On Fri, Oct 23, 2026 at 09:18:10AM +0100, Vincent Donnefort wrote:
> On Thu, Oct 22, 2026 at 11:48:17PM +0530, Amit Machhiwal wrote:
> > During early boot (early_trace_init() called from start_kernel()),
> > tracing allocates initial ring buffers (temp_buffer, array_buffer, and
> > snapshot_buffer) for CPU 0. Later, as secondary CPUs come online during
> > smp_init(), each CPU's ring buffer is allocated via trace_rb_cpu_prepare().
> >
> > Before attempting page allocation, __rb_allocate_pages() performs a
> > heuristic check using si_mem_available() to return early with -ENOMEM if
> > memory appears insufficient.
> >
> > However, on kernels built with CONFIG_DEFERRED_STRUCT_PAGE_INIT=y,
> > NR_FREE_PAGES only reflects the initial non-deferred pool across the entire
> > period from early_trace_init() through smp_init(). defer_init() leaves
> > only a single section per node (e.g. 16 MiB with 64 KB pages) initialized
> > up-front, and on kernels with large static binary footprints (debug
> > configurations enabling PAGE_OWNER, DEBUG_PAGEALLOC, KFENCE, or SLUB_DEBUG),
> > early core initialisations (SLUB caches, vmalloc, static ftrace records)
> > consume virtually all managed pages in this pool.
> >
> > si_mem_available() therefore returns a false negative (often 0-1 pages)
> > across both SYSTEM_BOOTING (for the initial CPU 0 buffers) and
> > SYSTEM_SCHEDULING (for secondary CPU buffers via trace_rb_cpu_prepare()),
> > causing __rb_allocate_pages() to prematurely abort with -ENOMEM.
> >
> > This failure is false: the buddy allocator is designed to replenish
> > ZONE_NORMAL on demand via deferred_grow_zone() when an actual allocation
> > is attempted. Checking si_mem_available() before calling alloc_pages_node()
> > short-circuits this on-demand growth entirely.
> >
> > deferred_grow_zone() is only permanently disabled after
> > page_alloc_init_late()
> > completes (it spawns per-node kthreads and blocks until done), which happens
> > inside kernel_init_freeable() — before system_state is set to
> > SYSTEM_FREEING_INITMEM. Skip the si_mem_available() check until
> > SYSTEM_FREEING_INITMEM, at which point all deferred pages are initialized
> > and si_mem_available() is accurate.
> >
> > Wrap the guard in #ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT so that on kernels
> > without deferred init the original check is preserved unchanged.
> >
> > Fixes: 2a872fa4e9c8 ("ring-buffer: Check if memory is available before
> > allocation")
> > Cc: [email protected]
> > Reported-by: Michal Suchánek <[email protected]>
> > Closes: https://lore.kernel.org/all/[email protected]/
> > Link: https://lore.kernel.org/all/[email protected]/
> > Signed-off-by: Amit Machhiwal <[email protected]>
>
> It is usually good practice to not use in-reply-to to link the series between
> each other.
>
> > ---
> > Changes since v2:
> > - Rebase on top of upstream ring_buffer.c changes from Vincent Donnefort;
> > update threshold to i < (nr_pages << subbuf_order). (Michal Suchánek)
> >
> > Changes since v1:
> > - Extend the guard to cover SYSTEM_SCHEDULING as well: smp_init() calls
> > trace_rb_cpu_prepare() -> __rb_allocate_pages() before
> > page_alloc_init_late()
> > completes, so the false negative window spans both SYSTEM_BOOTING and
> > SYSTEM_SCHEDULING. Use system_state >= SYSTEM_FREEING_INITMEM as the
> > correct
> > boundary. (Sashiko, Vincent Donnefort)
> > - Wrap in #ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT to preserve the original
> > check unchanged on non-deferred kernels. (Vincent Donnefort)
> > - Drop Tested-by: guard condition changed, requesting fresh test.
> >
> > kernel/trace/ring_buffer.c | 17 +++++++++++++++++
> > 1 file changed, 17 insertions(+)
> >
> > diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
> > index 9c4e40847e43..8e5a59a7b69b 100644
> > --- a/kernel/trace/ring_buffer.c
> > +++ b/kernel/trace/ring_buffer.c
> > @@ -2463,10 +2463,27 @@ static int __rb_allocate_pages(struct
> > ring_buffer_per_cpu *cpu_buffer,
> > * memory. It may not be accurate. But we don't care, we just
> > want
> > * to prevent doing any allocation when it is obvious that it is
> > * not going to succeed.
> > + *
> > + * With CONFIG_DEFERRED_STRUCT_PAGE_INIT, NR_FREE_PAGES only
> > + * reflects the initial non-deferred pool during early boot and
> > + * throughout SMP bring-up (smp_init() runs before
> > + * page_alloc_init_late() completes deferred initialization).
> > + * si_mem_available() returns a false negative across this
> > + * entire window while actual allocations succeed by growing the
> > + * zone on demand via deferred_grow_zone(). Only re-enable the
> > + * check from SYSTEM_FREEING_INITMEM onward, after
> > + * page_alloc_init_late() has finished and si_mem_available() is
> > + * accurate.
>
> This is quite a massive comment for something simple: When
> CONFIG_DEFERRED_STRUCT_PAGE_INIT, si_mem_available() will reflect the actual
> available memory only after SYSTEM_FREEING_INITMEM.
>
> > */
> > i = si_mem_available();
> > +#ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
> > + if (system_state >= SYSTEM_FREEING_INITMEM &&
> > + i < (nr_pages << subbuf_order))
> > + return -ENOMEM;
> > +#else
> > if (i < (nr_pages << subbuf_order))
> > return -ENOMEM;
> > +#endif
>
> Could we avoid the double i < (nr_pages << subbuf_order) with something like
>
> if (IS_ENABLED(CONFIG_DEFERRED_STRUCT_PAGE_INIT) &&
> system_state < SYSTEM_FREEING_INITMEM)
> i = LONG_MAX;
>
or even i = totalram_pages()?
> >
> > user_thread = current->mm != NULL;
> > }
> >
> > base-commit: 7be4bc52fd9e0e55dc9a2e0232b89427d2c4b614
> > --
> > 2.54.0 (Apple Git-157)
> >
>
> --
> Vincent
--
Vincent
