Re: Preliminary results - was: Re: Question on BIGGEST_ALIGNMENT in GCC on NetBSD/m68k
On Wed, 7 Jan 2026, Arnd Bergmann wrote:
> The more interesting bit then is what happens when you actually turn on
> -malign-int for the kernel itself. There are many drivers that change
> behavior, but mostly this is going to be a fix rather than a regression.
Sorry, Arnd, but wasting memory is a regression not a bug fix. And if you
do that on systems that have no memory to spare, you're probably going to
cause -ENOMEM failures.
If you want to fix issues with CONFIG_DETECT_HUNG_TASK_BLOCKER or
CONFIG_JUMP_LABEL, it's much better to simply align atomic types, as my
patches have already demonstrated.
Assumptions about atomics are prevalent. But I reject your pandering to
developers who make bad assumptions about integers, or those who would
prefer that all architectures were the same, as if all users were the
same.
Reply to: