[PATCH] mm, kasan: don't poison boot memory

[David Hildenbrand]

On 18.02.21 20:40, Andrey Konovalov wrote:
> On Thu, Feb 18, 2021 at 9:55 AM David Hildenbrand <david at redhat.com> wrote:
>>
>> On 17.02.21 21:56, Andrey Konovalov wrote:
>>> During boot, all non-reserved memblock memory is exposed to the buddy
>>> allocator. Poisoning all that memory with KASAN lengthens boot time,
>>> especially on systems with large amount of RAM. This patch makes
>>> page_alloc to not call kasan_free_pages() on all new memory.
>>>
>>> __free_pages_core() is used when exposing fresh memory during system
>>> boot and when onlining memory during hotplug. This patch adds a new
>>> FPI_SKIP_KASAN_POISON flag and passes it to __free_pages_ok() through
>>> free_pages_prepare() from __free_pages_core().
>>>
>>> This has little impact on KASAN memory tracking.
>>>
>>> Assuming that there are no references to newly exposed pages before they
>>> are ever allocated, there won't be any intended (but buggy) accesses to
>>> that memory that KASAN would normally detect.
>>>
>>> However, with this patch, KASAN stops detecting wild and large
>>> out-of-bounds accesses that happen to land on a fresh memory page that
>>> was never allocated. This is taken as an acceptable trade-off.
>>>
>>> All memory allocated normally when the boot is over keeps getting
>>> poisoned as usual.
>>>
>>> Signed-off-by: Andrey Konovalov <andreyknvl at google.com>
>>> Change-Id: Iae6b1e4bb8216955ffc14af255a7eaaa6f35324d
>>
>> Not sure this is the right thing to do, see
>>
>> https://lkml.kernel.org/r/bcf8925d-0949-3fe1-baa8-cc536c529860@oracle.com
>>
>> Reversing the order in which memory gets allocated + used during boot
>> (in a patch by me) might have revealed an invalid memory access during boot.
>>
>> I suspect that that issue would no longer get detected with your patch,
>> as the invalid memory access would simply not get detected. Now, I
>> cannot prove that :)
> 
> This looks like a good example.
> 
> Ok, what we can do is:
> 
> 1. For KASAN_GENERIC: leave everything as is to be able to detect
> these boot-time bugs.
> 
> 2. For KASAN_SW_TAGS: remove boot-time poisoning via
> kasan_free_pages(), but use the "invalid" tag as the default shadow
> value. The end result should be the same: bad accesses will be
> detected. For unallocated memory as it has the default "invalid" tag,
> and for allocated memory as it's poisoned properly when
> allocated/freed.
> 
> 3. For KASAN_HW_TAGS: just remove boot-time poisoning via
> kasan_free_pages(). As the memory tags have a random unspecified
> value, we'll still have a 15/16 chance to detect a memory corruption.
> 
> This also makes sense from the performance perspective: KASAN_GENERIC
> isn't meant to be running in production, so having a larger perf
> impact is acceptable. The other two modes will be faster.

Sounds in principle sane to me.

Side note: I am not sure if anybody runs KASAN in production. Memory is 
expensive. Feel free to prove me wrong, I'd be very interest in actual 
users.

-- 
Thanks,

David / dhildenb