[PATCH] ring-buffer: Do not WARN when trace_rb_cpu_prepare() fails to allocate

[PATCH] ring-buffer: Do not WARN when trace_rb_cpu_prepare() fails to allocate

[Christian Aurich Zanettini Martins]

Creating a tracing instance allocates the buffers of the other CPUs
through trace_rb_cpu_prepare(), which WARNs if the allocation fails.
But __rb_allocate_pages() bails out on purpose when si_mem_available()
is too low or the task gets killed, so a mkdir in instances/ with a
large trace_buf_size is enough to trigger it:

  WARNING: kernel/trace/ring_buffer.c:8094 at trace_rb_cpu_prepare+0xf1/0x100

The same failure for the current CPU, or in ring_buffer_resize(), just
returns -ENOMEM. Use pr_warn() here instead.

Fixes: b32614c03413 ("tracing/rb: Convert to hotplug state machine")
Reported-by: syzbot+fbe07bcf949966f3c00f@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/all/6a707c4f.13bfb6d0.1ecdd5.027e.GAE@google.com/
Signed-off-by: Christian Aurich Zanettini Martins <christian.aurichzm@gmail.com>
---

Notes:
    Tested in QEMU on 05441d697f7f (4 CPUs, 1G, trace_buf_size=350M): mkdir
    in instances/ WARNs every time before and only prints the message after.
    Same with a mkdir killed during the allocation.

 kernel/trace/ring_buffer.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index 1fd28a72e6ee..c018559ad61a 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -8091,8 +8091,7 @@ int trace_rb_cpu_prepare(unsigned int cpu, struct hlist_node *node)
 	buffer->buffers[cpu] =
 		rb_allocate_cpu_buffer(buffer, nr_pages, cpu);
 	if (!buffer->buffers[cpu]) {
-		WARN(1, "failed to allocate ring buffer on CPU %u\n",
-		     cpu);
+		pr_warn("failed to allocate ring buffer on CPU %u\n", cpu);
 		return -ENOMEM;
 	}
 
-- 
2.43.0

Re: [PATCH] ring-buffer: Do not WARN when trace_rb_cpu_prepare() fails to allocate

[Vincent Donnefort]

On Wed, Nov 04, 2026 at 07:25:58PM -0300, Christian Aurich Zanettini Martins wrote:
> Creating a tracing instance allocates the buffers of the other CPUs
> through trace_rb_cpu_prepare(), which WARNs if the allocation fails.
> But __rb_allocate_pages() bails out on purpose when si_mem_available()
> is too low or the task gets killed, so a mkdir in instances/ with a
> large trace_buf_size is enough to trigger it:
> 
>   WARNING: kernel/trace/ring_buffer.c:8094 at trace_rb_cpu_prepare+0xf1/0x100
> 
> The same failure for the current CPU, or in ring_buffer_resize(), just
> returns -ENOMEM. Use pr_warn() here instead.
> 
> Fixes: b32614c03413 ("tracing/rb: Convert to hotplug state machine")
> Reported-by: syzbot+fbe07bcf949966f3c00f@syzkaller.appspotmail.com
> Closes: https://lore.kernel.org/all/6a707c4f.13bfb6d0.1ecdd5.027e.GAE@google.com/
> Signed-off-by: Christian Aurich Zanettini Martins <christian.aurichzm@gmail.com>
> ---

I do not believe Steven is keen on this change [1]

[1] https://lore.kernel.org/all/20260806174638.6012fa24@gandalf.local.home/

> 
> Notes:
>     Tested in QEMU on 05441d697f7f (4 CPUs, 1G, trace_buf_size=350M): mkdir
>     in instances/ WARNs every time before and only prints the message after.
>     Same with a mkdir killed during the allocation.
> 
>  kernel/trace/ring_buffer.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
> index 1fd28a72e6ee..c018559ad61a 100644
> --- a/kernel/trace/ring_buffer.c
> +++ b/kernel/trace/ring_buffer.c
> @@ -8091,8 +8091,7 @@ int trace_rb_cpu_prepare(unsigned int cpu, struct hlist_node *node)
>  	buffer->buffers[cpu] =
>  		rb_allocate_cpu_buffer(buffer, nr_pages, cpu);
>  	if (!buffer->buffers[cpu]) {
> -		WARN(1, "failed to allocate ring buffer on CPU %u\n",
> -		     cpu);
> +		pr_warn("failed to allocate ring buffer on CPU %u\n", cpu);
>  		return -ENOMEM;
>  	}
>  
> -- 
> 2.43.0
> 

-- 
Vincent

Re: [PATCH] ring-buffer: Do not WARN when trace_rb_cpu_prepare() fails to allocate

[Christian Aurich Zanettini Martins]

Thanks Vincent. I took that reply as Steven explaining where the warning
comes from, but I may have misread it.

