r198858 - Disable LeakSanitizer in TableGen binaries, see PR18325

[Kostya Serebryany]

On Thu, Jan 9, 2014 at 3:15 PM, Alp Toker <alp at nuanti.com> wrote:

>
> On 09/01/2014 11:08, Chandler Carruth wrote:
>
>
>> On Thu, Jan 9, 2014 at 6:02 AM, Alp Toker <alp at nuanti.com <mailto:
>> alp at nuanti.com>> wrote:
>>
>>     Hi Kostya,
>>
>>     This doesn't look correct and fires warnings my my strict checks
>>     build.
>>
>>
>> I'm pretty sure the change works correctly, and it was definitely what
>> Kostya and I intended.
>>
>>
>>     Double-underscore prefixed identifiers are strictly reserved for
>>     the implementation.
>>
>>
>> Yes, and LeakSanitizer is something provided by the implementation? (the
>> toolchain's runtime libraries specifically) I don't see any problem with
>> double underscore here, any more than with __builtin_... functions or any
>> of the other sanitizer interfaces.
>>
>
> __builtin functions are introduced by the implementation (either the
> compiler or system header declarations) so there's no problem.
>
> Sanitizer is a system library so it's free to use the identifiers it wants.
>
> TableGen, on the other hand, is not a part of the implementation.
>
>
>
>> Can you clarify exactly what the error is and why? It's possible we could
>> provide a macro of some kind in the sanitizer headers to bundle this
>> functionality up if we need some special marking of this for checkers, but
>> without more information its hard to hell what the actual problem is.
>>
>
> These two commits _introduced_ names in the reserved namespace.
>
And that we the intention.
See the comments at compiler-rt/include/sanitizer/lsan_interface.h


>
> The checker I have is a strict warning flag I've been trying to encode the
> rules from the standard and in this case the diagnostic looks legitimate.


Can it be silenced for a particular line of code (like "// NOLINT")?

--kcc


>
>
> Alp.
>
>
>
> --
> http://www.nuanti.com
> the browser experts
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20140109/73956065/attachment.html>