[igt-dev] [PATCH i-g-t 2/2] tests: Do not use pread on userptr in gem_exec_parallel.

[Maarten Lankhorst]

Op 28-01-2021 om 18:45 schreef Chris Wilson:
> Quoting Chris Wilson (2020-12-15 09:00:21)
>> Quoting Maarten Lankhorst (2020-12-15 08:48:25)
>>> It's even silly to use pread, directly read from the pointer instead by
>>> executing gem_wait first.
>> Major GEM violation.
> And you pushed this without recording the nack.
>
> In what way does breaking the ABI and pretending it never happened help?
> -Chris

Hey,

I don't see any explanation for the nack, just a generic statement that this breaks abi. If it was more specific, I could have added a full discussion to the commits.

I have manually verified that the code bases using the userptr abi didn't break, both by running the automated tests, and inspecting the code base. On top of that,

both Dave Airlie and Jason Ekstrand acked the changes to the kernel.

~Maarten