[Bug report] hash_name() may cross page boundary and trigger sleep in RCU context

[Al Viro]

On Fri, Nov 28, 2025 at 05:35:37PM -0800, Linus Torvalds wrote:
> On Fri, 28 Nov 2025 at 17:01, Zizhi Wo <wozizhi at huaweicloud.com> wrote:
> >
> > Thank you for your answer. In fact, this solution is similar to the one
> > provided by Al.
> 
> Hmm. I'm not seeing the replies from Al for some reason. Maybe he didn't cc me.

<checks>
Sorry, I thought you've been somewhere in that Cc, should've verified that.

> That said, somebody should  definitely double-check me - because I
> think arm also did the vdso trick at high addresses that i386 used to
> do, so there is the fake VDSO thing up there.
> 
> But if you get a page fault on that, it's not going to be fixed up, so
> even if user space can access it, there's no point in looking that
> fake vm area up for page faults.

gate_vma is not inserted anywhere - it's special-cased in coredump_next_vma(),
proc_get_vma() and get_gate_page(); none of that can lead to insertion.

AFAICS its uses in mlock_fixup(), __mmap_complete() and should_skip_vam()
are pure paranoia - "if we somehow ended up running into gate_vma, let's
make sure we don't screw it over" and had always been that way.

So VMA lookup would get NULL.