Re: [syzbot] [mm?] WARNING in ep_write_iter

[Zi Yan <ziy@nvidia.com>]

On 17 Aug 2026, at 11:37, Alan Stern wrote:

> On Mon, Aug 17, 2026 at 11:22:47AM -0400, Zi Yan wrote:
>>> Again, I disagree with some of the details of this argument in this
>>> context.  For instance, if all kernel developers are supposed to know
>>> that they shouldn't ask kmalloc() for more than MAX_PAGE_ORDER at the
>>> risk of provoking a WARN, if this is such an important restriction, then
>>> shouldn't this requirement be mentioned in the kerneldoc for kmalloc()?
>>
>>
>> kmalloc is the normal method of allocating memory
>> for objects smaller than page size in the kernel.
>
> This means when you are allocating small objects, kmalloc is the normal
> way to do it.  It says nothing about what you should do when allocating
> large objects.

alloc_pages() or vmalloc or kvmalloc(). kmalloc() is used if you really
want physically contiguous memory.

>
> In particular, it doesn't say that if you ask kmalloc for more than 4 MB
> of memory, your system is likely to crash.

Crash because panic_on_warn is set. You just conflate a sysctl config with
a WARN. I understand panic_on_warn is set by many systems, but that does not
mean every system crashes.

Plus, if you use kvmalloc with a really large size, you also can get
a WARN[1][2].

[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/mm/slub.c?h=v7.2#n6912
[2] https://lore.kernel.org/all/CAHk-=wiRq+_jd_O1gz3J6-ANtXMY7iLpi8XFUcmtB3rBixvUXQ@mail.gmail.com/

>
>> See https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/include/linux/slab.h?h=v7.2#n1001
>>
>> In this case, kmalloc is used to request > 4MB memory.
>>
>>>
>>> Regardless, if it is important to let kernel developers know that their
>>> code is doing something wrong, why not make the WARN conditional on
>>> CONFIG_EXPERT or something similar?  In other words, prevent it from
>>> crashing production systems.
>
> You didn't respond to this.

I do not get your argument here. Why does this WARN need to be conditional
on CONFIG_EXPERT? What makes it special? Or you mean you want all WARNs
to be conditional CONFIG_EXPERT?

Anyway, if you do not want to check user input len and want core MM to
do the check. I will send a patch to make kmalloc return NULL when
the requested len is bigger than KMALLOC_MAX_SIZE without giving a warning,
based on my discussion with Vlastimil on IRC (he is out until next week).


Best Regards,
Yan, Zi