[clang][driver] Ensure we don't accumulate entries in -MJ files

[jansvoboda11]

Summary

Previously, each job would overwrite the -MJ file. This didn't quite work for Clang invocations with multiple architectures, which got fixed in D121997 by always appending to the -MJ file. That's not correct either, since the file would grow indefinitely on subsequent Clang invocations. This patch ensures the driver always removes the file before jobs fill it in by appending.

[MaskRay]

LGTM.

clang/test/Driver/compilation_database_multiarch.c

15

CHECK-EMPTY can match EOF.

[jansvoboda11]

Thanks for the review!

clang/test/Driver/compilation_database_multiarch.c

15

It can, but it's not really strict. It will match an empty line, but that might be followed by some non-empty lines, which I'd like this test to explicitly disallow.

[tschuett]

Thanks!

[MaskRay]

clang/test/Driver/compilation_database_multiarch.c

15

OK. I think CHECk-NOT: {{.}} works as well once the last byte before \n is matched.

[tschuett]

There are now several JSON fragments per file? Previously, there was one fragment per file.

This will fail now:
https://sarcasm.github.io/notes/dev/compilation-database.html#clang

[jansvoboda11]

In D128098#3594637, @tschuett wrote:

There are now several JSON fragments per file? Previously, there was one fragment per file.

This will fail now:
https://sarcasm.github.io/notes/dev/compilation-database.html#clang

There are on MacOS when building fat binaries (for both x86_64 and AArch64).

clang/lib/Driver/Driver.cpp

unsigned Model = llvm::StringSwitch<unsigned>(Name)
         .Default(~0U);
     if (Model == ~0U) {
       Diags.Report(diag::err_drv_invalid_value) << A->getAsString(Args)
                                                 << Name;
     } else
       BitcodeEmbed = static_cast<BitcodeEmbedMode>(Model);
   }
 
+  // Remove existing compilation database so that each job can append to it.
+  if (Arg *A = Args.getLastArg(options::OPT_MJ))
+    llvm::sys::fs::remove(A->getValue());
+
   // Setting up the jobs for some precompile cases depends on whether we are
   // treating them as PCH, implicit modules or C++20 ones.
   // TODO: inferring the mode like this seems fragile (it meets the objective
   // of not requiring anything new for operation, however).
   const Arg *Std = Args.getLastArg(options::OPT_std_EQ);
   ModulesModeCXX20 =
       !Args.hasArg(options::OPT_fmodules) && Std &&
       (Std->containsValue("c++20") || Std->containsValue("c++2b") ||

clang/test/Driver/compilation_database_multiarch.c

 // REQUIRES: system-darwin
 
 // RUN: rm -rf %t && mkdir -p %t
 // RUN: %clang -fdriver-only -o %t/out %s -mtargetos=macos12 -arch arm64 -arch x86_64 -MJ %t/compilation_database.json
+
+// Let's run that again and verify we're not accumulating redundant entries in the same file.
+//
+// RUN: %clang -fdriver-only -o %t/out %s -mtargetos=macos12 -arch arm64 -arch x86_64 -MJ %t/compilation_database.json
+// RUN: echo EOF >> %t/compilation_database.json
+
 // RUN: FileCheck --input-file=%t/compilation_database.json %s
 
 // CHECK:      { "directory": "{{.*}}", "file": "{{.*}}", "output": "[[OUTPUT_X86_64:.*]]", "arguments": [{{.*}}, "-o", "[[OUTPUT_X86_64]]", {{.*}} "--target=x86_64-apple-macosx12.0.0"]},
 // CHECK-NEXT: { "directory": "{{.*}}", "file": "{{.*}}", "output": "[[OUTPUT_ARM64:.*]]", "arguments": [{{.*}}, "-o", "[[OUTPUT_ARM64]]", {{.*}} "--target=arm64-apple-macosx12.0.0"]},
+// CHECK-NEXT: EOF