amdgcn -mcpu=tahiti < %s 2>&1 | FileCheck %s
-; RUN: not --crash llc -march=amdgcn -mcpu=tonga < %s 2>&1 | FileCheck %s
+; RUN: llc -march=amdgcn -mcpu=tahiti < %s -o /dev/null 2>&1 | FileCheck %s
+; RUN: llc -march=amdgcn -mcpu=tonga < %s -o /dev/null 2>&1 | FileCheck %s
 
 ; CHECK: lds: unsupported initializer for address space
 

diff  --git a/llvm/test/CodeGen/XCore/section-name.ll b/llvm/test/CodeGen/XCore/section-name.ll
index 4bf6aa1fc67a..5f91702a2847 100644
--- a/llvm/test/CodeGen/XCore/section-name.ll
+++ b/llvm/test/CodeGen/XCore/section-name.ll
@@ -1,4 +1,4 @@
-; RUN: not --crash llc < %s -march=xcore 2>&1 | FileCheck %s
+; RUN: llc < %s -march=xcore -o /dev/null 2>&1 | FileCheck %s
 
 @bar = internal global i32 zeroinitializer
 
@@ -6,4 +6,4 @@ define void @".dp.bss"() {
   ret void
 }
 
-; CHECK: LLVM ERROR: invalid symbol redefinition
+; CHECK: <unknown>:0: error: invalid symbol redefinition