[cfarm-users] Is there a list of funny C compilers?

[Jacob Bachmeyer]

On 4/23/25 01:17, Jing Luo wrote:
> On 2025-04-23 13:13, Jacob Bachmeyer wrote:
>> On 4/22/25 06:13, Jing Luo via cfarm-users wrote:
>>
>>> [...]
>>>
>>> Finally I also found a "we-violate-the-GPL-because-we-don't-care"
>>> compiler [5], a sourceless, prebuilt and customized gnu toolchain by
>>> Spacemit with "--build=x86_64-linux-gnu --host=riscv64-linux-gnu",
>>> which is linked from [6]. This one deserves a ((dis)honorary)
>>> mention because cfarm95 runs on the Spacemit CPU.
>>>
>>> [...]
>>> [5] https://archive.spacemit.com/toolchain/
>>> [6]
>>> https://milkv.io/docs/jupiter/development-guide/kernel-compilation
>>
>> Hmmm, a blatant GPL violation on GCC?  Have you contacted [A1] the FSF
>> about it?
>>
>> [...]
>>
>> [A1] https://www.gnu.org/licenses/gpl-violation.html
>
> Not yet: I thought I should at least ask for the source code before 
> reporting it to FSF. But unfortunately I can't find any email address 
> or postal address or phone number to contact them...

My understanding is that the FSF will start by asking nicely for 
sources, although that requires that the FSF be provided with some way 
to contact the distributor in question...

> On the other hand, I found a forum topic related to it [7]:
>
> (a rough translation from Chinese)
> [Q3] Is the source code of Spacemit's gcc available
> [A3] No. But [we] didn't make a lot of changes on gcc, basically it's 
> [the same as] the open source (sic.) version. After all, the only 
> thing added was assembler support, i.e. changing binutils. The patches 
> supporting the instruction will be ready to be released, so that 
> everyone who build their own toolchain can also patch [it] themselves.
>
> Oh well.
>
> [7] https://forum.spacemit.com/t/topic/387
>
That "No." is a GPL violation right there, as I understand it. Can the 
Chinese original be read as "GCC is unmodified from the GNU version" or 
are they attempting to minimize the fact that they distribute a modified 
*compiler*?

GNU binutils is another package, but could be conflated as part of GCC 
because GCC uses binutils to generate binary output, and binary 
distributions of GCC therefore normally include binutils.

This leads to a possible scenario:  they are distributing a build of 
unmodified GCC with their patched binutils bundled.  Is the promised 
binutils patch available?


-- Jacob