Re: [PATCH bpf-next v5 1/7] mm, bpf: Introduce try_alloc_pages() for opportunistic page allocation

[Shakeel Butt <shakeel.butt@linux.dev>]

On Wed, Jan 15, 2025 at 12:19:26PM +0100, Vlastimil Babka wrote:
> On 1/15/25 03:17, Alexei Starovoitov wrote:
> > From: Alexei Starovoitov <ast@kernel.org>
> > 
[...]
> > +
> > +	if (nid == NUMA_NO_NODE)
> > +		nid = numa_node_id();
> > +
> > +	prepare_alloc_pages(alloc_gfp, order, nid, NULL, &ac,
> > +			    &alloc_gfp, &alloc_flags);
> > +
> > +	/*
> > +	 * Best effort allocation from percpu free list.
> > +	 * If it's empty attempt to spin_trylock zone->lock.
> > +	 */
> > +	page = get_page_from_freelist(alloc_gfp, order, alloc_flags, &ac);
> 
> What about set_page_owner() from post_alloc_hook() and it's stackdepot
> saving. I guess not an issue until try_alloc_pages() gets used later, so
> just a mental note that it has to be resolved before. Or is it actually safe?
> 

stack_depot_save() does not seem safe here.

> > +
> > +	/* Unlike regular alloc_pages() there is no __alloc_pages_slowpath(). */
> > +
> > +	trace_mm_page_alloc(page, order, alloc_gfp, ac.migratetype);
> > +	kmsan_alloc_page(page, order, alloc_gfp);
> > +	return page;
> > +}
>