Re: [RESEND PATCH] mm/sparsemem: use wrapped macros instead of open-coding

[John Hubbard <jhubbard@nvidia.com>]

On 3/11/20 1:01 AM, chenqiwu wrote:
...
> 
> Hi John,
> I tested this patch only on arm64 build, not sure if any other build
> breakings. But I think any architecture is worth to support virt_to_pfn()


If a patch causes a build break (on any architecture), it normally should not
be merged into any maintainer's git tree. That's why the patch submission
checklist, here:

     https://www.kernel.org/doc/html/latest/process/submit-checklist.html

...requests "Builds on multiple CPU architectures".

We don't all always do all of the steps there, but it is still a good baseline
checklist. And conversely, if you routinely skip key steps such as cross-compilation,
then reviewers have a lot more to worry about when looking at your patch submissions,
so that potentially slows down approvals.


> API. Could you please send your compile-tested patch to x86 upstream?
> 


I don't think that will help in this case, unless of course there is something
wrong with my build setup, and I'm seeing a problem that no one else sees.


thanks,
-- 
John Hubbard
NVIDIA