Re: [patch] target triplet for Microblaze.

[Michael Eager <eager@eagerm.com>]

Pedro Alves wrote:
> On Thursday 18 November 2010 18:44:40, Michael Eager wrote:
>> Pedro Alves wrote:
>>> On Thursday 18 November 2010 17:43:50, Michael Eager wrote:
>>>
>>>> MicroBlaze is a proprietary processor developed by Xilinx and, AFAIK,
>>>> they have not licensed it to other manufacturers.  
>>>
>>>
>>>> Xilinx is the only possible specification which meets the definition
>>>> in Autobook.
>>> That can't be right.
>>>
>>> " cpu
>>>      The type of processor used on the system. (...)
>>>
>>>  manufacturer
>>>      A somewhat freeform field which indicates the manufacturer of the system. (...)"
>>>
>>> It's manufacturer of the system, not of the CPU.
>>> Note "used on the system", therefore cpu != system.  And "system"
>>> is really a broad term.  AFAIK, it's really a freeform field in
>>> practice, so you should be able to put anything there.  For example,
>>> if one takes xilink's toolchain sources (the gpl bits), fork / patch
>>> them for some reason, and build / distribute them oneself, one
>>> might want to legitimately change the manufacturer field to
>>> avoid confusion.
>> System, in this case, is the FPGA, manufactured by Xilinx.
> 
> Or "system" is a board that includes an FPGA manufactured by Xilinx,
> yet assembled and sold by some other vendor, that happens to
> decide to add some spice or to do some incompatible changes to Xilink's
> version of the toolchain.  In the case of an intel x86 cpu, what you
> call the "system"?  That answer to that should reveal that your
> interpretation of system cannot be the only one possible.

Perhaps you are arguing that the patch should be rejected,
so that the current behavior that only "microblaze-xilinx-gnu"
would is accepted, and the change would make "microblaze-megacorp-gnu"
also match.  If so, then I agree that this would be an undesirable
consequence.

-- 
Michael Eager	 eager@eagercon.com
1960 Park Blvd., Palo Alto, CA 94306  650-325-8077