[llvm] r203138 - [Support/LockFileManager] Make the LockFileManager more robust against races.

[Argyrios Kyrtzidis]

On Mar 11, 2014, at 10:29 AM, Chandler Carruth <chandlerc at google.com> wrote:

> On Tue, Mar 11, 2014 at 10:23 AM, Argyrios Kyrtzidis <akyrtzi at gmail.com> wrote:
>> On Mar 11, 2014, at 10:13 AM, Kostya Serebryany <kcc at google.com> wrote:
>> 
>> Can this fix be accompanied with a threaded test?
>> 
>> 
>> This may or may not be more work than the fix itself, could you file a
>> bugzilla request ?
> 
> While it is actually fairly common that producing a test is more work
> than fixes, it seems reasonable to expect multithreaded code to have
> multithreaded test cases. It shouldn't be a hard blocker for the fix
> going in, but I would assume that it is essentially being worked on
> immediately and doesn't need a PR to track?

The real use for LockFileManager is multi-process (parallel clang invocations) module building, not multi-threaded.
A synthetic multithreaded test could be constructed though not sure how usefully it covers the real use-case and when considering my other tasks I don't consider it immediate priority.