On Wed, 2021-08-04 at 14:46 -0700, Sean Christopherson wrote:
> Take a signed 'long' instead of an 'unsigned long' for the number of
> pages to add/subtract to the total number of pages used by the MMU. This
> fixes a zero-extension bug on 32-bit kernels that effectively corrupts
> the per-cpu counter used by the shrinker.
>
> Per-cpu counters take a signed 64-bit value on both 32-bit and 64-bit
> kernels, whereas kvm_mod_used_mmu_pages() takes an unsigned long and thus
> an unsigned 32-bit value on 32-bit kernels. As a result, the value used
> to adjust the per-cpu counter is zero-extended (unsigned -> signed), not
> sign-extended (signed -> signed), and so KVM's intended -1 gets morphed to
> 4294967295 and effectively corrupts the counter.
>
> This was found by a staggering amount of sheer dumb luck when running
> kvm-unit-tests on a 32-bit KVM build. The shrinker just happened to kick
> in while running tests and do_shrink_slab() logged an error about trying
> to free a negative number of objects. The truly lucky part is that the
> kernel just happened to be a slightly stale build, as the shrinker no
> longer yells about negative objects as of commit 18bb473e5031 ("mm:
> vmscan: shrink deferred objects proportional to priority").
>
> vmscan: shrink_slab: mmu_shrink_scan+0x0/0x210 [kvm] negative objects to delete nr=-858993460
>
> Fixes: bc8a3d8925a8 ("kvm: mmu: Fix overflow on kvm mmu page limit calculation")
> Cc: stable@xxxxxxxxxxxxxxx
> Cc: Ben Gardon <bgardon@xxxxxxxxxx>
> Signed-off-by: Sean Christopherson <seanjc@xxxxxxxxxx>
> ---
> arch/x86/kvm/mmu/mmu.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
> index b4b65c21b2ca..082a0ba79edd 100644
> --- a/arch/x86/kvm/mmu/mmu.c
> +++ b/arch/x86/kvm/mmu/mmu.c
> @@ -1700,7 +1700,7 @@ static int is_empty_shadow_page(u64 *spt)
> * aggregate version in order to make the slab shrinker
> * faster
> */
> -static inline void kvm_mod_used_mmu_pages(struct kvm *kvm, unsigned long nr)
> +static inline void kvm_mod_used_mmu_pages(struct kvm *kvm, long nr)
> {
> kvm->arch.n_used_mmu_pages += nr;
> percpu_counter_add(&kvm_total_used_mmu_pages, nr);
I am almost sure that I seen this bug as well (I do test 32 bit KVM hosts,
even with nested 32 bit guests once in a while), but I didn't dare to investigate
it due to the fact the 32 bit KVM host is a very rare thing these days.
Reviewed-by: Maxim Levitsky <mlevitsk@xxxxxxxxxx>
Best regards,
Maxim Levitsky
