Re: [PATCH] UML: add support for KASAN under x86_64

[Richard Weinberger <richard@nod.at>]

Patricia,

----- Ursprüngliche Mail -----
> Von: "Patricia Alfonso" <trishalfonso@google.com>
> An: "Johannes Berg" <johannes@sipsolutions.net>
> CC: "Dmitry Vyukov" <dvyukov@google.com>, "Jeff Dike" <jdike@addtoit.com>, "richard" <richard@nod.at>, "anton ivanov"
> <anton.ivanov@cambridgegreys.com>, "Andrey Ryabinin" <aryabinin@virtuozzo.com>, "Brendan Higgins"
> <brendanhiggins@google.com>, "davidgow" <davidgow@google.com>, "linux-um" <linux-um@lists.infradead.org>,
> "linux-kernel" <linux-kernel@vger.kernel.org>, "kasan-dev" <kasan-dev@googlegroups.com>
> Gesendet: Dienstag, 31. März 2020 18:39:21
> Betreff: Re: [PATCH] UML: add support for KASAN under x86_64

> On Mon, Mar 30, 2020 at 1:41 AM Johannes Berg <johannes@sipsolutions.net> wrote:
>>
>> On Mon, 2020-03-30 at 10:38 +0200, Dmitry Vyukov wrote:
>> > On Mon, Mar 30, 2020 at 9:44 AM Johannes Berg <johannes@sipsolutions.net> wrote:
>> > > On Fri, 2020-03-20 at 16:18 +0100, Dmitry Vyukov wrote:
>> > > > > Wait ... Now you say 0x7fbfffc000, but that is almost fine? I think you
>> > > > > confused the values - because I see, on userspace, the following:
>> > > >
>> > > > Oh, sorry, I copy-pasted wrong number. I meant 0x7fff8000.
>> > >
>> > > Right, ok.
>> > >
>> > > > Then I would expect 0x1000 0000 0000 to work, but you say it doesn't...
>> > >
>> > > So it just occurred to me - as I was mentioning this whole thing to
>> > > Richard - that there's probably somewhere some check about whether some
>> > > space is userspace or not.
>> > >
> 
> Yeah, it seems the "Kernel panic - not syncing: Segfault with no mm",
> "Kernel mode fault at addr...", and "Kernel tried to access user
> memory at addr..." errors all come from segv() in
> arch/um/kernel/trap.c due to what I think is this type of check
> whether the address is
> in userspace or not.

Segfault with no mm means that a (not fixable) pagefault happened while
kernel code ran.

>> > > I'm beginning to think that we shouldn't just map this outside of the
>> > > kernel memory system, but properly treat it as part of the memory that's
>> > > inside. And also use KASAN_VMALLOC.
>> > >
>> > > We can probably still have it at 0x7fff8000, just need to make sure we
>> > > actually map it? I tried with vm_area_add_early() but it didn't really
>> > > work once you have vmalloc() stuff...
>> >
> 
> What x86 does when KASAN_VMALLOC is disabled is make all vmalloc
> region accesses succeed by default
> by using the early shadow memory to have completely unpoisoned and
> unpoisonable read-only pages for all of vmalloc (which includes
> modules). When KASAN_VMALLOC is enabled in x86, the shadow memory is not
> allocated for the vmalloc region at startup. New chunks of shadow
> memory are allocated and unpoisoned every time there's a vmalloc()
> call. A similar thing might have to be done here by mprotect()ing
> the vmalloc space as read only, unpoisoned without KASAN_VMALLOC. This
> issue here is that
> kasan_init runs so early in the process that the vmalloc region for
> uml is not setup yet.
> 
> 
>> > But we do mmap it, no? See kasan_init() -> kasan_map_memory() -> mmap.
>>
>> Of course. But I meant inside the UML PTE system. We end up *unmapping*
>> it when loading modules, because it overlaps vmalloc space, and then we
>> vfree() something again, and unmap it ... because of the overlap.
>>
>> And if it's *not* in the vmalloc area, then the kernel doesn't consider
>> it valid, and we seem to often just fault when trying to determine
>> whether it's valid kernel memory or not ... Though I'm not really sure I
>> understand the failure part of this case well yet.
>>
> 
> I have been testing this issue in a multitude of ways and have only
> been getting more confused. It's still very unclear where exactly the
> problem occurs, mostly because the errors I found most frequently were
> reported in segv(), but the stack traces never contained segv.
> 
> Does anyone know if/how UML determines if memory being accessed is
> kernel or user memory?

In contrast to classic x86, without KPTI and SMAP/SMEP, UML has a strong
separation between user- and kernel-memory. This is also why copy_from/to_user()
is so expensive.

In arch/um/kernel/trap.c segv() you can see the logic.
Also see UPT_IS_USER().

Thanks,
//richard