On Mon Dec 15, 2025 at 9:34 AM UTC, Yeoreum Yun wrote:

quoted

Hi Brendan,

quoted

On Sun Dec 14, 2025 at 9:13 AM UTC, Yeoreum Yun wrote:

quoted

quoted

I don't have the context on what this code is doing so take this with
a grain of salt, but...

The point of the _nolock alloc is to give the allocator an excuse to
fail. Panicking on that failure doesn't seem like a great idea to me?

I thought first whether it changes to "static" memory area to handle
this in PREEMPT_RT.
But since this function is called while smp_cpus_done().
So, I think it's fine since there wouldn't be a contention for
memory allocation in this phase.

Then shouldn't it use _nolock unconditionally?

As you pointed out, I think it should be fine even in the !PREEMPT_RT case.
However, in case I missed something or if my understanding is incorrect,
I applied it only to the PREEMPT_RT case for now.

Hmm, I don't think "this code might be broken so let's cage it behind a
conditional" is a good strategy.

1. It bloats the codebase.

2. It's confusing to readers, now you have to try an understand why this
   conditional is here, which is a doomed effort. This could be
   mitigated with comments but, see point 1.

3. It expands the testing matrix. So now we have code that we aren't
   really sure is correct, AND it gets less test coverage.

Overall I am feeling a bit uncomfortable about this use of _nolock, but
I am also feeling pretty ignorant about PREEMPT_RT and also about this
arm64 code, so I am hesitant to suggest alternatives, I hope someone
else can offer some input here...

I understand. However, as I mentioned earlier,
my main intention was to hear opinions specifically about memory contention.

That said, if there is no memory contention,
I don’t think using the _nolock API is necessarily a bad approach.
In fact, I believe a bigger issue is that, under PREEMPT_RT,
code that uses the regular memory allocation APIs may give users the false impression
that those APIs are “safe to use,” even though they are not.

--
Sincerely,
Yeoreum Yun