Re: linux-next: build failure after merge of the mm-nonmm-unstable tree

[Finn Thain]

On Fri, 2 Jan 2026, Ard Biesheuvel wrote:

> 
> In this case, I'd prefer it if we added a helper, rather than 
> duplicating the same check 3 times.

Yes, and the next patch in that series did add that helper. I didn't want 
to rewrite Peter's patch and drop his authorship credit, so I added the 
helper in my own patch... but being that Peter still hasn't sent a 
signed-off-by tag, I should probably fold those patches together and take 
the authorship credit/blame myself.

> But in this check, testing for __DISABLE_EXPORTS is perfectly 
> reasonable: it is already used in this manner across architectures.
> 

I think Sasha's objection was valid, in that bug table entries are said to 
be emitted, whereas symbols and interfaces are exported (and imported). 
But I agree that he may have overlooked the precendent for such use/abuse 
of that macro e.g. in arch/x86/include/asm/ibt.h.

Ard, what do you think about __DISABLE_BUG_TABLE? Shall I change it back 
to __DISABLE_EXPORTS if/when I resubmit this series? I'm ambivalent about 
it.