Steven, the same si_mem_available() check fails silently for the current
CPU in alloc_buffer(), so whether mkdir WARNs or just returns -ENOMEM
depends on which CPU runs out first. If you'd rather keep the WARN I'll
drop this.

Em seg., 9 de nov. de 2026 às 06:29, Vincent Donnefort
<vdonnefort@google.com> escreveu:
>
> On Wed, Nov 04, 2026 at 07:25:58PM -0300, Christian Aurich Zanettini Martins wrote:
> > Creating a tracing instance allocates the buffers of the other CPUs
> > through trace_rb_cpu_prepare(), which WARNs if the allocation fails.
> > But __rb_allocate_pages() bails out on purpose when si_mem_available()
> > is too low or the task gets killed, so a mkdir in instances/ with a
> > large trace_buf_size is enough to trigger it:
> >
> >   WARNING: kernel/trace/ring_buffer.c:8094 at trace_rb_cpu_prepare+0xf1/0x100
> >
> > The same failure for the current CPU, or in ring_buffer_resize(), just
> > returns -ENOMEM. Use pr_warn() here instead.
> >
> > Fixes: b32614c03413 ("tracing/rb: Convert to hotplug state machine")
> > Reported-by: syzbot+fbe07bcf949966f3c00f@syzkaller.appspotmail.com
> > Closes: https://lore.kernel.org/all/6a707c4f.13bfb6d0.1ecdd5.027e.GAE@google.com/
> > Signed-off-by: Christian Aurich Zanettini Martins <christian.aurichzm@gmail.com>
> > ---
>
> I do not believe Steven is keen on this change [1]
>
> [1] https://lore.kernel.org/all/20260806174638.6012fa24@gandalf.local.home/
>
> >
> > Notes:
> >     Tested in QEMU on 05441d697f7f (4 CPUs, 1G, trace_buf_size=350M): mkdir
> >     in instances/ WARNs every time before and only prints the message after.
> >     Same with a mkdir killed during the allocation.
> >
> >  kernel/trace/ring_buffer.c | 3 +--
> >  1 file changed, 1 insertion(+), 2 deletions(-)
> >
> > diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
> > index 1fd28a72e6ee..c018559ad61a 100644
> > --- a/kernel/trace/ring_buffer.c
> > +++ b/kernel/trace/ring_buffer.c
> > @@ -8091,8 +8091,7 @@ int trace_rb_cpu_prepare(unsigned int cpu, struct hlist_node *node)
> >       buffer->buffers[cpu] =
> >               rb_allocate_cpu_buffer(buffer, nr_pages, cpu);
> >       if (!buffer->buffers[cpu]) {
> > -             WARN(1, "failed to allocate ring buffer on CPU %u\n",
> > -                  cpu);
> > +             pr_warn("failed to allocate ring buffer on CPU %u\n", cpu);
> >               return -ENOMEM;
> >       }
> >
> > --
> > 2.43.0
> >
>
> --
> Vincent

Re: [PATCH] ring-buffer: Do not WARN when trace_rb_cpu_prepare() fails to allocate

[Steven Rostedt]

On Mon, 9 Nov 2026 07:38:55 -0300
Christian Aurich Zanettini Martins <christian.aurichzm@gmail.com> wrote:

> Thanks Vincent. I took that reply as Steven explaining where the warning
> comes from, but I may have misread it.
> 
> Steven, the same si_mem_available() check fails silently for the current
> CPU in alloc_buffer(), so whether mkdir WARNs or just returns -ENOMEM
> depends on which CPU runs out first. If you'd rather keep the WARN I'll
> drop this.

Actually, because of the CPU coming on line (not the instances) I
pulled it into my test queue. I'm currently traveling so I haven't
pushed things up yet. But patchwork has updated the status[1] of this to
"Under review" which means it's in the testing process.

-- Steve

[1] https://patchwork.kernel.org/m/20261104222558.863948-1-christian.aurichzm@gmail.com/