Providing 'hsdis' binaries not possible because of GPLv2/GPLv3 license clash

[Volker Simonis]

On Mon, Jun 1, 2015 at 6:35 PM, dalibor topic <dalibor.topic at oracle.com>
wrote:

> On 28.05.2015 16:01, Volker Simonis wrote:
>
>> Chris has pointed out a real problem of the current project: it is not
>> possible to provide binaries of a part of the project (i.e. 'hsdis')
>> because of licensing issues.
>>
>
> Hi Volker,
>
> I appreciate your concern here, and accordingly spent some time doing a
> bit of research over the weekend.
>
> As far as I understand, the third party library in question has been
> traditionally distributed under a license arrangement very similar to that
> of hsdis. While I understand that at some point in time the vendor of that
> third party library changed their license preference for future releases, I
> have not been able to deduce from this thread why that change of preference
> makes it impossible to use an older release of that library published under
> a similar license arrangement to hsdis for your purpose.
>
> Better understanding this would help with the justification of the
> relicensing request as it appears on the surface that the status quo is if
> not fine, then at most mildly inconvenient.
>

Hi Dalibor,

thanks a lot for looking into this issue.

The problem with older versions of binutils which are properly licensed is
that they do not support modern and upcoming processor architectures. As
far as I know, binutils 2.17 was the last version licensed under GPLv2
while the support for Power 8 and AARCH 64 for example was only introduced
around binutils 2.23. But not only these 'exotic' processor architectures
require new versions of binutil. Even for x86 you won't be able to
disassemble modern AVX or transactional memory machine instructions
produced by the HotSpot VM.

 @Chris: reimplementing  the hsdis functionality is probably the most
>> pragmatic solution but I'd rather like to avoid code duplication here
>> to avoid problems with different incompatible version in the future.
>>
>
> Fwiw, I think LLVM MC might be (more?) interesting to consider in that
> regard in general. [0]
>

I understand your pain with relicensing the files, but I really don't think
that another disassembler is the solution here. I'm not aware of any other
disassembler which has the same, huge platform coverage like the GNU
disassembler from the binutils package and is still actively maintained.
And there's more than Intel x86:) Just think of all the Oracle-supported
architectures (open and closed). I don't think anybody wants to supports
different versions of hsdis which target different disassemblers.

Hopefully these arguments help to justify the relicensing request.

Thank you and best regards,
Volker


>
> cheers,
> dalibor topic
>
> [0] http://blog.llvm.org/2010/01/x86-disassembler.html
>
> --
> <http://www.oracle.com> Dalibor Topic | Principal Product Manager
> Phone: +494089091214 <tel:+494089091214> | Mobile: +491737185961
> <tel:+491737185961>
>
> Oracle | Kühnehöfe 5 | 22761 Hamburg
> <http://www.oracle.com/commitment> Oracle is committed to developing
> practices and products that help protect the environment
>