[Caching] Encoding cache key for input file with index instead of path

Avoid path encoding difference (for example, real_path vs. path from
symlink) by eliminating the path from cache key. Cache key is now
encoded with the index of the input file from all the input files from
the command-line, reguardless if those inputs will produce output or
not. This is to ensure stable ordering even the batching is different.

Add a new cache computation API that is preferred for using input index
directly. Old API for cache key is deprecated but still updated to
fallback to real_path comparsion if needed.

As a result of swift scan API change, rename the feature in JSON file to
avoid version confusion between swift-driver and libSwiftScan.

rdar://119387650
This commit is contained in:
Steven Wu
2023-12-11 12:05:11 -08:00
parent 443cac20fb
commit 76bde39ee7
16 changed files with 246 additions and 87 deletions

View File

@@ -145,10 +145,12 @@ bool replayCachedCompilerOutputs(
Optional<OutputEntry> DiagnosticsOutput;
auto replayOutputsForInputFile = [&](const std::string &InputPath,
unsigned InputIndex,
const DenseMap<file_types::ID,
std::string> &Outputs) {
auto lookupFailed = [&CanReplayAllOutput] { CanReplayAllOutput = false; };
auto OutputKey = createCompileJobCacheKeyForOutput(CAS, BaseKey, InputPath);
auto OutputKey =
createCompileJobCacheKeyForOutput(CAS, BaseKey, InputIndex);
if (!OutputKey) {
Diag.diagnose(SourceLoc(), diag::error_cas,
@@ -199,7 +201,8 @@ bool replayCachedCompilerOutputs(
}
};
auto replayOutputFromInput = [&](const InputFile &Input) {
auto replayOutputFromInput = [&](const InputFile &Input,
unsigned InputIndex) {
auto InputPath = Input.getFileName();
DenseMap<file_types::ID, std::string> Outputs;
if (!Input.outputFilename().empty())
@@ -225,27 +228,33 @@ bool replayCachedCompilerOutputs(
Outputs.try_emplace(file_types::ID::TY_CachedDiagnostics,
"<cached-diagnostics>");
return replayOutputsForInputFile(InputPath, Outputs);
return replayOutputsForInputFile(InputPath, InputIndex, Outputs);
};
auto AllInputs = InputsAndOutputs.getAllInputs();
// If there are primary inputs, look up only the primary input files.
// Otherwise, prepare to do cache lookup for all inputs.
if (InputsAndOutputs.hasPrimaryInputs())
InputsAndOutputs.forEachPrimaryInput([&](const InputFile &File) {
replayOutputFromInput(File);
return false;
});
else
llvm::for_each(InputsAndOutputs.getAllInputs(), replayOutputFromInput);
for (unsigned Index = 0; Index < AllInputs.size(); ++Index) {
const auto &Input = AllInputs[Index];
if (InputsAndOutputs.hasPrimaryInputs() && !Input.isPrimary())
continue;
replayOutputFromInput(Input, Index);
}
if (!CanReplayAllOutput)
return false;
// If there is not diagnostic output, this is a job that produces no output
// and only diagnostics, like `typecheck-module-from-interface`, look up
// diagnostics from first file.
if (!DiagnosticsOutput)
replayOutputsForInputFile(
InputsAndOutputs.getFirstOutputProducingInput().getFileName(),
"<cached-diagnostics>",
InputsAndOutputs.getIndexOfFirstOutputProducingInput(),
{{file_types::ID::TY_CachedDiagnostics, "<cached-diagnostics>"}});
// Check again to make sure diagnostics is fetched successfully.
if (!CanReplayAllOutput)
return false;