On 8 December 2015 at 08:50, Ard Biesheuvel <[email protected]> wrote:
> On 7 December 2015 at 14:32, Cohen, Eugene <[email protected]> wrote:
>> Ard,
>>
>>> As far as patch #4 is concerned, let's wait for confirmation from Eugene, 
>>> also
>>> regarding the issue of reporting defect against RVCT (if there is a point in
>>> doing so)
>>
>> You're referring to "[PATCH 4/4] CryptoPkg/OpensslLib: ignore more false 
>> positive warnings for RVCT" and the idea of reporting bugs to the RVCT team, 
>> right?
>>
>> My thinking is that since we're using RVCT 4 which is two generations behind 
>> now that we probably shouldn't expect to get support beyond something being 
>> a gross correctness problem.  I think someone who has spent some money on 
>> ARM Compiler 5 or 6 would be in a better position to ask for these updates.
>>
>> As for the practicality of fixing these compiler "bugs": sure from an edk2 
>> perspective these stand out because the other supported toolchains don't 
>> exhibit the artifacts.  For typical closed systems people who see this issue 
>> would just adjust the source a bit and move on.  Since I know support comes 
>> at a cost (and for an old compiler that cost may be higher) I don't think I 
>> would want to use up a silver bullet to get these fixed.
>>
>> Does OpenSSL support some sort of continuous integration concept?  If so it 
>> would be great if ARM could donate some RVCT compiler licenses so the 
>> maintainers could build against it.
>>
>
> As it turns out, I will have to move the suppression of 4017 from this
> .inf to the global RVCT definitions, since it is emitted spuriously in
> other places as well.
>
> So will proceed and commit 4/4 of this patch with the override for 4017 
> removed.
>

Correction: I will *not* proceed and commit 4/4. Instead, I will post
a new version with additional changes that are required to build
platforms using OpensslLib with RVCT.

Thanks,
Ard.
_______________________________________________
edk2-devel mailing list
[email protected]
https://lists.01.org/mailman/listinfo/edk2-devel

Reply via email to