[lld] r194290 - [mach-o] fix uninitialized variable

Shankar Easwaran shankare at codeaurora.org
Sun Nov 10 18:20:16 PST 2013


On 11/8/2013 7:02 PM, Chandler Carruth wrote:
> On Fri, Nov 8, 2013 at 4:15 PM, <shankare at codeaurora.org> wrote:
>
>> Hi Sean,
>>
>> I was trying to setup the buildbots for running Asan/Tsan on lld.
>>
>> Its currently under review and Rick foos is working on it, as we have some
>> build machines which could be used for running lld tests.
>>
>> MSan is not being handled currently as libstdc++ need to be compiled with
>> that flag, which I have not yet done.
>>
> Essentially, we don't yet have a great story for producing a complete
> toolchain for use with MSan. To get there, we'll need to do something like
> when building libc++ with a compiler that supports MSan, build archives and
> DSOs for use with MSan, and then adjust compiles and links that enable MSan
> to explicitly use libc++ and these special libraries.
>
> I'm not sure if we need to support "out of the box" libstdc++ use with MSan
> as there really is no effective "bootstrap" point at which to build a
> sanitized library.
We could have an explicit builder that could build libstdc++ with msan, 
run the libstdc++ tests and make it available for other builders to use 
this library if needed.

I am not sure if there is an environment similiar to this currently, 
that libraries and tools created by a different builder and reused by a 
totally different builder.

This way we get to the latest built libstdc++ with msan that could be 
used by the buildbots, it would be great.

What do you think ?

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by the Linux Foundation




More information about the llvm-commits mailing list